Clay-Containing Polymeric Nanocomposites Volume 1
L.A. Utracki
Clay-Containing Polymeric Nanocomposites Volume 1 L.A. Utracki C. Vasile
Rapra Technology Limited Shawbury, Shrewsbury, Shropshire, SY4 4NR, United Kingdom Telephone: +44 (0)1939 250383 Fax: +44 (0)1939 251118 http://www.rapra.net
First Published in 2004 by
Rapra Technology Limited Shawbury, Shrewsbury, Shropshire, SY4 4NR, UK
©2004, Rapra Technology Limited
All rights reserved. Except as permitted under current legislation no part of this publication may be photocopied, reproduced or distributed in any form or by any means or stored in a database or retrieval system, without the prior permission from the copyright holder. A catalogue record for this book is available from the British Library.
Every effort has been made to contact copyright holders of any material reproduced within the text and the authors and publishers apologise if any have been overlooked.
ISBN: 1-85957-437-8
Typeset, printed and bound by Rapra Technology Limited Cover printed by The Printing House, Crewe, UK
Preamble
Preamble L.A. Utracki
During the last few years terms like nanomaterials, nanocomposites and nanosystems have become fashionable. It seems that anything with ‘nano’ attached to it has nearly a magical effect – not so much on performance as on expectations. There is an extensive worldwide effort to introduce nanotechnology for the production of materials with specific functional characteristics, e.g., semiconducting, electromagnetic, optical, etc. New magneto-resistance materials with nanometre-scale spin-flip mean free path of electrons have been commercialised. The National Science Foundation (NSF) has solicited collaborative research proposals in the area of nanoscale science and engineering, including: nanoscale biosystems; nanoscale structures; novel phenomena and control; nanoscale devices and system architecture; nanosystems-specific software; nanoscale processes; multi-phenomena modelling and simulation at the nanoscale level; studies on societal implications of nanoscale science and engineering, etc. Nanostructures are of interest to many technologies. The potential of precise control of impurities and defects in a crystal and the ability to integrate perfect inorganic and organic nanostructures may lead to a new generation of advanced materials. To electronics, they offer quantum devices (resonant tunnelling transistors; single electron transistors; cellular automata based on quantum dots) and new processor architectures. To catalysis, they form the templates for catalytic activity, zeolite pores, etc. In biology, nanostructures are components of the mitochondrion, the chloroplast, and the ribosome. The advances in the synthesis and fabrication of isolated nanostructures range from colloidal synthesis of nanocrystals to the growth of epitaxial quantum dots. The techniques of molecular biology have made a wide range of biological nanostructures readily available through cloning and overexpression in bacterial production systems. Furthermore, work has begun on the use of self-assembly techniques to prepare complex and designed spatial arrangements of nanostructures. Techniques derived from microlithography in microelectronics (viz. photo, X-ray, and e-beam lithography) offer the potential to economically generate new types of 3D-structures. In short, there is a great potential for the wide use of nanotechnology for functional materials and devices. The central theme of this book is the use of nanotechnology for the development of new structural polymeric systems, the polymeric nanocomposites (PNCs), and particularly the clay-containing polymeric nanocomposites (CPNCs). These mass produced materials are dispersions of inorganic, nanoscale platelets in a polymeric matrix. Economics preclude the use of most of the manufacturing methods developed for ceramic and metallic nanocomposites. The key to the success of the CPNC industry is to provide new materials significantly outperforming the old ones at a marginal incremental cost increase per unit volume. Considering that at present the nanoclay is a natural mineral with welli
Clay-Containing Polymeric Nanocomposites recognised variability of composition, the secondary concerns focus on the consistency of performance, not only at the batch-to-batch level, but also on a long-term basis. Nanostructures are intermediate in size between molecular and micron-size systems, such as blends and composites. There is no doubt that structures with architecture controlled on the molecular level may lead to refined properties or even new sets of performance characteristics. Chemists have known this for centuries, viz. more recent developments in nanosize structures such as fullerenes, buckytubes, dendrimers and complex block copolymers. The unexpected behaviour of adsorbed monolayers of organic molecules on a high-energy crystal surface was discovered many decades ago. In the meantime the advances of microscopy reached atomic-scale, providing images of crystalline unit cells and bioactive macromolecules. Thus, it is legitimate to ask what, if anything, is so different about the nanomaterials that warrants the distinction. It is known that within the nanometre scale such properties as the melting temperature, the remanence of a magnet, and the band gap of a semiconductor depend upon the size of the component crystals. Furthermore, it has been shown that the mechanical properties of metallic alloys hyperbolically increase with the reduction of domain size. It is customary to define nanocomposites in terms of the size of the dispersed particles and the specific behaviour they engender. Thus, at least one dimension of these particles must be less than 10 nm. Since these particles are usually crystalline, the size and high surface energy leads to high surface area to volume ratio and strong orientational forces that may lead to high packing densities and quantum behaviour (explored as electronic, magnetic or optic elements in microelectronics technology). In the most popular nanofiller in the CPNC industry, montmorillonite, over 40% of atoms rest on the surface – the clay lamellae should be treated either as giant inorganic molecules or at least as hybrids occupying the grey zone between molecules and particles. This is not mere semantics, but has profound consequences as far as the fundamentals of CPNC are concerned, viz. miscibility or flow behaviour. This book summarises the pertinent developments in the area of the science and technology of clay-reinforced polymeric nanocomposites. There are several reasons for using clays, viz. availability, cost, and aspect ratio. The theory and experiments show that to maximise the benefits of nanotechnology the clay must be fully decomposed into individual crystalline lamellae (exfoliated) and these must be uniformly dispersed in a given matrix material. Furthermore, considering the large aspect ratio of clay platelets, their orientation must be controlled – for some applications perfect alignment is desirable, whereas in others isotropicity of reinforcement is essential. For example, aligning clay lamellae perpendicular to the flux direction may increase barrier properties by a factor of 100, whereas orienting them in the flux direction will hardly change the barrier properties over those of the matrix. Considering that the relaxation time of standard clay platelets (aspect ratio of 200 to 300) is of the order of one hour and that their dimensions are of the nanometre scale, the dispersion and orientation of clay during polymer processing is challenging. The main difficulties for CPNC technology rest in the hygroscopic character of clay and strong solid-solid interactions. It is a relatively simple task to disperse clay platelets or lamellae (i.e., to exfoliate them) in water or in water-soluble,
ii
Preamble polar monomers or oligomers (e.g., amino acids or glycols). However, preparation of CPNC in a hydrophobic, non-polar high molecular weight polymer, e.g., a polyolefin or polystyrene, is difficult. The most sensible way to approach the problem is to consider the process of preparation of CPNCs as blending two highly immiscible ingredients, i.e., from the perspective of polymer blending and compatibilisation. As in polymer blends here also one is obliged to ensure good interaction between the two antagonistic components: hygroscopic clay and hydrophobic polymer. The clay of preference is montmorillonite (MMT) with micron-sized particles formed by stacks of three-layer sandwiches: a layer of Mg and Al oxides inbetween the silicate layers. These sandwiches of 0.96 nm thickness and an average diameter of about 100 to 500 nm are the desired reinforcing entities for CPNC. The chemical constitution of the MMT unit cell offers three types of reactive sites: anions on the silicate surface, hydroxyl (–OH) groups, and (few) cations on the narrow edges. Historically, compatibilisation of clay involved forming an ionic bond between the clay surface and organophilic onium cations, especially quaternary ammonium ones. The advantage of this is that the chemical reaction not only changes the hydrophilic clay character into hydrophobic, but also it causes the clay particles to expand, i.e., to intercalate as a first step to the total dispersion of the clay platelets, i.e., to exfoliation. The disadvantage is that this chemical equilibrium process is diffusion controlled hence it may require an excess of intercalatant and it may take a long time to complete! ‘Compatibilisation’ with at least partial utilisation of the –OH groups has been carried out using their ability to react with epoxy or acid anhydride groups. However, since the –OH groups are mostly located on the peripheries of clay particles, there are few of them readily accessible and the reaction does not necessarily lead to intercalation/exfoliation. Since the solid-solid interactions are 100 times stronger than liquid-liquid ones, it is imperative that there is good miscibility between the pre-intercalated clay and the polymeric matrix – if not, even the initially exfoliated clay platelets may reassemble during processing. The most reasonable strategy is to prepare exfoliated CPNC using multiple steps, for example: 1. Swelling sodium montmorillonite in warm water which causes the interlayer spacing to expand from the initially dry state of 0.96 to about 1.3 nm. 2. Intercalation with cations suitable for the envisaged CPNC organophilic molecules, onium or Lewis-base types that increase the interlayer spacing to about 3 to 4 nm and will improve miscibility between the clay and the matrix. 3. Reactive compatibilisation of the organoclay/matrix polymer system, which results in stable exfoliation of the clay platelets (interlayer spacing larger than 8.8 nm). This step may not be necessary for highly polar polymers, such as water-soluble polymers (e.g., polyvinyl pyrrolidone or polyvinyl alcohol), or even for polyamides, but it is crucial for polyolefins or styrenics. 4. Melt compounding the pre-intercalated clay with matrix polymer. An alternative strategy is to disperse the product of step (2) in a monomer and polymerise it. This method has been particularly successful when the intercalatant used in step (2) can be incorporated into the macromolecular chain, thus forming what has became known as ‘hairy clay platelets’ with over one thousand macromolecules end-tethered to the clay surface.
iii
Clay-Containing Polymeric Nanocomposites To provide condensed information on the essential elements of CPNC technology, the book is divided into parts: Part 1. Introduction – presents a general overview of nanocomposites with polymeric as well as non-polymeric matrices. Part 2. Basic elements of PNC technology – focuses on the general methods and principles of polymeric nanocomposites. It starts with a brief description of PNC comprising non-clay nanoparticles (e.g., carbon nanotubes, polyhedral oligomeric silsesquioxanes (POSS), etc.), and then focuses on the clay-containing polymeric nanocomposites. The individual elements of CPNC technology are discussed, namely the general characteristics of clays, methods of purification, and the diverse methods used for intercalation and exfoliation. Part 3. Fundamental aspects – discusses the pertinent aspects of the thermodynamics, thermal stability, rheology, crystallisation and mechanical behaviour. Part 4. Technology of CPNC – reviews the evolution of CPNC technology for specific polymeric matrices, primarily using the patent literature. Thus, CPNCs with individual polymer matrices are reviewed, starting in historical order with polyamide (PA), polyolefin (PO) and other thermoplastics, then epoxies, polyurethanes and other thermosets. Part 5. Performance – discusses selected properties of CPNCs, viz. mechanical, flame retardancy, and barrier. Part 6. Closing remarks – summarises the information. Part 7. Appendices – provides explanations of abbreviations, symbols, and concepts used in the book. Part 8. References – contains well over 1,000 references to open and patent literature up to the beginning of 2004. Polymeric nanotechnology is in statu nascendi. In consequence, there is a bit of confusion and uncertainty about its value and the most suitable applications. Hopefully, this book will help answer some of these questions and in a small way accelerate wider introduction of this technology. In 1953, after getting a chemical engineering degree and spending the obligatory six-month stage as plant engineer, I started graduate studies in the field of phase equilibria and flow of polymer solutions. To emphasise the objectiveness of scientific writings it is expected to use the impersonal form. However, after 50 years in the profession I wish to revert to a more personal style, dedicating this latest creation (and, as Hermann Mark used to say, the dearest) to Czeslawa, my Wife, Friend and Supporter of nearly as many years. Curiously, this book was not planned, but rather it evolved in response to questions, comments and stories told to me by my colleagues from the Americas, Asia and Europe. There are too many people to whom I owe my thanks to list them all, but I wish to express my thanks to a very special trio: to Robert Simha who has been my mentor and brilliant star to follow for all these 40-odd years of my post-doc’ing with him, to Osami Kamigaito for introducing me to the fascinating world of clay-containing polymeric nanocomposites, and to my colleague and friend of many years, Jørgen Lyngaae-Jørgensen. Leszek Utracki Montreal, March 2004 iv
Contents
Contents Volume 1 Preamble .................................................................................................... i
Part 1 Introduction 1.1 General ............................................................................................... 1 1.2 NCs with Ceramic or Metallic Matrix .............................................. 2 1.2.1 1.2.2 1.2.3 1.2.4
Metallic Nanoparticles in Amorphous Matrix ................................ 2 Magnetic Oxides in Silica Nanocomposites .................................... 2 Optoelectronics ............................................................................... 3 Summary on Non-Polymeric NC .................................................... 3
1.3 NCs with Polymeric Matrix .............................................................. 3 1.3.1 PNC Definitions .............................................................................. 6 1.3.2 Methods of Characterisation of CPNCs .......................................... 8 1.3.2.1 X-Ray Diffraction (XRD) .................................................. 8 1.3.2.2 Small Angle Neutron Scattering (SANS) .......................... 11 1.3.2.3 Transmission and Atomic Force Electron Microscopy (TEM and AFM) ............................................................. 14 1.3.2.4 Fourier Transform Infrared Spectroscopy (FTIR) ............ 15 1.3.2.5 Nuclear Magnetic Resonance Spectroscopy (NMR) ........ 16 1.3.2.6 Other Methods ................................................................ 17 1.3.3 Determination of PNC Properties ................................................. 18 1.3.4 PNC Types and Methods of their Preparation .............................. 18 1.3.5 PNCs of Commercial Interest ........................................................ 18 1.3.6 Journals and Research Groups ...................................................... 29 1.3.7 Historical Perspective .................................................................... 30
Part 2 Basic Elements of Polymeric Nanocomposite Technology 2.1 Nanoparticles of Interest to PNC Technology ................................. 35 2.1.1 General .......................................................................................... 35 2.1.2 Layered Nanoparticles .................................................................. 35 2.1.3 Fibrillar Nanoparticles .................................................................. 38 2.1.3.1 Carbon Nanotubes (CNTs) ............................................. 38 2.1.3.1.1 2.1.3.1.2 2.1.3.1.3 2.1.3.1.4 2.1.3.1.5 2.1.3.1.6
Origin, Characteristics and Structure .................. Computation of Potential CNT Properties .......... Non-Polymeric Applications of CNTs ................. Sources ................................................................. PNC with CNTs for Electrical Conductivity ....... Graphite ..............................................................
38 41 44 46 46 47
v
Clay-Containing Polymeric Nanocomposites 2.1.3.1.7 PNC with CNTs – Thermoset Matrix ................. 48 2.1.3.1.8 PNC with CNTs – Thermoplastic Matrix ............ 50
2.1.3.2 Rod-Like CdSe Nanocrystals ........................................... 54 2.1.3.3 Imogolite ......................................................................... 54 2.1.3.4 Vanadium Pentoxide, V2O5 ............................................. 54 2.1.3.5 Inorganic Nanotubes ....................................................... 55 2.1.4 Other Nanoparticles ...................................................................... 56 2.1.4.1 Spherical or Nearly-Spherical Particles ............................ 56 2.1.4.2 Sol-Gel Hybrids ............................................................... 56 2.1.4.3 Polyhedral Oligomeric Silsesquioxanes (POSS) ............... 58 2.1.4.3.1 2.1.4.3.2 2.1.4.3.3 2.1.4.3.4
Origin and Structure ............................................ Properties ............................................................. Sources ................................................................. Applications .........................................................
58 60 66 67
2.2 Clays ............................................................................................... 73 2.2.1 General Characteristics ................................................................. 73 2.2.2 Crystalline Clays ........................................................................... 74 2.2.2.1 Kaolins ............................................................................ 74 2.2.2.2 Serpentines ...................................................................... 74 2.2.2.3 Illite Group (Micas) ......................................................... 74 2.2.2.4 Chlorites and Vermiculites .............................................. 76 2.2.2.5 Other Clays ..................................................................... 76 2.2.2.5.1 Glauconite ........................................................... 76 2.2.2.5.2 Sepiolite, Palygorskite and Attapulgite ................ 76 2.2.2.5.3 Mixed-Layer Clay Minerals ................................. 76
2.2.2.6 Smectites or Phyllosilicates .............................................. 76 2.2.2.6.1 Bentonite ............................................................. 79 2.2.2.6.2 Montmorillonite (MMT) ..................................... 80
2.2.3 Purification of Clay ....................................................................... 84 2.2.4 Reactions of Clays with Organic Substances ................................. 85 2.2.4.1 Clay in Aqueous Medium ................................................ 90 2.2.4.1.1 2.2.4.1.2 2.2.4.1.3 2.2.4.1.4 2.2.4.1.5
General ................................................................ Reactions with Edge Cations ............................... Reactions with –OH Groups ............................... Reaction with the Silicilic Surface Anions ........... Stabilisation by Polyelectrolytes ..........................
90 91 91 91 92
2.2.4.2 Clay Dispersion in Polar Organic Liquids ....................... 93 2.2.4.3 Absorption of Organic Molecules by Organoclay ........... 93
2.3 Intercalation of Clay ....................................................................... 97 2.3.1 2.3.2 2.3.3 2.3.4 2.3.5 vi
Introduction .................................................................................. 97 Intercalation by Solvents and Solutions ....................................... 100 Intercalation by Organic Cations ................................................ 102 Intercalation by Organic Liquids ................................................. 124 Intercalation by Monomers, Oligomers or Polymers ................... 126
Contents
2.3.5.1 Intercalation of Purified Clay by Hydrophobic Compounds ................................................................... 126 2.3.5.2 Intercalation of Purified Clay by Hydrophilic Compounds ................................................................... 127 2.3.6 Two-Step Intercalation ................................................................ 135 2.3.6.1 Intercalation by Silylation ............................................. 136 2.3.6.2 Intercalation Utilising Epoxy Compounds .................... 138 2.3.6.3 Intercalation Utilising Organic Anions .......................... 139 2.3.6.4 Intercalation Utilising Macrocyclic Oligomers (Cyclomers) ................................................................... 139 2.3.7 Intercalation by Inorganic Intercalants ........................................ 140 2.3.8 Melt Intercalation ....................................................................... 142 2.3.8.1 Quiescent (or Static) Melt Intercalation ........................ 143 2.3.8.2 Dynamic Melt Intercalation .......................................... 149 2.3.8.2.1 2.3.8.2.2 2.3.8.2.3 2.3.8.2.4 2.3.8.2.5 2.3.8.2.6
Melt Mixing ...................................................... Mixing Equipment ............................................. Mixing in an Extensional Flow Field ................. Melt Intercalation in a PA Matrix ..................... Melt Intercalation in PEG Matrix ..................... Melt Intercalation in PO Matrix .......................
149 150 158 160 165 165
2.3.9 Temperature and Pressure Effects on Interlamellar Spacing ........ 183 2.3.10 Layered Nanofillers, other than Montmorillonite ....................... 185 2.3.10.1 Kaolinite ........................................................................ 186 2.3.10.2 Micas and Synthetic Micas ............................................ 189 2.3.11 Summary of the Intercalation Methods ....................................... 198
2.4 Exfoliation of Clays ...................................................................... 201 2.4.1 Principles ..................................................................................... 202 2.4.2 Polymerisation in the Presence of Organoclay ............................. 204 2.4.2.1 Monomer Intercalation – PA-6 Nanocomposites .......... 204 2.4.2.2 Monomer Modification – Acrylic-Based Nanocomposites ............................................................ 206 2.4.2.3 Non-Reactive Intercalated Clays ................................... 209 2.4.2.4 Co-Vulcanisation ........................................................... 210 2.4.2.5 Common Solvent Method – Polyimide Based Nanocomposites ............................................................ 210 2.4.2.6 Other Methods – Epoxy-Based Nanocomposites .......... 218 2.4.2.7 Other Methods – PU-Based Nanocomposites ................ 224 2.4.2.7.1 2.4.2.7.2 2.4.2.7.3 2.4.2.7.4
Metal Particles ................................................... Silica .................................................................. Cadmium Sulfide Particles (CdS) ....................... Organoclays .......................................................
224 225 225 225
2.4.3 Melt Exfoliation .......................................................................... 232 2.4.3.1 PA-Based CPNCs ........................................................... 233 2.4.3.2 PO-Based CPNCs .......................................................... 237 vii
Clay-Containing Polymeric Nanocomposites
2.4.3.3 PCL-Based CPNCs ........................................................ 242 2.4.3.4 Other Systems ............................................................... 245 2.4.4 Functional CPNC ........................................................................ 245 2.4.4.1 Liquid Crystal/Clay Composite (LCC) .......................... 245 2.4.4.2 Biodegradable CPNC with Polylactic Acid (PLA) ......... 246 2.4.4.3 Poly(N-Vinyl Carbazole)/MMT .................................... 252 2.4.4.4 Polydiacetylene .............................................................. 254 2.4.4.5 Clay-Functional Organic Molecules .............................. 254 2.4.4.6 Super-Absorbent CPNC ................................................ 255 2.4.4.7 Emulsion Polymerisation of CPNC ............................... 255
Part 3 Fundamental Aspects 3.1 Thermodynamics ............................................................................ 257 3.1.1 Glass Transition in Thin Films .................................................... 257 3.1.2 Nanothermodynamics ................................................................. 260 3.1.3 Vaia’s Lattice Model for Organoclay Intercalation by Molten Polymer ........................................................................... 263 3.1.3.1 Introduction .................................................................. 263 3.1.3.2 Entropic Contributions ................................................. 264 3.1.3.3 Interactions ................................................................... 266 3.1.3.4 Consequences of the Model ........................................... 267 3.1.3.5 Model Prediction versus Static Intercalation Results ..... 270 3.1.4 Computations of Polymeric Brushes ............................................ 271 3.1.5 Balazs Self-Consistent Field Approach ........................................ 272 3.1.5.1 Numerical Simulation ................................................... 273 3.1.5.2 Analytical Self-Consistent-Field Theory for Compatibilised Systems ................................................. 278 3.1.5.3 Phase Behaviour ............................................................ 280 3.1.5.4 Contribution and Potential of the SCF Method ............ 285 3.1.6 Scaling Theory for Telechelic Polymer/Clay Systems ................... 287 3.1.7 Solid Surface Effects on Molecular Mobility ............................... 291 3.1.7.1 Surface Energy of Solids ................................................ 291 3.1.7.2 Polymer Adsorption on Solid Particles .......................... 293 3.1.7.3 Nanoscale Rheology ...................................................... 294 3.1.7.4 Molecular Modelling of Nanoconfined Molecules (Intercalation) ................................................................ 298 3.1.8 Kinetics of Polymer Intercalation ................................................ 302 3.1.8.1 Macromolecular Diffusion ............................................ 302 3.1.8.2 Stationary Intercalation ................................................. 304 3.1.8.3 Simulation of Melt Intercalation Kinetics ...................... 306 3.1.9 Pressure-Volume-Temperature Dependence for CPNC ................ 309 3.1.9.1 Equations of State (eos) ................................................. 309 viii
Contents
3.1.9.2 3.1.9.3 3.1.9.4 3.1.9.5
Simha-Somcynsky (S-S) Equation of State ..................... 310 Extension of S-S eos to Binary Miscible Systems ........... 315 Extension of S-S eos to Suspensions .............................. 317 Extension of S-S eos to Nanocomposites ....................... 318 3.1.9.5.1 Diluted, Exfoliated CPNC – Simplified Approach ....................................... 3.1.9.5.2 Dilute, Exfoliated CPNC – Gradient Mobility Approach .......................... 3.1.9.5.3 Intercalated CPNC – Concentration Gradient ... 3.1.9.5.4 PVT – Concluding Notes ...................................
319 324 327 331
3.2 Thermal Stability ........................................................................... 333 3.2.1 Thermal Stability During Processing ........................................... 333 3.2.2 Flame Retardancy and High Temperature Stability ..................... 339 3.2.3 Photo-Oxidative Stability ............................................................ 340
3.3 Rheology ....................................................................................... 341 3.3.1 3.3.2 3.3.3 3.3.4 3.3.5 3.3.6 3.3.7
Introduction ................................................................................ 341 Multi-Phase Flow Behaviour – An Overview .............................. 342 Rheology and Microrheology of Disc Suspensions ...................... 344 Similarity Between CPNC and Liquid Crystal Flow .................... 347 End-Tethered versus Non-Tethered CPNC .................................. 350 Fourier-Transform Rheology of CPNC ....................................... 356 Rheology of CPNC with PA Matrix ............................................ 356 3.3.7.1 Effects of Moisture ........................................................ 360 3.3.7.2 Strain Effects ................................................................. 363 3.3.7.3 Dynamic Flow Curves ................................................... 363 3.3.7.4 Apparent Yield Stress .................................................... 368 3.3.7.5 Zero-Shear Viscosity and the Clay Aspect Ratio ........... 369 3.3.7.6 Flow-Induced Orientation ............................................. 370 3.3.7.7 Steady-State Flow Curves – Shear History Effects ......... 372 3.3.7.8 Fourier Transform Analysis of CPNC ........................... 376 3.3.8 Rheology of CPNC with PO Matrix ........................................... 376 3.3.9 Foaming of CPNC ....................................................................... 384 3.3.10 Rheology of CPNC with PS and Styrenics Matrix ...................... 387 3.3.11 Rheology of CPNC with Other Polymer Matrix Types ............... 390 3.3.12 Rheology of CPNC – A Summary ............................................... 392
3.4 Nucleation and Crystallisation ...................................................... 395 3.4.1 3.4.2 3.4.3 3.4.4 3.4.5
Introduction ................................................................................ 395 Fundamentals of Crystallisation .................................................. 396 Effects of Clay on Crystallisation of PA-6 Matrix ....................... 399 Clay Effect on Crystallisation of Other Polyamides .................... 408 Crystallisation of PO Matrix ....................................................... 409 ix
Clay-Containing Polymeric Nanocomposites
3.4.6 Crystallisation of PEST Matrix ................................................... 414 3.4.7 Crystallisation of Syndiotactic PS Matrix .................................... 416
3.5 Mechanical Behaviour ................................................................... 417 3.5.1 Micromechanics of CPNC ........................................................... 417 3.5.2 Prediction of Tensile Strength ...................................................... 426 3.5.3 Fatigue Resistance of CPNC ........................................................ 428
Contents Volume 2 Part 4 Technology of Clay-Containing Polymeric Nanocomposites 4.1 Thermoplastic CPNC .................................................................... 435 4.1.1 Polyamides (PA) .......................................................................... 435 4.1.1.1 PA-Type Nanocomposites from Toyota ......................... 436 4.1.1.2 PA-Type Nanocomposites from AlliedSignal Inc. .......... 441 4.1.1.3 AMCOL Technology for PA .......................................... 445 4.1.1.4 Other Technologies for the Production of CPNC with PA Matrix ............................................................. 452 4.1.1.5 Mechanical Exfoliation of PA-Type CPNC ................... 462 4.1.1.6 PA-6/Kaolinite Nanocomposites .................................... 469 4.1.2 Polyolefins (PO) .......................................................................... 470 4.1.2.1 Toyota Patents on PO-Based CPNC .............................. 472 4.1.2.2 Dow Patents on CPNC Technology for PO ................... 476 4.1.2.3 Sekisui Chemical Patent on PO-Based CPNC ................ 481 4.1.2.4 Diverse Technologies for the Preparation of CPNC with PO-Matrix ................................................. 483 4.1.3 General Methods of CPNC Preparation ...................................... 498 4.1.3.1 Hudson’s Clay Grafting Method ................................... 499 4.1.3.2 Hasegawa et al. Method with Functionalised Compatibilisers ............................................................. 500 4.1.3.3 CPNC with Amino-Aryl Lactam Clays ......................... 503 4.1.3.4 Ishida’s Method ............................................................. 503 4.1.3.5 Edge Reactions of Clay Platelets ................................... 505 4.1.4 Vinyl Polymers and Copolymers ................................................. 506 4.1.4.1 Polymerisation in the Presence of Clay .......................... 507 4.1.4.1.1 Bulk Polymerisation by the Free Radical and Coordination Methods ............................... 508 4.1.4.1.2 Emulsion and Suspension Methods ................... 516 4.1.4.1.3 Solution Polymerisation Methods ...................... 526
4.1.4.2 Other CPNC Prepared by Solution Method .................. 530
x
Contents
4.1.4.3 Vinyl-Type CPNC Prepared by Melt Compounding ...... 533 4.1.4.4 Vinyl Polymer Matrix – A Summary ............................. 542 4.1.5 CPNC in Water-Soluble Polymeric Matrix .................................. 543 4.1.6 Thermoplastic Polyesters (PEST) ................................................. 553 4.1.7 Polycarbonate (PC) ..................................................................... 565 4.1.8 Liquid Crystal Polymers (LCP) .................................................... 567 4.1.9 Fluoropolymers ........................................................................... 569 4.1.10 CPNC with High Temperature Polymers .................................... 573 4.1.11 Electroconductive CPNC ............................................................. 576
4.2 Thermoset CPNC .......................................................................... 579 4.2.1 4.2.2 4.2.3 4.2.4
Epoxy Resins ............................................................................... 579 Unsaturated Polyester Resin ........................................................ 588 Polyurethanes .............................................................................. 590 Other CPNC with Thermoset Matrix ......................................... 599
4.3 Elastomeric CPNC ........................................................................ 601
Part 5 Performance 5.1 Mechanical Properties ................................................................... 611 5.2 Flame Retardancy of CPNC .......................................................... 611 5.3 Permeability Control ..................................................................... 618
Part 6 Closing Remarks 6.1 Summary ....................................................................................... 625 6.2 The Future .................................................................................... 627 6.2.1 Composition ................................................................................ 627 6.2.2 Method of Preparation ................................................................ 628 6.2.3 Characterisation and Testing ....................................................... 629
Part 7 Appendices 7.1 General and Chemical Abbreviations ............................................ 631 7.2 International Abbreviations for Polymers ...................................... 640 7.3 Abbreviations for Organic Cations Used as Clay Intercalants ....... 646 7.4 Notations ...................................................................................... 649 7.4.1 7.4.2 7.4.3 7.4.4 7.4.5
Notation Roman Letters ............................................................. 649 Notation – Greek Letters ............................................................. 652 Subscripts .................................................................................... 655 Superscripts ................................................................................. 655 Mathematical Symbols ................................................................ 655
xi
Clay-Containing Polymeric Nanocomposites
7.5 Dictionary ..................................................................................... 656 Dictionary References .......................................................................... 692
7.6 Companies Active in Organoclay, and/or CPNC Technology ....... 694
Part 8 References References ..................................................................................... 697
Part 9 Index Index .................................................................................................... 765
xii
Introduction
Part 1 Introduction
Clay-Containing Polymeric Nanocomposites
Introduction
1
Introduction
1.1 General Nanocomposites (NCs) are materials that comprise a dispersion of nanometresize particles in a matrix. The matrix may be single or multicomponent. It may contain additional materials that add other functionalities to the system (e.g., reinforcement, conductivity, toughness, etc.). The matrix may be either metallic, ceramic or polymeric - only the latter type is of interest at present. Depending on the matrix nature, NCs may be assigned into one of the three categories: • Polymeric (PNC), • Ceramic (CNC), • Metallic (MNC). The nanoparticles are classified as: 1. Lamellar, 2. Fibrillar, 3. Tubular, 4. Spherical, and 5. Others. For the enhancement of mechanical and barrier properties anisometric particles, especially lamellae are preferred. However, for rigidity and strength fibrillar are preferred, while for functional NCs (e.g., optical, electrical conductivity) spherical or other particles have also been used. Since the aspect ratio of exfoliated, mineral MMT is p = 50 to 2000, the specific surface is in the order of 750 to 800 m2/g. The reinforcing effect of nanoparticles is related to the aspect ratio (p) (ratio of the length or thickness to that of the diameter) and to the particle-matrix interactions. Independent of the actual dimensions, for p > 500 the reinforcing effects are the same as those of an infinitely large particle. Furthermore, the anisometric particles start overlapping when the volume fraction exceeds the ‘maximum packing volume fraction (φm)’ [Utracki, 1995]: for discs: for rods:
1/φm = 1.55 + 0.0598p 1/φm = 1.38 + 0.0376p1.4
(1)
For example, Equation 1 predicts that for discs with an aspect ratio of p = 500 the overlapping volume fraction is φm = 0.00032 (for rods φm = 0.00004). The overlapping generates a 3D network that in melts is responsible for the yield stress and in solid state for significant reinforcing effects. Hence a small amount of anisometric particles leads to large effects. 1
Clay-Containing Polymeric Nanocomposites Because of the small size, the nanoparticles are invisible to the naked eye; hence, they may be used to engender reinforced, but transparent composites (polymeric or ceramic). On a molecular level, the surface energy of clay particles is high. As a result, adsorbed molecules have a tendency to be strongly bonded in the layer adjacent to the clay surface [Horn and Israelachvili, 1998]. This results in a solid-like behaviour of the 5 to 6 nm thick surface layer and progressive reduction of viscosity with distance to bulk liquid viscosity at about 100 to 120 nm. In the absence of antagonistic interactions (such as hydrophilic nanoparticles in a hydrophobic matrix) the clay/organic liquid system may not require interface modification. PNCs can be used as a matrix for traditional multiphase systems (viz. blends, composites or foams), replacing neat polymers.
1.2 NCs with Ceramic or Metallic Matrix Despotakis [2001] published a review of nanotechnology, where he discussed the importance of the technology to diverse applications, types and sources of nanoparticles and commercial developments.
1.2.1 Metallic Nanoparticles in Amorphous Matrix Iron, cobalt and nickel nanoparticles are obtained either by thermal reduction under hydrogen of silica-based matrices containing 0.1-20 wt% of metal cations or by ionic implantation. Their morphologies and properties (electric or magnetic) depend on the processing conditions. In the case of hydrogen reduction, two kinds of matrices have been used, dense sodalime silicate glasses and porous silica gels. In the first case, reduction is controlled by diffusion of hydrogen inside the glass and yields a broad size distribution. In the second case, much smaller size distributions occur with particles exhibiting either a super-paramagnetic or -ferromagnetic behaviour according to the reduction temperature. Differences are observed between nickel on one part and iron and cobalt on the other. The last two have a greater tendency to form silicates and are more sensitive to re-oxidation. In the case of ionic implantation, the implanted element appears to be both in metallic and ionised states. The metallic fraction gives rise to nanoparticles with a super-paramagnetic behaviour. The ratio of metal:oxide depends on the implantation parameters (dose and energy). Due to the small implantation depth, further thermal treatments can easily lead to total reduction or to a re-oxidation.
1.2.2 Magnetic Oxides in Silica Nanocomposites In the area of thin layer magneto-optical recording, apart from metallic multilayer devices, barium hexaferrite and yttrium garnet are good candidates. An alternative is to disperse particles inside a transparent matrix. Preliminary studies have been performed on the system of iron oxide/silica with 20 wt% Fe2O3. Iron containing gels were prepared from Fe(NO3)3, tetraethyl orthosilicate (TEOS), ethanol, HNO3, H2O and formamide. After gelation and drying, the magnetic properties were measured. Below 700 °C, paramagnetic behaviour was observed, while at higher temperature, ferromagnetic γ-Fe2O3 particles were formed. Near 1000 °C a new phase, closely related to ε-Fe2O3, was observed with a coercive field of ca. 7000 Oe. 2
Introduction
1.2.3 Optoelectronics Semiconductors with carrier lifetime of the order of one picosecond (ps) are necessary for ultrafast optoelectronic applications. NC semiconductors of GaAs grown by molecular-beam-epitaxy at low temperatures (LT-GaAs) show a unique combination of electronic and optoelectronic properties. Implantation of GaAs can result in properties that are similar to those of LT-GaAs: sub-ps carrier lifetime, high resistivity, and high electron mobility. Carrier lifetimes as short as 30 fs were determined for the samples annealed at 500 °C. The short carrier lifetime in high energy ion implanted GaAs seems to be controlled by the nanoscale morphology, i.e., by the carrier capture/ recombination at the intrinsic point and/or by extended defects. Another nanocomposite optoelectronic material system comprises Cu-diffused InP crystals. These contain nanometre-size metallic precipitates that radically modify the dynamics of photoconductivity and photoluminescence decay. Furthermore, they significantly reduce the electron and hole trapping time. Spectral photoresponse measurements evidenced the presence of a well-exposed Urbachtail in the sub-band gap radiation absorption, whereas a Z-scan type experiment has shown an enhancement in both the intensity dependent IR absorption and the nonlinear refractive index.
1.2.4 Summary on Non-Polymeric NC The principles of reinforcement and the strategies discussed for the polymer based NC are valid for the other matrices. Non-polymeric NC systems studied are listed in Table 1.
1.3 NCs with Polymeric Matrix In consequence of the outlined fundamentals, PNCs normally require 1-3 vol% of nanoparticles. They behave as a single phase and single component material. PNCs exhibit transparency, low density, reduced flammability, low permeability and enhanced mechanical properties. Furthermore, they may be easily modified by additives and used replacing neat polymers in polymer blends, traditional composites or foams. There are several methods of classification of polymeric nanocomposites. For example, one may consider how many dimensions of the dispersed particles are in the nanometre range: 1. One dimension. The nanoparticles are in the form of sheets of one to a few nanometres thick to hundreds to thousands of nanometres long and wide, hence they can be named as polymer-layered crystal nanocomposites [Pinnavaia and Lan, 2000a]. These systems are of principal industrial interest and the main object of this publication. There are a wide variety of both synthetic and natural crystalline materials that can be used. 2. Two dimensions. These nanoparticles are elongated, viz. fibres, nanotubes or whiskers, e.g., carbon nanotubes [Ebbesen, 1997] or cellulose whiskers [Favier et al., 1997, Chazeau et al., 1999]. The polymeric nanocomposites containing single-walled or multi-walled carbon nanotubes (CNTs) have been extensively studied. At low loading, they show low density, high mechanical properties, and electrical conductivity. 3. Three dimensions. These are mainly iso-dimensional spherical particles, for example, obtained by the sol¯gel methods [Reynaud et al., 1999], and by polymerisation promoted directly from their surface [von Werne and Patten, 1999]. 3
Clay-Containing Polymeric Nanocomposites
Table 1 NCs with either a ceramic or metallic matrix
4
Matrix
Nano-particles
Properties
Silica, SiO2
NiFe2O4 spinel
Magnetic properties
Silica, SiO2
Fe
By electro deposition
Fe
Zn
Reaction milled
α-Fe
R2Fe14B
Magnetic properties
ZrO2
Al2O3
Pore structure stability
Al2O3
ZrO2
Pore structure stability
Metal particles
Gold shells
Blocking properties
Fe-oxides
Cr-oxides
Magnetic properties
Al2O3
SiC
Creep, strength, toughening
Di-alkyl amine
Exfoliated MoS2
High electrical conductivity
Al2O3-ZrO2
SiC
Microstructure and performance (electrical conductivity, refractoriness, wear and impact resistance) depend on the calcinating temperature
SiO2
ZnO
Sol-gel method
Alumina, Al2O3
SiC
Improved abrasion resistance
Fe/Cr
Cr2O3
Magnetic properties
Al2O3
SiC or Y2O3 (5 vol%)
Sintering
SiC
BNx
SiC grain boundary, oxidation resistance and mechanical properties control by BNx nanotubes
γ-Al2O3
Amorphous SiO2 coated
Reaction sintering
Al2O3
ZrO2
Room temperature electrostatic self assembly
Matrix (e.g., anodic alumina)
Carbon nitride nanofibres and nanospheres, C3N4
Preparation for use in lubricants, catalyst supports, gas storage, and drug delivery
Alumina, Al2O3
C
Structure and superplasticity
Pb-zirconatetitanate
PbS
Piezoelectric and pyro-electric behaviour
γ-Fe2O3
Ag
Negative magnetization
Si3N4
20 vol% SiC
Strength, structure, properties
Introduction
Table 1 Continued... Matrix
Nano-particles
Properties
Al2O3
Ni or Ni-Co
Pressureless sintering, processing, properties
CoPt
Ag
For high density recording media
Metal matrix
Nanosize C-fibres
Structure-related mechanical performance
Al
Cu
Jet vapour deposition
Al2O3ZrO2
BaTiO3
Bio-applications for bone grafting
The materials are either structural or functional. In this book the focus is on the former. To this date, it is primarily the structural nanocomposites based on layered silicates, i.e., clays (CPNC), which have been commercially produced. Clays are easily available and their intercalation methods have been known since the 1930s [Theng, 1974]. Owing to the nanometre-size particles, CPNC show markedly improved mechanical, thermal, optical, and physicochemical properties when compared with neat polymers or their composites [Kojima et al., 1993a]. The improvements include moduli, strength, heat resistance, barrier properties, flammability, etc. Nanometre-scale structures are frequently found in biological materials with impressive performance [Mark, 1996]. For example, bone has a structure of 4 nm thick hydroxyapatite crystals dispersed within a collagen matrix. In the desire to synthesise analogues to biological systems several methods for constructing synthetic composites with a degree of nanometre-scale organisation have been tried. Thus, elongated ceramic particles have been precipitated within polymer matrices by drawing the polymer during the precipitation reaction, silica and CdS have been precipitated in liquid crystal polymer (LCP) [Kovar and Lusignea, 1988; Nelson and Samulski, 1995], metals have been electrodeposited inside the pores of commercial nanopore membranes [Martin, 1996], and polymers have been grown within the cavities of layered inorganic structures and zeolites [Okada and Usuki, 1995; Frisch and Mark, 1996]. Furthermore, an organic-inorganic nanocomposite was formed by dissolving the inorganic polymer (LiMo3Se3)n in a conventional monomer acting as the solvent, then polymerising the matrix in situ [Golden et al., 1996]. The above mentioned methods are well suited for the preparation of a particular composition, but they are not versatile enough to offer good control over nanometre-scale architecture and composition in all resins. Several strategies have been developed for constructing ordered nanocomposites with well-defined, tunable nanoarchitecture and the ability to be used in a wide variety of polymers: 1. Addition of nanoparticles, especially anisometric ones, viz. nanofibres, nanotubes or nanoplatelets. Owing to the expanding industrial interest, the present text will focus on this type of PNC, in particular the ones with the clay platelets. 2. Copolymerisation or grafting polymeric chains with monomers having bulky groups, viz. the polyhedral oligomeric silsesquioxanes (POSS) [Lichtenhan 5
Clay-Containing Polymeric Nanocomposites et al., 1995; Lichtenhan, 1996; Haddad and Lichtenhan, 1996; Lichtenhan and Schwab, 2001; Le´sniak, 2001]. 3. Preparation of ordered nanoscale structures using the LCP technology [Gin et al., 1998]. 4. Sol-gel methods [Mauritz et al., 1995; Wen and Wilkes, 1996; Deng et al., 1998]. 5. Hydrothermal method [Quian et al., 2000]. 6. Others. Polymer nanocomposites are emerging as a new class of industrially important materials. At loading levels of 2-3-vol%, they offer similar performance to conventional polymeric composites with 30-50 wt% of reinforcing material. Note that high filler loading in the latter materials causes an undesirable increase of density; hence heavy parts, decreased melt flow, and increased brittleness. Furthermore, the classical composites are opaque with often a poor surface finish – these problems are absent in PNCs. The clay-containing PNC, a CPNC, offers several advantages over the matrix polymer or classical composites. The main improvements are in: modulus, impact strength, heat resistance, dimensional stability, barrier properties (for gases and liquids), flame retardance, optical properties, ion conductivity, thermal stability, etc. Since these advantages are achieved at low clay loading the density is virtually unaffected and the CPNC may be used to replace the neat polymer in blends, composites or foams. Synergistic effects in these applications have been reported. Current consumption of PNC is only a few kilotons per annum, projected to increase by 2009 to 500 kton/y. The cost differential between the neat matrix and its PNC is about 10 to 15%. CPNC’s main market is in the transport industries, with growing demand for packaging, appliances, building and construction, electrical and electronic, horticulture, power tools, etc. Several reviews on nanoparticles, clays and polymeric nanocomposites are available, viz. on the chemistry of clays [van Olphen, 1977; Wittingham and Jacobson, 1982; Newman, 1987], a short one on clay-containing PNC [Giannelis, 1998], on flammability [Gilman et al., 1998; Gilman et al., 1998], a large and well written one [Alexandre and Dubois, 2000], on clay-containing PNC [Ruban et al., 2000], on synthetic clays and resulting PNC [Carrado, 2000], on nanoparticles [Shipway et al., 2000; Ajayan et al., 2003], and more recent ones on CPNC [Utracki and Kamal, 2002; Okamoto, 2003], etc. Several edited proceedings of nanocomposite symposia are also available, viz. [Utracki and Cole, 2002; Krishnamoorti and Vaia, 2002; Hahn, 2002; Komarneni, 2002; Laine, 2001; Lyon, 2001; Nakatani, 2001; Benedek et al., 2001; Komarneni et al., 1997; Komarneni et al, 2000; Pinnavaia and Beall, 2000], etc.
1.3.1 PNC Definitions Clay-containing polymeric nanocomposite (CPNC): a polymer or copolymer having dispersed exfoliated individual platelets obtained from an intercalated layered material. Compatibilisation: Process of modification of the interfacial properties in CPNC, resulting in formation of the interphase, formation and stabilisation of the desired morphology. 6
Introduction Exfoliated layered material: individual platelets (of an intercalated layered material) dispersed in a carrier material or a matrix polymer with the distance between them > 8.8 nm. The platelets can be oriented, forming short stacks or tactoids or they can be randomly dispersed in the medium. Exfoliation: converting intercalate into exfoliate. Intercalant: material sorbed between platelets (of the layered material) that binds with their surfaces to form an intercalate. Often the intercalant is an onium salt, viz. C18H37-NH3+Cl-, that bonds ionically with platelet anion. Intercalation may involve organic or inorganic salts, monomers, polymers, etc. Intercalated material: layered material with organic or inorganic molecules inserted between platelets, thus increasing the interlayer spacing between them to at least 1.5 nm. Intercalating carrier: a carrier comprising water with or without an organic solvent used to form an intercalating composition capable of achieving intercalation of the layered material. Intercalating composition: a composition comprising an intercalant, an intercalating carrier for the intercalant polymer and layered material. Intercalation: forming an intercalate. Layered material: synthetic or mineral inorganic compound, such as smectite clay, formed of adjacent layers with a thickness, for each layer, of 0.3-1 nm. Interlayer spacing: also known as d-spacing, d001 or basal spacing is the mineral thickness (in MMT = 0.96 nm) and the galley thickness, i.e., the thickness of the repeating layers as seen by XRD. Interlayer thickness = d-spacing less the mineral layer thickness. Li-MMT: lithium montmorillonite. Miscibility: polymer system, homogeneous down to the molecular level, associated with the negative value of the free energy and heat of mixing, ΔGm ≈ ΔHm ≤ 0, and ∂2ΔGm/∂φ2>0. Operationally, it is a thermodynamically stable CPNC, where the clay platelets are well dispersed into a homogeneous polymeric matrix. MMT: montmorillonite. Na-MMT: sodium montmorillonite. H-MMT: protonated MMT, etc. Matrix polymer: thermoplastic, thermoset, or elastomeric polymer in which the intercalate or exfoliate is dispersed to form a clay-based polymeric nanocomposite (CPNC). Nanocomposite (NC): a matrix material (metallic, ceramic or polymeric in nature) having dispersed particles, with at least one dimension that does not exceed 10 nm. Platelets: individual layers of the layered material. Polymeric nanocomposite (PNC): a polymer or copolymer having dispersed in it nanosized particles, viz. platelets, fibres, spheroids, etc. Short stack or tactoid: intercalated or exfoliated clay platelets aligned parallel to each other. Spacing: Two measures of spacing are used: interlayer spacing, (also known as d-spacing, d001 or basal spacing) and interlamellar spacing. The former 7
Clay-Containing Polymeric Nanocomposites comprises the latter plus the platelet thickness. For example for MMT: d001 = interlamellar spacing + 0.96 (nm).
1.3.2 Methods of Characterisation of CPNCs 1.3.2.1 X-Ray Diffraction (XRD) The key to CPNC performance is the extent of intercalation and exfoliation, XRD is the principal method that has been used to examine this. One example of the wide angle X-ray scattering data obtained for polystyrene (PS)/clay systems [Tanoue et al., 2003] is shown in Figure 1 as the scattering intensity versus 2θ, where θ is the angle of diffraction. Within the range of 2θ ≤ 10° the XRD spectrum of PS is featureless. Incorporation of 4.8 wt% of intercalated organoclay shows a distinct peak and a shoulder. Their positions and shapes provide information on the structure of the diffracting species, the organoclay. The presence of multiple peaks in XRD spectra is quite common - it often originates from different organoclay structures and its incomplete change during incorporation in a polymeric matrix [Polzsgay et al., 2004]. The instruments that measure X-ray scattering are divided into the more common wide angle and newer small-angle X-ray scattering machines, WAXS and SAXS, respectively. It is common to consider the scattering angle 2θ = 2° as a boundary between these two, but newer WAXS instruments frequently are able to provide reliable scattering profile down to 2θ = 1°. The interlayer spacing, d001, is commonly determined from the XRD spectrum as arbitrary intensity versus 2θ. The spacing is then calculated from Bragg’s law: d00n = nλ/(2sinθ)
(2)
where n is an integer, θ is the angle of incidence (or reflection) of the X-ray beam, and λ is the X-ray wavelength – most X-ray machines use Cu-Kα1 radiation with λ = 0.1540562 nm. For the principal reflection, n = 1, the dependence given by Equation 2 is shown in Figure 2. It is worth noting that, within the range of interest for CPNC (2θ = 1-12°), there is a straight-line relation between d001 and 1/(2θ): or:
1/2θ = –0.00012773 + 0.11331d001 d001 = 0.0011273 + 8.8253/2θ; R = 1.0000
(3)
where R is the correlation coefficient. The peak position and the interlayer spacing related to it is one part of the information provided by XRD measurements. The intensity of the diffraction peak and its dependence on the concentration of scattering particles yields another. Cullity and Stock [2001] in their monograph on X-ray diffraction derived the following relation for the intensity (I) of the diffraction peak of α-substance mixed with β-substance: Iα = Kφα/[φα(μα - μβ) + μβ]
(4)
where φα is the volume fraction of the diffracting substance α, and μ is the mass absorption coefficient. Depending on the relative magnitude of μα and μβ within the full range of concentration Equation 4 predicts additivity, as well as positive or negative deviation from it. However, within the limited range of clay concentrations used in CPNC, the relation may be simplified to read: 8
Introduction
Figure 1 XRD of PS and PS with 4.8 wt% of Cloisite® 10A (clay treated with dimethyl benzyl hydrogenated tallow ammonium chloride (2MBHTA)) [Tanoue et al. 2003].
Figure 2 Bragg’s law dependencies for Cu-Kα1 radiation with λ = 0.1540562 nm. The straight line is given by Equation 3; correlation coefficient R = 1.0000.
Iα = K´wα Ineat α
(5)
where wα is the weight fraction of substance α. It has been frequently observed that during exfoliation (especially during the mechanical exfoliation of intercalated clay in a PO matrix) the position of the XRD peak remains at the same angular position 2θ, but it broadens and its 9
Clay-Containing Polymeric Nanocomposites intensity decreases. Parallel with these changes there is an enhancement of the CPNC performance. It can be postulated that in this case the exfoliation process involves breakage of intercalated stacks and/or peeling of individual platelets or short stacks. On this basis Ishida and co-workers [2000] used XRD to calculate the degree of exfoliation XE: XE = 100 × [1 – A/A0]
(6)
where A and A0 are the area under the XRD peak for the PNC and for the mixture with intercalated clay, respectively. Replacing in Equation 6 the intensity ratio of Equation 5 by the area under the peak ratio, is motivated by an additional assumption that during the progressive dispersion process many stacks slightly change the interlayer spacing hence the observed broadened peak is an envelope over a family of peaks having similar interlayer spacings. However, there are several possible sources for the XRD peak broadening in a CPNC – one being the assumed above existence of a variety of clay stacks with a range of similar d001 spacing. Another mechanism of peak broadening is based on the imperfections in the crystalline lattice of m-layers of clay platelets forming a stack t = (m – 1) × d001 thick. Another mechanism of peak broadening is based on the imperfections in the crystalline lattice of m-layers of clay platelets forming a stack t = (m – 1) × d001 thick and scattering the X-rays at angles θ1 and θ2. Because of the small angle difference between θ1 and θ2 the destructive interference of reflected beams is incomplete [Cullity and Stock, 2001]. Calculation leads to the following formula credited to Scherrer: t = kλ/(B1/2cosθ) ; k ≅ 0.9
(7)
where θ ≅ (θ1 + θ2)/2 is the angle of X-ray beam incidence corresponding to the peak position, λ is the X-ray wavelength, and B1/2 ≅ θ1 - θ2 is peak width (in radians) at half peak height (Imax/2). From Equation 7 the number of clay platelets per average stack with the interlayer spacing d001 is: m = 1 + t/d001
(8)
Note that in this interpretation the peak broadening is caused by crystalline defects in individual platelets within the stack having about constant interlayer spacing, and not by overlapping peaks that correspond to stacks with different interlayer spacing. The development of technology often requires more precise information on the interlayer spacing than that provided by WAXS. With growing frequency SAXS and small angle neutron scattering (SANS) are being used within the effective scattering angle down to 2θ = 0.05 or the characteristic diffracting distance of about 180 nm. Thus, for example, Bafna et al. [2003] used SAXS and WAXS to study the effects of addition of maleated polyethylene (PE-MA) on PE/MMT structural parameters. By performing scattering experiments on specimens oriented in three orthogonal directions, the authors managed to determine not only the interlayer spacing, but three-dimensional (3D) orientations of six structural features, viz. size of tactoids (ca. 120 nm), organoclay (d002 ≅ 2.4 to 3.1 nm), spacing in undispersed clay (d002 ≅ 1.3 nm), clay (110) and (020) planes, thickness of PE crystalline lamellae (d001 ≅ 19 to 26 nm), and polymer unit cell (110) and (200) planes.
10
Introduction 1.3.2.2 Small Angle Neutron Scattering (SANS) Small-angle neutron scattering (SANS) may also be used to determine d001 spacing in PNC. The method is more sensitive, permitting the range of measurements to be extended to small angles, thus to large spacings. Furthermore, it is readily adapted to different specimens and provides additional structural information not available from XRD. As far as nanocomposites are concerned, SANS was initially used to study structure and interparticle interactions in aqueous dispersions of MMT, hectorite and kaolinite [Ramsay and Lindner, 1993; Brown et al., 1998]. Measurements were carried out on a water dispersion of MMT under static conditions and shear flow in a Couette-type cell. At concentration w = 5 to 65 g/L and low ionic strength, the suspensions were thixotropic. The self-organised structures were more extensive when the MMT particles were more anisotropic. Here, under no-flow condition, SANS detected preferential platelet alignment at distances larger than d001 = 10.3 nm. Under low shear stress, MMT platelets showed preferential alignment in the direction of flow. This orientation occurred over a large range of deformation rates, giving rise to anisotropic scattering; furthermore, spatial correlations persisted. At shear rates exceeding 104 s-1 the ordered platelet structure was destroyed and only preferential alignment was observed. Shear alignment was also observed in dispersions of kaolinite. As before, particle alignment increases with flow rate. More recently, SANS has been used to study the structural aspects of clay dispersions in water-soluble polymers. Jinnai and co-workers [1996] investigated a four-component clay-polymer-salt-water system, consisting of polyvinyl methyl 1/ 2 ether (PVME; Mn = 18 kg/mol, radius of gyration rg2 ≅ 5 nm), Na-vermiculite [Si6.13Mg5.44Al1.65Fe0.50Ti0.13Ca0.13Cr0.01K0.01O20(OH)4Na1.29], n-butyl-ammonium chloride, and heavy water. A single vermiculite crystal was placed in the cell and the concentration of PVME changed. It has been known [Walker, 1960] that depending on the temperature the system has two structures, separated by Tc = 14 °C: below Tc vermiculite shows uniform swelling (gel phase) with d001 of about 12 nm while above this temperature the gel structure degenerates into local tactoids with d001 of about 2 nm (tactoid phase). The neutron diffraction experiments were carried out at the Japan Atomic Energy Research Institute using the SANS-J instrument. The study showed that addition of PVME had no effect on the phase transition temperature between the tactoid and gel phases. Furthermore, in the tactoid phase the spacing of 1.94 nm indicated absence of polymer diffusion into the interlamellar galleries. However, in the gel phase the clay plates were found to be better aligned and more regularly spaced than in the system without the polymer. The diffraction pattern from the polymer-containing sample was sharper. It showed pronounced first-order and strong second-order diffraction peaks, which is rare for an aqueous sample. In the gel phase, d001 decreased with polymer content, from 12 to 8 nm. The conformation and location of the polymer chains in these mixtures were not unequivocally determined. In the gel phase PVME macromolecules can fit in the d001 = 12 nm interlayer space. They are either: (A) adsorbed onto the surface of a single plate, in a flattened configuration, (B) in the supernatant fluid surrounding the gel, or (C) they form bridges between two adjacent clay platelets. 1/ 2 In the tactoid phase it is not possible for a polymer with rg2 ≅ 5 nm to exist as free chains inside the tactoids. 11
Clay-Containing Polymeric Nanocomposites Similar results were obtained when PVME was replaced by polyethylene glycol 1/ 2 (PEG; Mn = 18 kg/mol, radius of gyration rg2 ≅ 5 nm) [Hatharasinghe, 1998]. Thus, the new system consisted of n-butyl ammonium vermiculite, PEG, n-butyl ammonium chloride, and heavy water. In analogy to the system with PVME, here also as PEG volume fraction increased from 0 to 0.04 the d001 spacing (obtained by SANS) decreased from 12 to 6.5 nm (see Figure 3). However, the addition of polymer had no effect on the phase transition temperature between the tactoid and gel phases of the clay system, Tc = 13 ± 1 °C. As for the other system, the PEG presence in the gel phase made the clay plates more parallel and more regularly spaced. The d001 spacing versus PEG volume fraction, φ, was fitted to the hyperbolic relation with three empirical parameters ai: d001 = a0 + a1/(a2 + φ)
(9)
The least squares fit to PEG data yielded the following parameter values: ao = 3.896 ± 1.793; a1 = 0.1363 ± 0.0995; and a2 = 0.0163 ± 0.0091 with the standard deviation σ = 0.571 and the correlation coefficient squared, r2 = 0.9983. In Figure 3 the data from the previously studied system with PVME are also shown. The dependence is similar, indicating that the two chemically different polymers bring about a similar contraction of the gel phase. It is noteworthy that the authors generalised their d001 vs. φ data using an exponential, instead of hyperbolic dependence: d001 = a0 + a1 exp{-a2φ}
(10)
Figure 3 Interlayer spacing of n-butyl ammonium vermiculite vs. volume fraction of the added water-soluble polymer: circles - PEG, squares - PVME. The solid line was calculated by the least-squares fit of PEG data to Equation 9. [Data: Jinnai et al., 1996; Hatharasinghe et al., 1998].
12
Introduction The least squares fit to PEG data yielded the following parameter values: ao = 6.123 ± 0.743; a1 = 6.148 ± 0.737; and a2 = 74.825 ± 24.304 with the standard deviation σ = 0.513 and the correlation coefficient squared, r2 = 0.9986. Thus, Equations 9 and 10 seem to provide similarly good descriptions for the observed phenomenon. However, they give quite different limiting values for infinite dilution of clay platelets at φ → 0, viz. d001 = 12.26 and 80.97 nm, respectively. As can be judged from the data in Figure 3, the limiting value predicted by Equation 10 is not acceptable. The explanation for the mechanism responsible for this reduction of interlamellar spacing should be consistent with the structure of macromolecules in suspension as well as with the rapid phase transition from gel to tactoid. In aqueous medium, n-butyl ammonium vermiculite is expected to form hydrogen bonds with surrounding water. In the absence of polymer, the interlamellar spacing is about 11 to 12 nm. Since water monolayer thickness is about 0.28 nm, the large interlamellar spacing most likely originates from the electrostatic repulsion of butyl ammonium chloride ions associated head-to-tail with butyl ammonium vermiculite salt. Addition of a polymer seems to cause progressive extraction of the ionic species from the interlamellar species. It has been assumed that, because a large number of solvent molecules must be desorbed to accommodate a single polymer molecule, the translational entropy so gained by the system provides a strong driving force for polymeric adsorption. However, this assumption may not be correct in the aqueous media, where the hydrogen bonding of water molecules is energetically preferred. SANS has been used by Carrado and co-workers [1996] to monitor the structural changes in synthetic hectorite upon hydrothermal crystallisation in the presence of polyvinyl alcohol (PVAl). It was found that the PVAl coats the small initially formed silicate particles, hindering their further growth. However, upon removal of the polymer no change has been observed in the extended inorganic network. Similarly, Muzny and co-workers [1996] monitored the dispersion of synthetic hectorite in a polymer matrix. The organically modified clay platelets were dispersed in polyacrylamide (PAA). The studies showed that a homogeneous dispersion was achievable only with a large excess (equivalent to five times the cationic exchange capacity (CEC) of the silicate or higher) of the organic cationic intercalant. Krishnamoorti and Giannelis [1997] studied the linear viscoelastic behaviour of end-tethered poly-ε-caprolactone (PCL) with MMT intercalated with ω-amino dodecyl acid (ADA). The dynamic shear moduli, storage (G´) and loss (G´´), were determined at small amplitude on either freshly loaded specimens, or pre-sheared at large amplitude. The large-amplitude oscillatory shear significantly reduced the linear viscoelastic response, decreasing the low frequency value of G´´ by one and G´ by two decades. Such a sharp decrease of the signal means that in the sheared specimens the polymeric matrix controls the CPNC flow behaviour; hence the MMT platelets are oriented. The alignment was confirmed by SANS. Similar effects have been reported in several rheological studies (see Section 3.2) as well as for injection moulded CPNCs [Kojima et al., 1994; 1995]. Ho and co-workers [2001] studied SANS of clay suspensions in non-aqueous media. The stated aim was to understand and optimise the potential processing conditions for clay dispersed in a polymeric matrix. As a model system the authors used Na-MMT (Cloisite ® Na + ) and MMT intercalated with 2-methyl 2-hydrogenated tallow ammonium chloride (MMT-2M2HT; Cloisite® 15A or C15A) from Southern Clay Products (SCP). Prior to experiments, Na-MMT was 13
Clay-Containing Polymeric Nanocomposites fractionated and from Cloisite® 15A excess of the intercalant was extracted with hot ethanol. The purified samples were dried in a vacuum oven at room temperature for 3 days. The clays were then dispersed in chloroform, benzene, toluene, p-xylene, cyclohexane and octane. Both as-received and purified samples were studied. The scattering profiles and dispersion behaviour in organic solvents of the as-received and purified C15A were significantly different, confirming that the organic modifier was present in excess. However, in both cases the clay platelets were fully exfoliated in chloroform while they were only swollen in benzene, toluene and p-xylene. The scattering profiles indicated that the swollen tactoids of purified clay were thinner, and therefore more numerous. Neither the purified nor the as-received clay showed any temperature effect on scattering. According to atomic force microscopy (AFM) the average diameter of C15A platelets was in the range of 0.4-1.0 μm, giving the aspect ratio: p = 400 to 1000. The SANS experiments at the wave vector q ≡ (4π/λ) sin(θ/2) = 0.004 to 0.517 Å-1 (θ is the scattering angle) were carried out at the National Institute of Standards and Technology (NIST). The measurements were conducted using either dry clay powders or dispersions of C15A in deuterated organic solvents, with Na-MMT in a solution of deuterated/protonated water (D2O/H2O). For well-dispersed, individual thin circular disks, theory predicts that isotropic, total coherent scattered intensity depends mainly on the platelet thickness, diameter, orientation, concentration, and the scattering contrast between the solvent and the clay. The presence of the intercalant as well as formation of stacks could also be accounted for. In agreement with the WAXS data the interlayer spacing of purified C15A was determined by SANS as d001 = 2.43 nm. The d001 spacing in dry Na-MMT was found to be d001 = 0.99 nm. Both SANS and WAXS data indicated that at a concentration of ca. 0.5 to 4 wt% in deionised water, Na-MMT platelets were fully exfoliated. The scattering intensity varied with 1/qn, with n ≅ 2.2. This value is consistent with calculations assuming total exfoliation of platelets, which are 1 nm thick and have diameter of about 600 nm. For C15A in an organic solvent the thickness of the tallow layer depended on the solvent, increasing with the solvent solubility parameter from 1.61 nm (for toluene or xylene) to 1.86 nm (for chloroform). Correspondingly, the calculated average number of platelets in the stack varied from 3 to 1, hence in chloroform the organoclay was fully exfoliated. 1.3.2.3 Transmission and Atomic Force Electron Microscopy (TEM and AFM) At limiting low scattering angle, 2θ ≅ 2°, the XRD/WAXS scattering intensity and resolution decrease, i.e., the method is not useful for spacing: d > 8.8 nm. Within this range TEM may be used to determine the extent of intercalation/exfoliation. However, TEM also offers a direct method for confirming the XRD data (micrograph in Figure 4(a)) and with growing frequency it is being used at low magnification (micrograph in Figure 4(b)) to check on the uniformity (or lack) of clay stack dispersion in polymeric matrix. The low magnifications are also useful to check the purity of clay, e.g., the presence of non-layered particles such as quartz. Dennis and co-workers [2000; 2001] proposed a method for quantification of TEM images. Micrographs with magnification of 130,500 were cover with a mask in which twelve squares of 1 × 1 inch (2.5 cm × 2.5 cm) were cut out. The degree of dispersion was expressed as the number of clay platelets per square inch. This dispersion measure was found to correlate well with the tensile modulus. 14
Introduction (a)
(b)
Figure 4 TEM of intercalated organoclay in PS matrix: (a) magnification 150k, (b) 4.1k [Tanoue et al., 2003a].
Atomic force microscopy (AFM) has been used mainly in a tapping mode at the cantilever’s frequency of 300 kHz and amplitude of 50-100 nm. The difference between the clay modulus and that of a polymer results in good image resolution. Starting in 1990, high resolution TEM (HRTEM) became the preferred tool for the determination of structure, in particular of crystalline nanoparticles, viz. carbon (CNT) or boron nitride nanotubes (BN-NT). The magnification in HRTEM is x106 or better, with resolution of 0.1 nm (JEOL 4000EX, 400 kV, Cs = 1 mm, focus spread = 8 nm, divergence angle = 0.7 mrad) [Gavillet, 2001]. 1.3.2.4 Fourier Transform Infrared Spectroscopy (FTIR) Since 1964 FTIR has been used to study the hydration of bentonite, and the formation of an electrical double layer between the platelets. Significant differences in the silicate stretching region, νSi-O = 1150 to 950 cm-1, are related to water (H2O or D2O) and hydrated cation (Na+, K+,Ca+2, etc.) content, which in turn is related to the interlayer spacing between the clay platelets [Shrewring et al., 1995]. Yan and co-workers [1996] have shown that in hydrated MMT νSi-O exponentially decreases with the clay-to-water ratio, all the way to exfoliation. Oxidation or reduction of the metallic ions within the octahedral clay layers introduces changes to CEC clay hydration and swellability, hence the FTIR spectrum [Yan and Stucki, 1999]. The use of FTIR for characterisation of CPNCs is more recent. Initially, it has been used in studies of the polymer matrix morphology, e.g., conformation and crystallisation behaviour of sPS [Wu et al., 2001] or α- to γ-crystalline form transition of PA-6 [Wu et al., 2002]. However, the Si-O stretching vibration has been found to be very sensitive to long range interactions caused either by imposed 15
Clay-Containing Polymeric Nanocomposites stress [Loo and Gleason, 2003], or expansion of the interlayer spacing [Cole, 2003]. The advantage of the spectroscopic methods, FTIR and Raman, is that along with XRD they are applicable to the intercalated system, and stretch to exfoliated CPNC that do not scatter X-rays. To examine the interactions between the clay platelets, the intercalating agent and polymer, FTIR provides important information. By comparing the experimental and calculated spectra the type and intensity of interactions can be identified [Aranda and Ruiz-Hitzky, 1999]. The method has also been used to analyse the thermal decomposition of ammonium intercalants during the melt compounding method of CPNC preparation [Tanoue et al., 2003]. More details are provided in Section 3.2. 1.3.2.5 Nuclear Magnetic Resonance Spectroscopy (NMR) Solid-state 1H, 13C, 15N and 29Si magic angle spinning (MAS) NMR spectroscopy at frequencies 100.40, 30.41 and 79.30 MHz, respectively has been used. The tensile modulus (E) was found [Usuki et al., 1995] to be proportional to the chemical shift (Cs; in respect to tetramethyl silane) of 15N in ammonium-clay complex with the slope: dE/d(Cs) = 0.1097 (GPa/ppm). The chemical shifts provide information about the degree of clay hydration, the interactions engendered by intercalation and the structure of clay-organic matrix complexes. NMR has also been used to determine PEG chain dynamics within the interlayer spacing of synthetic mica/MMT [Schmidt-Rohr and Spiess, 1994]. According to Aranda and Ruiz-Hitzky [1992] the intercalation of MMT by PEG increased the interlayer spacing by Δd001 = 0.8 nm. Thus, the polymer conformation within these interlamellar galleries may be either a 0.8 nm diameter helix, or two chains in a planar zigzag conformation. The former was deemed to be more probable. Measurements of the 1H NMR line widths and relaxation times across a large temperature range were used to determine the effect of bulk thermal transitions [Kwiatkowski and Whittaker, 2001]. The 13C cross-polarity/magic angle spinning NMR spectra of PEG within the nanocomposite showed that the type of motion being experienced by these chains is the helical jump motion of the α-transition, thus the same as within the crystalline phase of neat PEG. The short proton spinlattice relaxation time in the rotating frame, 1H T1ρ, measured across a wide range of temperatures by 1H NMR provided additional evidence that these chains undergo helical jump motion. Measurements of 1H NMR spectra across a wide temperature range have confirmed that large amplitude motion of the PEG chains within the montmorillonite nanocomposite persist below the Tg of neat PEG. There was no observed change in the rates of relaxation at the transition temperatures expected for neat PEG. Solid-state NMR, both proton and 13C, was used to study CPNC with PA-6 as matrix [VanderHart et al., 2001]. The nanocomposites with 5 wt% organoclay were generated either by blending or by in situ polymerisation. The systems contained mineral MMT having non-stoichiometric amounts of Mg2+ and Fe3+ ions substituted into the octahedral, central layer of the clay platelet. The presence of Fe3+ ion contaminants in the MMT induced paramagnetic properties. The paramagnetic contribution to the proton longitudinal relaxation time (T1H) is a function of the field and Fe3+ concentration in the clay. These paramagnetic properties can be used to determine the hard-to-get information on CPNCs, viz. the degree of dispersion, the stability of intercalant, etc. Evidently, NMR can also
16
Introduction provide information on the preponderance of α- and γ-crystalline phases of PA-6 in CPNCs. The α-crystallites are characteristic of the neat PA-6, while γ-crystallites are formed in the presence of clay platelets. The Fe3+ induced paramagnetism of MMT and the resulting spin-diffusionmoderated reduction in longitudinal proton relaxation time, T1H, may be used to rank the degree of clay dispersion in CPNCs, and to investigate morphological stratification of the PA-6 α- and γ-crystallites with respect to the clay surface. It was found that variations in T1H correlate well with TEM measurements of the clay dispersion. The chemical stability of dimethyl dihydrogenated tallow ammonium ion (2M2HTA) used as MMT intercalant was also investigated. During the organoclay compounding with PA-6 at 240 °C most of the intercalant decomposed, releasing a free amine with one methyl and two hydrogenated tallow substituents. According to the authors, the combination of temperature and shear stress in blending caused decomposition. However, judging by T1H, the CPNCs with the best dispersion of clay also had the most extensively degraded intercalant. The polarity of PA-6 macromolecules well compensated for the loss of the 2M2HTA intercalant. Solid state NMR has also been used to quantitatively determine the degree of clay dispersion in PS/MMT nanocomposites [Bourbigot et al., 2003]. A new method, similar to the one described above, was developed. In both, paramagnetic Fe3+ within the octahedral layer of MMT has been used. The results correlate with XRD and TEM data. The new method is significantly faster than TEM, but with the difference that the information pertains to the bulk of the specimen, not to its surface. Evidently, both these methods are applicable only to clays containing paramagnetic Fe3+. 1.3.2.6 Other Methods MAS-NMR is particularly valuable for the reactive systems, i.e., those PNC that are prepared by dispersing intercalated clay into a monomer, which is then polymerised. Direct evidence that the reaction involves Si atoms has been reported [Sellinger et al., 1998; Komori et al., 1999a,b]. Diverse calorimetric methods have also been used. For example, a differential scanning calorimetry (DSC)/dynamic thermal analysis (DTA) method was used to characterise the layered material, e.g., to study clay hydration [Lagaly et al., 1975]. Thermogravimetric analysis (TGA) has been used for examining interlayer packing density [Vaia et al., 1994]. Chromatography was used to determine the heat of reaction involved in the formation of intercalate complex [Bruno et al., 1999]. The cone calorimeter is an indispensable tool for flammability studies. The measurements are conducted according to ASTM E 1354-92 at an incident flux of 50 kW/m2 using a cone shaped heater. Exhaust flow is set at 24 L/s and the spark continues until the sample ignites. All samples should be run in duplicate and the average value reported. The results are reproducible to within about 10% [Gilman et al, 1999; Gilman et al., 2000a,b; Zhu et al., 2001a,b]. Electrophoretic mobility provides an important insight into the surface charge of clay particles. As expected, large changes have been reported as a function of pH. This information is crucial for the optimisation of the intercalation process [Wilson et al., 1999].
17
Clay-Containing Polymeric Nanocomposites
Copper sulfide
1.3.3 Determination of PNC Properties Standard test methods have been used to determine PNC properties. CPNC usually contains low clay concentration, which results in increased melt viscosity, significant improvement of barrier properties, reduction of flammability, improved optical properties and increased rigidity. Thermodynamic and flow properties and the diverse effects of nanoparticles on the nucleation and crystallinity of polymeric matrices are discussed later in this book. For example, it has been reported that the impact strength of crystalline polymer is reduced by incorporation of nanoparticles. For example, addition of 2.2 wt% of organoclay to PA-6 reduced the impact strength by a factor of 4.3. However, recent data suggest that the problem is related to the influence of clay on the crystallisation kinetics. By proper selection of process variables, the same impact strength of PNC as of the matrix resin was obtained [Graff, 1999]. The fundamentals of the mechanical behaviour, details on the measurements of the mechanical and barrier properties of CPNCs, as well as on the flame retardancy are included in later sections of this book.
1.3.4 PNC Types and Methods of their Preparation As shown in Table 2, during recent years many different polymers have been used as a matrix for CPNC with a diversity of nanoparticles and for a variety of applications, from enhancement of mechanical performance, reduction of permeability, good flame retardancy, improved optical properties, engendering magnetic, electric or light-transmitting performance, biocompatibility, thermochromic effects, etc.
1.3.5 PNCs of Commercial Interest As shown in Table 3, the commercially important PNCs all contain exfoliated clays, in particular montmorillonite (MMT). The dimensions of the latter particles are typically: thickness 0.96 nm, with width and length from ca. 50 to about 1500 nm. Several large plastics producing companies have heavily invested in the development and production of CPNC (Toyota Central Research and Development Laboratories (Toyota – for short), Unitika, Ube, AlliedSignalHoneywell, RTP Co., Basell, GM, Showa Denko, Bayer, BASF, Solutia, Dow Plastics, Magna International, Mitsubishi, Eastman, etc.). At least one company (Nanocor) is totally dedicated to the manufacture of organoclays and polymer concentrates for use in CPNC. Ube Industries, RTP Co. and Bayer have started production of the experimental PA-based PNCs with clay nanoparticles for moulding or film blowing. GE Plastics has developed conductive grades of PC and PPE/PA experimental alloy with carbon nanotubes. Additional information is included in Table 4. After 2.5 years of development, recently, General Motors Corporation (GMC) announced that in collaboration with Basell Polyolefins (50-50 BASF-Shell joint venture created in 1990 by merging Elenac, Basell and Targor), Southern Clay Products, and Blackhawk Automotive Plastics Inc., it has developed thermoplastic olefin nanocomposite (or TPO-based CPNC) for step-assist use in 2002 GMC Safari and Chevrolet Astro vans. The key to success is the exfoliation of MMT achieved during polymerisation.
18
Nanoparticles
Copper sulfide
Au-nanoparticles
Lyophilized smectic clays
Amorphous Fe2O3 nanoparticles
TiO2
Organo-platinum or organo-silver spherical nanoclusters
CdS
NiO
α-RuCl3
Ni
PbS
Au or Ag nanoparticles
Polymer
Polyacrylic acid
Poly(2-oxazoline)
PS, polymethyl methacrylate (PMMA), styrene-acrylonitrile copolymer (SAN)
PVAl
Polyaniline (PANI)
Poly(styrene-bacrylic acid)
Polyacrylonitrile or polyacrylamide
Polyimine
A polymer
Polyacrylamide
Polyvinyl acetate (PVAc)
Polyethylene (PE)
Optical anisotropy
Synthesis under gamma radiation
Synthesis under gamma radiation
For magnetic properties
Synthesis under gamma radiation
The bound organo-metal was reduced by hydrogen
Polymerisation in the presence of nanoparticles
Uniformly dispersed super paramagnetic nanoparticles
Free radical emulsion or solution polymerisation
Polymerisation on functionalised Au
Optical properties of films
Process and/or key characteristics
Table 2 Examples of PNCs with functional properties
[Dirix et al, 1999a,b]
[Qiao et al., 2000]
[Liu et al., 2000]
[Wang et al., 2000]
[Castro et al., 2000]
[Qiao et al., 2000 a,b,c,d]
[Brown et al., 2000]
[Su and Kuramoto, 2000]
[Kumar et al., 2000]
[Hoffman et al., 2000a,b]
[Jordan et al., 2001]
[Gotoh et al. 2001]
References
Introduction
19
20
Nanoparticles
Tetraethoxysilane
Tetraethoxysilane
Complexing and immobilisation of dendrimer particles with (CuS)15
Carbon nanotubes
Cellulose whiskers
Core-shell MeO-PEG/PLAmethacryloyl latex particles
Fe
Intercalated clay
Aromatic PA fibers
MMT from Milos
Polymer
PCL
Hydroxyethyl methacrylate (HEMA)
Poly(amido-amine) dendrimer
PVAl
PVC
Poly(ethylene glycol-b-lactide) with a methoxy group at PEGend and methacryloyl group at PLA
Polymer solution
PET
Aromatic PA
Poly(4-vinylpyridine) (P4VP) (neutral, salt and quaternised forms)
Protonation by acidity of MMT
Polymer blending
[Fournaris et al., 1999]
[Akita and Kobayashi, 1999]
[Frisk and Laurent, 1999]
[Yu et al., 1999]
Nanocomposites by reduction of Fe2+ Polycondensation
[Kim et al., 1999]
[Dubief et al., 1999]
[Shaffer and Windle, 1999a,b]
[Beck Tan et al., 1999]
[Hajii et al., 1999]
[Tian et al., 1999]
References
Stable core-shell type spherical, stable nanoparticles with a number-averaged diameter of about 30 nm
Mechanical performance
Thermomechanical and electrical properties
SAXS and SANS indicate CuS at periphery of dendrimer molecules.
In situ acid-catalysed sol-gel in free radical polymerisation
Chemical synthesis of a biocompatible silica-aliphatic polyester hybrid
Process and/or key characteristics
Table 2 Continued...
Clay-Containing Polymeric Nanocomposites
Nanoparticles
Colloidal nanometer size manganese dioxide (MnO2)
Iron oxide nanoparticles
TiO2
Nano particles of ZnS
Zirconia
Mn-Zn ferrites
Fe2O3
CdS
Cu2+- complex and Cu
Cadmium sulfide
Layered double hydroxide, Mg6Al2(OH)16CO3 × 4H2O
Ag
Polymer
Poly-N-vinylcarbazole polypyrrole or polyaniline (PANI)
PANI
Styrene-maleic anhydride copolymer (SMA)
Poly(styrene-co-acrylic acid)
Polydimethyl siloxane (PDMS)
A polymer
Poly(maleic monoester)
Polycarbonate-co-poly(p-ethyl phenol)
Poly(itaconic acid-co-acrylic acid)
Polydiacetylene
Polystyrene sulfonate, polyvinyl sulfonate
Polyacrylamide
10
Synthesis of nanocomposites at RT
Surface and interphase properties
Nonlinear optical properties
Enzyme-catalysed synthesis
Thin films
Room temperature synthesis
Sol-gel method
Simultaneous polymerisation and precipitation
Covalently bonded, stable 3D nanocomposites.
Cast film preparation of conductive NC
DC conductivities improved 10 to 10 fold
7
Process and/or key characteristics
Table 2 Continued...
[Zhu et al., 1998]
[Wilson et al., 1999]
[Schwerzel et al., 1998]
[Huang et al., 1998]
[Wang et al., 1999]
[Lee and Kang, 1998]
[Mathur et al., 1999]
[Dirè et al., 1998]
[Cheng et al., 1999]
[Wang et al., 1999]
[Tang et al., 1999]
[Biswas et al., 1999]
References
Introduction
21
22 Intercalation Strong effects of processing on NC performance
V2O5
Mo 2 S
Cellulose whiskers, 5 × 300 nm
Emulsion ultrasonication followed by polymerisation
CdS
PP was blended with smectite
NbSe2
LiMoO3
Ag
Ag
Au
SiO2-TiO2 nanoparticles
Polypyrrole or polythiophene
PEG
Latex
Carboxylated-PS
Thiol-containing polymers
PP
A polymer
A polymer
Polyacrylamide
Polydiacetylene
PMMA or PS
Perfluoro sulfonate Ionomer
[Wang et al., 1997]
[Tsai, 1997]
[Kurokawa et al., 1997]
[Premachandran et al., 1997]
[Mei et al., 1997]
[Favier et al., 1997]
[Gonzales et al., 1998a,b]
[Goward et al., 1998]
[Massam et al., 1998]
References
Sol-gel process
Radical polymerisation
Optical properties
[Shao et al., 1996]
[Gonsalves et al., 1996]
[Zhou et al., 1996]
Synthesis under gamma radiation synthesis [Zhu et al., 1997]
Encapsulative precipitation
Superconducting polymers
X-ray diffraction and IR spectroscopy studies performed
Enzymatic synthesis
Stable nanoparticles
Interleaved nanocomposites for lithium batteries
Improved tensile strength and modulus at T < Tg
Dispersed clay with alkylammonium ions
Epoxy
Process and/or key characteristics
Nanoparticles
Polymer
Table 2 Continued...
Clay-Containing Polymeric Nanocomposites
[Lemmon and Lerner, 1995] [Jarjayes et al., 1995]
V2O5
V2O5
SnO2
TiS2
γ-Fe2O3 particles
MoS2
Ag, 20 to 50 nm diameter
CaCO3, 15 to 40 nm diameter
PANI
Polypyrrole
PEG
Polypyrrole
Polyethers
A polymer
PVC
Ca(OH)2 + CO2 in rotating bed reactor + surface treatment
At 0.1 to 2 wt% long-term antimicrobial effect
Preparation and characterisation
Electrical and magnetic characterisations
Polymerisation in the presence of nanoparticles
Redox polymerisation of aniline in V2O5 Xerogel
Synthesis, characterisation, and transport properties
[Chen et al., 2002]
[Aymonier et al., 2002]
[Lemmon and Lerner, 1994]
[Maeda and Armes, 1995]
[Wu et al., 1996]
[Kloster et al., 1996]
[Liu et al., 1996]
Polymer electrolyte Xerogel
Synthesis, structure, and reactions have been studied
V2O5 intercalative nanocomposites
References
PEG
Process and/or key characteristics
Nanoparticles
Polymer
Table 2 Continued...
Introduction
23
24
Toyota, AlliedSignal, Ube, Nanocor, Unitica, Showa Denko, Bayer, BASF, and SolutiaDow. Ube CPNC is commercially available; RTP offers to supply PNC on demand
Basell, GM, Southern Clay Products Inc., Toyota, Ford, Dow Plastics, and Magna International
Toyota
Eastman, Bayer, BASF, Tetra-Laval
Polyolefins, esp. polypropylene (PP) and copolymers
Polystyrene and blends with poly(styrene-covinyl methyl oxazoline)
Polyethylene terephthalate (PET)
Major players
Polyamide, esp. PA-6
Matrix
Solution-expanded clay particles present during the polycondensation
Transparency, low permeability, high strength and stiffness, low density
High strength, tensile modulus increased by 37% (over PS), 43% smaller thermal expansion
Low density (0.91 g/ml) with stiffness equivalent to composites with 35 wt% talc, dimensional stability, low-T impact strength, ductility to – 35 °C, high heat ageing, 75% lower flammability, excellent surface finish
Exfoliated smectite ≤5wt%. Compatibilizer is needed; e.g., Toyota uses maleic anhydride modified PP
Surface-treated clay particles; 4.8 wt%
Compared to PA: similar density, transparency, 70 °C higher HDT, 70% higher tensile modulus, 130% higher flexural modulus, 50% lower oxygen permeability, 70% lower flammability, etc.
Properties
Exfoliated silicates, e.g., 1-3 vol% Namontmorillonite. Exfoliation is crucial
Nanoparticles
Table 3 Examples of commercially important PNCs
Food packaging, esp. as central layer in co-formed beer, juice and soft drink bottles
Aerospace, automotive
Automotive: body panels, door panels, interior trim, instrument panels, pillars, consoles, etc.
Automotive (e.g., truck mirror housing, engine covers), tool housing, garden equipment, telecommunication, aerospace, speciality application, barrier film for food packaging, etc.
Applications
Clay-Containing Polymeric Nanocomposites
Nanocor, Mitsubishi
Showa Denko, Bayer
Toyota, Mitsui, Showa Denko, Mitsubishi
Polyoxymethylene or acetal (POM)
Polyolefins (PO)
Major players
Ethylene-vinyl alcohol copolymer (EVAl)
Matrix
Polypropylene columnar crystals
Montmorillonite (?)
Exfoliated smectite clay particles; loading ≤5-wt%
Nanoparticles Properties
Low density, high rigidity and strength
Low warpage, low shrink, high surface quality, HDT increased by 45 °C, 40% higher modulus
NC to be the central layer in multilayer, co-formed products
Table 3 Continued...
Automotive
Automotive under-hood applications, electronics, etc.
Packaging films, for moisture and oxygen sensitive foods and electronics
Applications
Introduction
25
Clay-Containing Polymeric Nanocomposites
Table 4 Examples of companies active in the nanotechnology field Company
Comments
Advanced Refractory Technologies, Inc., Buffalo, NY
Diamond-like nanocomposites as a protective wear resistant thin film for commutator bars in DC motors.
AlliedSignal Corp. (now Honeywell)
Toughened PA nanocomposites for films, injection and blow moulding. Compared with PA, the nanocomposites have 50-80% higher stiffness and up to 70 °C higher HDT with ca. 1% increase of density. The nanocomposites retain PA toughness, clarity, hot-fill heat resistance, and oil/grease resistance. PA-6 with 2 or 4 wt% organoclay has three or six-fold reduction of O2 transmission rate (OTR).
Applied Sciences
Manufacturer of vapour-grown carbon nanotube made by pyrolysis of coal. Pyrograf-III comes in 100- and 200-nm diameter and has potential as an electrically conductive additive and modifier of plastics' coefficient of thermal expansion. The nano-tubes enhance electrical conductivity over a broad resistivity range and boost mechanical properties – at 0.5% loading the volume resistivity is ca. 104 ohm-cm.
Argonide
Manufactures alumina ceramic nanowhiskers by electroexplosion of metal wire. The NanoCeram whiskers (2 nm diameter, ca. 100 nm length) are used as reinforcements and thermally conductive additives.
Basell
In collaboration with General Motors and Southern Clay Products developed nanoclay/TPO compound for automotive step assist vans of 2002. The step is moulded by Blackhawk Automotive Plastics.
Bayer AG
Producer of PA-6 nanocomposites for transparent barrier film packaging Durethan LPDU 601-1 and LPDU 601-2, which offer different degrees of barrier improvement (LPDU is data link layer Protocol Data Unit delivered through a network.
Carbolex
Carbon nanotube sales for research and industry.
Clariant Masterbatches of Milford, Delaware
CPNC with PP matrix for packaging applications. Supplier of Nanomer masterbatches (with PO matrix) for use in making nanocomposite plastics.
Crenova (Vestamid)
Manufacturer of electrically conductive PA-12 with carbon nanotubes.
Dow Plastics
Developing in-reactor compounding of nano-PP by using nanoclays as the catalyst support for in situ polymerisation of PP. The focus is on highly loaded (up to 10% clay) systems for semi-structural automotive uses.
26
Introduction
Table 4 Continued... Company
Comments
Eastman Chemical
Jointly with Nanocor developing nanocomposites of PET via the in-reactor approach. Its initial focus is on rigid containers, coinjected with the nanocomposite barrier layer in the middle.
GE Plastics
Manufacturer of electrically conductive PPE/PA blends with CNT for automotive painted parts.
Hybrid Plastics
Pioneered the development of polyhedral oligomeric silsesquioxanes (POSS) for plastics. The technology bridges the property space between hydrocarbon-based plastics and ceramics. It imparts new or improved properties to materials through the controlled reinforcement of polymer chains at the molecular level.
Hyperion Catalysis
Producer of carbon nanofibers, viz. Graphite Fibril™ nanotubes. Its hollow carbon nanotubes 10 nm in OD diameter are 10 to 100 microns long, for at least a 1000:1 aspect ratio and are primarily used to provide high conductivity at 0.5 wt% loading. Hyperion supplies masterbatches containing 15-20% fibrils in, e.g., PA-6, -12 and -66, PBT, PC, PEEK, PEI, PPS, and PS, PP, PET, and EVA.
IBM
R&D activities on nanocomposites in applications such as polymer light emitting diodes.
Kabelwerk Eupen (Belgium)
EVAc/organoclay compositions for wire & cable applications. Drastic reduction of heat release at 3 to 5 wt% clay loading.
Nanocor; subsidiary of the AMCOL International Corp
Formed to capitalise on the patented technology for chemically modified clays into plastic resins, improving strength, heat stability and barrier properties. Supplier of organoclay and its concentrates PA-6, MDX6 and PP. Licensed by Toyota Central R&D Labs to use their technology.
Nanomaterials Research Corporation
Precision manufacturer of powders and devices.
NexTech Materials, Ltd.
Manufacturer of innovative ceramic products (nano powders) to meet the needs of the chemical, electrochemical, electronics and consumer markets.
Physical Sciences Inc.
Research contract from the National Science Foundation to develop an advanced carbon nanotube membrane for direct methanol fuel cells.
27
Clay-Containing Polymeric Nanocomposites
Table 4 Continued... Company
Comments
Polymeric Supply
Thermoset CPNC with unsaturated polyester for marine and transport applications.
RTP Co.
Manufacture organoclay-reinforced PA-6 and PP, as well as polyether ether ketone (PEEK) (and other polymers) with carbon nanotubes for static-dissipative bodies, e.g., in semiconductor applications.
Shenzhen Chengyin Technology Co., Ltd.
Focused on the R&D of nanostructured titanium dioxide in anti-ultraviolet, antibacteria, antistatic and photocatalysis.
Showa Denko
Commercial production of PA-66 and POM nanocomposites by melt compounding with organoclay for improved flame retardancy and rigidity. The FR grades: Systemer FE 30600 and 30602 provide UL 94V-0 and V-2 performance, respectively, at 0.4 mm thickness. Their flex moduli are 30-80% higher, and HDT 30 to 80°C higher, than those of neat PA.
Southern Clay Products, Inc, (SCP) Gonzales, TX
Manufacturer of clay and clay products, e.g., intercalated clays for use in CPNC.
Toyota Central R&D Laboratories, Nagakute
Hold the first patents on nanocomposites. Licenses for the technology of PA-6 nanocomposites given first to Ube International and Nanocor. Other licensees for this as well as PP-based CPNC technology followed. Toyota succeeded in producing CPNC by melt compounding in a twin-screw extruder.
Ube International
Manufactures the nylon-clay hybrids (NCH), of PA-6, PA66, PA-12, and PA-6/66 for extrusion and injection moulding. Compared to neat PA-6, NCH has 68% higher tensile modulus and 126% higher flexural modulus. At 2 wt% clay the oxygen permeability is reduced by a factor of two.
Unitika Co.
PA-6 nanocomposites for injection moulding (e.g., Nylon M2350) produced using a proprietary reactive exfoliation technology.
Yantai Haili Ind. & Commerce of China
Ultrahigh molecular weight polyethylene (UHMWPE)/organoclay for earthquake-resistant pipes.
28
Introduction A TPO with 2.5 wt% clay is as stiff as PP with 10 times the amount of talc. The PNC is much lighter; the weight savings can reach 20%, depending on the part and the material that is being replaced by the TPO nanocomposite. The nanocomposites are stiffer, less brittle at low temperatures, more durable and more recyclable than currently used materials. Parts made of nanocomposites cost about as much as conventional TPO, but it takes less material to manufacture them. GMC expects the price to improve as the volume of CPNC used by the transport industry will increase. However, for use in body panels the TPO CPNC requires additional research.
1.3.6 Journals and Research Groups During the last few years there has been an explosion of scientific journals related to nanotechnology: 1. Carbon Nanotubes: Synthesis, Structure, Properties and Applications, Eds., M.S. Dresselhaus, G. Dresselhaus and P. Avouris, Applied Physics Series, Springer, Berlin; 2001. 2. Fullerenes, Nanotubes and Carbon Nanostructures, Marcel Dekker, Inc., New York, 2002. 3. International Journal of Nanoscience, World Scientific, Singapore, 2002. 4. Journal of Nanoparticle Research, Kluwer Academic, Dordrecht, 1999. 5. Journal of Vacuum Science & Technology. B, Microelectronics and Nanometer Structures Processing, Measurement and Phenomena, Published for the American Vacuum Society by the American Institute of Physics, 1991. 6. Lab on a Chip, Royal Society of Chemistry, Cambridge, UK, 2001. 7. Materials Science & Engineering. C, Biomimetic and Supramolecular Systems, Elsevier, Amsterdam, 1999. 8. Microscale Thermophysical Engineering, Taylor & Francis, London, 1997. 9. Nano Letters, American Chemical Society, 2001 10. Nanostructured Materials, Pergamon Press, c1992-c1999; absorbed by Acta Materialia. 11. Nanotechnology, Pergamon Press, 1990. 12. Physica E. Low-Dimensional Systems & Nanostructures, North-Holland, Amsterdam, 1997. 13. Virtual Journal of Nanoscale Science & Technology [electronic], American Institute of Physics, 2000. While there is no monograph on PNC, there are several volumes of proceedings published [Komarneni, 1996; Komarneni et al., 2000]. Several industrial consortia have been established for the sole purpose of advancing R&D on PNC, viz. DowMagna International, Toyota-Ube-Mitsui-Mitsubishi, Basell-GM-Southern Clay, etc. In Canada and the USA, there are also research consortia on CPNC at, e.g., the National Research Council Canada (NRCC), the National Institute for Science and Technology (NIST) and Edison Polymer Innovation Corporation (EPIC). NRCC activities focus on CPNC development, NIST concentrates on the inflammability of CPNCs. To the EPIC consortium five Ohio-based universities have submitted project proposals on development of PNC for thermoplastics, coatings, elastomers, advanced composites, and for core technology (rheological behaviour, nanoparticle/polymer 29
Clay-Containing Polymeric Nanocomposites interface, and particle surface chemistry) where fundamental issues will be investigated that apply to all of the application technologies. Activities focus on coatings, electro-optical behaviour, barrier properties, elastomers, synthesis and processing, advanced composites and core technologies. NSF announced the Nanoscale Science and Engineering (NSE) program for the USA universities with US$74 million budget for FY 2001. The number of international conferences on PNC is growing exponentially – in 2001 there were at least six major conferences in Europe and North America. Guided by the evident industrial interests, this monograph will focus on the PNCs developed as structural materials comprising exfoliated clay particles.
1.3.7 Historical Perspective Intercalation of clays started in the 1930s. For hydrophilic applications in paper coating, H2O + Na4P2O7 + compounds with -OH groups (e.g., PEG) were used at high stresses [Maloney, 1939]. Dry smectite galley spacing of Δd001 = 0.35 nm, increased upon swelling with 3 molecular layers of H2O to 1.2-1.4 nm. In the early 1950s, for hydrophobic applications clay started to be intercalated with quaternary alkonium chlorides. For example, fatty quatenary alkonium chlorides were used, viz. methyl and/or benzyl hydrogenated tallow ammonium chloride (containing 2.0% C14, 0.5% C15, 29.0% C16, 1.5% C17, 66.0% C18 and 1.0% C 20 alkyl groups). The organoclays, commercialised by NL Industries as Bentone™, were found useful for thickening lubricating greases, oil-based muds, packer fluids and paint-varnish-lacquer removers [Hauser, 1950; Jordan, 1950]. Soon, the process was described in the textbooks [Grim, 1968]. The first use of an organoclay in polymers was to reinforce elastomers [Carter et al., 1950]. The patent describes MMT intercalation using diverse onium compounds, viz. ammonium, phosphonium, arsonium, stannonium, etc. The organoclay was combined with elastomeric latex, and then processed by standard methods. In 1961, Blumstein reported polymerisation of vinyl monomers (e.g., methylmethacrylate) in the presence of intercalated MMT. In 1963 Nahin and Backlund patented low density polyethylene (LDPE)-organoclay 1:1 compositions. In the organoclay the amount of onium salt (e.g., hexamethylene diamine) varied from 4 to 16 wt%. The aim was to produce rigid, γ-ray crosslinkable compound. The patent describes methods of intercalation, hot mill blending, moulding and irradiation. Nanocomposites with PVC and PS were also discussed. In the latter case, the authors stress the difference in behaviour of PS melts compounded with organoclay and polymerised in its presence. In 1976 Fujiwara and Sakamoto (from Unitika) filed a patent application for the use of ammonium-salt intercalated clays in a hydrophobic matrix. In particular, the organoclay was added to monomer prior to its polymerisation into a polyamide – this led to the first CPNC. A few years later, Toyota obtained the first US patent for the polymerisation of several vinyl monomers, e.g., styrene, in the presence of clay [Kamigaito et al., 1984]. The composition contained 85 wt% of clay, hence exfoliation was impossible, but in the presence of styrene the interlayer spacing of MMT expanded from d001 = 1.25 to 1.5 nm. Upon addition of dichloro-dimethyl silane rapid polymerisation resulted in good mechanical performance. In the following years attention shifted to dispersion of small quantities of MMT in PA-6 – the work resulted in several US patents [Okada et al., 1988; Kawasumi et al., 1989]. The process consisted of caprolactam polymerisation in the presence of organoclay.
30
Introduction The melt exfoliation process for PA-based PNCs was developed in the early 1990s at AlliedSignal [Maxfield et al., 1995, 1996; Christiani and Maxfield, 1998]. The later patents describe a three step process: 1. Treating a suspension of clay with peptising Na6P6O18 in aqueous solution at 50-90 °C with organosilanes, organotitanates or organozirconates that increase the interlayer spacing to at least d001 = 5 nm. 2. Saturation of the dried clay with precursor monomer then polymerising it. 3. Compounding the modified clay with matrix polymer until the desired level of exfoliation is reached. The complexes of organosilanes, organotitanates and/or organozirconates (with or without onium salt) showed prolonged thermal stability at temperatures above 300 °C. Owing to the high hydrophobicity and non-polar nature of polyolefins, preparation of PO-based CPNC is more difficult; hence historically it followed that of PA-based ones. The first patents from Toyota were for elastomers reinforced with clay and carbon black (CB) [Usuki et al., 1989]. To prepare a PP-based CPNC Kawasumi and co-workers [1997] intercalated NaMMT using an oligomeric PO with polar telechelic hydroxyl groups. Next, a general method for the preparation of PNC with diverse polymers was disclosed [Usuki et al., 1999]. The process involves three steps: (i) Intercalation of clay with an onium ion rendering it compatible with a ‘guest molecule’. (ii) Contacting the organoclay with the ‘main guest molecule’ at T ≤ 250 °C. (iii) Transferring the modified clay to a reactor for PO polymerisation (preferred) or blending with a synthetic resin. Since the guest molecule and the polymer added in the 3rd step may not be the same, it is possible to form CPNC with a miscible blend as matrix. The patent listed vinyl-based polymers, thermosetting resins and rubbers (e.g., PE, PP, PS, polyisobutylene (PIB), acrylics, thermoplastic polyurethane (TPU), styrenebutadiene-styrene terpolymer (SBS), liquid butadiene rubber (BR), polybutadiene (PB), etc.) as the matrix. For example, maleated-PP (PP-MA) with high acid value hence low MW) was used as the ‘guest molecule’ that resulted in good exfoliation [Kato et al., 1997]. In a patent from Dow, PO-based CPNC was prepared by Nichols and Chou [1999]: (i) Intercalation with an organic, polymeric or inorganic intercalant (e.g., Si(OC2H5)4, Si(OCH3)4, Ge(OC3H7)4, Ge(OC2H5)4) that resulted in Δd001 = 0.5-60 nm. (ii) After drying at 50-80 °C or calcination at 450-550 °C, the intercalated material was dispersed in a monomer or melt-blended until at least 80 wt% of the layers were exfoliated (aspect ratio p = 10-2000). The claims mention LDPE and linear low density PE (LLDPE) copolymers with density ρ = 850-920 kg/m3 and a melt index (MI) = 0.1-10 g/min. Another method for the preparation of PO-type PNCs followed similar steps [Hudson, 1999]: (i) MMT was functionalised in H2O/EtOH with aminosilane, e.g., amino ethyl-dimethyl ethoxysilane;
31
Clay-Containing Polymeric Nanocomposites (ii) Carboxylated or maleated PO was grafted to the functionalised MMT by amine-carboxyl bond; and (iii) 0.1-50 wt% of the modified particles was dispersed in PE or PP. The key is the physical bonding through co-crystallisation between the grafted PO and the main PO resin. One of the drawbacks of the ammonium salt intercalated NC is the low temperature stability. To solve the problem, Ellsworth [1999] used 10-80 phr organophosphonium R1P+(R2)3 cations (where R1 is a C8-C24 alkyl or arylalkyl group and each R2 is an aryl, arylalkyl, or a C1-C6 alkyl group) with a melt processable fluoroplastic. The dispersion was carried out by melt blending in a twin-screw extruder (TSE). The resulting interlayer spacing was d001 ≥ 3.5 nm. While organoclays intercalated with conventional, quaternary ammonium cations are stable only up to about 250 °C, the one intercalated with tributyl-hexadecylphosphonium bromide (3BHDP) was reported stable to about 370 °C, making it possible to prepare CPNC with high temperature engineering or speciality polymers. Another method for PNC preparation uses bentonite intercalated with primary C12- or C16-ammonium chloride and treated with an epoxy [Ishida et al., 2000]. The authors consider the method applicable to any matrix polymer. The key factor has been the addition of 2 wt% epoxy (e.g., Epon 825; see Figure 5). In the first step of the process, MMT was treated with C12- or C16-ammonium chloride, filtered then dried overnight at 100 °C. The modified clay and epoxy were added to molten polymer and melt mixed for 10 to 120 min. The epoxy was capable of swelling the ammonium-treated clay, allowing virtually any monomer or polymer to interdiffuse. Intercalated or exfoliated composites have been prepared with 24 different polymers, using a clay loading of 10 wt%. Comparing the XRD spectra of samples containing epoxy with those for samples without epoxy led to the conclusion that there was an increased dispersion of clay in the former systems. Several factors affect the efficiency of nanocomposite formation as well as its structure. Therefore, while intercalation and/or exfoliation are not complete for all the polymers tested, lower clay loading, longer mixing time, and higher swelling agent concentrations might have resulted in a better exfoliation. Table 5 shows that for several polymers (with 10 wt% clay (MMT-ADA) and 2 wt% epoxy (Epon 825)) mixing for 30 min resulted in a partial nanocomposite structure.
Figure 5 Chemical composition of Epon-series of epoxy resins from Shell. Epon 828 contains three species: 88% with n = 0, 10% with n = 1 and 2% with n = 2 (hence an average = 0.14). In Epon 825 n = 0; in Epon 826 = 0.07; in Epon 1001 = 2.3 and in Epon 1004 = 4.8. Similar resins are available from other manufacturers, e.g., Dow (DER-series) or Ciba (Araldite-series).
32
Introduction
Table 5 Results of CPNC formation after melt mixing a polymer for 30 min with 10 wt% of MMT-ADA and 2 wt% epoxy [Ishida et al., 2000] Polymer
Exfoliated (%)
MW (kg/mol)
Result
Polyethyl methacrylate (PEMA)
50
515
exfoliated
Polyisobutylene
39
500
exfoliated
Poly(1-butadiene)
23
420
exfoliated
Polyisoprene
26
410
exfoliated
Polybutyl methacrylate (PBMA)
66
337
intercalated
Chloroprene (CR)
81
188
exfoliated
Poly(1-butene)
26
185
exfoliated
SAN
85
185
intercalated
Polyoxadecyl methacrylate (PODMA)
0
170
intercalated
PE
68
125
intercalated
PMMA
50
120
exfoliated
PVAc
70
60
exfoliated
PS
54
45
exfoliated
PC
64
29
exfoliated
PCL
54
15
exfoliated
PEG
82
10
intercalated
PVAl
80
2
exfoliated
Polytetrafluoroethylene (PTFE)
89
unknown
exfoliated
PP
77
unknown
intercalated
PA-12
99
unknown
exfoliated
PVC
100
unknown
exfoliated
POM
70
unknown
exfoliated
33
Clay-Containing Polymeric Nanocomposites In principle, owing to the polarity of thermosetting monomers or pre-polymers, the exfoliation of clays in these resins is simpler. The low molecular weight monomeric components may diffuse more easily to the interlayer galleries. For example, Pinnavaia and co-workers [Pinnavaia and Lan, 1998; Wang and Pinnavaia, 1998] prepared epoxy nanocomposites with layered silicilic acid (magadiite). The CPNC showed great improvement in tensile modulus and strength. Using alkylammonium, dimethyl di-octadecyl amine (MMT-2M2ODA) the clay layers were found spaced at about 8 nm. In another study Wang and Pinnavaia [1998b] used MMT intercalated with protonated octadecylamine (MMT-ODA) and PEG, then crosslinked it with diisocyanate. Thus prepared elastomeric polyurethane nanocomposites had the clay dispersed in the form of tactoids. Detailed discussion on these systems may be found in Part 4.2.
34
Introduction
Part 2 Basic Elements of Polymeric Nanocomposite Technology
Clay-Containing Polymeric Nanocomposites
Nanoparticles of Interest to PNC Technology
2.1
Nanoparticles of Interest to PNC Technology
2.1.1 General Nanoparticles used in polymeric nanocomposites have been divided into three categories defined in terms of the number of dimensions of their nanometre size, viz. one dimension (platelets), two dimensions (fibres and whiskers), and three dimensions (nearly spherical particles). Layered nanoparticles that can be exfoliated into a dispersion of individual platelets are of main interest. Out of these, in industry the mineral or synthetic clays have a dominant position, hence these will be discussed in greater detail.
2.1.2 Layered Nanoparticles The layered materials of interest to CPNC technology have an average platelet thickness from 0.7 to about 2.5 nm. A partial listing is given in Table 6. The average
Table 6 Layered nanoparticles for use in CPNC Smectite clays
Montmorillonite, Bentonite, Nontronite, Beidellite, Volkonskoite, Hectorite, Saponite, Stevensite, Sauconite, Sobockite, Svinfordite
Other clays
Vermiculite; Illite, Ledikite, Attapulgite, Magadiite, Kenyaite, Mica
Synthetic clays
MgO(SiO2)s(Al2O3)a(AB)b(H2O)x (where AB is an ion pair, e.g., NaF); synthetic Magadiite; Lucentite and Somasif (from CoOp); Laponite (Mg4Li2Si8O20OH2F2+2, from SCP); Fluorohectorite (Corning)
Layered 2nhydroxides
Mg6 Al3.4 (OH)18.8 (CO3)1.7 H2O Zn6 Al2 (OH)16 CO3 nH2O
Chlorides
FeCl3, FeOCl
Chalcogenides
TiS2, MoS2, MoS3, (PbS)1.18(TiS2)2
Cyanides
Ni(CN)2
Oxides
H2Si2O5, V6O13, HTiNbO5, Cr0.5V0.5S2, W0.2V2.8O7, Cr3O8, MoO3(OH)2, VOPO4-2H2O, CaPO4CH3-H2O, MnHAsO4-H2O, Ag6Mo10O33, etc.
Others
Graphite, oxidised graphite, nanotalc, etc.
Smectite clays Chalcogenides
Synthetic clays Hydroxides Cyanides Oxides
Chlorides
35
Clay-Containing Polymeric Nanocomposites interparticle spacing between layers of the layered material or fibrils of the fibrillar material may depend on concentration – the higher the loading the smaller the spacing. Since PNCs are mainly used as structural materials, the preferred layered materials are phyllosilicate clays of the 2:1 type, more precisely smectites, and in particular montmorillonite (MMT). The layer surface has 0.25 to 0.9 negative charges per unit cell and a commensurate number of exchangeable cations in the interlamellar galleries. The amount of this high aspect ratio nanomaterial that needs to be added to a polymeric matrix to engender CPNC with improved performance can be as little as 5 ppm. The aim is to totally exfoliate the platelets, but frequently doublets and short stacks (tactoids) may also be present. Considering the importance of MMT a more detailed description of its structure and properties is given below. While MMT is abundant and inexpensive its main drawback is that it is a mineral with variable composition, which is impossible to totally purify. Variability of CPNC has frequently been blamed on structural (particle size distribution and aspect ratio) as well as chemical (surface reactivity) variability. Consequently, there is a growing interest in synthetic or at least semi-synthetic layered materials with well-controlled physical and chemical properties. While experiments with these model systems are essential for development of a basic knowledge of intercalation and exfoliation, the hope is that the technology can be developed to produce synthetic layered materials on a large-scale for the manufacture of CPNC with consistent performance characteristics. To prepare functional PNCs other layered materials (than MMT) as well as non-layered materials have been used, e.g., metallic particles, metal oxides, metal sulfides, etc. The structure and size of these particles depends on the method of preparation, which in turn is dictated by the principal application of these materials. There is a growing interest in synthetic layered nanofillers (see Table 7). Hectorite has a significantly smaller aspect ratio (p) than MMT, viz. p < 100, but their synthetic homologues have even smaller diameter flakes, viz. p = 10 to 30. By contrast with hectorite, synthetic fluoromica (FM) can be prepared with high aspect ratio, e.g., p < 2,000 [Yano et al., 1993]. Their advantages/disadvantages are summarised in Table 8. A simple method for producing synthetic fluoromica (FM) at a relatively low temperature was described by Tateyama et al. [1993]. Accordingly, a powdery mixture of 10 to 35 wt% of an alkali silicofluoride (e.g., Na2SiF6) as the main component (optionally with an alkali fluoride) and natural talc is heated for about one hour at 700 to 900 °C. The resulting material has the general formula: αMFxβ(aMgF2xbMgO)xγSiO2 where M is an alkali metal (Li, Na, K), and 0.1 ≤ α ≤ 2: 2 ≤ β ≤ 3; 3 ≤ γ ≤ 4; a + b = 1 are coefficients. For example, swellable FM may have the composition: talc/LiF/Na2SiF6 = 80:10:10; or talc/Na 2 SiF 6 /Al 2 O 3 = 70:20:10. The heating temperature greatly affects swellability and the interlayer spacing. For example, the fluoromica produced at 700-750 °C shows the XRD peak at d001 = 0.91 nm, while that produced at 780-900 °C has the XRD peak at d001 = 1.61 nm. A one-step method for the preparation of synthetic grafted smectite clays has been described [Carrado et al., 2001]. By contrast with the older methods that start with a mineral precursor, e.g., talc, to produce Li-hectorite (CEC ≅ 0.8 meq/g), the new process involves sol-gel hydrothermal transformation of an organotetraethoxy silane (TEOS) or organotrialkoxy silane, viz. phenyltriethoxy silane (PTES). Aqueous slurries of LiF, magnesium hydroxide, and the silane are refluxed for 2-5 days. 36
Nanoparticles of Interest to PNC Technology
Table 7 Synthetic clays Company
Synthetic clay
Clay type
Laporte Industries, Ltd., Luton UK/Southern Clay Products Inc., P.O. Box 44, Gonzales, TX 78629 - (888) Fax (210) 672-7206; http://www.laponite.com/
Laponite LRD or RD was developed by Laporte more than 30 years ago, and has been manufactured at the Widnes site since 1985
SYnL-1; hectorite; 25 x 10 nm
NL Industries; Baroid Division,
[email protected]
Barasym SMM-100, composition (%): SiO2: 49.7 Al2O3: 38.2, TiO2: 0.023, Fe2O3: 0.02, MgO: 0.014, Na2O: 0.26, K2O: 0.043. 2.1.3.1.2 Computation of Potential CNT Properties To explore the possibilities of CNTs the theoretical computation of properties of individual SWNTs, their aligned crystalline bundles and interactions with polymeric matrix have been carried out. These lead to sometimes surprising results that require experimental confirmation. Structural properties of a CNT crystal were computed by Tersoff and Ruoff [1994]. The calculations used a valence force model for computing atomic interactions within each tube, and the Lennard-Jones 6-12 potential to calculate interactions between tubes in the bundle. The computations were carried out for 41
Clay-Containing Polymeric Nanocomposites bundles of CNTs of uniform diameter, ranging from d = 1 to 6 nm. A local maximum on the elastic modulus (M) versus d was found for dmax ≅ 1.5 nm. Similarly, the cohesive energy, Ea, versus d was found to go through a local maximum. The computed value of the equivalent bulk modulus in the radial direction, Mmax ≅ 0.3 eV/Å3 = 0.69 TPa, is comparable to the tensile modulus in the axial direction, E = 1 to 1.5 TPa. The computations also indicated that while the small diameter tubes, d < 2 nm, are perfectly cylindrical, those with larger diameters are flattened against each other by van der Waals forces. Using molecular dynamics (MD) Yakobson et al. [1996] studied deformation of CNTs under load in axial compression, bending and torsion. Depending on the stress, the nanotubes reversibly transformed into different morphologies, with each shape change associated with an abrupt release of energy and a singularity in the stress-strain curve. The authors also noted that the MD computations resulted in behaviour predicted by the classical continuum shell model with properly chosen parameters (tensile modulus, E = 5.5 TPa, Poisson ratio, ν = 0.19, and SWNT wall thickness h = 0.066 nm). The model accurately predicted the SWNT behaviour beyond the linear response. The simulations showed that CNTs sustain large deformations without signs of brittleness, plasticity, or atomic rearrangements. Lordi et al. [1999] studied the distribution of pentagons on CNT tips. Five pentagons are located at the tube end. From high-resolution TEM and computerimage simulation, the authors identified the pentagon distribution, which, as it was shown, controls the oxidation. However, the applicability of the continuum model to CNT deformation is not universal [Harik, 2001]. The author derived two non-dimensional parameters that control SWNT buckling: 1. The SWNT aspect ratio, pCNT ≡ dCNT/LCNT 80%)
OD = 10-30; L = 1,000-10,000
US$1.5 to 2.0
€500 €250
45
Clay-Containing Polymeric Nanocomposites they are slow. Furthermore, there are serious difficulties in developing interactions at the nanotube/matrix interface. Thus, the use of CNTs is limited to high value added applications and niche markets, for example, in electronics and space. 2.1.3.1.4 Sources As shown in Table 9, numerous organisations on three continents offer CNTs. Depending on purity, method of synthesis and geographical location, the price varies from US$1.5 to 1,500 per gram. There is rapid progress in the preparative methods, which results in increasing purity of the ‘as received’ CNTs. If the presence of residual catalyst is not detrimental to the desired performance of nanocomposites, purification may not be necessary and the cost immediately drops by a factor ranging from 2 to 20 (for MWNT and SWNT, respectively). Nanocyl offers functionalised CNTs for an additional €50/g. The functionalities include: -H, -OH, -Cl, -CO, -COOH, -NH2, -SH, -SCH3, etc. Hyperion Catalysis commercialised unpurified CNTs at US$44/kg (15-20% purity). These are also available in masterbatches with PC, PBT, PET, PS and PEEK. GE Co. has developed a polyphenylene ether (PPE)/PA alloy (Noryl GTX-990EP) with ca. 2-wt% of CNT for electrostatic painting [Scobbo, 1998; Scobbo et al., 1998]. The alloy has low density (1080 kg/m3), HDT 158 °C, notched Izod impact strength 600 J/m2, flexural modulus of about 2.4 GMPa, tensile strength 60 MPa, elongation at break 22%, bulk resistivity of ca. 30 Ωm and MI = 4.8. The alloy is being used for automotive mirror housings, door handles, gas tank caps, fuel lines and plastic fenders. The technology is being adapted to acrylonitrile-butadiene-styrene terpolymer (ABS), PC, thermoplastic polyester (PEST) and PA automotive compounds. Toray Industries has developed a catalytic CVD method of producing dualwall carbon nanotubes (DWNT), which when commercially produced in 2004 will be available at US10¢/g. According to the Mitsubishi Research Institute, by 2020 the market for fullerenes and CNT will expand to US$3.6 billion. The projected value of nanostructured materials in 2005 and 2010 is US$96 and 208 billion, respectively [Despotakis, 2003]. 2.1.3.1.5 PNC with CNTs for Electrical Conductivity The earliest use of CNTs in polymeric nanocomposites was for engendering functional properties, e.g., a semi-conducting, non-linear, light-sensitive current injection (light-sensitive electrical conductivity) [Romero et al., 1996]; for molecular optoelectronics consisting of CNT dispersed poly(m-phenylenevinylene-co-2,5-di-octoxy-p-phenylene-vinylene) (PmPV) with the electrical conductivity increased by up to eight orders of magnitude [Curran et al., 1998]; carbon nanotubes helically wrapped by poly(phenyl acetylene) chains for reduced photodegradation [Tang and Xu, 1999]; polymerisation of pyrrole onto CNT for enhanced electrical, thermal and magnetic properties [Fan et al., 1999]. MWNTs were dispersed in polyphenylene vinylene (PPV) by spin coating for high quantum photovoltaic efficiency (ca. twice as large as that of indium-tinoxide) [Ago et al., 1999]. A novl PNC of an electroluminescent, conjugated PmPV and MWNTs was prepared by sonication of the two constituents in toluene [McCarthy et al., 2000]. As observed under TEM, the polymer crystallised on the CNT, coating it uniformly and growing out in the form of tree branches. MWNTs have the ends normally closed, which requires structural defects, viz. dislocations, sp3-hybridised carbon, or the presence of either pentagons or 46
Nanoparticles of Interest to PNC Technology heptagons. Defects may also occur along the tube body. These nucleate the crystalline growth of the semiconjugated PmPV. Sandler et al. [1999] reported that a CNT loading of 0.1 vol% in epoxy resulted in an increase in electrical conductivity of four orders of magnitude (to 0.01 S/m). A nonionic surfactant has been used to investigate the effects of the degree of CNT dispersion on conductivity [Gong et al., 2000]. PNCs of MWNTs (d = 80-90 nm) with polypyrrole (PPY) have been prepared using in situ polymerisation [Chang et al., 2000]. The PNCs were characterised using SEM, TEM, XRD, Raman scattering, thermogravimetric analysis (TGA), conductivity, and magnetic susceptibility measuring devices. The PPY adhered to the CNT exterior, without chemically bonding. The process affected polaron density and the orientation of PPY macromolecules. The magnetisation of the composite was found to be the sum of the magnetisations of the components. The semiconductor-like conductivity of the PNC was larger than that of PPY. The thermal stability resembled that of PPY. CNT modification, their conversion into carbides: SiC, WC, etc., coating with metals or organic electrically conductive polymers has been carried out for a decade. Recently, one-dimensional PPY/CNT nanocomposites were produced by electrochemical deposition of PPY onto CNT in well-aligned large arrays [Chen et al., 2001a]. The coating thickness was controlled by the film-formation charge. Uniform thickness from 10 to 93 nm was obtained, changing the morphology from coated individual tubes to filling the gap between all CNT in an array, forming uniform conductive material. Unpurified SWNTs with a broad distribution of diameters (dav ≅ 1.1 nm) and lengths was dispersed in Epon 862 to increase its thermal conductivity [Biercuk et al., 2002]. At 1 wt% loading the PNC showed a 70% increase in thermal conductivity at 40 K, rising to 125% at room temperature. The percolation threshold for electrical conductivity was between 0.1 and 0.2 wt% of the SWNT loading. These results suggest that the thermal and electrical properties of SWNT-epoxy nanocomposites may be improved without the need to purify or chemically functionalise the nanotubes. 2.1.3.1.6 Graphite When electrical conductivity is desired, CNT can be replaced with exfoliated graphite. Chen et al. [2001b] applied this strategy. The authors used natural flake graphite with diameters ranging from 50 to 1000 μm. The graphite was first expanded by treating it with a mixture of concentrated sulfuric acid and nitric acid for about 16 h. After washing and drying at 100 °C the material was heat treated at 1050 °C for 15 s, thereby expanding graphite particles in the c-axis (or orthogonal) direction by a factor of about 350 (compared to the original graphite). The expanded graphite was mixed with styrene (St) and methyl methacrylate (MMA) mixture (St/MMA = 70/30), in the presence of benzoyl peroxide, then heated at 150 °C for 30 min and cooled to room temperature (RT). A black solid was crushed, rolled on a twin-roll mixer for 5 min, and moulded into 4 mm thick rectangular plates. TEM showed that the graphite was dispersed in the form of exfoliated sheets forming 10-40 nm thick stacks. The transition from electrical insulator to semiconductor occurred when the expanded graphite content was 1.8 wt% while at 3.0 wt% loading the electrical conductivity increased from 10-14 to 10-2 S/cm hence by 12-decades. This enhancement may be attributed to the high aspect ratio (p), of the dispersed graphite – from the percolation threshold concentration 47
Clay-Containing Polymeric Nanocomposites its value, p ≅ 99, was calculated. To preserve the high value of p, hence the electrical conductivity, extensive roll-milling should be avoided. Addition of up to 5 wt% of exfoliated graphite linearly increased the tensile strength from 24 to 29 MPa and reduced the notched Izod impact strength from 29 to 19 J/m. Similarly, Xiao et al. [2001] prepared PNC based on exfoliated graphite with PS as matrix. Benzoyl peroxide was dissolved in styrene then exfoliated graphite was added to the reaction vessel. The reaction mixture was stirred for 4 h at 85 °C then for 2 h at 150 °C. Increasing the amount of exfoliated graphite caused the molecular weight (MW) and molecular weight distribution (MWD) of PS to increase. The glass transition temperature (Tg) increased to ca. 124 °C. As a control, the same compositions were prepared by compounding dry blends at 170 °C in a two-roll mill. The in situ prepared PNCs had a higher thermal stability than either PS or PS/exfoliated graphite prepared by melt blending. The volume resistivity versus exfoliated graphite content is shown in Figure 8. It is noteworthy that incorporation of 5 wt% exfoliated graphite into PS, reduced the volume resistivity by about 3 or 17 decades, respectively for the PNC prepared by melt blending or by polymerisation. The difference in conductivities originates in the structural differences between these two types of PNC. 2.1.3.1.7 PNC with CNTs – Thermoset Matrix The more recent applications of CNT are for enhancement of mechanical performance. For the optimum effect CNTs should be dispersed within a polymeric matrix and bonded to it. Furthermore, since the enhancement is related to the effective aspect ratio, the process must minimise the CNT attrition. The nanotubes are dispersed by mechanical means or by ultrasonication (1) in a monomer(s), which is then polymerised, or (2) in a polymer solution.
Figure 8 Reduction of volume resistivity of PS as a function of incorporation of exfoliated graphite. The upper dependence is for melt compounded PNC, whereas the lower one is for nanocomposites prepared by polymerisation of styrene in the presence of graphite [Xiao et al., 2001].
48
Nanoparticles of Interest to PNC Technology The former, reactive approach is preferred for PNC with thermoset matrix. Most of the early work has been done on epoxy systems. For example, Schadler et al. [1998] dispersed 5 wt% of MWNT in Epon 828, and then cured it with triethylene tetraamine. In standard tension and compression tests only relatively small increases of moduli were found, viz. tension increased from 3.10 (for neat epoxy) to 3.71 GPa, while compression rose from 3.63 to 4.50 GPa. The Raman peak position that indicates strain in the C-C bond under load, significantly shifted under compression, but not under tension. Lourie and Wagner [1998a,b,c; 1999] reported similar results for SWNT/epoxy PNC. These results indicate a basic problem with load transfer in tension, particularly severe for PNCs containing MWNT. Only the outermost layer of each MWNT may be bonded to the matrix. The interactions between individual layers in MWNT are relatively weak, a van der Waals-type, and the shear strength between layers is small, similar to that in graphite (σ12 ≈ 0.48 MPa), confirmed by pullout experiments in AFM [Yu et al., 2000a]. These authors measured the tensile strengths of MWNT using a ‘nanostressing stage’. The outermost layer showed a Young’s modulus of E = 270 to 950 GPa, and a tensile strength of 11 to 63 GPa. As the straining continued, the outer layer broke first via the ‘sword-in-sheath’ failure. Another reason for the relatively low enhancement of epoxy properties by incorporation of CNT is the tendency of nanotubes to aggregate into bundles. A nearly constant value of the Raman peak in tension was related to tube sliding within the nanotube bundles and, hence poor interfacial load transfer between nanotubes and the matrix [Ajayan et al., 2000]. It is the low modulus of the bundles that controls the PNC performance, and not the axial modulus of individual tubes. To enhance the reinforcing effects the authors suggested three methods: 1. Breaking the bundles into individual tube fragments and randomly dispersing them in the matrix; 2. Radiation or chemical crosslinking of the tubes within bundles to increase the bundle rigidity and eliminate the inner tube slippage; and 3. Obtaining strong carbon CNT/matrix interfacial interactions. Cooper et al. [2001] dispersed SWNTs or MWNTs in epoxy and studied the micromechanical properties using Raman spectroscopy. SWNTs have been deformed in a diamond anvil pressure cell. Upon hydrostatic compression the disordered-related peak at wavenumber 2640 cm-1 up-shifted at an initial rate of 23 cm-1/GPa. However, the band downshifted upon application of a tensile stress. These Raman peak displacements provided evidence of the stress transfer from the matrix to CNT, thus reinforcement effects. The authors calculated the effective modulus of SWNT and MWNT dispersed in epoxy as > 1 TPa and about 0.3 TPa, respectively. Raman spectroscopy combined with mechanical testing was used to probe the alignment of CNT in PNC [Wood et al., 2001]. SWNT (0.1 wt%) was dispersed in a UV curable urethane acrylate oligomer by ultrasonication, mixed with the curing agent and then sheared to induce flow orientation. The UV cured films (~150 μm thick) were evaluated, recording the Raman spectral shifts with strain in the longitudinal and transverse direction. The shifts obtained, in parallel and perpendicular to the flow direction, were significantly different. The adhesion between CNT and the polymer exceeded the shear yield strength of the matrix. Purified MWNTs were dispersed by sonication in a 1,2-dichloroethane solution of epoxy, polyethersulfone (PES) and 4,4-diaminodiphenylsulfone at 84 °C [Qiao 49
Clay-Containing Polymeric Nanocomposites et al., 2002]. Then, the solvent was evaporated and the film crosslinked. The range of MWNT loadings explored was from 0 to 40 wt%. The bending strength increased by ca. 136% to ca. 42 MPa. The volume resistivity decreased by six orders of magnitude from 8.59 × 105 to 0.1415 MΩcm. For the purified MWNT the percolation threshold of electrical conductivity was high, ca. 20 wt%, indicating that the CNT were isolated from each other by a layer of polymer. 2.1.3.1.8 PNC with CNTs – Thermoplastic Matrix As in the case of the thermoset matrices, here there are also two main routes to formation: (1) by dispersing CNT in a polymer solution, and then evaporating the solvent, or (2) dispersing CNT in monomer(s) and then polymerising it (them). However, for thermoplastics a third method, melt compounding has also been tried. Evidently, the melt compounding method would be preferred in an industrial environment, thus it is being explored with growing frequency. 1. The Solution Method Whereas the reactive approach is preferred for thermoset matrices, the solution method has been favoured for thermoplastics. The method involves preparation of a polymer solution and mixing it with a dispersion of CNT in the same solvent. For example, Jin et al. [1998] first ground MWNT, then dispersed it in chloroform, and sonicated. The suspension was then added to a chloroform solution of polyhydroxyamino ether (PHAE; a thermoplastic reaction product of an amine with diglycidyl ether and epoxy; BLOX from Dow). To align the nanotubes, cast films (with up to 50 wt% CNT) were stretched at T = 95 to 100 °C up to 500%. The alignment was confirmed by XRD. Shaffer and Windle [1999b] dispersed CNT in aqueous PVAl solutions (MW = 85-146 kg/mol), with CNT. Stirring at 480 rpm was required to prevent aggregation, and then the adsorbed polymer sterically stabilised the system. The films were cast under controlled water evaporation conditions. The tensile elastic modulus and damping versus CNT concentration and temperature were examined. The data could be fitted to the theoretical expression if E = 150 MPa was assumed for the MWNT modulus and the effective length of 35 nm. A threshold for the electrical conductivity was obtained at about 20 vol%. Unpurified SWNTs (d = 5 to 20 nm, l ≈ 1 μm for >70%), produced by the arc discharge method were used in another study. The CNT and PMMA were mixed together in toluene in an ultrasonic bath for 24 h. Films (about 200 nm thick) were obtained by spin casting [Stéphan et al., 2000]. The Raman spectra suggested that PMMA intercalated into CNT bundles. At low concentration, the quantity of intercalated PMMA may lead to a destruction of bundles, causing a uniform dispersion of CNT in the solution. The films were prepared for use in multilayer diodes. PS (MW = 48 or 280 kg/mol) was dissolved in toluene, and MWNT was dispersed in it using high energy ultrasonication for 0.5 to 120 min [Qian et al., 2000]. Next, the PS solution and MWNT suspension were mixed in an ultrasonic bath for 30 min. The mixture was cast producing uniform, ca. 0.4 mm thick film for tensile tests. The optimum sonication time increased with the nanotube length, viz. 30 and 60 min for 15 and 50 μm long tubes (d = 33.6 nm), respectively. A homogeneous dispersion was obtained without attrition. Tensile tests of the films showed that addition of 1 wt% MWNT increases the modulus by 36-42% and
50
Nanoparticles of Interest to PNC Technology stress at break by 25%, indicating significant load transfer across the nanotubematrix interface. The observed enhancement of stiffness was found to be in good numerical agreement with the values calculated from the relation [Hill et al., 2002]:
⎡ 3 1+ 2pK1φ 5 1+ 2K2φ ⎤ Ec /E p = ⎢ x + x ⎥ 8 1 − K2 φ ⎦ ⎣ 8 1− K1φ
p ≡ l f /d f ; K1 ≡
(E /E ) − 1 : (E /E ) + 2p f
f
p
p
K2 ≡
(E /E ) − 1 (E /E ) + 2 f
p
f
p
(11)
where E is tensile modulus of a composite, polymer or fibre (subscript c, p or f, respectively), p is the fibre aspect ratio and φ is the fibre volume fraction (calculated as 0.00487). The modulus of MWNTs depends on diameter; here Ep = 450 GPa was used. Substituting these parameters into Equation 11 yields the PNC modulus (1 wt% of MWNT) for p = 446 or 1167 as Ec/Ep = 1.48 or 1.62, respectively. These values are 10% higher than the experimental data of 1.36 and 1.42, respectively. As in conventional fibre composites, the crack propagation shows MWNT pull-out, as well as crack bridging by the nanotubes. Hill et al. [2002] produced SWNTs and MWNTs using the arc discharge and CVD methods, respectively. Dispersing into HNO 3, refluxing for 48 h, centrifuging, washing and drying under vacuum purified the CNTs. Purified samples were refluxed in thionyl chloride and then treated with poly(styrene-cop-(4-(4´-vinylphenyl)-3-oxabutanol)). The functionalised CNTs were soluble in organic solvents (e.g., toluene, THF, or chloroform), making it possible to intimately mix them with PS. A transparent cast film ca. 50 μm thick, was prepared with 5 vol% CNT. Its transparency indicated that miscibility of functionalised CNT with PS resulted in homogeneous dispersion. A similar approach was used by [Mitchel et al., 2002], who dissolved anionic PS in toluene at room temperature, and then added appropriate quantities of either pristine SWNT or its functionalised version. The latter contained 4-(10hydroxy decyl) benzoate groups attached to 1 in 66 carbon atoms. The solutions were dried at room temperature, and then annealed at 180 °C under vacuum and tested. Nonlinear viscoelastic dynamic melt flow behaviour was absent for PNCs containing 0.5 wt% of non-functionalised SWNT, but present for PNCs containing 0.35 wt% of the functionalised CNT. Furthermore, the functionalised PNC had the percolation threshold at 1 vol% SWNT, while that for the pristine SWNTbased composites was twice this amount. Functionalisation resulted in better dispersion and significant enhancement of performance. Poly(p-phenylene benzobisoxazole) (PBO) has been synthesised in the presence of 0, 5 and 10 wt% SWNT (diameter of ca. 0.95 nm) [Kumar et al., 2002]. The reaction was conducted in polyphosphoric acid (PPA) at 100, 160 and 190 °C for 36 h, and then the product was spun into fibres using dry-jet wet spinning. Dried fibres were heat-treated at 40 MPa tension in N2 at 400 °C for 2 min. No CNT aggregates were observed under a polarised microscope. The tensile properties of the PBO fibre containing 10 wt% SWNT indicated that the modulus increased by 21%, strain at break by 40%, and the tensile strength by 61%. The compressive strength was also higher by 43%. The PBO/SWNT systems
51
Clay-Containing Polymeric Nanocomposites demonstrated less thermal shrinkage and less creep under stress. The morphology of these fibres suggests a co-alignment of the SWNT and PBO fibrillae within the oriented composite fibre. 2. The Reactive Method Jia et al. [1999] reported on the reactive preparation of PNC with MWNT dispersed in a PMMA matrix. Two CNTs were used: (1) as prepared (purity > 98%), and (2) the former CNT ground in a ball mill for 20 min, boiled for 0.5 h in concentrated HNO3 then washed and dried. TEM of the latter sample showed significant fragmentation of the original CNTs. The CNTs were dispersed in methyl methacrylate (MMA), the free radical initiator (2,2´-azobisisobutyronitrile (AIBN)) was added and the reaction was conducted at T = 358 to 363 K. For the untreated CNT the required amount of AIBN was three times greater than what was needed to polymerise MMA alone. The performance of the resulting PNC was poor. For the ball milled CNT there was no need for extra initiator and reasonable enhancement of performance was obtained (at 5% CNT tensile strength increased 30%, toughness 11%, hardness 42% and HDT by 39 °C (compared to PMMA)). The authors postulated that about 24% of the untreated CNTs were attacked by free-radicals that open their π-bonds. A C-C bonding may be generated between the nanotube and the matrix. When the π-bonds are at the curved points of the nanotube, the free radical attack may result in breaking the CNTs, which become shorter and with opened ends. The emulsion polymerisation method was used for the preparation of PNC with polyaniline (PANI) as a matrix in which 0.2 to 10 wt% of CNT was dispersed [Deng et al., 2002]. TEM and XRD showed a network made of nanotubes and PANI fibres. However, according to FTIR there was no direct interaction between these two components. The conductivity and thermal properties depended on CNT content, viz. PANI conductivity increased 25-fold by incorporation of 10 wt% CNT. 3. The Melt Compounding Method Considering the high cost of CNT and the need to work with small quantities of material, the early publications revealed quite unorthodox melt processing methods [Haggenmueller et al., 2000]. The authors prepared SWNT/PMMA nanocomposites with enhanced mechanical and electrical properties by engendering high CNT alignment. First, PMMA was dissolved in DMF and combined with the SWNT dispersion in this solvent (DMF was used for purification of the CNT). The cast film was subsequently folded and broken into pieces, then hot pressed at 180 °C. The procedure was repeated 30 times. The final dispersions (containing 1 to 8 wt% of SWNT) were melt spun to draw down ratio: DR = 20 to 3600. The elastic modulus and yield strength of the PNC fibres increased with CNT content and DR. Raman spectroscopy indicated that the nanotubes were well aligned. The electrical conductivity increased with CNT content, e.g., for 1.3 to 6.6 wt% SWNT, from 0.118 to 11.5 S/m in the flow direction and from 0.078 to 7.0 S/m in the perpendicular direction. Lozano and Barrera [2001] used an internal mixer for dispersing 2 to 60 wt% of CNT in PP. The CNT was MWNT-type prepared by CVD using large diameter catalyst particles, which resulted in large nanotubes, having diameter of ca. 80 to 200 nm and length of 30 to 100 μm. Melt compounding eliminated CNT 52
Nanoparticles of Interest to PNC Technology agglomerates yielding isotropic, non-porous PNC. Incorporation of the nanofiller increased the rate of crystallisation as well as the total crystallinity (from 49 for PP to 69 for PP with 60 wt% CNT). The tensile yield strength reached maximum at 5 wt% loading – for higher CNT content the strength was reduced below the neat PP value. By contrast, the dynamic tensile modulus at room temperature increased from ca. 5 (for PP) to ca. 18 GPa (for PP with 60 wt% CNT), i.e., by 350%. The authors associated the lack of reinforcement by the CNT with the increased brittleness of the PP, caused by its inability to further crystallise on deformation, ‘a property brought on by the molecular restrictions caused by the fiber dispersion’. High purity MWNT (purity >95%, diameter 20 to 30 nm, length 20 to 100 μm) was uniformly dispersed in PP, ABS, PS or HIPS in an internal mixer [Andrews et al., 2002]. The samples from the mixer were crushed, and then compression moulded into films containing v = 0.5 to 5 vol% of CNT. The percolation threshold for surface resistivity varied with the matrix; for PP, PS and ABS it was 0.05, 0.2 and about 9 vol%, respectively. The mechanical properties showed small improvements. Addition of 12.5 vol% of MWNT to PP resulted in a doubling of the modulus and a decrease of tensile strength by a factor of 1.7. For PS systems the modulus increased with 25 vol% loading from 1.9 to 4.5 GPa, along with a linear increase of the glass transition temperature: Tg (°C) = 96 + 0.48v (vol%). For the other two systems there was little change in the mechanical properties. The observed failure (nanotubes pull-out) suggested that surface treatment might improve interfacial bonding and increase tensile strength. Pötschke et al. [2002] used a commercial PC compound from Hyperion, with 15 wt% of MWNT (d = 10 to 15 nm; L = 1 to 10 μm). The masterbatch was prepared by melt compounding in a Buss Kneader. The authors diluted it with PC in a short TSE (L/D = 10) at 240 °C, to obtain concentrations of 0.5, 1, 2 and 5 wt%. The percolation threshold for electrical conductivity was somewhere between 1 and 2 wt% – the volume resistivity decreased from 1014 to about 102 Ωcm. The dynamic melt flow of compression moulded PNC was studied at 260 °C and the frequency range from 0.1 to 100 rad/s. Nonlinear viscoelastic behaviour was evident for a MWNT loading above 1 wt%, coinciding with the electrical percolation threshold. The increase of melt viscosity with nanotube composition was higher than that reported for either nanofibres with larger diameters or for carbon black. This difference is caused by the high aspect ratio of the MWNT, p > 67. Melt dispersion of SWNTs and MWNTs in PC has also been described [Sennett et al., 2003]. The MWNT prepared by CVD had d = 20 to 50 nm and L ≅ 20 μm; hence a respectable aspect ratio of about 1000. A conical micro-TSE was used at 250 to 266 °C, 80 rpm under N2, and the compounding time was varied from 1 to 120 min. To induce orientation, the extrudate was spun at a fibre draw speed of 10 to 70 m/min. Good dispersion was obtained even at short residence time (1 to 5 min), but longer mixing further improved the dispersion. Similarly, the CNT orientation improved as the fibre draw rate increased, but to obtain good alignment draw rates > 70 m/min were required. SWNT were found to be more difficult to disperse than MWNT. The use of lower molecular weight resin was reported to facilitate the dispersion. Unfortunately, this short and interesting note does not contain any data on the electrical conductivity or mechanical performance. 53
Clay-Containing Polymeric Nanocomposites 2.1.3.2 Rod-Like CdSe Nanocrystals These nanocrystals show size- and shape-dependent optical and electrical properties. Quantum rods of CdSe were prepared by pyrolysis of dimethyl cadmium and selenium tributyl phosphine solution in phosphonic acid. The rods had 3.0 to 6.5 nm diameter and were 7.5 to 40 nm long. As expected, the band gap shifted to lower energies with increase of both the diameter and the length of these crystals [Li et al., 2001a]. 2.1.3.3 Imogolite Imogolite (from imogo = volcanic ash in Japanese) is weathered pumice, discovered in Murakasami, Japan and described in 1962 by Yoshinaga and Aomine. This para-crystalline phyllosilicate has a composition written either as: Al2SiO3×(OH)4; SiAl2O5 · 2.5 H2O or Al2O3×SiO2×nH2O, with molecular weight per crystalline unit cell of 4754. However, while the nominal Si:Al ratio is 2, the measured one varies from 1.05:1 to 1.15:1. Imogolite occurs as tubes of several micrometers in length, having inside diameters of 1.0 and outside diameters of 2.0 – 2.52 nm; occasionally the tubes may branch out. The nanotubes have good rheological, adsorptive, and surface properties due to their unique structure and functional groups on the surface. For example, the tubes may form raft-like structures with honeycomb-shaped cross-section [Kajiwara et al., 1986; Donkai et al., 1993]. Imogolite may also be synthesised at T = 25-100 °C from a dilute solution of AlCl3 and Na4SiO4 via precipitation reactions. The imogolite structure has Si in tetrahedral coordination and Al in octahedral coordination. Imogolite has been used for reinforcing water-soluble polymers, viz. polyvinyl alcohol (PVAl) [Hoshino et al., 1992a] and hydroxy propyl cellulose (HPC) [Hoshino et al., 1992b]. Imogolite has several rather broad, low XRD peaks. Conclusive identification can be made by FTIR and TEM. Its CEC is 1.7 mol/g at pH 7.0, the calculated specific surface is 1000, while its measured value is 900 to 1100 m2/g. The density is 2.7 g/ml, hardness = 2 to 3, refractive index (nD) = 1.47 to 1.51. The crystals are transparent and fragile [Gabriel and Davidson, 2000]. 2.1.3.4 Vanadium Pentoxide, V2O5 Since CNTs are expensive and inhomogeneous (because of a wide range of chirality), V2O5 nanofibres have been proposed as an alternative. These may be formed by polycondensation of vanadic acid (HVO3) in water. The individual flat fibres are ca. 10 nm wide, ca. 1.5 nm thick and up to several microns long. A single fibre has a double layer structure, each consisting of two V2O5 sheets. Consequently, each fibre consists of only four layers of vanadium atoms and represents a wire of molecular dimensions. Owing to the disassociation of surface V-OH groups (V-OH + H2O ↔ V-O– + H3O+) the fibre surface is negatively charged. The V2O5 fibres are being investigated for electronic applications [Sinn, 1999]. Furthermore, V2O5 can also be formed into ribbons, sheets or nanotubes. V2O5 is available in a diversity of morphologies. Commonly, it is an orangeyellow to rust-brown orthorhombic crystalline solid with molecular weight of 304.15, density of 3,350 kg/m3, melting point of 670-685 °C and boiling point of ca. 1800 °C (it starts losing oxygen at 700 °C) [Macintyre, 1992; Cotton et al., 1999]. V2O5 is used as a catalyst, e.g., in the oxidation of SO to SO2, of alcohol to acetaldehyde, of aniline to aniline black, for the manufacture of yellow leaded glasses with inhibited UV light transmission, for depolarisers, etc. Its main 54
Nanoparticles of Interest to PNC Technology commercial use is as colorant, with colour strengthened by tin and zirconia. Although yellows can be prepared with antimony, vanadium is stable at higher temperatures. Wider V2O5 fibres, e.g., 1×25×(200 to 600) nm, are known as molecular ribbons. These may also be synthesised at pH2 from VO(OH)3x2H2O. The ribbon thickness is determined by the molecular structure. A magnetic field aligns the ribbons in the field direction. The ribbons are promising new materials that may be used as an important catalyst, a building block for electronic devices and reinforcing material with a strongly anisotropic structure [Gabriel and Davidson, 2000]. During the hydrothermal synthesis, organic templates control the crystallisation mode. This method has been used to prepare layered structures of V2O5 in the presence of coordination compounds, viz. 2,2´-bipyridine and ethylene diamine [Nesper et al., 2001; Ollivier et al., 1998]. Muhr et al. [2000] prepared V2O5 and mixed vanadium oxide nanotubes (VONT), reacting either primary n-alkyl amines (CnH2n+1NH2; 4 ≤ n ≤ 22) or α,ω-diamines (H2N-CnH2n-NH2; 14 ≤ n ≤ 20) with vanadium (V) alkoxide, followed by hydrothermal treatment. The monoamine templates tend to form wide tube openings and tube walls consisting of 2 to 10 layers, whereas the diamines lead to tubes with comparatively thick walls with >10 layers. The amines get intercalated into VONT structures that have 2 to 30 layers, resulting in an outer diameter (OD) = 15 to 100 nm and tube length L = 500 to 15,000 nm. The interlayer distances (1.7-3.8 nm) increase with the length of the alkyl chain of the amines. The VONT collapse into amorphous form at about 250 °C. In another publication from the same laboratory [Krumeich et al., 2000] VONT was similarly prepared from vanadium alkoxides and amines by a sol-gel reaction and a subsequent hydrothermal treatment. The product showed bent VOx layers rolled into tubes. The layer structure inside the tube walls is frequently disordered, and several types of defects were identified. Vanadium oxide in the form of scroll-like nanorolls was prepared for use as a cathode material for rechargeable lithium batteries [Edström et al., 2001]. The rolls consisted of several layers, separated by templates, viz. hexadecyl amine (C16H33NH2) or dodecyl amine (C12H25NH2). 2.1.3.5 Inorganic Nanotubes Nanotubes are not limited to carbon and V2O5. In 1992 inorganic fullerene-like nanotubes of tungsten disulfide were described [Tenne et al., 1992]. It seems that any compound that forms stable two-dimensional sheets can be rolled into a nanotube. Different synthetic methods may lead to nanotubes having different type, size and yield. The methods range from substitution of atoms in an already fabricated tube [Hang et al., 1998], laser heating [Laude et al., 2000] and arc discharge [Cumings and Zettl, 2000] to template-assisted synthesis [Shenton et al., 1999; Zhang et al., 2000a]. For example, 2-D tungsten disulfide (WS2) sheets transformed into nanotubes are extremely inert and durable. They show potential for novel scanning probe microscope (SPM) tips. A standard AFM tip will rarely remain sharp for more than a few hours under normal operating conditions. WS2 nanotube tips are so rigid and inert that they have been used for months with no sign of wear. Boron nitride (BN) provides another example. These inorganic nanotubes may be synthesised in a discharge using a tungsten electrode hollowed and filled with boron nitride powder [Chopra et al., 1995]. In contrast to CNTs, the ones 55
Clay-Containing Polymeric Nanocomposites from BN should all be insulating with a band gap of about 5 volts. Another method of preparation is by means of a solid-state process. Graphite and hexagonal boron nitride powders were ball milled at room temperature, then annealed below the melting points for both graphite and boron nitride, at T ≤ 1300 °C. The latter stage leads to the nucleation, re-ordering and crystal growth of hexagonal C and BN nanotubes of both cylindrical and bamboo-like morphology. High energy ball milling was found to result in solid-state phase transformations and chemical reactions [Chen et al., 1999a]. Gole et al. [2003] prepared and characterised silica-based nanospheres (monodispersed, with d ≤ 30 nm diameter), nanotubes, nanofibre arrays, and nanowires. The agglomeration of small nanospheres (d ≤ 10 nm) into wire-like groupings has suggested a possibility of growing silica nanotubes (SiNT) by a more efficient agglomeration process. The SiNT were generated from silicon particles at 1400 oC. Their outer diameter was, OD = 70 to 80 nm, with wall thickness of ca. 20 nm. Biaxially structured SiC-SiOx nanowires have now also been generated from C/Si/SiO2 mixtures. Depending on the size, the Young’s modulus of these structures is 50 to 100 GPa.
2.1.4 Other Nanoparticles 2.1.4.1 Spherical or Nearly-Spherical Particles For several decades carbon black and fumed silica with particle size below 50 nm have been used in tyres, tubings, sealants, etc. More recently, functionalised silica nanoparticles were copolymerised with methyl methacrylate to generate transparent films with good surface finish and controllable (by silica content, x) refractive index: nD = 1.509 – 0.00035x [Yu et al., 2003]. Newer, nanosized particles are metal, metal oxide, nitride, sulfide or carbide powders. They find use in the preparation of materials with specific electrical and/or magnetic properties. CdSe, ZnS, CdS, PbS, and many others have been used for the preparation of functional PNC, for example showing non-linear optical quantum-size or other semiconductive properties. For example, nanosized dispersions of ZnS and CdS in PVAl were prepared by a hydrothermal method. Thus, ZnCl2 or CdCl2 were dissolved in an aqueous solution of PVAl, then CS2 was added and the mixture was heated in an autoclave for 8 h at 120 °C. Uniform particles of ZnS and CdS had a diameter of ca. 60 to 100 and 80 to 120 nm, respectively [Qian et al., 2000]. The same solvo-thermal method was used for the preparation of nanocrystalline particles of tin chalcogenide or chromium nitride [Qian et al., 1999a]. Electrolytic precipitation of Cu in pores of solvent crazed polymers (e.g., PVC or PP) resulted in the formation of ca. 15 to 20 nm diameter particles [Arzhakowa et al., 2003]. The size and concentration of metal particles depended on the extent of the crazing. 2.1.4.2 Sol-Gel Hybrids Sol-gel hybrids are produced by methods which are based on the reaction of precursors (e.g., metal alkoxides M(OR)n) that result in the formation of a nanometric dispersion of an inorganic phase [Mauritz et al., 1995; Wen and Wilkes, 1996; Deng et al., 1998; Hand et al., 1998]. Hydrolysis and condensation leads to the formation of metal oxopolymers. The sol-gel process takes place 56
Nanoparticles of Interest to PNC Technology under relatively mild, well controlled conditions, hence inorganic and organic components can be formed at the nanometre scale, at a wide range of composition. The hybrids usually have ≥ 10% dispersed phase. Two types of hybrids are recognised: 1. Those where only weak interactions between the organic and inorganic species exist (viz. van der Waals, hydrogen bonding or electrostatic interactions). 2. Where the inorganic and organic components are chemically bonded by either covalent or ionic-covalent bonds. Examples of PNCs prepared by the sol-gel methods are provided in Table 2. Several methods of PNC preparation by the sol-gel method have been developed [Brinker and Scherer, 1990]:
The principal sol-gel process involves the formation of a colloidal suspension (sol) and gelation of the sol to form a network in a continuous liquid phase (gel). Often the hydrolysis is catalysed using either acids (acetic acid, HCl, HF, etc.), bases (ammonia, amines, KOH, etc.) or salts (e.g., KF). The rate and extent of the hydrolysis reaction depend on the strength and concentration of the catalyst. From a homogeneous solution of an oligomer or polymer with inorganic precursor, chemical reactions lead to dispersion of the inorganic phase in the polymeric matrix. The precursors consist of a metal or metalloid compound with reactive ligands. Metal alkoxides are most popular because they readily react with water. However, other compounds, viz. aluminates, titanates, and borates are also used, often mixed with silicon alkoxides. After the reaction, the solvent must be removed before the dispersed phase has time to aggregate. Control of the interface is crucial. The use of electrostatic charge or hydrogen bonding interactions may be required. For example, silica nanoparticles can be generated by hydrolysis and condensation of silicon tetraalkoxides, Si(OR)4, in polyoxazoline ethanol solution. Hydrogen bonding between Si-OH silanol groups and the carbonyl and amide functionalities of the polymer ascertained homogeneous and stable dispersion. It is noteworthy that these advantageous properties could be enhanced by grafting the polymer with -Si(OR)3 groups, which after hydrolysis increase the chemical affinity between organic and inorganic components. From a homogeneous solution of an inorganic precursor and a monomer, chemical reactions lead to the formation of inorganic gels, followed by polymerisation. For example, hydrolysis and polycondensation of silicon alkoxides in the presence of methyl methacrylate (MMA) leads to PNC with superior optical and mechanical properties. 57
Clay-Containing Polymeric Nanocomposites The pre-formatted building blocks (e.g., oxometallic clusters, CdS or CdSe nanoparticles, metallic or oxides colloids) are functionalised, saturated with an oligomer or monomer, and polymerised. A review of this strategy has been presented [Anonymous, 1999a]. The sol-gel methods lead to the preparation (often at room temperature) of PNC with inorganic nanoparticles that engender improved hardness, optical transparency, chemical durability, tailored porosity, and thermal resistance. The materials are used in optics, protective and porous films, optical coatings, window insulators, dielectric and electronic coatings, high temperature superconductors, reinforcement fibres, fillers, and catalysts [Zeigler and Fearon, 1990]. The phase separated morphologies of certain copolymers and ionomers can act as 3-dimensional templates during the sol-gel polymerisation of silicon alkoxide and organo-alkoxysilane monomers. In the presence of these templates the inorganic oxide or organically modified silicate nanophases grow within specific nanoscopic domains. For example, Mauritz et al. [2001] prepared a variety of organic/inorganic nanocomposites via in situ sol-gel polymerisation of metal alkoxide and organo-alkoxysilane monomers as well as copolymerisation of their mixtures. Thus, poly(styrene-b-isobutylene-b-styrene) was first sulfonated and neutralised, then tetraethoxy silicate was dissolved into the ionic domains. Nafion® perfluorosulfonic acid membrane and Surlyn® ionomer have been used as a templates for several silicate and titanate compounds. After hydrolysis and condensation of the ≡Si-OH groups, PNC with highly regular structure of spherical dispersions, d = 2 to 5 nm, was obtained. The polymer ionic groups are entrapped within the silicate or titanate nanoparticles. 2.1.4.3 Polyhedral Oligomeric Silsesquioxanes (POSS) 2.1.4.3.1 Origin and Structure Ladenburg first described the synthesis of silsesquioxanes in 1875. In 1915 Meads and Kipping studied the hydrolysis and condensation reactions of trifunctional silanes, concluding that polycondensation of ‘siliconic acids’ leads to complex mixtures of little synthetic value. This delayed investigation on silsesquioxanes until 1955, when Barry et al. [1955] described crystallisable organosilsesquioxanes, defining the atomic ratio as: O:Si = 3:2, as for example, in H8Si8O12 – the siloxy analog of cubane (C8H8) in which C-C bonds are replaced by Si-O-Si. However, silsesquioxane does not have to be cubic – amorphous and ladder-like structures are also known. In 1965 Brown and Vogt conducted a controllable synthesis of silsesquioxanes. According to Marcolli and Calzaferri [1999] three synthetic routes have been used: cohydrolysis of trifunctional organo- or hydro-silanes, substitution reactions with retention of the siloxane cage, and corner-capping reactions. Murugavel et al. [1999] discussed the chemistry of silanetriols and triaminosilanes as useful synthons for the generation of three-dimensional metallosiloxanes. Starting from stable N-bonded silanetriols and triaminosilanes, metallosiloxanes and iminosilicates with aluminium, gallium, indium, titanium, zirconium, tantalum, tin and rhenium incorporated, a heterosiloxane framework could be prepared. Since some of these contain hydrolysable functionalities they may be used as starting materials for the preparation of supramolecular cage structures. Functionalised POSS may be synthesised by polycondensation of trifunctional RSiY3, where R is a hydrocarbon and Y is a hydrolysable functionality such as 58
Nanoparticles of Interest to PNC Technology chloride, alkoxide or silanol. However, this route does not control the placement of the functionality on the cage. Functionalisation of fully substituted silane compounds by hydrosilylation or chlorination also does not lead to controlled substitution. Silylation of anionic species has been reported to produce functionalised species, e.g., via: (a) the addition of hydrogen atoms bonded directly to silicon atoms onto aliphatic unsaturated compounds, or (b) the addition of sulfur onto aliphatic unsaturated compounds. The type of organic functionality that can be incorporated on the cage is limited. These methods have low yield and a large percentage of impurities from side reactions that must be removed. The reaction between trisilanol with a variety of compounds of the type R´´MX3 (R´´ = alkyl, alkenyl, aryl, H; M = Si, Ge, or Sn; X = halogen or alkoxide) leads to a variety of monomeric Σ8 silsesquioxanes. In the early 1990s, Lichtenhan reacted trisilanol with R´SiCl 3 or R´Si(OCnH2n+1)3 starting a new class of monomers called polyhedral oligomeric silsesquioxanes, or POSS (see Figure 9). In this case the R´ group was polymerisable or useful for grafting reactions, or sol-gel processing, e.g., acidic, acrylic, alcohol, amine, α-olefin, epoxy, ester, halides, isocyanate, organo-halides, phenol, silanes, silanols, styrenic, vinyls, etc. The process that leads to functional POSS was patented [Lichtenhan et al., 1999]. In the presence of a metathesis catalyst an effective amount of olefins of alkyls, cyclics, aryls, siloxyls or their isomersis able to provide POSS with 1-8 reactive functionalities. The process has been used in the US Air Force Laboratory to synthesise dozens of types of POSS. The products are soluble in common solvents, e.g., chloroform, hexane or THF. The chemical structure of POSS is: (RSiO1.5)Σn, thus as the name indicates (sesqui = ‘one and a half’) intermediate between silicas (SiO2)n and silicones (R2SiO)n. For most POSS: Σn = 8, but Σn = 6, 10 or 12 are also available. The size of the cage varies from about 0.7 to 3 nm. Polyhedral siloxanes or silsesquioxanes sensu stricto are topologically equivalent to a sphere and are also called spherosiloxanes. Most POSS compounds are crystalline, but changing the R-group may result in liquid crystal or liquid-like behaviour. The octasilsesquioxanes (Σn = 8) have cube-shaped molecules that consist of a Si8O12 core and eight reactive sites that may be differently functionalised. From the point of view of polymer technology the mono-substituted species, R´R7Si8O12
Figure 9 Reaction between trisilanol and a compound of the type R´SiX3 (R´ = reactive group; X = halogen or alkoxide) leads to a variety of monomeric Σ8 silsesquioxanes, or POSS monomers [Lichtenhan et al., 1999].
59
Clay-Containing Polymeric Nanocomposites (where R´ ≠ R are substituents) are the most interesting – the group R´ may provide an ability to the octasilsesquioxane to enter polymerisation, copolymerisation or grafting reaction, whereas the other functionalities, R, make the system miscible. It should be noted that incorporation of POSS into crystallisable macromolecules reduces the crystallinity, which in turn tends to lower the chemical and solvent resistance. A new type of PNC formation was recently proposed by Ricardo et al. [2001]. The authors started with octakis(hydrido dimethyl siloxy) octasilsesquioxane Q8M8H, converting it to octa-ethylbenzyl chloride. Using bromoester as an initiator and CuCl as a catalyst, atom transfer radical polymerisation (ATRP) was employed to produce star-type PMMA with arms having controlled molecular weights. The concentration of catalyst and initiator control MW, MWD and the star-star coupling reactions. The arms had Mn = 4 to 15 kg/mol (varying with conversion), and the polydispersity MW/Mn = 1.2 to 1.5. The number of arms per Q8M8H-unit was found to vary from 6.2 to 7, hence less than the maximum of 8. The lower number of arms may result from star-star coupling reactions. The new PNCs should have unique properties, combining those of nanocomposites with those of the controlled star-branched polymers [Roovers, 1985]. Mechanical properties of these PNCs have not yet been reported. 2.1.4.3.2 Properties POSS with reactive groups may be incorporated into virtually any polymer either by compounding (as a nanofiller), copolymerisation or blending [Lichtenhan et al., 1993; Lichtenhan, 1995; Haddad and Lichtenhan, 1996; Shockey et al., 1999; Feher et al., 1999]. POSS may provide a variety of property enhancements to existing resin systems. Owing to its chemical nature, POSS can be used to upgrade the thermal and physical properties of most plastics. The following isotropic enhancements have been reported for POSS-copolymers and blends: higher decomposition temperature, Tg increased by 100-200 °C, lower flammability (delayed combustion and reductions of heat evolution), bulk density reductions of up to 10%, increased O2 permeability, reduced thermal conductivity, improved resistance to oxidation, increased modulus and hardness while maintaining the stress and strain characteristics of the base resin, good processability and mouldability (viscosity reduction of up to 24% was reported) [Lichtenhan, 1996; Tsuchita et al., 1997]. Substituted POSS have been used as Wittig reagents, precursors to SiC powders, low dielectric constant materials, alumino-and gallio-silicates, silica-reinforced composites, a variety of microporous materials, etc. A variety of functional groups have been attached to POSS, viz. acrylates, silanes, silanols, olefins, epoxies, amines, esters, phenols, styrenics and thiols. Functionalised POSS was polymerised to yield hybrid inorganic-organic homopolymers or copolymers [Tsuchita et al., 1997].The free radical reaction of propyl-methacryl-POSS gives POSS macromers, which can undergo hydrosilylation into oligomers and polymers with improved mechanical properties, increased thermal stability to oxidation and resistance to degradation by UV [Lichtenhan, 1995]. Properties and performance of POSSbased polymers continue to be explored [Zheng et al., 2001; 2002a,b; Li et al., 2002]. The modern POSS is only about 10 years old, but between 1993 and 2002 over 160 articles on POSS were published. POSS have been successfully incorporated into a number of thermoplastic matrices such as styrenics, acrylics, LCP, siloxanes, polyamides and more recently 60
Nanoparticles of Interest to PNC Technology PO. The resins are inherently reinforced by the presence of the inorganic cages of size between a molecule and a macromolecule. By controlling the nature of the substituted groups, R and R´, POSS has controllable reactivity, miscibility, low density, neutral pH and low VOC. Their polymers or copolymers are isotropic, free of metals, and transparent. In Figure 10, the compositional variation of Tg and the decomposition temperature for 10 wt% mass loss (Td) is shown for a series of copolymers. These were prepared by free-radical polymerisation of α-methyl styrene with styryl-based POSS. The latter contained one styryl-ethyl polymerisable group and seven inert R-groups, either cyclohexyl (-c-C6H11) or cyclopentyl (-c-C5H9). Thus, the macromer had a spherical (Si8O12) core, surrounded by seven inert groups for solubility and one reactive. It was found that the cyclohexyl derivative is about twice as soluble as cyclopentyl [Haddad and Lichtenhan, 1996]. The difference in behaviour between these two types was small, but cyclohexyl-substituted POSS showed better performance. Figure 11 shows the variation of Tg for two series of isobutyl-styryl-POSS copolymers with either vinylpyrrolidone (P4VP-POSS) or acetoxylstyrene (PASPOSS). The Figure demonstrates that POSS may increase the Tg, but at a relatively high POSS content – for both systems incorporation of POSS initially leads to a reduction in the HDT. Since the ratio of molecular weight of POSS to P4VP is 8.3 the observed minimum at about 2 mol% is equivalent to 14 wt% of POSS. The authors [Xu et al., 2002] rationalised this behaviour by noting that the POSScontaining copolymer had reduced molecular weight and narrower molecular weight distribution. Furthermore, the Tg depends on:
Figure 10 Composition dependence of the glass transition (Tg) and decomposition (Td) temperatures for copolymers of 4-methyl styrene with styryl-based POSS. In the latter C5 and C6 stand, respectively, for R = -C5H9 and -C6H11 [1addad and Lichtenhan, 1996].
61
Clay-Containing Polymeric Nanocomposites
Figure 11 Variation in glass transition temperature for two series of vinylpyrrolidone random copolymers with POSS (P4VP- POSS) and acetoxylstyrene with POSS (PAS-POSS). Data [Xu et al., 2002].
1. A diluent role of POSS that reduces the dipole-dipole interaction of the matrix monomer, 2. The dipole-dipole interaction between POSS siloxane and the polar carbonyl of organic polymer, and 3. The POSS-POSS intermolecular interaction. At a relatively low POSS content, the diluent role dominates, thus Tg decreases. At a high POSS concentration, the POSS-PAS, POSS-P4VP and POSS-POSS interactions cause the Tg to increase. Figure 12 illustrates the variation of Tg for copolymers of butyl methacrylate (BM) with up to 50 mol% of cyclopentyl-POSS-methacrylate – a significant increase from 50 to 245 °C was observed. The experimental data may be described by the dependence:
[
3/2
Tg = ( 1+ K * w 1w 2 ) ⋅ w 1Tg1
3/2
+ w 2Tg 2
]
(12)
where wi and Tgi are, respectively, the weight fraction and glass transition temperature of component i, and K* is the binary interaction parameter between them. The relation was derived for strongly interacting miscible blends [Utracki, 1989], thus the agreement implies that there are strong thermodynamic interactions between the two types of mers expressed by the high value of the binary interaction parameter: K* = 6.39 ± 0.52. Since such an effect has not been observed in copolymers of BM with methyl methacrylate (MM), it is evident that stiffening of the copolymer macromolecules involves the POSS units. The correlation coefficient squared for the dependence in Figure 12 is r2 = 0.999. The dependence predicts that Tg should decrease at POSS loadings over 50 wt%. 62
Nanoparticles of Interest to PNC Technology
Figure 12 Glass transition temperature of polybutyl methacrylate and its copolymers with up to 50 mol% of heptacyclopentyl methacrylate (CpPOSS-MA) [Lichtenhan, 1995]. Points – experimental, line calculated from Equation 12.
Figure 13 illustrates the effect of melt compounding PP with 10 wt% of POSS = [CH3SiO1.5]S8. At T < Tg ≈ 0 °C the stiffening effect is negligible, whereas at Tm > T > Tg the complex tensile modulus (E*) increased by a factor of up to two, or the usage temperature window increased by 45 °C, i.e., the POSS compound at 145 °C has the same value of E* as neat PP at 100 °C [Schwab et al., 2001]. In other words, POSS is not effective in the glassy state, but is able to ‘reinforce’ PP at T > Tg. (Similar effects were observed for norbornyl elastomer with cyclopentyl or cyclohexyl POSS [Bharadwaj et al., 2000]). Schwab et al. [2001] also provided numerical values characterising the mechanical performance of these nanocomposites (see Table 10). Improvement of the tensile and flexural properties was modest, but of impact strength and HDT were significant. It is noteworthy that POSS is expected to reside in the amorphous PP phase, which constitutes ca. 40% of the polymer. If so, the local POSS concentration is higher by a factor of about 2.5. Using a metallocene catalyst, Zheng et al. [2001] copolymerised ethylene or propylene with up to 10.4 mol% of norbornylethyl cyclopentyl-POSS macromer (see Figure 14). As shown in Figure 15, incorporation of the POSS sharply reduced the crystallinity, while the decomposition temperature and the temperature for 5 wt% mass loss in TGA increased to a plateaux by ca. 20 and 100 °C, respectively. The latter two properties reached their respective plateaux at POSS ≈ 20 wt% [Zheng et al., 2001]. Zheng et al. [2002b] carried out copolymerisation of styrene with the POSSstyryl macromer 1-(4-vinylphenyl)-3,5,7,9,11,13,15-heptacyclopentylpentacyclo octasiloxane. The catalyst CpTiCl3 with methylaluminoxane (MAO) was used. Random syndiotactic copolymers have been obtained with POSS content up to 24 wt% or 3.2 mol % (see Figure 16). 63
Clay-Containing Polymeric Nanocomposites
Figure 13 Complex tensile modulus versus temperature for PP and its copolymer with 10 wt% of [CH3SiO1.5]Σ8 [Schwab et al., 2001].
Table 10 Mechanical properties and HDT of PP and its composites with [CH3SiO1.5]Σ8 [Schwab et al., 2001] Property (improvement)
ASTM
Flexural modulus (GPa)
[CH3SiO1.5]Σ8 content (wt%) 0
2
5
10
D790A
1.655
1.730 (4.5%)
1.757 (6.2%)
1.80 (8.8%)
Tensile strength (MPa)
D638
34.5
34.5 (0%)
35.1 (1.7%)
35.8 (3.8%)
Izod impact strength (kJ/m)
D256A
26.7
29.3 (9.7%)
33.1 (24%)
40.0 (50%)
D468
99
105 (6°)
115 (16°)
124 (25°)
HDT (°C)
Incorporation of 24 wt% POSS increased the T g by 4 °C, destroyed sPS crystallinity, did not affect the decomposition temperature under N2, but increased that in air by 37 °C (reduced oxidative degradation). Li et al. [2002] copolymerised vinyl ester (VE) and 50% styrene (St) with 0, 5 or 10 wt% of POSS (MW = 1305: (C6H5CHCHO)4(Si8O12)(CH=CHC6H5)4). At 5 wt% of POSS the system was homogeneous, but at 10 wt% silicon-rich crosslinked domains of irregular shape were observed – these ranged in size from a few to about 75 nm. Incorporating POSS into the VE/St resin had almost no influence on Tg or on the width of the loss peak in the glass transition range. 64
Nanoparticles of Interest to PNC Technology
Figure 14 Norbornenylethyl cyclopentyl-POSS, C44H76O12Si8, with MW = 1021.77 g/mole; soluble in THF, chloroform or hexane [Zheng et al., 2001].
Figure 15 Thermal properties of (norbornenyl-ethyl-cyclopentyl)-POSS-coethylene. Incorporation of POSS sharply reduced crystallinity. The decomposition temperature, Td, and the temperature for 5% wt loss during TGA, T5%, reached a plateaux at POSS content of about 20 wt% [Zheng et al., 2001].
65
Clay-Containing Polymeric Nanocomposites
Figure 16 Coordination copolymerisation of styrene with the POSS-styryl macromer 1-(4-vinylphenyl)-3,5,7,9,11,13,15-heptacyclopentylpentacyclo octasiloxane Reproduced from Zheng et al. [2002b], copyright 2002, with permission from John Wiley & Sons, Inc.
Similar absence of effect on Tg was also observed in specimens prepared by blending in non-reactive POSS units into VE resin. Copolymerisation with 10 wt% POSS resulted in a two-phase, crosslinked system. There was no measurable effect of POSS incorporation on Tg ≅ 131 ± 1 °C. At 40 °C the bending storage modulus, E´, of the VE/styrene resin and its copolymer with 10 wt% POSS was E´ = 1.24 and 1.58 GPa, respectively. The flexural modulus of these specimens was 1.89 to 2.14 GPa, respectively. However, addition of POSS reduced the flexural strengths by about 21%. The VE and composite samples showed poor solvent resistance to THF. The following property improvements have been cited for POSS in thermoplastic or thermoset systems: • • • • • • • • •
Controlled miscibility for mixing and blending. Maintained processability and mouldability of neat polymer. Viscosity reductions of up to 24% (relative to silica-filled composites!). Low density, by eliminating the need to use common fillers. For example, by replacing silica as filler the bulk density may be reduced by up to 10%. Extended temperature range, and resistance to oxidation and wear. Reduced flammability by delayed combustion and low heat evolution rate. Increased modulus and hardness while maintaining the stress and strain characteristics of the base resin. Increased O2 permeability and low thermal conductivity. Simple disposal as of silica.
2.1.4.3.3 Sources POSS is available from Sigma-Aldrich Chemical Co. (www.sigma-aldrich.com), Gelest Inc. (www.gelest.com), The Mather Group at UCONN (University of Connecticut), Hybrid Plastics (www.hybridplastics.com), etc.
66
Nanoparticles of Interest to PNC Technology Sigma-Aldrich Fine Chemicals is a global supplier. The company offers 68 POSS compounds (monomers, polymers, silanols, reagent and precursors) in quantities of 1 to 10 g. Gelest was founded in 1990 to serve the advanced technology applications market. The company offers a dozen POSS compounds at a cost of 100 to 450 US$/kg. The main source of POSS is Hybrid Plastics, a spin-off from the US Air Force Research Labs (AFRL). Lichtenhan and Schwab launched the company in 1998, on a license from AFRL and an Advanced Technology Program (ATP) startup grant from the US National Institute of Standards and Technology (NIST). The continuing cooperation with AFRL is for the application of POSS-technology to rocket, aero, and space vehicle systems. Starting with six products, in 2003 Hybrid Plastics has a list of nearly 300 (guaranteed 97% pure) products, grouped into four categories: 1. POSS® Molecular Silicas™ with Si–O core surrounded by non-reactive organic groups providing miscibility with an organic matrix. These may be added (up to 50 wt%) to polymers, yielding nanocomposites with nanoscale reinforcements. 2. POSS® Silanols have 1 to 4 ≡Si–OH silanol groups. These may react with metal or glass surfaces (or with inorganic fillers) rendering them hydrophobic. Silanols containing epoxide, methacrylate, and olefinic groups are available for copolymerisation or grafting. 3. POSS® Functionalised Monomers have 1 to 8 reactive groups, such as amines, esters, epoxies, methacrylates, olefins, silanes, styryls and thiols. In most cases they are available as: 1-R-3,5,7,9,11,13,15-heptacyclopentyl cyclo octasiloxane (R-POSS). 4. POSS® Polymers possess a hybrid inorganic-organic composition and can be either thermoplastic or thermoset. They are either (1) co-polymers with standard monomers, or (2) neat POSS resins. The types available include: silicones, styrenics, acrylics and norbornenes. The price depends on quantity: viz. for R&D gram quantity the price is US$200 to 2,000/kg, for semi-bulk it is US$60 to 200/kg, and for production level it is > US$20/kg. ‘Sampler Kits’ contain 5 to 7 types of a specific type of POSS (e.g., POSSmethacrylate, POSS-epoxy, or POSS-silanol, 20 g each) at US$650 to US$1200 per set. Multi-ton quantities can be produced using a continuous process. In 2002 POSS cost was ca. US$400/kg. New plant for making POSS may push the price of some POSS to about US$33/kg. Hybrid Plastics’ sales and marketing arm is Divex, Inc. 2.1.4.3.4 Applications The method of POSS incorporation depends on the category (see above) and the expected application. Evidently, it is advisable that the molecular silicas are dissolved in a monomeric or polymeric liquid. The dissolution should be carried out in a suitable mixer, e.g., a twin-screw extruder (TSE), as dispersing [CH3SiO1.5]Σ8 in molten PP [Schwab et al., 2001]. The functionalised POSS monomers are mainly copolymerised, e.g., as butyl methacrylate with heptacyclopentyl methacrylate (CpPOSS-MA) [Murugavel et al., 1999], but homopolymerisation, grafting and crosslinking may also be carried out. However, since the cost of POSS is high (even in comparison to engineering resins) the recommended use is < 50% incorporation into high performance resins. In the case of POSS® Polymers, blending with other resins in a TSE may provide a suitable solution.
67
Clay-Containing Polymeric Nanocomposites POSS can be used in many guises as: • Additives – for heat/abrasion resistant paints and coatings. Used as crosslinking agents, viscosity reducers, fire retardants, also to increase mechanical properties, HDT and gas permeability, to decrease dielectric properties (in photoresists, interlayer dielectrics), etc. Thus, for example, Weidner et al., from Wacker-Chemie used POSS as a crosslinking agent. • Plastics – in aerospace, electronic, optical, medical, biomedical and pharmaceutical applications, for catalysis, packaging, coatings, liquid crystal display elements, magnetic recording media, as coupling agents, fire retardants, nanofillers in speciality polymers, dendrimers, etc. Optical disks, microelectronics, and medical products are target niches. Takamuki et al., of Konica Corporation used POSS for transparent printing plates; Zank and Suto of Dow Corning Asia for electronic materials; Nguyen of American Dye Source as hot melt inks; Canon for LED, etc. • Pre-ceramic coatings that, depending on the oxidising conditions, convert to silicon dioxide, silicon oxy-carbide or silicon carbide. These POSS may be used as ablative materials (for nozzles, insulations, etc.), for cladding and coatings in the electronics industry, etc. • Reagents, POSS have been used for catalysis supports, as monomers, crosslinkers, biological scaffolds, reactive ion-etch resistant layers; they find application in microelectronics, for drug delivery, etc. For example, POSS silanols are formulated at room temperature. At T > 40°C the resin becomes tacky, and then a viscous liquid at 120 °C. The cure is catalysed by dibutyl tin diacetate, zinc acetate or zinc-2-ethylhexanoate. After condensation the resin becomes tough binder or film. Shell Oil Co. and Solvay used POSS as an epoxidation catalyst (US Patents 5,750,741 and 6,127,557), etc. • Surface Modifiers, as silane replacement in corrosion and abrasion resistant coatings, lubricants and compatibilisers, for controlled drug release, and optical fibre coatings. Table 11 summarises the relative merits of three types of PNC systems: CPNCs with organoclay, those with POSS, and other PNCs with other nanofillers. By the end of 2002, Pentron Clinical Technologies had introduced the NanoBond Universal Bonding System, a POSS-reinforced adhesive. The resin infiltrates etched surfaces and provides a strong bond between the tooth and dental restorative material. The kit includes a primer, the adhesive and a dual cure activator. Finally, it is worth mentioning that not only cage-type POSS are known. A ladder-like variety (L-POSS):
shows superior heat, radiation, water and fireproof resistance, outstanding electrical properties as well as the formation of high-strength film (for coatings, electronic and optical devices). Changing the side and/or end groups may modify the molecular structure of L-POSS.The compound is used in photoresists, interlayer dielectrics and protective coating films for semiconductor devices. 68
Nanoparticles of Interest to PNC Technology
Table 11 Comparison of the three basic PNC technologies (after Hybrid Plastics, 2002) Properties
Organoclay
POSS
Nanofillers
Characteristics
Mineral or synthetic clays: nanosized organically treated platelets, indirectly interacting with polymers
Nanosized chemical molecules that directly interact and bond with polymers
Nanosized, fillers (fibres, whiskers, tubes, spheres, etc.) for polymeric matrices
Size
1 by ≥ 100 nm, anisotropic
0.7-3 (mostly 1.2) nm
≥ 20 nm
Size distribution
Polydispersed
Mono or polydispersed
Polydispersed
Transparent oils or solids
Solid
Physical appearance Solid Density
Moderate
Low
High
Hygroscopic
No
No
Possible
High T stability
No
Yes
Yes
Trace metals
Possible
None
Possible
Biocompatibility
Possible
Yes
Yes
O2 permeability
Low
High
Poor
Recyclability
Yes
Possible
Limited
Dispersability
Possible
Dissolves molecularly
Limited loading
Interfacial properties
Possible
Adjustable
Limited
Melt compounding
Yes
Ye s
No
Glass transition, Tg
Small change
Significant in amorphous
Minor change
Viscosity
Increased
Reduced
Increased
Hardness
Increased
Increased
Variable
Modulus, E
Increased, anisotropic1
Increased
Increased
HDT
Increased
Increased
Increased
Creep
Improved
Improved
Poor
Increased
Reduced
Improved
Improved
Impact/toughness
Anisotropic
Fire retardance
Improved
1
Impact/toughness
Variable
Improved
Variable
Cost
Moderate
High
Low
Notes: 1 Nanoclays may be flow-oriented, which will impart anisotropy to the modified polymer system
69
Clay-Containing Polymeric Nanocomposites Applications for liquid crystal display elements, magnetic recording media and optical fibre coatings have also been disclosed. L-POSS may also be used for gas separation membranes, binders for ceramics and controlled release drugs as well as additives in cosmetics and resins. The ladder-like poly(methyl silsesquioxane)s (PMSQ) are used for coatings, particularly in electronics and optical devices. To improve the barrier, mechanical and thermal properties Ma et al. [2002] tried to disperse MMT in PMSQ. The aim was to develop high heat-resistant coating. The Na-MMT was pre-intercalated with trimethyl hexadecyl ammonium bromide (3MHDA), dried and ground. Dispersing the organoclay in chloroform, then adding methyl trimethoxysilane, hydrolysing it, neutralising and polymerising, resulted in the PMSQ/MMT nanocomposite:
XRD and TEM showed that MMT was only intercalated with d001 = 3.42 nm. Annealing the nanocomposite at 150 °C for 3 h did not change the spacing, but annealing at 300 °C for 3 h did reduce d001 to 3.0 nm. Furthermore the relative intensity of the peak significantly increased. The stability of interlayer spacing at T ≤ 150 °C was explained by the molecular structure of the in situ polymerised PMSQ. The decrease of spacing at 300 °C is most likely due to decomposition of the quaternary ammonium ion complex on MMT. Unfortunately, the authors did not report on the performance of these interesting materials. The commercial success of a material depends on delivery of enhanced performance at low cost. Currently, the PNCs available on the market contain 2 to 5 wt% of organoclay. The market price for PA-6 and PP is about US$3,000 and 1,000/ton, respectively. Assuming that a 10% increase of cost would be acceptable (if performance warrants it), the cost of compounding and organoclay must not exceed US$300 and 100/ton, respectively. At the current cost of compounding (at least US$50/ton) and clay the incremental cost is ca. US$300 to 400, leaving little room for profit. Thus, in the case of CPNC the commercial success of PNC hinges on the intercalation and compounding costs. Applying a similar algebra to POSS with the lowest projected price of US$33,000/ton the situation is difficult. Furthermore, to achieve an interesting enhancement of properties relatively large amounts of POSS must be added (from 10 to 50 wt%). For applications as a structural material the cost would be prohibitive. Obviously, the economics look better when more expensive
70
Nanoparticles of Interest to PNC Technology engineering or speciality resins are modified by POSS. However, by the same token POSS use is automatically diverted from structural (large production volume) to functional speciality products. From the point of view of structural performance, incorporation of POSS resulted in relatively minor gains, viz. mechanical performance about 10 times smaller than that obtained by incorporation of the same quantity of organoclay. The effect on the transition temperatures, Tg or Tm, was found to depend on the type of POSS, the polymeric matrix and the method of its incorporation – variations from a decrease by 100 to an increase by 200 °C were reported. Finally, the ‘lightweight’ of POSS systems is in comparison with mineral-filled composites and it does not seems to offer much above the virtually ‘neutral’ effect on matrix density of organoclay. In consequence, CPNC and POSS find applications in different domains, the former as structural material, e.g., for the automotive or packaging industries, the latter as a functional material in electronic, space or bio-applications. Vive la différence!
71
Clay-Containing Polymeric Nanocomposites
72
Clays
2.2
Clays
2.2.1 General Characteristics Clays originate from the hydrothermal alteration of alkaline volcanic ash and rocks of the Cretaceous period (85-125 million years ago). The airborne ash carried by winds formed deposits characterised by high volume bedding of ash, deposited in seas and alkaline lakes. Different opinions have been expressed regarding the mechanism of the ash to clay transformation. Probably the change began in marine water in reactions involving sufficient amounts of Mg+2 and Na+. Several geological processes may have lead to the formation of clays during millions of years [Keller, 1979; Giese and van Oss, 2002; Drits, 2003; Lagaly and Zismer 2003]. Clays are distinctive from rocks in several aspects: 1. Wet clays can be formed by application of light force and after release of the pressure they retain the imposed shape. 2. Clays are composed of extremely fine crystals, usually plate-shaped, less than 2 μm in diameter and less that 10 nm thick. They are mostly phyllosilicates, i.e., hydrous silicates of Al, Mg, Fe, and other elements. Having at least one small dimension and large aspect ratio they have large specific surface areas. This in turn makes clays physically sorptive and chemically surface active. Several clay types carry an excess negative electric charge owing to internal substitution by lower valency cations, viz. Mg2+ substituted for Al3+, which makes clay slightly acidic. A clay deposit usually contains non-clay minerals as impurities, viz. quartz, sand, silt, feldspar, mica, chlorite, opal, volcanic dust, fossil fragments, heavy minerals, sulfates, sulfides, carbonate minerals, zeolites, and many other rock and mineral particles ranging in size from colloidal to pebbles. Clays are classified on the basis of their crystal structure and the amount and locations of charge (deficit or excess) per basic cell. In the context of PNCs, the amorphous clays are a great nuisance as they are difficult to remove from the crystalline ones. The crystalline clays range from kaolins, which are relatively uniform in chemical composition, to smectites, which widely vary in their composition, cation exchange properties, and the ability to expand. The ease of separation of the individual layers is related to the interlamellar charge, x. The latter parameter changes from zero (talc) to x = 0.2 to 0.6 for smectites, to x = 0.6 to 0.9 for vermiculites, and to x = 1 to 2 for micas [Giese and van Oss, 2002]. Clay particles are usually plate-shaped, less often tubular or scroll-like. Individual clay particles are nanometre-sized at least in one dimension. Aqueous suspensions of clays are thixotropic and sensitive to ion concentration.
73
Clay-Containing Polymeric Nanocomposites The synthesis of clays has been extensively studied [de Kimpe et al., 1961; Roy, 1962; Weaver and Pollard, 1973; Velde, 1977]. For example, organic compounds facilitate the synthesis of kaolin at low temperature by condensing aluminum hydroxide into octahedrally coordinated sheets [Linares and Huertas, 1971]. More recently, Carrado [2000] published an excellent review on synthetic clays and the resulting CPNCs.
2.2.2 Crystalline Clays Most clays are crystalline, composed of fine, usually plate-shaped crystals about 1 nm thick with high aspect ratio, and have large specific surface areas. They absorb up to a 30-fold amount of water, and when wet, can be easily shaped – pottery is as old as human civilisation. 2.2.2.1 Kaolins These include kaolinite, dickite, nacrite and halloysite-endellite. The structural formulae for kaolinite and endellite are A1 4 Si 4 O l0 (OH) 8 and A14Si4Ol0(OH)8.4H20, respectively. The kaolinite lattice consists of one sheet of tetrahedrally coordinated Si (with O) and one sheet of octahedrally coordinated Al (with O and OH), hence a 1:1, or a two-layer structure. A layer of OH completes the charge requirements of the octahedral sheet. Adjacent cells are spaced about 0.71 nm across the (001) plane. When solvated in ethylene glycol, endellite expands to 1.0 nm in the c-direction. Halloysites are usually tubular or scroll-shaped; they may be differentiated from kaolinite and dickite by treatment with potassium acetate and ethylene glycol. For example, kaolin (KGa-1) from Georgia [van Olphen and Fripiat, 1979]: •
• • • •
Contains (wt%): SiO2 = 44.2, Al2O3 = 39.7, TiO2 = 1.39, Fe2O3 = 0.13, FeO = 0.08, MnO = 0.002, MgO = 0.03, Na2O = 0.013, K2O = 0.05, F = 0.013, P2O5 = 0.034. Loss on heating to 550 °C is 12.6 wt%. CEC = 0.02 meq/g, the specific surface area = 10.05 ± 0.02 m2/g. DTA: endotherm at 630 °C, exotherm at 1015 °C, dehydroxylation weight loss 13.11% (theory 14%). The unit cell composition is: (Mg0.02 Ca0.01 Na0.01 K0.01)[Al3.86 Fe(III)0.02 Mntr Ti0.11][Si3.83l0.17]O10(OH)8, octahedral charge: 0.11, tetrahedral charge: -0.17, interlayer charge: -0.06.
2.2.2.2 Serpentines Substituting Mg for Al in the kaolin structure results in the serpentine, Mg3Si2O5(OH)4. Here all three possible octahedral cation sites are filled, yielding a tri-octahedral group carrying a charge of +6. In kaolinite only 2/3 of the sites are occupied by Al, yielding a di-octahedral group, also with a charge of +6. Most serpentines are tubular or fibrous. Chrysotile occurs in both clino- and ortho-structures. 2.2.2.3 Illite Group (Micas) ‘Mica’ is a generic term applied to a group of complex aluminosilicates having a sheet or plate like structure with a wide range of chemical compositions and 74
Clays physical properties. All micas form flat six-sided monoclinic crystals with a remarkable cleavage in the direction of the large surfaces, which permits them to split easily into optically flat films, as thin as one micron. When split into thin films, they remain tough and elastic even at high temperature. The dictionary defines mica as ‘a class of silicates having a prismatic angle 120o, eminently perfect basal cleavage, affording thin tough laminae or scales, colorless to jet black, transparent to translucent, of widely varying chemical composition, and crystallising in the monoclinic system’. Illites or micas are not pure minerals. The mica structure consists of a pair of tetrahedral sheets enclosing an octahedral sheet. Between each such sandwich there are interlayer sites, which can contain large cations. Considerable variation exists in the composition and polymorphism of the illites. A basal spacing exhibited in XRD, d001 ≅ 1.0 nm, is somewhat broad and skewed toward wider spacings. Muscovite derivatives are typically dioctahedral; phlogopite derivatives are trioctahedral. The cation-exchange capacity of illite is CEC = 0.2-0.3 meq/g of dry clay. The interlayer potassium exerts a strong bond between adjacent clay structures. Thus, mica possess a 2:1 sheet structure, similar to MMT, except that the maximum charge deficit in mica is typically in the tetrahedral layers and contains potassium held tenaciously in the interlayer space. As a result, micas are difficult to exfoliate. However, once exfoliated they form dispersions of platelets with the highest aspect ratio, thus they are particularly useful for the control of gas or liquid permeability. Several examples of CPNC with mica are provided in this book. Pironon et al. [2003] proposed the use of FTIR of clay-NH4+ to distinguish illite from smectite clay. The coordination of the octahedral sheet is completed by OH anions. The general formula of mica group minerals is XY2-3Z4O10(OH)2, where X represents the interlayer site, Y the octahedral sites and Z the tetrahedral sites. The octahedral sheet can be made up in two ways: either dominantly of divalent cations such as Mg2+ or Fe2+, in which case all three sites are filled (trioctahedral mica), or else dominantly trivalent cations such as Al3+, in which case one of the three sites is left vacant (dioctahedral mica). If solely Si occupies the tetrahedra, the sandwich is charge-balanced and there is no need for interlayer cations – the resulting minerals are talc (trioctahedral) or pyrophyllite (dioctahedral). In true micas Al substitutes for Si in the tetrahedra, and charge balance is maintained by K, Na or Ca, in the interlayer site. The important rock-forming micas are the trioctahedral phlogopite and the dioctahedral ‘white’ micas: • Phlogopite: KMg3[AlSi3]O10(OH)2 • Wonesite (or sodium phlogopite): NaMg3[AlSi3]O10(OH)2 • Annite: KFe3[AlSi3]O10(OH)2 • Eastonite: K[Mg2Al][Al2Si2]O10(OH)2 • Muscovite: KAl2[AlSi3]O10(OH)2 • Paragonite: NaAl2[AlSi3]O10(OH)2 • Margarite: CaAl2[Al2Si2]O10(OH)2 • Mg-Al-celadonite: K[MgAl][Si4]O10(OH)2 • Fe-Al-celadonite: K[FeAl][Si4]O10(OH)2 The average composition of mineral mica is (wt%): SiO2 45.57; Al2O3 33.10; K2O 9.87; Fe2O3 2.48; Na2O 0.62; TiO2 trace; CaO 0.21; MgO 0.38; moisture 75
Clay-Containing Polymeric Nanocomposites at 100 °C 0.25; P 0.03; S 0.01; graphite C 0.44; loss on ignition (H2O) 2.74. The physical properties of a typical phlogopite and muscovite are listed in Table 12. 2.2.2.4 Chlorites and Vermiculites Chlorite was identified as a mineral yielding a 1.4 nm basal spacing in clays. Chlorite is a three-layer phyllosilicate separated by a Mg(OH)2 interlayer. Chloritelike structures have been synthesised by precipitating Mg and Al between MMT sheets [Slaughter and Milne, 1960]. The interlayer sheet in vermiculite is octahedrally coordinated, 6H2O about Mg2+. The basal spacing of vermiculite varies from 1.4-1.5 nm with the nature of the interlayer cation and its hydration. The cation-exchange capacity of vermiculite is relatively high and it may even exceed that of MMT. Vermiculites are known to have high aspect ratio (p ≤ 2,500), exceeding that of MMT by nearly one order of magnitude. 2.2.2.5 Other Clays 2.2.2.5.1 Glauconite Glauconite is green, dioctahedral, micaceous, rich in Fe3+ and K+ ions. It has many characteristics common to illite. Glauconite may contain randomly placed expandable layers of the montmorillonite-type. The glauconitic green sands of New Jersey have been used in ion exchange, water-softening installations, and as a source of slowly released potassium in soil. 2.2.2.5.2 Sepiolite, Palygorskite and Attapulgite Sepiolite and palygorskite contain a continuous two-dimensional tetrahedral sheet and thus differ from the other layer silicates by absence of the octahedral sheet. Details of the structures were described by Jones and Galan [1988]. The attapulgite structure is similar to palygorskite minerals resembling cardboard, paper, leather, cork, or even fossil skin. These clays have distinctive properties, not shown by platy clays. Attapulgite and palygorskite sorb both cations and neutral molecules. Typical CEC is about 0.2 meq/g of dry clay. Sepiolite and attapulgite are best identified by their 110 reflections in XRD, 1.21 and 1.05 nm, respectively. 2.2.2.5.3 Mixed-Layer Clay Minerals In mixed-layer clay sheets illite may be interspersed with MMT or chlorite. Corrensite has regular alternation of chlorite and vermiculite layers. 2.2.2.6 Smectites or Phyllosilicates Smectites are the most frequently used clays for a variety of non-ceramic applications. These 2:1 phyllosilicates have a triple layer sandwich structure that consists of a central octahedral sheet dominated by alumina, bonded to two silica tetrahedral sheets by oxygen ions that belong to both sheets (see Figure 17). Smectites are structurally derived from pyrophyllite [Si8Al4O20(OH)4], or talc [Si8Mg6O20(OH)4] by substitutions mainly in the octahedral layers, viz. Al can be substituted by Mg, Fe, Cr, Mn or Li. When substitutions occur between ions of unlike charge, deficit or excess charge develops on corresponding parts of the structure. The charge imbalance is compensated by the presence of cations (usually 76
Clays
Table 12 Physical properties of micas Characteristic
Unit
Colour Density
g/ml
Specific heat
Phlogopite
Muscovite
Amber/ yellow
Ruby/ green
2.6-3.2
2.6-3.2
0.21
0.21
Hardness
Moh’s Scale
2.3-3.0
2.8-3.2
Hardness
Shore Test
70-100
80-105
5-25
55-75
Optic axial angle Tensile strength
MPa
≈ 100
≈ 1.75
Shear strength
MPa
100-130
220-260
Compression strength
MPa
-
190-280
Modulus of elasticity
GPa
140-210
140-210
Thermal expansion x106 ⊥ to cleavage plane
1/°C
30-60
9-36
Calcination temperature
°C
900-1000
700-800
Maximum operating temperature
°C
800-900
500-600
Thermal conductivity, ⊥ to cleavage planes
J/s/cm/°C
≈ 0.00042
≈ 0.0054
Thermal conductivity cleavage planes
J/s/cm/°C
≈ 0.050
-
%
3.0
4.5
Very low
Very low
to
Water of constitution Moisture absorption Apparent electric strength; thin
KV/mm
-
120-200
RMS at 15 °C
KV/mm
30-60
40-80
Permittivity at 15 °C
5-6
6-7
Power factor (loss tangent) ×10-3 at 15 °C
1-5
0.1-0.4
1-100
400200,000
Sulfuric
Hydrofluoric
Volume resistivity ×10-12 at 25 °C Chemically affected by acid
Ω cm
77
Clay-Containing Polymeric Nanocomposites
Figure 17 Idealised structure of dry phyllosilicate with the unit cell: [Al2(OH)2(Si2O5)2]2 + 5 wt% H2O. The unit cell molecular weight is 720 + water and counterions (e.g., Na+); the minimum d001 spacing for dry MMT is 0.96 nm (the interlamellar gallery height 0.30 nm), the surface area of the unit cell is 0.458 nm2. After van Olphen and Fripiat, 1979.
Na+, Ca2+, K+) sorbed between the three-layers. The cations are stoichiometric, but held relatively loosely and are readily exchanged by other cations. The triplesheet layers form stacks with the interlamellar gallery between them. Their cationexchange capacity (CEC) is high, 0.8-1.2 meq/g of air-dried clay, and may be used as a diagnostic criterion of the group. Characteristically, smectites expand in H 2O or alcohol. The size and composition of the interlamellar gallery is highly variable – its minimum thickness is 0.26 nm, which corresponds to a monolayer of water molecules. With each of the three sheets 0.22 nm thick, the minimum thickness of the phyllosilicate interlayer spacing as read by XRD is d001 = 0.92 nm. The flat thin sheets of smectite crystals have irregular shape and can be up to 1,500 nm in the largest dimension. Thus, the aspect ratio of smectites is: p ≤ 1500. Owing to counterbalancing ions present, the nominal value for d001 is taken as 0.96 nm. As shown in Figure 18, the van der Waals interaction energy (EA) that holds the stacks together depends very much on the distance between the platelets (h), or on the interlamellar spacing. In the early literature, the term montmorillonite was used for this group. In the natural state these minerals are partially hydrated, with XRD basal spacing d001 = 1.2-1.4 nm. Solvating them in ethylene glycol expands d001 to 1.7 nm, while heating to 550 °C collapses it to 0.96 nm. The DTA curves for smectites show three endothermic and one exothermic peak, within the ranges 150-320, 78
Clays
Figure 18 Van der Waals attraction energy between two phyllosilicate layers separated by a distance h (nm) (after van Olphen and Fripiat, 1979).
695-730, 870-920, and 925-1050 °C, respectively. The crystal lattice is weakly bonded. The smectite lattice is expandable between the silicate layers, hence when soaked in water it may swell to several times its dry volume (e.g., bentonites of Badlands). A broken surface of these clays typically shows a ‘corn flake’ or ‘oak leaf’ texture. There are several species of smectite clay, but the two of greatest commercial importance and value are montmorillonite (MMT) and hectorite (HT). MMT tends to have sheet morphology whereas hectorite has a lath or strip morphology. Commercial availability of hectorite is limited whereas MMT deposits are large and widely spread around the globe, e.g., 7 in Canada, 6 in the USA, two in South America, 15 in Europe, 7 in Africa, 3 in Australia, and 8 in Asia. Typical chemical formulae of the smectite clays are listed in Table 13. 2.2.2.6.1 Bentonite Bentonite (named after Ford Benton, Wyoming) is rich in MMT (usually > 80%). Its colour varies from white to yellow, to olive green, to brown to blue. Grades of this mineral show a broad spectrum of properties and consequently find a variety of applications and uses. Its origin is a hydrothermal alteration of volcanic ash deposited in a variety of freshwater (e.g., alkaline lakes) and marine basins (abundant marine fossils and limestone), characterised by low energy depositional environments and temperate climatic conditions. The deposits date from Jurassic to as recent as the Pleistocene epoch, but most are from the Cretaceous period (85-125 million years ago). Bentonite beds range in thickness from several centimetres to tens of metres (most 0.3 to 1.5 m) and can extend for hundreds of kilometres. Bentonite is widely distributed on all continents. In Canada large deposits are located in British 79
Clay-Containing Polymeric Nanocomposites
Table 13 Typical smectite clays Mineral Montmorillonite (MMT)
Chemical composition of 1/2 crystalline unit cells [All.67Mg0.33(Na0.33)]Si4Ol0(OH)2
Hectorite
[Mg2.67Li0.33(Na0.33)]Si4O10(HOOFF)2
Beidellite
A12.17[Al0.33(Na0.33)Si3.17]O10(OH)2
Nontronite
Fe(III)[Al0.33(Na0.33)Si3.67]O10(OH)2
Saponite
Mg3[Al0.33(Na0.33)Si3.67]O10(OH)2
Sauconite
[Znl.48Mg0.l4Al0.74Fe(III)0.40][Al0.99Si3.01]O10(OH)2X0.33
Volkhonskoite
contains Cr2+
Medmontite
contains Cu2+
Pimelite
contains Ni2+
Note: Na0.33 or X0.33 refers to the exchangeable cation in the interlamellar gallery, of which 0.33 equivalent is typical
Columbia (French Bar, Hat Creek, Princeton, Quilchena, etc.), Alberta (Rosalind), Saskatchewan (three ca. 1 m thick beds of Na-MMT near Truax), and Manitoba (Morden). In the USA, it occurs mainly in Wyoming (overburden: 12 m thick; 1997 price: US$25 to 40 a short ton; estimated deposits: ≥ 1 billion ton), Georgia, Florida, Mississippi, South Dakota, Montana, Utah, Nevada, and California. Other known deposits are located in Algeria, Tasmania, Mildura and Scone (Australia; estimated deposit of 70 million tons), San Juan (Argentina), Sarigyugh (Armenia), Campina Grande (Brazil), Guangxi and Hangzhou Linan (P. R. China), Hamza (Egypt), Gewane, Mille and Ounda Hadar (Ethiopia), Landshut (Germany), Milos (Greece), Kutch (Gujarat, India, estimated deposits: 15 million ton of good quality sodium bentonite), Java - (Indonesia; estimated deposits 8 million ton), Sardinia (Italy), Annaka and Kuroishi (Japan), Pershwar (Pakistan), Holy Cross Mountains, Upper Silesia Coal Basin and Sudetes (Poland; deposits: 3 million ton), Eskisehir and Biga Peninsula (Turkey), Tam Bo (Vietnam; deposits: 3 million ton), etc. Main uses for bentonite are in foundry sands, drilling muds, iron ore pelletising, absorbents, as a variety of composite liners, food additive for poultry and domestic animals, in filtration, foods, cosmetics and pharmaceuticals. Bentonite has been used for clarification of liquids (especially white wine and juice). Bentonite is part of most adsorbent, bleaching and catalyst clays. About 6 million tons of bentonite is produced annually [Harben and Bates, 1990; Carr, 1994]. 2.2.2.6.2 Montmorillonite (MMT) Montmorillonite (MMT) is the name given to clay found near Montmorillonite in France, where MMT was identified by Knight in 1896. It is the most common 80
Clays phyllosilicate used for the production of commercial CPNC. MMT has been known under several names, viz., smectite; sodium montmorillonite; sodium bentonite or swelling bentonite (Wyoming bentonite (US)); sodium-activated bentonite (UK); sodium-exchanged bentonite, etc.); the non-swelling (in water) bentonite is calcium montmorillonite or bentonite (Mississippi bentonite (US)); Sub-bentonite (Texas bentonite, US)). Also known are magnesium montmorillonite (Saponite & Armargosite), potassium montmorillonite (Metabentonite), lithium montmorillonite (Hectorite), etc. The idealised structure of Na-MMT is shown in Figure 17. The unit cell is usually written as:
[
]
− ( 0.67 ) Triple layer sandwich of two silica tetrahedron Al3.33 Mg0.67 Si8 O20 (OH )4 sheets and a central octahedral sheet with 0.67 ⇓ negative charge per unit cell Aqueous interlamellar layer containing 0.67 Na+ (n × H2 O) Na0+.67 cations per unit cell
Thus, an idealised MMT has 0.67 units of negative charge per unit cell, in other words, it behaves as a weak silicilic acid. Since the molecular weight of a unit cell is Mu = 734 + water, the CEC of idealised MMT is: CEC = 0.915 meq/g (one ion per 1.36 nm2), i.e., the anionic groups are spaced about 1.2 nm apart. The charge is located on the flat surface of the platelets. As Thiessen demonstrated in 1947, a small positive charge is also present at the platelet edges. MMT composition varies across a relatively wide range not only with the geographic location but also with the deposit strata [Ross and Hendrics, 1945; Ross, 1960]. The data for 100 samples gave the following ranges of composition: 1. The octahedral layer: Al3.0 - 4.0 Mg0 - 1.4 Fe3+0 - 1.0 2. The tetrahedral layers: Al0 - 0.8 Si7.2 - 8.0 3. Exchangeable cation in the aqueous layer: Na0.67 - 0.8 Chemical analysis of a typical MMT yields: SiO2 = 51.14, Al2O3 = 19.76, Fe2O3 = 0.83, ZnO = 0.1, MgO = 3.22, CaO = 1.62, K2O = 0.11, Na2O = 0.42 and water 22.8 wt%. Drying at 150 °C eliminates 14.81 wt% of water, with 7.99 wt% remaining. Another chemical analysis of MMT gave: Al 9.98 wt%, Si 20.78 wt%, H 4.10 wt%, and O 65.12 wt%. Depending on composition, MMT colour varies from brick red (due to Fe+3) to pale yellow or blue-grey. The CEC ranges from 0.8 to 1.2 meq/g. In the montmorillonite-nontronite series, as the Fe3+ content increases from 0 to 28%, the refractive index ranges from 1.523 to 1.590. XRD patterns of a hydrated MMT yield typically d001 = 1.2-1.4 nm basal spacing. These values, along with expansion to ca. 1.8 nm in glycerol, have been used for MMT identification. The specific surface area of MMT is Asp = 750-800 m2/g (theoretical value is 834 m2/g). From the cited values it follows that the density of the triple sandwich is 4.03 g/ml and that the interlamellar gallery thickness is 0.79 nm, hence the interlayer thickness of hydrated MMT should be d001 = 1.45 nm and the average density ρ = 2.385 g/ml. Drying MMT at 150 °C reduces the gallery height to 0.28 nm (which corresponds to a water monolayer), hence the interlayer spacing decreases to d001 = 0.94 nm and the average density increases to ρ = 3.138 g/ml. Assuming that MMT platelets are fully exfoliated and that locally they are parallel 81
Clay-Containing Polymeric Nanocomposites to each other, the interlayer spacing should be inversely proportional to the clay volume fraction, φ, d001 = h/φ where h ≅ 0.96 nm is the thickness of a single (exfoliated) clay platelet. This simple relation predicts that the interlayer thickness, e.g., of PA-6 containing, respectively, 4 and 25-wt% of clay should be 32 and 4.6 nm, which is not far from the measured values (see Figure 3). Commercially, MMT is supplied in the form of powder with about an 8 μm particle size, each containing about 3000 platelets with a moderate aspect ratio p = 10 to 300. Typical properties and applications are listed in Table 14. For example [van Olphen and Fripiat, 1979] Na-MMT from Wyoming, SWy-1, has the composition (wt%): SiO2 = 62.9, Al2O3 = 19.6, TiO2 = 0.090, Fe2O3 = 3.35, FeO = 0.32, MnO = 0.006, MgO = 3.05,CaO = 1.68, Na2O = 1.53, K2O = 0.53, F = 0.111, P2O5 = 0.049, S = 0.05. Weight loss on heating to 550 °C = 1.59 and to
Table 14 Properties and applications of montmorillonite [Grimshaw, 1980] Physical Constants
Applications
Unit cell molecular wt. (g/mol)
540.46
To slow down water flow through soil
Density (g/ml)
(2.5) 2.3 to 3.0
To produce nanocomposites
Crystal system and d-spacing (nm)
Monoclinic; 1.47x0.442x0.149
To de-colour & purify liquids, viz. wines, juices, etc.
Moh’s hardness @ 20 °C
1.5- 2.0
As filler for paper or rubber
Appearance
White, yellow or brown with dull luster
Cleavage
Perfect in one direction, lamellar
In drilling muds to give the water greater viscosity As a base for cosmetics and drugs As an absorbent As a base for pesticides and herbicides
Characteristic
In H2O its volume expands up to 30-fold
Field indicators
Softness, and soapy feel
DSC endothermic peaks, T (°C)
140, 700, 875
DSC exothermic peak, T (°C)
920
MMT swells in water more than any other mineral
Largest for Na-MMT, Absorption of ammonia, smaller for multivalent proteins, dyes and other polar, aromatic or ionic compounds counter-ions
82
As food additive for poultry and pets For thickening of lubricating oils and greases For binding foundry sands To generate thixotropy
Clays 1000 °C = 4.47 wt%. CEC = 76.4 meq/100 g, principal exchange cations are Na+ and Ca2+. DTA endotherms occur at 185 °C (shoulder at 235 °C), desorption of water occurs at 755 °C, dehydroxylation; shoulder at 810 °C, exotherm at 980 °C. Weight loss in dehydroxylation = 5.53 wt% (theory = 5%). Cell structure is: (Ca0.12 Na0.32 K0.05)[Al3.01 Fe(III)0.41 Mn0.01 Mg0.54 Ti0.02][Si7.98 Al0.02]O20(OH)4; octahedral charge = -0.53, tetrahedral charge = -0.02, interlayer charge = -0.55, unbalanced charge = 0.05. The early treatments of MMT involved acid-treatment and preparation of organoclays. The aim of the former was purification, replacement of Ca2+ for H+ and dissolution of some Fe, Al and Mg ions from the octahedral layers. Manufacture of organoclays started in the 1940s (NL Industries) for use as thixotropic additives to control flow behaviour of oils, greases, suspensions, paints, printing inks, cosmetics, etc. As described earlier in this text, PNC literature recognises four types of dispersion of layered silicates in a polymer matrix (see Figure 19): (A) conventional dispersion of non-intercalated clay particles with the basic dry structure, (B) intercalated form where the interlayer spacing d001 < 8.8 nm, and (C or D) exfoliated structures where d001 > 8.8 nm with the individual platelets either ordered (because of stress field or concentration effects) or not, respectively. In aqueous suspension the clay platelets may adopt more complex structures (see Figure 20). The platelet association (flocculation) may take place by face-to-face (FF), face-to-edge (FE) or edge-to-edge (EE) interactions [Qian et al., 2000]. Evidently, each clay structure results in a different set of suspension properties. It suffice to note that from the hydrodynamic point of view the encompassed volume of suspended particles dramatically increases from structure (a) to (g), the increase that is reflected in rheological properties. The EE and EF structures are especially significant.
Figure 19 Four types of clay platelet dispersions in a polymeric matrix
83
Clay-Containing Polymeric Nanocomposites
Figure 20 Structures encountered in clay suspensions [1Qian et al., 2000]
Recently, Okamoto et al. [2001a] reported formation of a ‘house of cards’ structure in PP/clay nanocomposite melt under extensional flow. The authors considered that high strain hardening and rheopectic effects originate from the perpendicular alignment of the silicate layers to the stretching direction. When the stress vanishes, the platelets form the complex structures.
2.2.3 Purification of Clay The names bentonite, smectite and montmorillonite are often used interchangeably. However, industrially these terms represent different minerals with different degrees of purity. Bentonite is the ore that comprises smectite clay and impurities, such as gravel, shale, limestone, etc. Purification of the ore that results in extraction of MMT is a complex and expensive process. Clay is usually mined in open pits. The depth of a deposit can be from a few centimetres to several metres with a length of up to hundreds of metres. After the overburden is removed, layers of clay are disked and allowed to sun dry. The clay is removed from the pit in layers, and stockpiled in multiple layers. Next, the dry clay is transported to processing facilities in trucks that are loaded from the stockpile in such a way that each ‘swipe’ of the front-end loader goes through every layer of the stockpile, insuring homogenisation. Polymer-grade clay should have < 5 wt% non-smectite impurities. Typical contaminants include silica, feldspar, gypsum, albite, anorthite, orthoclase, apatite, halite, calcite, dolomite, sodium carbonate, siderite, biotite, muscovite, chlorite, stilbite, pyrite, kaolinite, hematite and many others. MMT usually has > 50 wt% calcium montmorillonite, Ca-MMT. Since Ca-MMT (as well as H-MMT) show non-uniform expansion in water, purification of clay often involves cation exchange into Na-MMT [Norrish, 1954]. The purification often involves reduction of clay particle size either by mechanical means (milling, grinding, comminuting, etc.) or by application of hydrodynamic means [Cohn, 1967]. 84
Clays Clocker et al. [1976] provided a detailed description of the clay purification process. Thus, clay is mixed with water, then heated by steam to pressurise (T ≤ 243 °C; P ≤ 3.5 MPa) and hydrate the clay. Next, the slurry is rapidly expanded to cause some delamination, non-clay particles removed and large clay particles recycled through the steaming and expansion cycle. The slurry is left in a holding zone for up to 2 h, at optimised temperature (e.g., T = 80 to 130 °C) and pressure (P = 0 to 0.2 MPa) for the intercalation, and again separated (e.g., by hydrocyclone) from the non-colloidal particles. Finally, the purified clay with expanded interlayer spacing is treated with an onium intercalant (e.g., octadecyl ammonium chloride (ODA)), filtered out, washed and dried. The resulting organoclay has been used for thickening solutions in various solvents and paints. Organoclay suitable for use in CPNC must be purified and uniformly intercalated with great care. For example, one of the principal advantages of CPNC is to control diffusion of fluids through a polymeric membrane – if the non-colloidal mineral content exceeds ca. 0.5 wt%, holes can be formed around such a foreign particle (e.g., quartz) dramatically increasing the penetrant flux. A patent from AMCOL [Clarey et al., 2000] describes modern purification technology. The clay (see Figure 21) is ground to > 90 wt% particles having diameter d < 200 μm. Dry clay is fed from loader (12) into storage tanks (22) and (24). Using a blower (36) and air injectors (32 and 34) the clay is transported into a receiving vessel (38), and through an auger (48) into a blunger (50) where it is mixed with water. The slurry, containing 5 to 50 wt% clay + impurities, is sedimented to remove stones to a waste hopper (56), while the rest is conveyed to an attrition scrubber (60). The washed clay is pumped by pump (80) into a feed tank (84) for a series of hydrocyclones (86, 98, 110, 142, 170 and 181) that remove impurities of a size > 50 μm. The purified slurry containing 3 to 7 wt% clay is fed to the ion-exchange column feed tank (180). At this stage > 90 vol% clay particles have diameter below 40 μm, with mean particle size of < 7 μm. The suspension is directed upward through cation exchange columns (212 to 222) that replace Ca2+ by Na+, giving > 95 wt% Na-MMT. After the ion exchange, the clay is conveyed through a feed tank (240) to a high speed centrifuge (250) operating at centrifugal force of 2.5 to 3.5 kG. The product is conveyed to a tank (258) and then through a vibrating dewatering unit (266) which produces the clay slurry containing about 12 wt% solids. The slurry is transported to a spray dryer (272), where it is dried to about 9 wt% water. Finally, the material is conveyed to an air filtering baghouse (276), holding tanks (282 to 286), and then to bagging apparatus (292 and 294) [Harben and Bates, 1990]. Statistical process control is used to maintain the product within the accepted limits. The important process control parameters are: concentration, counterion level, purity, particle size, moisture ratio, dispersive characteristics, etc.
2.2.4 Reactions of Clays with Organic Substances MMT crystals are made of flat sheets (individual clay platelets), 0.92 nm thick and up to 1,500 nm in the largest dimension. Their specific surface area is 750 to 800 m2/g. The crystals form large particles or aggregates, even after purification the Na-MMT particles are ca. 8 μm in diameter, each containing about 3000 platelets with an aspect ratio of p = 50 to 300. To be incorporated into a polymeric matrix these particles must be dispersed into individual stacks of platelets, then delaminated into a uniform dispersion of individual platelets in the matrix. The 85
Clay-Containing Polymeric Nanocomposites
Figure 21 Process for clay purification (see text) Reproduced from Clarey et al. [2000], with permission from AMCOL.
86
Clays process of delamination usually goes through two stages: intercalation and exfoliation. As evident from the definition of these terms, they differ not in character, but in the magnitude of the achieved platelet separation. The attraction energy (Uattraction) of two platelets of equal thickness is given by Equation 13 [Stokes and Evans, 1997]: U attraction = −
⎤ ⎡1 1 2 − ⎥ ⎢ 2 + 2 2 12π ⎢ h (h + 2δ) (h + δ) ⎥⎦ ⎣ A11
(13)
where, A11 is the Hamaker constant, h is the separation distance between plates, and δ is the platelet thickness. It is noteworthy that the surface energy of inorganic solids is about 100 times higher than that of organic liquids. According to Equation 13 the interactions between two crystalline planes decrease with the inversed square of the separation distance (see Figure 18). Thus, the logical approach to the process of delamination is to increase the gallery space in several steps, inserting progressively larger molecules – this is indeed the most common intercalation/ exfoliation strategy. However, before discussing the methods developed for the intercalation and exfoliation that lead to CPNC it is advisable to summarise the chemistry of clays. Several books and reviews are available on the topic, viz. the ever popular early text by Weiss [1969], Mortland’s review [1970] and other such publications [Weaver and Pollard, 1973; van Olphen and Fripiat, 1979; Grimshaw, 1980; Theng, 1974; Raussell-Colom and Serratosa, 1987; Mark et al., 1995], a review on synthetic clays [Carrado, 2000], etc. Several active sites on the MMT crystal surface have been identified. In the absence of water, for molecules not fully coordinated Al constitutes an electronpair accepting site. Ions of Fe+3 and Fe+2 provide, respectively, oxidising (electronaccepting) and reducing (electron-donor) sites (similar activity in the presence of other metallic impurities is expected). Interactions between the clay surface and organic molecules by van der Waals forces and by entropic rearrangement effects have also been observed. From the point of view of intercalation/exfoliation that may lead to CPNC the following three active sites are the most important [Brown et al., 1952]: 1. The platelet edges have a few positive charges that attract negatively charged ions or molecules. These sites have been used to attach organic molecules making MMT organophilic. However, the reactions involving edge ions do not increase the interlayer spacing, hence they are not useful for intercalation. Evidently, once MMT is intercalated, these sites can be used, e.g., to enhance platelet miscibility with polymeric matrix or mechanical dispersability during melt compounding. For example, Na-MMT is able to react with weak organic acids, viz. tannin or ligno-sulfonates. This method has been used for dispersing clay particles in drilling mud. The reaction does not increase the interlayer d001 spacing. 2. The –OH groups (four per unit cell) may participate in hydrogen bonding and chemical reactions. These are mainly located at the crystal edges, bound to Si, Al or other octahedral ions. The face surface location of –OH groups has also been postulated [Deuel, 1952; Uytterhoeven, 1960; 1962]. The concentration of –OH groups can also be enhanced by digesting MMT with NaOH solution. 87
Clay-Containing Polymeric Nanocomposites Hydrogen bonding to the surface oxygen atoms provides the mechanism responsible for adsorption of organic molecules by clay particles. The reaction depends on the suspending medium. In water, hydrogen bonding by a watersoluble polymer is much less likely to take place than in an organic medium. H-bonding is significantly weaker than that with exchangeable ions. It depends on the polarity of the interacting group, hence its strength decreases from –NH3+, to –OCH2- to –CH3 [Gonzales-Carreño et al., 1977] as well when the charge is tetrahedrally located [Farmer and Russel, 1971]. The H-bonding depends on the pH of the suspension, thus H-MMT is more likely to involve H-bonding than Na-MMT [Mortland, 1966]. Often H-bonding is a competitive process, e.g., H-bonding between the alcohol molecules themselves moderates adsorption of alcohol by MMT [Annabi-Bergaya et al., 1981]. It has been reported that (after cation exchange) clay containing either UO22+ or Fe3+ can be made hydrophobic by acid-catalysed reaction in n-heptane with octadecyl-trimethoxy silane [Giaquinta et al., 1997; Wasserman et al., 1998]. The silane film inhibited free exchange of water in and out of the interlamellar galleries, thus after exposure to water the interlayer spacing remained stable, similar in magnitude to the dry state. The authors did not elaborate on the reaction mechanism. 3. The anionic group on the MMT face surface; MMT in aqueous medium behaves as weak silicilic acid. These sites are of principal interest for diverse intercalation methods. It has been observed that organic compounds may diffuse into galleries and coordinate to alkaline cations. The strength of the complex as well as expansion of the d001 spacing depends on the cation and organic compound. For example, for a series of crown ethers absorbed by MMT, expansion by Δd001 = 0.4 to 0.8 nm was reported [Ruiz-Hitzky and Casal, 1978]. The cation exchange reactions in aqueous medium have been extensively studied. It has been known for 70-odd years that the surface cations can be exchanged for other cations. A measure of this capacity is given by the cation exchange capacity (CEC). For example, the CEC of MMT is 0.8-1.2 meq/g hence one negative ion per 1.36 nm2 is located on the flat platelet surface. These values are considered to be optimal – sufficient concentration of ions to obtain a good level of chemical activity, and at the same time not too much of them to engender too strong solidsolid interactions. To facilitate the ion exchange, an aqueous slurry of sodium montmorillonite is mechanically mixed either in the shear field (e.g., in a colloid mill) or subjected to ultrasonics [Pérez-Maqueda et al., 2003]. It is noteworthy that in aqueous medium clay interlayer spacing expands and platelets interact with each other, increasing the suspension viscosity. The increase depends on the clay counterion, concentration, and flow field [Malfoy et al., 2003]. Thus, for practical reasons the clay suspension during the purification and ion-exchange steps is usually kept below ca. 7 wt%. The cation exchange reaction is reversible, hence to obtain a high conversion of Na-MMT into organoclay, an excess of intercalating organic cation RH+ is used: RH+ + Na+ – MMT RH+ – MMT + Na+ The reaction rate depends on the type of clay, the medium, the type of cation to be exchanged, the reaction conditions, viz. temperature (T), pH, concentrations and geometry of clay particles, etc. For example, recently, natural smectite (NaMMT) was intercalated with 3-methyl hexadecyl ammonium bromide (3MHDA) with a molar excess ranging from 0 to 3 CEC. The interlayer spacing increased 88
Clays from the original value of d001 ≅ 1.4 nm to 1.9, 3.2 and 3.9 at a 3MHDA loading of 1, 2, and 3 CEC, respectively. When heated at a rate of 2 oC/min the interlayer spacing started to collapse at T ≥ 200 oC [Lee and Kim, 2003]. To ensure that free cations are in the system, the pH is usually adjusted to at least one unit lower than pK; a too acidic medium may cause cations to leach out from the clay and interfere with the desired exchange reaction. The type of cation present within the interlamellar galleries affects the magnitude and uniformity of interlayer spacing. For example, H+ and Ca2+ show a non-uniform expansion in water, whereas Na+ and Li+ provide easy expansion up to total exfoliation. Under ambient humidity the d001 depends on the cation, viz. for Na+ it is 1.4 nm, for Ca2+ it is 1.5 nm and for UO22+ it is 1.48 [Giaquinta, 1997]. The fastest rates of exchange are reported for Na-MMT. The presence of less hydrated monovalent ions (e.g., K+, Rb+, Cs+) and of multivalent ions (e.g., Ca2+, Mg2+, etc.) reduces the rate. It seems that multivalent cations are capable of simultaneous interaction with anions on adjacent MMT platelets, making intercalation more difficult. However, since the thickness of the triple sandwich is 0.66 nm (see Figure 17) compounds having the smallest dimension below 0.6 nm may diffuse into the interlamellar gallery when d001 > 1.4 nm. The reaction rates in water are faster than those in aqueous solutions of organic liquids, viz. alcohols, but there are exceptions to this rule, e.g., during intercalation with organic cations. Increased T accelerates the process, hence the recommended range is T = 60-80 °C. Similarly, an increase of pressure (P) has been also reported to speed up the reaction. The particle shape and size also affect the kinetics. The reaction is diffusion controlled. It starts at the rim and the rate of the linear diffusion down the slit of diameter d, is proportional to the diffusion time, td, viz. δd/δtd∝td , hence the diffusion time is proportional to the square root of d. This prediction is almost confirmed by data from Mackintosh et al. [1971]. As shown in Figure 22 the experimental exponent a = 0.45, instead of 1/2 was found.
Figure 22 Diffusion time of K+ in exchange reaction for dodecylammonium cation as a function of the biotite clay diameter. Data [Mackintosh et al., 1971]. The broken line follows the empirical relation: td = aoda, with ao =15.63, a = 0.45 and the correlation coefficient, r = 0.9967.
89
Clay-Containing Polymeric Nanocomposites Thus, modification of clay involves mainly a reaction with the anionic silicilic groups on the platelet surface. In the presence of strong, but small in size cations the interaction is mainly ionic, e.g., silicilic acid with Na+. However, when the latter ion is replaced with a quarternary ammonium one, the van der Waals interactions become more important. Their importance increases with the hydrocarbon chain length [Theng, 1964]. Another type of interaction involves the disruptive effects large cations have on the hydration shell around the interlayer cation – the entropic effects. It has been found that the effect is more important for Na+ than for Ca2+ or Mg2+ [Theng, 1967]. 2.2.4.1 Clay in Aqueous Medium 2.2.4.1.1 General Clays are hydrophilic hence the first step in their modification involves reactions in the aqueous medium. The small and highly mobile water molecules easily diffuse into interlamellar species, causing a lateral expansion of the clay crystals. The process is diffusion controlled, hence to swell platelets that are twice as large requires four times longer. The rate depends on many factors and the time required to reach equilibrium swelling varies from minutes to days. Since water monolayer thickness is about 0.28 nm, the first step involves expansion of d001 from 0.96 to about 1.25 nm. Further expansion up to complete exfoliation can be accomplished by judicious control of conditions, especially the ionic strength and counterion type. MMT exposed to water vapour expands its basal spacing in steps (which correspond to one to four molecular layers of water) to d001 = 1.25 to 2.0 nm. The origin of the water absorption by MMT is on the one hand hydration of the counterions in the interlamellar galleries and on the other hydrogen bonding to the clay surface. The work to remove the last monolayer of H2O is about 0.1 J/m2 at the equivalent pressure of 400 MPa. Dispersing Na-MMT in distilled water causes the ions that are associated with the clay surface to diffuse out. The osmotic pressure pulls the ions away from the clay surface, whereas the electrostatic charge tends to hold them near the surface. Eventually a steady state is achieved and an electrostatic double layer is formed that keeps the clay platelets apart. The counterions in the double layer are fairly mobile and can readily be exchanged. The reaction constant depends on the ionic concentration in the double layer and in the supernatant solution. The double layer is formed by adsorption of solvated ions. Their concentration and the associated electric potential decrease with the distance from the platelet surface, following the Nernst equation:
Φ = (kBT/νe)ln(c/co)
(14)
where: kB is the Boltzmann constant, v is the valence, e is the electronic charge, c and co are the ion concentration in the solution and where the potential Φ = 0, respectively. Thus, the thickness of the double layer depends on the counterion concentration and valency. Clay platelets can be fully exfoliated in aqueous media, but on standing, since the surface and edges have different charge, the EF interactions may lead to formation of a ‘house of cards’ structure. If enough clay is present, all the water will be tied resulting in a gel formation. Shearing may 90
Clays disrupt the structure, dramatically reducing the viscosity. The gelation is reversible, as evidenced by the thixotropic effects [van Olphen and Fripiat, 1979]. Another type of structure that may be formed is a tactoid – a parallel alignment of clay platelets, up to 10 nm distant from each other, forming high clay concentration regions in suspensions. In some systems there is a sharp transition between the gel and tactoid phase. For example, an aqueous suspension of nbutyl ammonium vermiculite, n-butyl ammonium chloride and either PVME or PEG has a phase transition temperature between the tactoid and gel phases of the clay system, Tc = 13 ± 1 °C [Jinnai et al., 1996; Hatharasingle et al., 1998]. Tactoid formation is caused on the one hand by the presence of a repulsive double layer and on the other by the attractive van der Waals forces. Tactoids have smaller effects on flow behaviour than the ‘house of cards’ structures. 2.2.4.1.2 Reactions with Edge Cations Tannates have been used as dispersing agents in drilling fluids. The most popular tannate is quebracho tannin, a red-coloured tannin extracted from the South American quebracho tree. Since the tannin is a weak acid, the sodium salt solution is alkaline. Clay suspensions are dispersed by the addition of a small amount of a tannate. The tannate anions are adsorbed at the edge surfaces of the clay particles by complexing with the exposed octahedral aluminium ions. Consequently, the edge charge is reversed and a negative double layer is created by which EF and EE association is prevented. The anion adsorption on the clay edges is absent on the faces of the clay plates. As a result, the basal spacing does not increase after tannate adsorption. To be effective, a phenol must contain at least three phenolic groups in the molecule, two of which should be adjacent. Thus, 1,2,3-hydroxy benzene is effective for dispersing MMT, but 1,3,5-hydroxy benzene is not. Similarly alizarin dyes react with the platelet edges, causing dispersion [Schott, 1968; Freudenberg and Maitland, 1934]. 2.2.4.1.3 Reactions with –OH Groups The vibration stretching frequency of structural –OH groups in MMT are slightly lower than those of unperturbed –OH groups, but easily distinguishable from the broadband of hydrated minerals [Farmer, 1971]. Thus, in aqueous medium these groups form hydrogen bonds with water molecules. The strength of these interactions is expected to be higher than those between water molecules and =Si=O surface groups. 2.2.4.1.4 Reaction with the Silicilic Surface Anions As stated before, anions are adsorbed on the edges of the clay particles and organic cations on the anionic inner surfaces of the clay platelets. This is evident from the high adsorption capacity of the clay as well as the resulting expansion of the d001 spacing. The exchange reactions between Na-MMT and ammonium ions, R-NH3+Cl–, or R4N+Cl– were studied in the early 1930s [Smith, 1934; Gieseking 1939]. When onium salt is added to aqueous clay suspension, the organic cations replace the cations present on the clay surfaces. There is a strong preference for the less mobile organic cations. Often they are adsorbed quantitatively until all the exchange positions are exhausted. The ammonium groups become ionically attached to the clay surface while the hydrocarbon chains interact with the clay 91
Clay-Containing Polymeric Nanocomposites surface and displace the adsorbed water molecules [Hendricks, 1941; Jordan, 1949; Jordan et al., 1950]. When the chains are too long to lie flat in the available space they may tilt, crowding within the interlamellar space. Depending on the conditions, the hydrocarbons may crystallise, hence the spatial organisation of the hydrocarbon segments is given by the appropriate crystalline cell unit and the performance of such organoclay depends on temperature – whether below or above the hydrocarbon melting point. When the hydrocarbon chains cover the clay surface, it precipitates from the aqueous suspensions. However, the formed hydrophobic clay can be homogeneously dispersed in organic medium. Often the concentration of the organic cations exceeds the amount equivalent to the clay CEC. For example, commercial organoclays contain from 98 to 153% of intercalant per nominal CEC of MMT. The reason for the excess is to force the reversible reaction of ion exchange towards the organocomplex and to increase the interlayer spacing to maximum. The orientation of the interlayer cations, hence the d001 spacing, is usually determined by the projected surface area of the cation (in a given orientation) provided that it does not exceed the available surface area per surface anion. Aromatic cations assume either a parallel or an upright position between the layers, depending on the available space. Weiss and Kantner [1960] proposed a method of estimating the surface charge density from the d001 spacing of complexes with mono- and di-alkylammonium cations of different chain lengths. When the modifier molecules are large and bulky the steric effect precludes total cation exchange. The average distance between the silicilic anions on the MMT surface is about 1.2 nm. However, when the concentration of the organic cations is high, their adsorption may exceed the amount equivalent to the clay CEC. For example, quaternary ammonium compound having long hydrocarbon chains can be absorbed in the amount equivalent of two-and-half times the CEC. The excess enters the interlamellar galleries in a head-to-tail configuration, profiting from the hydrocarbon/hydrocarbon chain solubility and the tendency to crystallise. Under these circumstances, the excess ammonium compounds form a layer with the cationic groups facing the water phase, forming a diffused electric double layer that prevents the particles from precipitation. Thus the resulting organoclay is hydrophilic, hence unsuitable for use in organic medium, e.g., for the preparation of CPNC [Cowan and White, 1958; McAtee, 1962; Diamond and Kinter, 1963]. Often, the excess modifier can be removed by washing with hot water or alcohol solution, provided that the cations are not too large [Furukawa and Brindley, 1973]. To avoid formation of the ionic double layer the ionic intercalant is used to stoichiometry, and then its amine is added to form the head-to-tail, non-ionic complex. The perfidy of Nature also provides for an opposite effect – adsorption of less than the stoichiometric number of organic cations. This situation has been observed for large molecules (e.g., codeine on MMT) adsorbed from dilute solutions to form a monolayer on the clay surface, hence shielding adjacent anionic sites on the clay surface by steric hindrance [Weiss, 1963]. 2.2.4.1.5 Stabilisation by Polyelectrolytes Addition of water-soluble polyelectrolytes is a powerful method for controlling clay dispersion [Hesselink, 1971; Hesselink et al., 1971]. Polyelectrolyte consists 92
Clays of long-chain molecules with ionic groups usually located along the entire length of the chain. The polyion may be a polycation with amino groups, or a polyanion with carboxyl, sulfate, sulfonate, or other negative groups. A single polymer molecule may contain both positive and negative groups, as is the case for proteins with amino and carboxyl groups [Hauser, 1950]. Bio-polyelectrolytes (e.g., gum arabic, gelatin, alginates, pectin), modified biopolymers (e.g., oxidised starch, carboxymethylcellulose) and synthetic polyelectrolytes are becoming available, viz. polyacrylic acid [Warkentin and Miller, 1948], polyacrylonitrile [Mortensen, 1962], polyvinyl alcohol [Greenland, 1963], etc. Addition of polyelectrolytes improves the stability of clay dispersion in the aqueous medium. 2.2.4.2 Clay Dispersion in Polar Organic Liquids Like water molecules, polar organic compounds may be adsorbed on the clay surface. The adsorption energy of many of these compounds is comparable with that of water. Depending on the concentration, they can displace adsorbed water from clays, and be removed from the clay by washing with water. When MMT is dispersed in a polar organic liquid (e.g., alcohol, glycol or amine) the suspending liquid molecules penetrate into the interlamellar galleries and displace water. The basal spacing of the complex depends on the size of the organic molecules and on their orientation and packing geometry. The extent of d001 expansion in polar organic liquids is so well defined that it can be used to identify MMT, e.g., ethylene glycol that gives d001 = 1.7 nm has been used to identify MMT [MacEwan, 1948; Bradley, 1945]. The exact mechanism of association between clay and polar molecules is not known (similarly for water!). They may interact through the ionic groups of clay and/or through hydrogen bonding. However, there is little spectroscopic evidence for the interaction between -CH2- or -CH3 and =SiO groups. FTIR indicates that hydrogen bonding is less important [Greene-Kelly, 1955; Brindley and Rustom, 1958]. For adsorption of ammonium ions it was shown that entropic effects provide the driving force [Vansant and Uytterhoeven, 1972]. When the clay surface becomes covered with polar molecules containing a substantial proportion of hydrocarbon groups, the surface becomes oleophilic, and under these conditions the organoclay can be used as an oil or grease thickener. However, since the exchange cations are still present, the complexes are usually sensitive to water. For example, as shown by XRD, the MMT-pyridine complex at low water concentration hydrates stepwise and then exfoliates when a large amount of water is added [van Olphen and Deeds, 1962]. Similarly low molecular weight alkane ammonium complexes (e.g., n-butyl ammonium) show a spectacular interlayer swelling in water [Garrett and Walker, 1962]. However, pyridinium and other large organic cation exchange complexes are not sensitive to H2O. Apparently here the steric shielding of unused silicilic anions provides sufficient protection. Polar, water-soluble macromolecules such as polysaccharides or polyethylene glycol are readily sorbed by MMT, which leads to expansion of the interlamellar space, d001. Edge sorption may also take place [La Mer and Healy, 1963]. 2.2.4.3 Absorption of Organic Molecules by Organoclay The initial adsorption of organic molecules (as described above) may lead to change of the clay character from hydrophilic to hydrophobic and to an expansion of the 93
Clay-Containing Polymeric Nanocomposites interlayer d001 spacing. Both factors tend to facilitate further absorption of organic substances by the interlamellar organic phase. The absorption may take place from the vapour phase, from solution or from melt. It constitutes a vital element in the strategy of CPNC manufacture. The mechanism of the secondary absorption is based on the miscibility principle – only the substances that either can chemically react or are miscible with the interlamellar organic phase can be absorbed. Furthermore, the thermodynamic free energy of mixing must be able to compensate for the energy of increasing separation of the clay platelets. Thus, for example, absorption of a base requires that the interlamellar phase comprises active acidic functionalities, while absorption of nonionic compounds requires that their solubility parameter is comparable to that of the interlamellar phase [Utracki, 1989; 2002b; Utracki and Kamal, 2002a; Utracki, 2004]. The absorption leads to further expansion of the interlamellar space, proportional to the total number of -CH2- groups within the space. The process is diffusion controlled, hence a long residence time and intensive mixing may be required. Maximum swelling is facilitated by the organophilic nature of treated clay and by the high polarity of the organic penetrant. Thus, the initial modification (in the aqueous phase) should result in the exchange of at least 50% of Na+ by organic cations and the interlamellar space should be at least 0.8 nm. The best secondary intercalant should have high polarity and be organophilic, e.g., nitrobenzene or benzonitrile. When the secondary intercalant is hydrocarbon, unsaturation is a definite asset otherwise swelling may have to be aided by addition of a well miscible (with it) polar organic liquid [Grim, 1968]. Certain organic molecules may also penetrate the interlamellar space to form coordinated compounds by not directly interacting with the clay surface, but rather with the cations ionically bound to it. Two types of molecules are suitable, with either polar or aromatic functionalities. Since this secondary intercalation is usually conducted using air-dried organoclay (often suspended in an alcohol), the process must (1) replace the adsorbed water, and (2) form the coordination shell around the cation. The presence of residual moisture enlarges the interlamellar space, hence easier diffusion, but at the same time the energetics of step (2) must be sufficient to compensate for it. When the secondary intercalant was either ethanol or acetone the number of molecules in a complex depended on the interlamellar cation. Thus there were 2 for each K+, 3 for each Na+, 8-10 for each Ca2+ or Ba2+, etc. The interlamellar spacing also varied, respectively from 1.3-1.4 to 1.7 nm [Bruque et al., 1982]. The complexes of Cu-, Ni-, Zn-, Cd-, Hg- or Ag-MMT with PA or thiourea are highly stable [Peigneur et al., 1978]. Complexes with water molecules forming bridges between the metal cation and the organic molecules are known for, e.g., pyridine, nitrobenzene, benzoic acid, aniline, nitriles, ketones or PEG. The latter compound forms complexes with M+-MMT even in aqueous suspensions [Parfitt and Greenland, 1970]. The presence of H2O on the one hand causes expansion of the interlamellar space, but on the other it makes the system less stable at higher temperatures. Complex formation of M+-MMT with neutral organic molecules results in their protonation and stronger bonding. Protonation may originate directly from the cations present in the system, viz. H+, NH4+ or in general M+, or from the water molecules that are coordinated to a cation, viz. [M(H2O)x]n+. FTIR or
94
Clays NMR readily detects protonation. The molecules that undergo these reactions are amines, azoles or phenols [Raussell-Colom and Serratosa, 1987]. In non-aqueous systems hydrogen bonding of organic molecules to the clay surface takes place. The positive charge deficiency in the tetrahedral layer results in negative charge that is spread out on several surface oxygen atoms coordinated to Al3+ or Si4+ ions [Farmer and Russel, 1971]. The molecules capable of forming the hydrogen bonds are mainly alcohols, amines, ketones, etc. Identification of these reactions by spectroscopic means is difficult. In MMT at the platelet edges there is a limited number of silanol groups that can be reacted with a diversity of organic compounds, viz. alcohols, organochlorosilanes, isocyanates, epoxies, diazomethane: R − OH → ≡ Si − OR + H 2 O ⎧ ⎪ Cl − Si(R)3 → ≡ Si − OSi(R)3 ⎪ ≡ Si − OH + ⎨ ⎪ O = C = N − R → ≡ Si − O − CO − NH − R ⎪⎩ R − CH (O) = CH 2 → ≡ Si − O − CH (CH 2 − OH ) − R
Since the number of silanol groups in MMT is small, only a limited number of organic molecules can be grafted. The grafting does not affect the interlayer spacing, but may provide for better miscibility with the organic matrix. However, the silanol groups can also be formed by treating a suspension of MMT in alcohol with HCl [Lentz, 1964; Zapata et al., 1972]. The treatment removes surface cations from octahedral layers: ≡ Si − O − Mg − O − Si ≡ + 2H + → 2(≡ Si − OH ) + Mg2 + The silanol groups in turn can be reacted with a diversity of compounds, e.g., having groups susceptible for subsequent polymerisation: polycondensation or polyaddition.
95
Clay-Containing Polymeric Nanocomposites
96
Intercalation of Clay
2.3
Intercalation of Clay
2.3.1 Introduction The first step in the preparation of clays for use in PNC is purification of the mineral (see Section 2.2.3). Depending on the ultimate use of the resulting NaMMT, before intercalation the powder may be subjected to further preparatory steps, either to reduce the particle size and/or to reduce the particle size distribution. Since the time (t) that intercalant needs to diffuse a distance (l) is given by the proportionality: t ∝ l2, a decrease of clay particle diameter by 30% results in reduction of the intercalation time by half. Thus, as described in patents from Southern Clay [Knudson and Jones, 1986; 1992a,b], the interdiffusion of intercalating molecules is facilitated by reduction of the clay particle size – high stresses not only reduce the diameter but may also cleave the stack, reducing the required force to bend the middle layers during intercalation. Earlier patents similarly focused on the reduction of clay particle size either by mechanical grinding, comminuting or by hydrodynamic forces [Cohn, 1967; Clocker et al., 1976]. Since the exchange reaction of the inorganic cation (such as Na+) for the organic one proceeds from the clay platelet edge toward the centre as a regular front, reduction of platelet size reduces the time required for the intercalation (increasing the reaction temperature to ca. 70 oC also helps) [Newman, 1987]. Ion exchange strongly depends on pH – the optimum is about one unit below the pK-value of the organic salt. It is noteworthy that reduction of particle diameter is detrimental for the control of barrier properties, but it may not be essential for the other performance criteria. Intercalation of clay that has a wide distribution of platelet size often results in an uneven degree of intercalation, evidenced by broad XRD diffraction peaks [Ho et al., 2001]. The peak is much sharper when more uniform size NaMMT particles are used. The aims of intercalation are to: 1. Expand the interlayer spacing, 2. Reduce solid-solid interaction between the clay platelets and 3. Improve interactions between the clay and the matrix. The first goal has been traditionally achieved by making use of the anionic charge within the interlamellar galleries. Since the van der Waals interactions between solid surfaces decrease with the square of the separating distance (see Figure 18), insertion of organic or inorganic molecules into the interlamellar space greatly helps to achieve the second aim. To reach the third goal, to compatibilise the system, the principles developed for compatibilisation of polymer blends should be used [Utracki, 1994; 1998; 2000; 2002a; Ajji and Utracki, 1997; Utracki and Kamal, 2002b; Ajji, 2002; Brown, 2002]. 97
Clay-Containing Polymeric Nanocomposites For successful intercalation the selected clay should have a cation-exchange capacity: CEC = 0.5-2.0 meq/g, as for CEC < 0.5 the ion exchange is insufficient, while for CEC > 2.0 meq/g, the interlayer bonding is too strong for easy intercalation, thus smectites and vermiculites have the optimum CEC – theoretically 1.39, experimentally 0.8 to 1.2 meq/g. By contrast, kaolin has a cation-exchange capacity < 0.1 meq/g, while mica, illites, attapulgite and sepiolite are about 0.2 meq/g. As a consequence, MMT, saponite and hectorite are the preferred clays for CPNCs, but since MMT is more abundant and it has a fairly large aspect ratio, p ≅ 300, (natural hectorite has the smallest) it became the main nanofiller for PNC technology. Owing to the large aspect ratio, p ≤ 2,000, of synthetic micas, natural vermiculites and natural micas, several attempts have been made to use them in CPNC as barrier material against permeation of gases, vapours and liquids. Purification and intercalation can be quite abrasive, significantly reducing the clay aspect ratio [Ferreiro et al., 2001; Jeon et al., 2004]. According to the former authors, synthetic hectorite (Ilaponite LRD) had p ≅ 30, whereas MMT had p ≅ 100. Jeon et al. estimated the aspect ratio of MMT-based organoclay to be ca. 80. Both teams reported that in solution organoclay tends to form spherical particles of diameter varying from about 20 to 240 nm. Traditionally, the main use of the intercalated clays has been to produce thixotropic effects in aqueous or non-aqueous systems, e.g., to improve paper coating, lubricant thickening or to prevent sedimentation of dispersed solids. During the last 20 years or so additional uses for clay for CPNC technology have emerged. Intercalation and exfoliation for fine chemical delivery systems is the most recent. Thus, intercalated organoclays have had numerous uses: • •
•
•
•
• •
98
As water-based or organic liquid-based (e.g., ethanol, acetone) thickener for insecticide paste [Hamilton, 1936]. ‘Bentonite’ intercalated with either ammonium (e.g., with dodecyl ammonium, octadecyl dienyl ammonium, dimethyl dicetyl) or phosphonium (e.g., with triphenyl lauryl) cation has been used, e.g., for thickening lubricating oils and greases, oil-based muds, packer fluids, paint-varnish-lacquer removers [Jordan et al., 1950]. Application of onium intercalated layered silicates as organophilic thickeners of organic liquids [Hauser, 1950]. A wide diversity of onium salts have been described in the patent literature, viz. ammonium, phosphonium, sulfonium, etc. Use of MMT intercalated with quaternary ammonium (commercially available as ‘Bentone®’ from the National Lead Co.) as an inherently thixotropic additive in nail enamel applications. The organoclay provided suitable rheological behaviour and prevented settling of nacreous pigments [Kuritzkes, 1969]. In CPNC to facilitate diffusion of monomers and/or macromolecules into the interlamellar space, that eventually would lead to exfoliation [Fujiwara and Sakamoto, 1976; Okada et al., 1988]. For delivery of cosmetic, medical or agro-active compounds, e.g., sunscreen, antibacterial, antifungal, and many others [Beall et al., 1998]. For delivery of fungicides, pesticides, insecticides, acaricides and other agriculturally active compounds [Beall et al., 1999a].
Intercalation of Clay Intercalation of the clay particles is diffusion controlled. Water is a ‘natural’ intercalant for clays, but it can hardly be considered unique. Its efficiency is most likely related to the good balance between the dipole moment that drives the process and the molecular size that restricts its motion. In aqueous systems, water molecules can easily diffuse in and the electrostatic double layer can push the individual platelets apart and keep them away from each other – at low clay concentration total exfoliation can be achieved. On a molecular scale, intercalation can be visualised as inserting peas between each pair of play-cards in a stack. Evidently, to be able to do that the cards must be pre-spaced and there must be a driving force for the peas to enter the narrow space bending the cards from the edges. Thus, the way to intercalate clay is to use progressively larger molecular species. To prevent re-assembly of layers it is desirable that at least some intercalant is bound to the clay surface. For the application of clays in CPNC technology intercalation should increase the interlamellar spacing to about 3-4 nm and make the clay organophilic. The goal is usually reached in stages. 1. In dry clay the solid-solid interactions keep the interlamellar gallery at a water monolayer level, of about 0.26 nm. The sheets are strongly bound to each other (see Figure 18). The traditional method for reducing the solidsolid interactions, hence to reduce the resistance to intercalant diffusion into interlamellar space, is to disperse the clay in water or an aqueous solution of water-soluble organic solvents, e.g., alcohol or glycol. 2. The second step usually involves exchange of Na+ for an organic cation. Since the pK of onium salts increases with the degree of substitution, in most cases the quaternary onium is used. However, for the use of organoclay as a reactive component (e.g., in thermosets) a primary or secondary onium salt may be preferred. Furthermore, depending on the expected application the onium salt may have a functional group (e.g., a vinyl functionality). 3. In the third step the organoclay may be further treated with reactive compound to compatibilise the clay/polymer systems. Three classes of reactive compounds have been used: known glass-fibre sizing agents (e.g., silanes, titanates, zirconates), known reactive compatibilisers (e.g., oligomeric or polymeric compounds with glycidyl, maleic, isocyanate and other reactive groups), and organometallic compounds. This third step may be a part of the last step in the preparation of CPNC, the exfoliation of organoclay and dispersion of individual platelets in the polymeric matrix. Several routes have been used to intercalate clay particles. They can be classified as follows: 1. Use of solvents or low MW solutions, such as water, alcohols, glycols, crown ether or monomer solutions. 2. Use of organic cations, viz. ammonium, phosphonium or sulfonium. 3. Silylation of clay platelets. 4. Incorporation of inorganic compounds that form interlamellar pillars. 5. Use of organic liquids, viz. monomers, macromers, oligomers, polymers (PEG, PVAl, PDMS, PVP), copolymers, and their solutions. 6. Melt intercalation. 7. Others. 99
Clay-Containing Polymeric Nanocomposites
2.3.2 Intercalation by Solvents and Solutions As was shown in Figure 17, the basal spacing of MMT is 0.96 nm. In the presence of ambient humidity it expands to about 1.25 nm and upon addition of ethylene glycol to 1.7 nm. Intercalation using aqueous solution of glycols, glycerol or sorbitol was reported to be facilitated by the addition of Mg(OH)2 in the amount that changes pH to about 9.4 [Buffett, 1965; Burns, 1974]. Vaia et al. [1995a] directly intercalated Na-MMT or Li-MMT with PEG by heating the suspension to 80 °C for 2 to 6 h. The PEG displaced water molecules, expanding the interlayer spacing to d001 = 1.77 nm, but the clay could not have been exfoliated. Thixotropic thermoset polyester resin was prepared for hand lay-up mouldings by incorporating clay intercalated with polyols. Similarly, Kornmann et al. [1998] produced CPNC based on MMT and unsaturated polyester (UP). Na-MMT was intercalated at 50 °C in MeOH with methacrylate then dried. The treated MMT was dispersed in UP (containing 42 wt% styrene and co-octanoate). The mixture, after curing at room temperature and post curing at 70 °C, was found to be exfoliated. In 1998 Katahira et al. [1998a,b,c,d] published a series of articles on the use of mica for the production of PA-6-based CPNC. First, Na-mica flakes were cleaved and dispersed into hydrolysed and protonated ε-caprolactam (ε-CL). Phosphoric acid has been used as a catalyst for the protonation of ε-CL. The intercalation was found to occur in two steps: a rapid (even at 20 °C) exchange of Na+ in mica galleries with protonated ε-CL, then a significantly slower exchange of water solvated Na+ ions – the latter step could be accelerated by heating to T > 60 °C. The intercalated mica had the interlayer spacing of d001 = 1.47 nm. Serrano et al. [1998a] patented exfoliated CPNCs containing ≤ 99.95 wt% EVAl. No onium ion or silane coupling agent was needed for the intercalation. The intercalant was selected from between water-soluble oligomers (degree of polymerisation (DP) = 2-15) or polymers (DP > 15) having dipole moment greater than that of H2O (1.85 D). For example, P4VP, PVAl, copolymers of vinyl acetate and vinyl pyrrolidone, or their mixtures could be used. The intercalant was absorbed between the silicate platelets increasing the interlamellar galleries by Δd001 = 0.5 to 10 nm. The authors speculated that on the clay surface the intercalant molecules were ionically complexed with interlayer cations. Depending on the intercalant concentration the adsorbed molecules formed from 1 to 5 layers on the clay surface. In the following patent from the same laboratory [Serrano et al., 1998b] it was reported that the best results were obtained using an oligomer with DP = 2-15 or a polymer containing 50-80 wt% of an oligomer. The optimum performance was achieved by expanding the interlayer spacing to d001 = 3.0-4.5 nm. The preferred intercalants were P4VP, PVAl, or their mixtures. Intercalation was conducted in a twin-screw extruder (TSE) using intercalant solution at T = Tm + 50 °C. A masterbatch containing 20-80 wt% of clay was prepared. The recommended presence of oligomers and relatively high mixing T indicate recognition of the critical role diffusion has in the process. The method was extended to non-aqueous systems, but using Na-MMT that contained 10-15 wt% H2O (exfoliation was not obtained when the water content was below 8 wt%). For example, when 80 wt% PA was compounded under N2 with Na-MMT at T = Tm + 50 ≈ 230 °C, the clay platelets were exfoliated. Similarly, mixing PET with 10 wt% of Na-MMT or PC with 50 wt% of Na-MMT resulted in exfoliation. 100
Intercalation of Clay N-Vinylcarbazole (NVC) was cationically polymerised by Biswas and Ray [1998; 1999], in the presence of well-dried MMT, either directly at T > Tm = 64 °C, or in benzene at 50 °C. The resulting CPNC of polyvinylcarbazole (PVK) was intercalated, but during 100 min of polymerisation the interlayer spacing increased from d001 = 0.98 to 1.46 nm. However, intercalation was not the aim of this work. The authors reported two important discoveries of PVK grafting to the MMT surface and of the possibility of achieving clay intercalation under anhydrous conditions. Evidently, the dipole moment and aromatic character of NVC provided a large driving force. Srikhirin et al. [1998] intercalated two polydiacetylenes: 14-amino-10, 12-tetradiynoic acid (a diacetylenic mono-amino acid, MADA) and 10, 12-docosadiyndiamine (diacetylenic diamine, DADA) into layered silicates: MMT from Wyoming (CEC = 0.33 meq/g) or from Arizona (CEC = 0.57 meq/g) and vermiculite (VMT). The intercalation was conducted in an aqueous ethanol suspension for MMT and VMT over 1 and 7 days, respectively. The intercalation was confirmed by XRD and FTIR. The d001 spacing and polymerisability of the diacetylenes depended on the length of the diacetylene molecule, the layer charge density of the clay, and the solvent treatment. The DADA dichloride was intercalated into both MMT and VMT. The d001 = 1.41 and 3.1 nm was obtained for MMT and VMT, respectively. Intercalation with MADA resulted in d001 = 1.91 and 3.68 nm for the two MMT types, and d001 = 4.5 nm for VMT. Irradiation produced polymer only in the latter case. In the final product the interlayer spacing changed but a little: d001 = 4.2 nm was obtained. The authors concluded that when the intercalated diacetylene in MMT lays flat on the clay surface it is unable not be polymerised owing to the lack of proper packing density. In VMT the intercalated MADA is tilted with respect to the clay surface and the diacetylene has proper packing for propagation of radical polymerisation. As was already mentioned, organic compounds may diffuse into galleries and coordinate to alkaline cations [Ruiz-Hitzky and Casal, 1978]. The strength of the complex as well as expansion of the d001 spacing depends on the nature of the cation and the organic compound. Recently, on separate occasions, Gilman et al. [2001] and Yao et al. [2001] used crown ethers and cryptands to intercalate clays that were subsequently dispersed in polymers (PA-6 and PS, respectively). Similarly, MMT has been intercalated with C60-fullerene [Ishikawa et al., 1997]. The process was conducted in a phenol-formaldehyde-N-vinyl-2-pyrrolidone mixture at 60 °C, in the presence of HCl, subjecting the system to ultrasonics. The resulting thermosets with good electrical conductivity were used in Li-batteries. In 2001 Yao et al. used Na-MMT and K-MMT swollen overnight in water, then added a 1~3 % solution of crown ether in acetone. The complexation was conducted for 24 h at 20 to 45 °C, then the clay complex was filtered, washed and finally dried in a vacuum oven. The crown ethers were found to increase the d001 spacing by Δd001 = 0.4 to 0.7 nm, hence they complexed with either K+ or Na+. PStype CPNCs were prepared by dispersing ca. 1 wt% of crown ether-modified clay in styrene (St). The radical polymerisation of St in the presence of these clays was conducted overnight at 80-120 °C. For Na-MMT-based systems XRD showed no significant change in the d-spacing, but CPNCs were formed from K-MMT intercalated with cis-di-cyclohexano-18-crown-6 and the interlayer spacing increased to d001 = 7.7 nm. It is noteworthy that the complexation constant of crown ethers with K+ is higher than that with Na+. The larger is the constant, the stronger is the complex and the more organophilic the clay. 101
Clay-Containing Polymeric Nanocomposites
2.3.3 Intercalation by Organic Cations This method of intercalation by organic cations has been dominant for 70 years or so. As mentioned, the fastest exchange rates have been reported for Na-MMT. The reactions in water are faster than those in aqueous solutions of organic liquids, especially at higher temperatures, viz. T = 60 to 80 °C. Increased pressure (P) also speeds up the reaction. The diffusion-controlled, reversible reaction starts at the rim and the distance it travels is proportional to the square root of time (see Figure 22). As shown in Table 17, there is a great diversity of intercalants for layered clays. Functionally, these compounds are similar to compatibilisers or emulsifiers – they must diffuse into interlayer space and ascertain good thermodynamic interactions between the modified clay and polymeric matrix. Furthermore, similarly like compatibilisers, intercalants must provide good stress transfer between the two principal CPNC components – clay and polymer. Thus, the three criteria for selecting intercalant are: (1) the kinetics for diffusion into the interlamellar galleries, (2) high bonding strength with clay platelets, and (3) strong interaction with the polymer matrix. To facilitate selection, Tanaka and Goettler [2002] used molecular dynamics (MD) computation. The authors built a model comprising a MMT platelet, an intercalant (12 were considered), and a matrix (PA-66). Next, using MD methods the binding energies were computed for 600 K within each of the three model components, as well as the binary interaction energies between them. Finally, the fracture toughness parameter, Gc, was equated with the computed binding energies. The simulation showed that the positive charge initially placed on R4N+ spreads over the intercalant molecule; the charge is not localised on the nitrogen atom, but primarily on the substituent groups bonded to it. The charge equilibration method showed that the binding energy between PA-66 and the clay platelet decreases almost linearly with the intercalant volume. Thus, pristine clay/PA-66 showed the highest binding strength (hence toughness). The second best performance was found for MMT intercalated with ω-amino dodecyl acid (ADA), and the worst for MMT with dimethyl di-octadecyl ammonium (2M2OD). Similar MD simulations were carried out for CPNC comprising MMT, intercalant, and PA-6 [Fermeglia et al., 2003]. The results on the one hand paralleled the previously data for PA-66, and on the other indicated that the intercalants originally selected in Toyota R&D laboratories were indeed the best. Clay reactions with ammonium ions were studied in the early 1930s [Smith, 1934; Gieseking, 1939]. Later on attention shifted to the whole class of onium salts defined as organic compounds of the type RXHv. The element X (in its highest valency) may be pentavalent, as in ammonium, phosphonium, arsonium and stibonium; tetravalent, as in oxonium, sulfonium, selenonium and stannonium; or trivalent, as in iodoium [Hauser, 1950]. Clays readily react with less mobile organic cations, ionically bonding them to the surface. The organic part is selected considering three aspects of intercalation: 1. Ability to non-ionically interact with the clay surface, 2. Ability to expand the interlamellar space, and 3. Miscibility with the polymeric matrix. In 2003, Lagaly and Ziesmer published an excellent review on clay chemistry [Lagaly and Ziesmer, 2003]. 102
Intercalation of Clay Traditionally, for the preparation of organophilic thixotropic additives used in lubricants, quaternary ammonium chlorides with aliphatic hydrocarbons were introduced. Later, for paints and varnishes, mixed aliphatic/aromatic ammonium salts were used. For latex systems, it has been found advantageous to use hydrocarbons with -OH groups, viz. hydroxyethyl, carboxyl, etc. Owing to low cost and suitable hydrocarbon lengths, tallow oil derivatives have frequently been used. Commercial hydrogenated tallow oil typically contains (in wt%): 2.0 C14, 0.5 C15, 29.0 C16, 1.5 C17, 66.0 C18 and 1.0 C20 alkyl groups. It is obtained from natural oils, e.g., corn, soybean, cottonseed, castor, or animal oils and fats. Other oils, e.g., coco or coconut, are also being used. The commercial organoclays are still intercalated with ammonium salts having these three types of organic chains. Unfortunately, as discussed in Part 3.2, quaternary ammonium ions decompose at the processing temperatures of most thermoplastics, viz. at T ≥ 200 °C. This decomposition leads to reduction of the interlayer spacing (reversed intercalation) and it may discolour the product, reduce mechanical properties, decrease impact strength, etc. Nevertheless, these quaternary cations are the principal intercalants used in the industry [Hamilton, 1936; Maloney, 1939]. It has been found that the d001 spacing of layered clays intercalated with alkylammonium salts depends on the length of the alkyl radical as well as on the CEC. As shown in Figure 23 the spacing changes in steps [Hackett et al., 1998]. Molecular modelling and the FTIR measurements led to the model shown in Figure 24 [Vaia et al., 1994]. One of the early applications of organophilic clays was to ‘thicken’ low viscosity liquids. Such gellants had to have good dispersability in aqueous or organic systems. Owing to the hydrophilic character of clays, development of gellants for organic solvents and solutions was more challenging. To minimise
Figure 23 The interlayer d001 spacing of clays with CEC = 80, 100 and 150 as a function of the alkyl chain length in a primary ammonium salt: RNH3+Cl-. After Hackett et al. [1998].
103
Clay-Containing Polymeric Nanocomposites the amount of clay that was necessary for creating the desired effects two conditions had to be met: 1. The clay had to have the largest possible aspect ratio; and 2. The organoclay had to be miscible in the matrix liquid. In other words, the clay had to be exfoliated. Knowing the aspect ratio the required concentration could be calculated from Equation 1 as φ > φm. The conditions for miscibility could be calculated considering the molecular structure of the matrix liquid and the intercalating radical using, e.g., Hansen’s solubility parameter approach [van Krevelen, 1993], or the equation of state approach [Utracki, 2004].
Figure 24 FTIR-based alkyl ammonium chain ordering model as a function of alkyl length: (a) isolated, short chains in a monolayer; (b) intermediate chains disordered, forming a quasi-bilayer; (c) long chains with increased order and multi-layer spacing. Reproduced from Vaia et al. [1994], copyright 1994, with permission from the American Chemical Society.
104
Intercalation of Clay For example, Finlayson and Jordan [1978] prepared organophilic smectite (CEC = 0.75-1.20 meq/g) by reacting it with methyl benzyl dihydrogenated tallow ammonium chloride (MB2HTA). The clays were hectorite and Wyoming bentonite (CEC = 1.0-1.2 meq/g), purified and converted to sodium form. Other quaternary ammonium intercalants, viz. dimethyl dihydrogenated tallow (2M2HTA), methyl trihydrogenated tallow (M3HTA), benzyl trihydrogenated tallow (B3HTA) and dimethyl benzyl hydrogenated tallow (2MBHTA), were described as suitable as well – selection of one or the other depends on the character of the liquid to be thickened. The intercalation was carried out at T = 66-77 °C by mixing an aqueous suspension of 3-7 wt% Na-MMT and the quaternary ammonium compound for sufficient time to achieve adequate intercalation. The product was filtered and washed at 60 °C, then dried and ground. When the organophilic clays are to be used in emulsions, the drying and grinding steps may be eliminated. These organophilic clay complexes were used as additives to lubricating greases, oil-based muds, oil-based packer fluids, paint-varnish-lacquer removers, paints, foundry moulding sand binders, etc. A simple test was devised to determine the thickening characteristics of the new organoclays. Thus, low viscosity oil was mixed with 4.5 wt% of organophilic clay for 0.5 min at 1800 rpm, then 0.12 wt% of water was added and mixing continued. After 6-9 min the viscosity η > 200 Poises was obtained. MMT reacted with dimethyl-dioctadecyl ammonium ion (2M2ODA) has been industrially produced as a thickener for coatings in solvents with low polarity (e.g., toluene and xylene). For polar solvents (e.g., DMF, methanol and ethanol), clay with a dimethyl-benzyl-octadecyl ammonium ion (2MBODA) has been used, but the gelling efficiency of this system was low. As a solution to the latter problem, Iwasaki et al. [1994] developed organophilic clay with hydroxy-polyoxy-ethylenealkyl ammonium ions. The authors used clay with CEC = 0.85-1.30 meq/g and ammonium salt of mono-hydroxy-polyoxy-ethylene-tri-alkyl-, mono-hydroxypolyoxy-ethylene-di-alkyl-, di-hydroxy-polyoxy-ethylene-di-alkyl-, di-hydroxypolyoxy-ethylene-alkyl-, tri-hydroxy-polyoxy-ethylene-alkyl- or tri-hydroxypolyoxy-ethylene-. The number of statistical segments (n) in polyethylene glycol, -(CH2CH2O)n- was n = 2-20. A three step method of preparation was used: 1. 1 to 15 wt% clay was dispersed in water, 2. The quaternary ammonium salt solution was added 0.5 to 1.5-fold on the clay CEC, 3. The organophilic clay was washed with water, dried and pulverised. According to XRD, the dehydrated smectite had the basal spacing d001 = 0.96 nm. Under ambient temperature and humidity the spacing increased to 1.2-1.6 nm. The organoclay intercalated with quaternary hydroxy-poly(oxy-ethylene)-alkyl ammonium ions had d001 ≥ 1.8 nm – the actual value depended on the degree of polymerisation of the PEG segment, -(CH2CH2O)n-. Since these groups have a high affinity to molecules of a polar solvent, the solvent expanded the interlayers and exfoliated the clay. The organoclay was used as gellant in the production of coatings, plastic products, films and adhesives containing highly polar organic solvents such as DMF or ethanol. Good film-forming ability was achieved. In a cited example 20 g of hectorite (d001 = 1.25 nm, CEC = 1.1 meq/g) was dispersed in 1 L of tap water, and 21.4 g of a quaternary ammonium salt. The mixture was stirred at room temperature for 2 hours. The product was separated, washed, dried and pulverised. XRD confirmed formation of organoclay with d001 = 2.1 nm. 105
Clay-Containing Polymeric Nanocomposites The product formed a transparent dispersion in N,N-dimethylformamide showing good affinity to highly polar organic solvents. In another patent from NL Industries [Finlayson, 1980], several quaternary ammonium salts were used, viz. methyl benzyl or dibenzyl dialkyl ammonium chloride, methyl benzyl dihydrogenated tallow ammonium chloride or methyl benzyl dicoconut fatty acid ammonium chloride. The intercalation was conducted as described in the previous patent. On an industrial scale the organoclay could be prepared in colloid mills or other high speed dispersers. The non-aqueous, self-activating organophilic clays were useful in paints, varnishes, enamels, waxes, epoxies, mastics, adhesives, cosmetics and the like. The use of benzyl or dibenzyl radicals in the intercalant indicates that the paints that needed thickening were dispersed in at least partially aromatic solvents. In a later patent [Finlayson and Mardis, 1983], the intercalating cation could be any quaternary onium, viz. ammonium, phosphonium, sulfonium or their mixtures. The intercalant should contain at least one alkyl group having 12-22 carbon atoms, derived from natural oils, e.g., tallow. Additional onium radicals may include methyl, ethyl, decyl, lauryl, stearyl, benzyl and substituted benzyl, phenyl (as in N-alkyl and N,N-dialkyl anilines), alkyl phenyl, naphthalene, anthracene and phenanthrene. The clay-onium complex was reacted with organic anion. The latter should have molecular weight < 1 kg/mol and be derived from an organic having a pKA < 11.0. Suitable acids include: •
Carboxylic acids (benzoic, phthalic acid, benzene tri-, tetra- and hexa-carboxylic acid); alkyl carboxylic acids: H-(CH2)n-COOH, wherein n is a number from 1-20; alkyl dicarboxylic acids: HOOC-(CH2)n-COOH, wherein n = 1-8; hydroxy alkyl carboxylic acids (citric, tartaric, or 12-hydroxystearic acid); unsaturated alkyl carboxylic acids (maleic, fumaric or cinnamic acid); fused ring aromatic carboxylic acids (naphthalenic or anthracene carboxylic acid); cycloaliphatic acids (cyclohexane-, cyclopentane-, or furan carboxylic acids). • Organic sulfuric acids: (a) sulfonic (benzene-, phenol-, dodecyl-, benzene dior tri-sulfonic, p-toluene sulfonic acid); alkyl sulfonic or di-sulfonic acids (e.g., of butane); and (b) half-esters of sulfuric acid with lauryl or octadecyl alcohol. • Organophosphorus acids including: phosphonic; phosphinic (e.g., dicyclohexyl phosphinic), thiophosphinic acids, phosphites, phosphates, e.g., dioctadecyl phosphate. • Phenols, viz. phenol, hydroquinone, t-butyl catechol, p-methoxy phenol, and naphthols. • Thio-acids (thio-salicylic, -benzoic, -acetic, -lauric, -stearic, etc.). • Amino acids, e.g., 6-aminohexanoic or 12-aminododecanoic. • Polymeric acids, e.g., low molecular weight acrylic acid polymers and copolymers; styrene-maleic anhydride copolymers. • Miscellaneous acids and acid salts, which form a water insoluble precipitate with an organic cation. The intercalation was carried out in aqueous medium by mixing the ingredients for 1 to 60 min, at T = 60 to 77 °C, and then filtering, washing, drying and grinding. The amount of anion should be 0.1 to 0.5 meq/g of clay, that of organic cation 1.0 to 1.6 meq/g. 106
Intercalation of Clay The preferred process involves: (a) Preparing water slurry with 1-80 wt% smectite and heating to T = 20-100 °C (b) While agitating the suspension adding organic anion and organic cation in a sufficient amount to satisfy the clay CEC and the cationic activity of the anion (c) Continuing the reaction for a sufficient time to form a reaction product comprising an organic cation-organic anion complex, which is intercalated into clay, and then recovering the product. In 1986, Knudson and Jones from Southern Clay Products (SCP) [Knudson and Jones, 1986] reported further improvements of the intercalation method. Prior to the reaction with the ammonium intercalant the authors subjected clay to high-energy pugmilling in a machine. The device had a short barrel (L/D = 4-10), a motor driven screw, and perforated die plate. The reasons why the extrusion improved the intercalation were not discussed. The authors observed that intercalation was improved if prior to the reaction with onium salt the clay is subjected to high energy milling, dispersing or grinding. However, pugmilling with energy of 30, 51 and 108 HP-hr/ton reduced the smectite clay particle size from 475 nm to, respectively, 391, 277 and 276 nm. Since, according to the diffusion mechanism the process depends on the diffusion path length, this would considerably reduce the diffusion time. The high shear stress may not only reduce the stack diameter, but also its height. In a typical procedure, the crude clay, e.g., bentonite, was wetted with 25 to 40 wt% water, and then passed through the pugmill under conditions which impart at least 40-50 HP-hr per ton of dry clay. When the crude clay is not Na-clay, Na2CO3 may be added during the pugmilling. The clay is then dispersed in water at a concentration below 10 wt%, screened and centrifuged to remove non-clay contaminants, such as quartz. The fine fraction from the centrifuge (d < 4 μm at 4-5 wt% solids), was reacted with a quaternary amine chloride, e.g., 2M2HTA, 2MBHTA, M3HTA, etc. The organoclays had high gelling efficiency so only a small amount was needed to achieve the desired result. Knudson and Jones [1992a] disclosed further improvements of this technology. The organoclay manufacture followed the standard steps: clay purification, conversion to Na-smectite, reduction of the platelet size by high stress shearing, and reacting it with quaternary ammonium salt. The key improvement was the use of a Manton-Gaulin colloid mill (also known as a ‘Gaulin homogeniser’). In the mill the clay suspension that entered the valve area at high pressure and low velocity was subjected to high acceleration, turbulence and cavitation. As it passed through a narrow orifice the velocity increased to about sonic level. Directing this high velocity stream to impact on a ring enhanced the milling effectiveness. As a result, the particle size was reduced from 756 to 438 or 352 nm for an energy input of 210 or 700 HP-hours per ton, respectively. Another patent describes the use of Cowles disperser and Greer mill for subjecting the clay slurry to high shear flow [Knudson and Jones, 1992b]. In the Cowles disperser the clay particles are subjected to laminar flow, while in Greer mill the clay slurry is forced to flow at high speed through a 0.254 mm gap and to impact on a deflector plate. The gelling capacity of such treated clays was improved. In Jordan’s patent [Jordan, 1994] another approach to intercalation was adopted. The organoclay was to be used as a thickener or a gelling agent, especially 107
Clay-Containing Polymeric Nanocomposites for paints. The author noted that intercalation in aqueous slurry is advantageous as ‘solvation relaxes the clay’s structure in order to permit penetration of the organic cations’. However, it is preferable to avoid too dilute a reaction because the slurry preparation takes time and space and dewatering of product is energyintensive. A new, waterless method was developed, based on reacting preblended clay with quaternary cations by mixing the dry materials in a high pressure reaction vessel. The process can be batch or continuous. MMT, hectorite, saponite, attapulgite, sepiolite and their combinations can be used. The onium cation has a formula R4M+ X-, wherein M is nitrogen or phosphorous, R is alkyl, aryl, or alkyl-aryl and X is halogen or methyl sulfate. To facilitate the process 3-15 wt% of a dispersant (neopentyl glycol, pentaerythritol, hydrogenated castor oil, sulfonated castor oil, a plasticiser, toluene sulfonamide, trialkoxyphosphate, etc.) might be added. It is preferred that the dispersant has beneficial effects on the final product. For example, a mixture of clay, quaternary salt and dispersant passes from a ribbon blender to a high pressure mixing auger-extruder with a perforated die. At the downstream end a mixing blade assembly stirs the highly compressed mixture at pressures of 20-55 MPa. The temperature should be near, but not above 80 °C. The mixture may be either cycled through the same or another mixer/reactor, or it may be transferred for further processing or milled to suitable particle size. Since the process obviates the need for slurring the clay, sand and other particulates may be present in the product. The material (10 wt% of organoclay with the remaining constituents) was successfully used in a cable filling application as an oil gelling agent. The dielectric constant was almost invariant at T = 21-148 °C. Two years later, Jordan [1996] published a refinement of the above method. Thus, a mixture of two quaternary cations was found to induce superior performance in a variety of applications. The mixture comprised at least 5% of each: X2R2N+ and XYR2N+ where X is methyl, Y is benzyl and R is an alkyl, derived from saturated tallow oil. The clay may include 3-15 wt% of a dispersant, e.g., hydrogenated castor oil. The organoclay comprised 50-90 wt% of a smectite (MMT, hectorite, saponite, attapulgite, sepiolite, or their mixture) having exchangeable, inorganic cations at least in part substituted by a mixture of 10-50 wt% of two organic X2R2N+ and XYR2N+ cations. The composition may include 3-15% (based on clay) of a dispersant (e.g., neopentyl glycol, pentaerythritol, hydrogenated castor oil, sulfonated castor oil, toluene sulfonamide, trialkoxyphosphate). As in the previous patent, the pre-blended mixture was processed using a blender and an augerextruder with perforated die. The product was milled to usable particle size. The interlaminar spacing was not determined. The modified clay was evaluated for paint and grease thickening applications. In the first case, the thixotropic index and the number of oversized particles were measured – the new organoclay was found to equal or exceed the best commercially available materials made using a high-cost slurry process. The greases were tested for consistency by utilising a Helipath instrument. Again, the results showed that the new organoclay offered significantly better performance than the standard materials. Eilliott and Beall [1989] patented an organoclay composition for use as a rheological additive in a variety of non-aqueous liquids, such as paints, varnishes, enamels, waxes, adhesives, inks, laminating resins, gel coats, etc. To obtain 108
Intercalation of Clay maximum dispersability and thickening (or gelling) efficiency, it was necessary to add a low molecular weight polar organic to the clay. Such compounds have been called polar activators, dispersants, dispersion aids, solvating agents, etc. An organosilane of the type RnSiX4-n was a new alternative. In this formula, n = 1-3, R is an organic radical having C-Si link and X is a hydrolysable alkoxy, acryloxy, amino or halogen group. The new organoclay of Na-MMT (CEC ≥ 0.75 meq/g) with a quaternary ammonium ion and 0.5 to 5 wt% organosilane has been used for preparing thixotropic thermoset compositions from unsaturated polyesters and unsaturated aromatic monomers, e.g., styrene. The composition was crosslinked by peroxide, and used in the preparation of glass fibre (GF) laminates. The smectite-type clays of CEC ≥ 0.75 meq/g, e.g., Na-MMT or hectorite were preferred. The onium was R1R2R3R4N+M-, where M is an anion, R1 is an alkyl containing 12-22 C-atoms and R2-R4 are alkyls containing 1-22 C-atoms, arylalkyl groups containing 7-22 C-atoms, aryl groups containing 6-22 C-atoms and their mixtures. The long chain alkyls can be derived, e.g., from hydrogenated tallow oil. Aryl groups include phenyl and substituted phenyl. Arylalkyl groups include benzyl and substituted benzyl groups. Examples of useful quaternary ammonium compounds are 2M2HTA (preferred), M3HTA, 2MBHTA, MB2HTA, etc. The organosilanes used in this patent are the same coupling agents as used in plastics composites [Plueddemann, 1982], viz. methyl-, ethyl-, or propyltrimethoxy silane, di-methyl- or diethyl- di-methoxy silane, tri-methyl- or triethyl- methoxy silane, phenyl triethoxy silane (preferred), etc. Furthermore, the thixotropic properties of the organophilic clays can be improved by the addition of ≤ 25 wt% of hydrogenated castor oil. Intercalation took place in a slurry process. A purified aqueous suspension of 1.5-5 wt% clay was heated to 60-75 °C, then quaternary ammonium salt and the organosilane (as an emulsion in water or alcohol) were added. To complete the reaction, agitation and heating continued for 15 to 120 min, then the organophilic clay was filtered, washed and dried at 90 °C for 24 hours. The dried product, ground and screened through a 170 mesh screen, was used as a rheological additive in a wide variety of non-aqueous liquids, viz. paints, varnishes, enamels, waxes, adhesives, inks, laminating resins, gel coats and the like. The new clays were particularly useful for preparing thixotropic crosslinkable compositions from unsaturated polyesters and styrene. The developed compositions were used in the ‘pre-gel’ and ‘directly add’ processes. MMT intercalated with 12-amino-dodecanoic acid (ADA) was used to prepare CPNC based on PA-6, which in turn could be blended with polypropylene/ ethylene-propylene rubber (PP/EPR) [Fukui et al., 1992]. For example, 100 g of MMT (d ≈ 100 nm) were dispersed in 10 l of water, then 51.2 g of ADA and 24 ml of HCl were added. The mixture was stirred for 5 min, filtered, washed and vacuum dried. Some degree of dissatisfaction with onium ion as the principal intercalant is evident from patents deposited by Toyota Central Research and Development Laboratories (Toyota – for short). For example, [Usuki et al., 1996] described the preparation of CPNC in three steps: (i) Intercalation of clay with an onium ion (having ≥ 6 carbons) rendering it compatible with a ‘guest molecule’ with a polar group in its chain and length equal to or larger than that of the onium alkyl; (ii) Contacting the organo-clay with the ‘guest molecule’ at T ≤ 250 °C; and 109
Clay-Containing Polymeric Nanocomposites (iii) Transferring the modified clay to a reactor (the preferred reactive exfoliation method) or compounding it with a resin. The patent listed thermoplastic and thermosetting resins as well as elastomers (e.g., PE, PP, PS, PIB, acrylics, TPU, SBS, liquid BR, PB or IR, etc.) as potential matrix polymers. However, the main interest was to produce CPNC in low polarity polymers, such as PO or rubbers. The guest molecule should have a polar group at the chain-end. The following groups might be used: hydroxyl (-OH), halogen (-F, -Cl, -Br, or -I), carboxyl (-COOH), anhydrous-carboxylic acid (maleic), thiol (-SH), epoxy radical, or primary, secondary or tertiary amine (-NH2, -NH, -N). In the described examples the authors used as onium salts 2M2ODA, 2M2TDA, or TDA; as a ‘guest molecule’ hydrogenated, low molecular weight polybutadiene with an -OH group at the chain end (HTBR), stearic acid or alcohol alone or with HTBR, maleated PP (PP-MA with high acid value hence low MW), etc. To ascertain good miscibility of organoclay with the matrix polymer a second ‘guest molecule’, e.g., low molecular weight polyisoprene (IR) or oligopropylene, could be used. The process resulted in a high degree of clay dispersion [Kato et al., 1997]. The patent on CPNC with PA as matrix [Okada et al., 1988] described the following method for uniformly dispersing clay platelets with interlayer distance d001 ≥ 2.0 nm. The manufacturing process comprised three steps: 1. Intercalation of a clay (CEC = 0.5 to 2.0 meq/g, e.g., smectite, vermiculite or halloysite); 2. Mixing the organoclay with a PA-monomer and 3. Polymerisation. Only the first step is of interest in the present discussion. The intercalation was accomplished by dispersing clay in an aqueous, acidified intercalant solution, followed by washing the organoclay with water to remove excess ions, or by mixing an aqueous suspension of clay with a cation-exchange resin previously treated with an intercalant. The easiest to treat were the sodiumsubstituted clays, viz. Na-MMT (CEC = 0.8-1.0) or Na-vermiculite (CEC = 1.8). Inorganic (viz. Cu2+, Al3+), or organic cations were used. In the latter case α,ωalkyl acids H3N+ CnH2n COOH were used, with n = 12 for the preferred dodecyl (ADA), n = 14 for tetradecyl (ATDA), n = 16 for hexadecyl (AHDA), or n = 18 for octadecyl (AODA). For example, to a suspension of 100 g Na-MMT in 10 L of water, 51.4 g of ADA and 24 ml HCl were added, and the mixture was stirred for 5 min. After filtration, the organoclay was thoroughly washed and vacuumdried. After execution of steps 2 and 3 (dispersion in a monomer, then polymerisation) the resulting CPNC contained fully exfoliated clay platelets and showed significant improvement of the mechanical and thermal properties. As displayed in Figure 25 the intercalation effect on the interlayer spacing depended very much on the paraffin chain length and the presence of monomer; only after polymerisation was full exfoliation obtained. A patent from Toyota [Kawasumi et al., 1989] reported that clays may be intercalated in an aqueous medium with ammonium salt of a primary, secondary or tertiary linear organic amine. However, owing to the envisaged use of the organoclay in polymerisation, the preferred intercalant has remained ADA. Several patents on the intercalation of clays with onium ions originate from SCP. Thus, Dennis [1998] described the preparation of organoclay compositions 110
Intercalation of Clay
Figure 25 Interlayer spacing after intercalation with ω-amino acid (AA) having n CH2 segments in the molecule. The lower curve is for intercalation with AA only, the upper curve is for AA with ε-caprolactam at 100 °C (2 g of lactam per 0.5 g of Na-MMT + AA). Data Okada and Usuki [1995].
useful in grease and ink formulations. The organoclay was the reaction product of a smectite-type clay having CEC ≥ 0.50 meq/g and a branched chain alkyl quaternary ammonium compound. Useful clays include bentonite, hectorite, as well as, synthetic smectite-type clays, such as MMT, beidellite, hectorite, saponite and stevensite. The branched quaternary ammonium intercalant has the formula: R1R2R3C(R4 R5R6)N+M–, where M– is an anion (preferably methyl sulfate, chloride or bromide), R1 = R2 = -CH3, R3 is a linear or branched saturated or unsaturated alkyl group having 12-22 carbon atoms, R4 is hydrogen or a saturated lower alkyl group of 1-6 carbon atoms; R5 is hydrogen or a linear or branched saturated alkyl group of 1-22 carbon atoms; R6 is a linear or branched saturated alkyl group of 5-22 carbon atoms; and M is the salt anion. Especially preferred are the quaternary ammonium salts wherein R3 is hydrogenated tallow and R4, R5 and R6 are together a 2-ethylhexyl. Organo-MMT was prepared using the following steps: 1. Aqueous bentonite slurry (2 wt% solids) was passed three times through a homogenising Manton-Gaulin mill (MG-mill). 2. The slurry was then heated to 66 °C and reacted for 30 min with dimethylhydrogenated tallow-2-ethyl-hexyl ammonium methyl sulfate (2MHTL8, a commercially available quaternary ammonium salt as Arquad HTL8), a blend of Arquad HTL8 with dimethyl dihydrogenated tallow ammonium chloride (2M2HTA, a commercially available quaternary ammonium salt as Arquad 2HT), or other quaternary amines. 3. The reacted slurry was sheared again through an MG-mill at 29 MPa. 4. The organoclays were vacuum filtered, fluid bed dried at 80 °C, and milled through a 0.2 mm screen. Several commercial organoclays from SCP, e.g., Cloisite®-25A, -6A, -15A and -20A comprise these compounds. 111
Clay-Containing Polymeric Nanocomposites The following patent [Gonzales et al., 1998a], specifically describes the preparation of MMT organoclays to be used for the production of CPNC. The organoclay is the reaction product of smectite-type clay having a CEC ≥ 0.50 meq/g and a mixture of a first quaternary ammonium salt (e.g., 2M2HTA) with either a second quaternary ammonium salt containing a -C=C- double bond, or a chain transfer agent. The second ammonium salt has the formula: R1R2R3R4N+M–, where M– is an anion (preferably methyl sulfate, chloride or bromide), and Ri are independently selected from the group consisting of (a) linear or branched, saturated or unsaturated alkyl groups having 1 to 22 carbon atoms, (b) aralkyl groups which are benzyl and substituted benzyl moieties, (c) aryl groups, (d) β,γ-unsaturated groups having ≤ 6 carbon atoms or hydroxy-alkyl groups having 2-6 carbon atoms, and (e) hydrogen, with the proviso that at least one of the substituents is a linear or branched unsaturated alkyl group. The chain transfer agent may be a thiol, DL-cysteine, α-methylketone, or a halogen compound. The intercalation process comprises the following steps: 1. Dispersing a clay in an aqueous medium, 2. Heating the dispersion to T ≥ 30 °C, 3. Adding a first quaternary ammonium salt, then the second quaternary ammonium compound or the chain transfer agent, and 4. Agitating the mixture to complete the reaction. For example, MMT was intercalated with 2M2HTA and either DL-cysteine, or N,N-dimethyl-amino-methacrylate. For the first system d001 spacing was determined as 2.3 nm, whereas for the second two peaks were found at d001 = 1.4 and 2.6 nm – evidently in the latter case not all MMT tactoids were intercalated. Another patent from SCP describes the preparation of organoclays for use as a rheological additive in an unsaturated polyester resin/styrene system [Farrow et al., 1998]. Thus, a mixture of two clays was treated with 0.35 to 0.65 meq of the alkyl quaternary ammonium salt per 1 g of the clay mixture. The mixture contained 70 to 90 wt% of clay (a) selected from between sepiolite, palygorskite and their mixtures; and 30 to 10 wt% of clay (b) selected from a smectite group consisting of hectorite, MMT, bentonite, beidellite, saponite, stevensite and their mixtures. The alkyl quaternary ammonium cation was either 2M2HTA, MB2HTA, 2MBHTA or 2MHTL8, with the counterion being chloride, bromide, methyl sulfate, nitrate, hydroxide, acetate, phosphate, etc. Production of organoclay followed five steps: 1. The clays (sepiolite, palygorskite, hectorite) were crushed, ground, slurried in water and screened to remove grit and other impurities. 2. Dilute (1 to 6% solids) aqueous slurry of the clays was then subjected to high shearing in a homogenising Manton-Gaulin mill, operated with a pressure differential across the gap from 14 to ≥ 55 MPa. 3. The clay slurries may be mixed together, e.g., 80 wt% sepiolite with 20 wt% hectorite or 70 wt% palygorskite with 30 wt% MMT. 4. The mixture was treated with a quaternary ammonium salt. 5. The resulting organoclay was dewatered, dried and ground. The products were found to display highly desirable properties when used as a thixotrope in the gelling of an unsaturated polyester resin. Thus, the organoclay (0.1 to 4 wt%) may be directly dispersed in an unsaturated polyester 112
Intercalation of Clay resin/monomer solution. The organoclays also showed excellent performance in high temperature drilling fluids, offering high gel strength at T ≤ 230 °C. When used with PA-66, they improved the mechanical properties, e.g., tensile strength by 20%, tensile modulus by 31% and flex modulus by 28%. In a patent from Ciba [Zilg et al., 1999] phyllosilicates were intercalated with a salt of a cyclic tertiary amidine or amidine mixture. In Figure 26 (a), the basic structure of the now patented intercalant family is presented. In the suitable compounds: R1 is the alkyl radical (C2 to C8), R2 is hydrogen or an aliphatic radical containing an unsaturated bond, which may be substituted by a carboxyl or carbonyl group, R3 is hydrogen or alkyl; each of A and B is –CH2–, or A and B together are the radical –(CH=CH)–, and X– is an anion. In Figure 26 (b) the radical R´´ = -CnH2n+1, where n = 0, 1, 2, 3 or 4. Owing to the presence of a reactive moiety in the second radical (–OH or –CH=CH–), the intercalant may be modified to bond with a matrix polymer. By contrast with ammoniumintercalated clays those with amidine have greater thermal stability during processing, higher interfacial adhesion and better dispersability. Another advantage of the amidine intercalants is that they do not change the stoichiometry of the thermoset matrix reaction. According to the patent, the phyllosilicate may be synthetic or mineral, preferably MMT and/or hectorite, with d001 = 0.7 to 1.2 nm and CEC = 0.5-2.0 meq/g. The cyclic tertiary amidine should have two substituted groups, one aliphatic with unsaturated bonds and/or functional groups, and the other either a hydrogen or a linear or branched aliphatic radical with one or more unsaturated and/or functional groups. The following amidine compounds have been synthesised: • • • •
ricinyl-4,5-dihydro-1-H-imidazole hydrochloride (RDI, from castor oil), hydroxyethyloleyl-4,5-dihydro-1-H-imidazolinium hydrochloride (HDI), aminoethyloleyl-4,5-dihydro-1-H-imidazolinium hydrochloride (ADI), and 1-methyl-2-nortallowalkyl-3-tallow-fatty acid amidoethylimidazolinium methosulfate (MNTFA).
(a)
(b)
Figure 26 (a) The basic structure of the cyclic amidine intercalant family; (b) Derivatives of (a) particularly suitable as intercalants (see text). Reproduced from Zilg et al. [1999], with permission.
113
Clay-Containing Polymeric Nanocomposites Synthetic MMT was intercalated and XRD showed that the interlayer spacing increased from d001 = 0.94 to 2.67 (for RDI), 3.27 (for HDI and ADI), and 4.01 nm (for MNTFA). The intercalation started with dissolution of the intercalant in acidified, hot water, followed by addition (while stirring) of a synthetic clay suspension. As a result of the cation exchange, flocculated clay precipitated, it was filtered, washed (until Cl– could not be detected with AgNO3 solution), and then dried under vacuum at 80 °C for 72 h. According to the Ciba patent, the amidine-intercalated clays may be used for the production of CPNC with virtually any polymeric matrix (thermoplastic, thermosetting or elastomeric), viz. PO, vinyls, styrenics, acrylics, PA, thermoplastic polyesters (PEST), PC, polyphenylene sulfone (PSF), polyarylethers, diverse polycondensates or polyadducts, rubbers, etc. Preferred polymers are PEST, PU, thermosetting epoxy and polyurethanes. A recent invention for polymeric (primarily PEST) composition with high barrier properties [Barbee and Matayabas, 2000; Matayabas et al., 2000] again uses layered clay that has been cation exchanged with onium salt represented by the formula [R1R2R3R4M]+ X-, where M is N or P; X- is a halide, hydroxide, or acetate anion; R1 is a straight or branched alkyl group having at least 8 carbon atoms; R2, R3 and R4 are independently hydrogen or a straight or branched alkyl group having 1 to 22 carbon atoms. The polymer barrier properties were improved by the addition of up to 30 wt% of a mixture of onium-intercalated clay (e.g., MMT-MT2EtOH) with an agent, which expands the interlayer spacing to d001 ≥ 3 nm and is miscible with the polymer. The ratio of clay to the ‘expanding agent’ varied from 1:4 to 4:1. Suitable ‘expanding agents’ are mainly low molecular weight compounds (from monomers to Mn = 25 kg/mol), viz. PCL, PDMS, polyepoxides, PS, polyacrylates, PC, PU, PSF, PA, PEST, polyethers, polyketones and vitamin E. The ‘expanding agents’ used in examples were: PETG 6763, PEG (MW = 3.35 kg/mol), PCL (MW = 2 kg/mol), carbinol-terminated PDMS, etc. The process recommended by these authors for manufacturing PET nanocomposites comprised: 1. Preparing the organoclay 2. Pre-swelling the organoclay with an ‘expanding agent’ and 3. Incorporating the expanded organoclay in a polyester. For example, an aqueous suspension of Na-MMT (CEC = 0. 95 meq/g) was mixed at 50 to 80 °C in a blender to form a 2 wt% slurry. The methyl tallow bis2-hydroxyethyl quaternary ammonium (MT2EtOH) (Ethoquad T/12) was added in sufficient amount to exchange most of the cations present in the galleries. The precipitate was filtered out, washed and dried. XRD of the product showed that d001 = 2.0 nm. Next, the expanding agent (PET modified with 30 mol% of 1,4cyclohexane dimethanol available as PETG 6763) was dissolved in methylene chloride, the organoclay was added and the mixture was blended at high speed. Next, the suspension was poured over PET pellets, followed by evaporation of the methylene chloride, and drying the coated pellets in a vacuum oven at 110 °C. The coated pellets were finally compounded in a TSE at 275 °C and 200 rpm. For the manufacture of polycarbonates, polyesters and polyphenylene ethers (PPEs) from low viscosity macrocyclic oligomers, layered minerals were intercalated with novel cations [Takekoshi et al., 1996]. The preferred type cations were guanadinium and amidinium, viz. hexa-butyl-guanidinium. However, more traditional onium cations were also acceptable, especially of pyrrolidine, 114
Intercalation of Clay piperidine, piperazine and morpholine as well as heterocations derived from cyclododecanes. Intercalation was performed by dispersing Na-MMT (10% H2O; CEC = 1.19 meq/g) into an aqueous methanol solution. To the resulting homogeneous dispersion a solution of either dodecylammonium (DDA) or hexadecylpyridinium (HDP) chloride was added. A white precipitate was recovered by filtration and subsequently washed with water and dried. There is no data on interlayer spacing, but upon incorporation of 5% of MMT intercalated with DDA or HDP the modulus of CPNC with PPE as the matrix increased by 15% or 29% (in respect to neat PPE), respectively. Quaternary ammonium ions with heterocyclic rings have also been used for rubbers [Weber and Mukamal, 1984]. It has been frequently shown that the size of the intercalating onium ion determines the resulting interlayer spacing. The most exhaustive list of these effects can be found in Miyanaga et al. [1999] patent. Four different clays were used, but for illustrating the effects only data for MMT (Kunipia F; CEC = 1.19 meq/g) will be cited in Table 15. Intercalation was performed by dispersing 15 g of clay in 1 l of deionised water at 70 °C. To the suspension a water/ethanol solution of onium salt was added in the amount of 1.05 times the equivalent of the clay CEC with vigorous stirring for 30 min, and then allowed to stand. The organoclay was filtered, washed, and then dried under vacuum at 80 °C for 72 hours. The d001 spacings as determined by XRD are listed in Table 15. The results indicate that the best correlation between d001 and the size of the onium salt involves the three largest radicals. To prepare CPNC with high temperature processable fluoroplastics as the matrix, MMT was intercalated with phosphonium cations of the general formula: R1P+(R2)3 [Ellsworth, 1999]. The fluoroplastics of interest have either melting point or glass transition temperature Tm, Tg > 220 °C. Thus, for example, 10 g of sodium hectorite (10 g) was dispersed in alcohol/water 1:1 mixture (200 ml) at 90 °C. To the suspension 50 ml hexadecyl-tributyl-phosphonium bromide in isopropyl alcohol was added. Next, the reaction mixture was heated at 90 °C for 8 h with stirring. The organoclay was filtered, washed, dried at 120 °C for 24 h, milled and screened through a 325 mesh (40 micron) sieve. To produce rubber compositions with enhanced mechanical performance intercalated clay was incorporated [Weber and Mukamal, 1984]. As the patent claim specifies, the intercalant can be a quaternary onium ion of the traditional aliphatic-substituted type or selected from between: imidazolium, pyridinium, pyrrolidinium, pyrrolium, pyrrolinium, pyrazolium, triazolium, pyrimidinium, pyridazinium, pyrazinium, triazinium, indolium, indazolium, benzimidazolium, quinolinium, isoquinolinium, cinnolinium, phthalazinium, quinazolinium, quinoxalinium, naphthyridinium, quinolizinium, carbazolium, acridinium, phenazinium, phenanthridinium, phenanthrolinium, benzo[H]isoquinolinium, purinium, porphinium and pteridinium. In these compounds the heterocyclic rings could be substituted by alkyl(s) or not, and they could be partially hydrogenated. Preparation of CPNC in PEST matrix requires high temperature thus thermally stable organoclay. Furthermore, the organoclay usually induces a brownish color that must be compensated for by addition of suitable dyes. A patent from Eastman elegantly solves these problems by using cationic dyes (e.g., optical, fluorescent brighteners) [Barbee et al., 2000c]. The preparation of these CPNCs is by melt compounding of the pre-intercalated clay with molten polymer. The clay is pre-intercalated with a dye, having the cation group (a quaternary ammonium) separated from the chromophore 115
Clay-Containing Polymeric Nanocomposites
Table 15 Interlayer spacing of MMT (CEC = 1.19 meq/g) intercalated with, R1R2R3R4N+. Data [Miyanaga et al., 1999] R1
R2
R3
R4
d001 (nm)
—
—
—
—
1.23
C10H21 C10H21
C10H21
C10H21
2.80
CH3
C18H37
C18H37
C18H37
4.11
CH3
C4H9
C18H37
C18H37
2.85
H
CH2-Ph
C18H37
C18H37
2.65
CH3
CH3
C18H37
C18H37
2.81
CH3
CH3
CH3
C18H37
2.13
CH3
CH3
C10H21
C10H21
1.98
CH3
C8H17
C8H17
C8H17
1.83
C4H9
C4H9
C4H9
C4H9
1.55
CH3
CH3
CH3
[CH2CH(CH3)O]30H
4.58
CH3
CH3
C12H25
[CH2CH(CH3)O]30H
3.87
CH3
[CH2CH(CH3)O]10H
[CH2CH(CH3)O]10H
[CH2CH(CH3)O]10H
5.19
CH3
CH2CH(CH3)OC8H17 CH2CH(CH3)OC8H17 CH2CH(CH3)OC8H17
CH3
CH3
(CH2CH2O)2(CO)C17 H35
(CH2CH2O)2(CO)C17H35 2.98
CH3
[CH2CH(CH3)O]10H
[CH2CH(CH3)O]10H
CH2CH2(COO)C12H25
3.46
CH3
[EO/PO(1/2)]10H
[EO/PO(1/2)]10H
[EO/PO(1/2)]10H
3.62
CH3
CH3
[C3H7CH(CH3)C4H8]2 [C3H7CH(CH3)C4H8]2 CHCH2 CHCH2
2.73
CH3
CH3
C9H19CH(CH3)CH2
C9H19CH(CH3)CH2
2.56
CH3
CH3
CH3
C14H29CH(C12H25)CH2
2.89
CH3
CH3
CH3
C18H37
2.13
CH3
CH3
C18H37
C18H37
2.81
CH3
C18H37
C18H37
C9H19CH(CH3)CH2
2.65
CH3
CH3
CH3
[C3H7CH(CH3)C4H8]2 CHCH2
2.45
3.69
Note: EO/PO (1/2) represents poly(ethylene glycol-co-propylene glycol) with the monomer ratio 1:2
116
Intercalation of Clay by at least two C-groups. Colourless CPNC was prepared by dry blending a mixture of clay pigments with Claytone‚ APA and PET. The well-dried mixture was extruded at 275 °C. Unfortunately, there is no information on the extent of intercalation/exfoliation in, or the properties of the CPNC. Similarly for PET, Imai et al. [2002] developed intercalant able to react with PET, with a cationic group to bind to clay and stable at least up to 275 °C. The selected compound was a dimethyl isophthalate substituted with a triphenyl-phosphonium group (dimethyl isophthalate triphenyl phosphonium, DIP). In spite of lack of exfoliation, 8 wt% of the pre-intercalated clay increased the matrix modulus by 80%. TNO patented the use of dyes as intercalants [Fischer et al., 2001], such as, e.g., methylene blue: N
H3 C
CH3 N
S
N+ CΓ
CH3
CH3
Virtually any polymer can be used as a matrix, but for the coating applications the preference goes to PU, acrylics, siloxanes, polyesters and polyethers. The new intercalants offer better thermal stability, controllable amount of intercalation, ability to introduce functional groups for end-tethering the matrix macromolecules, etc. The technology was commercialised for the production of Planomers®, PlanoCoatings® and PlanoColors®. The Planomer range is an organoclay tested in diverse polymeric matrices, viz. PO, PMMA, PA, PU, PS, PC, phenolics, and biopolymers. The second material is used as a transparent barrier material in packaging or for protection. The PlanoColors® are offered as highly UV stable, clay-based, metal-free nanopigments in a variety of colours. Conroy et al. [2002] prepared high temperature CPNC by dispersing up to 10-wt% of clay in molten phthalonitrile monomer or oligomer that in turn was polymerised into CPNC, stable to T ≥ 450 °C. The authors replaced the customary (and thermally unstable) ammonium intercalants by compounds with nitrile groups. This new type of organoclay may be used with a variety of polymers, such as, PA, PC, PO, PEI, PI, TPU, PVP, PVAl, PEG, epoxy, etc. Unfortunately, this patent application does not provide examples of CPNC preparation or performance characteristics. In summary, the standard method of clay intercalation with organic cations usually starts with purified bentonite or MMT. The clay has a layer of Al and Mg hydroxides between two layers of silica that are negatively charged and ionically balanced by Na+, K+, Ca2+and Mg2+. The repeating layers are 0.96 nm thick and under ambient conditions the basal or d-space is d001 = 1.2-1.5 nm, hence the interlamellar space in the gallery is about 0.2 to 0.5 nm. At this stage the clay is unsuitable for use in PNC technology - the interlamellar gap is too narrow for the diffusion of macromolecules, the solid-solid interactions between the clay platelets are too high. The oldest and still most frequently used method has been to replace the inorganic cations with onium ones. This increases the interlayer distance to d001 ≥ 2.5 nm, reduces the interlayer forces, makes the clay more hydrophobic and makes it amenable to dispersion in a monomer or polymer. However, the most common intercalant, quaternary ammonium has limited thermal stability. Ammonium-modified clay may be used to prepare CPNC by first dispersing it in a monomer that in turn is polymerised by UV, acid, base or heat. 117
Clay-Containing Polymeric Nanocomposites This approach is suitable, provided that the polymerisation is conducted at T < 250 °C. Alternatively, CPNC may be prepared by melt blending ammoniummodified clay with a polymer that can preferably be processed at T < 200 °C. With the exception of elastomers, there are few industrially interesting polymers that can be processed at these low temperatures. It is customary to process a polymer at temperatures about 40-50 °C above its transition temperature, in the case of an amorphous polymer the glass transition, Tg, or in the case of a semicrystalline resin, the melting point, Tm. In several patents besides quaternary ammonium, phosphonium and/or sulfonium salts are also listed. It is noteworthy that clay modified with organophosphonium cations may be thermally stable up to 370 °C hence it is suitable to prepare CPNCs with a high temperature matrix polymer. The most common cation-exchange method uses onium salt, the quaternary ammonium being the most common (the primary or secondary have been used in specific cases to ascertain chemical reactivity, e.g., with PA-6 or thermosets). The reason for using the quaternary ammonium salts originates from consideration of the binding strength to MMT. The strength increases with the number of substituents in the ammonium cation [Maes et al., 1980], viz. NH4+ < RNH3+ < R2 NH2+ < R3 NH + < R4 N +
The authors observed that the exchange reaction between either Na-MMT or Ca-MMT and ammonium salt is thermodynamically reversible, and that the ionic charge (usually assigned to the N-atom) is delocalised over the R-alkyl substituents. The delocalisation increases with the number of substituents (from 0 to 4) and their molecular weight. Most commercially available organoclays are intercalated with quaternary ammonium – examples are listed in Table 16. The typical dry particle size of these materials is ca. 5 μm, with the sieve analysis: 10 vol% less than 2 μm; 50% less than 6 μm; and 90% less than 13 μm. The quaternary onium ions can quantitatively replace Na+ in Na-MMT. It can be calculated that stoichiometric replacement by means of 2M2HTA would result in organoclay containing 34 wt% of organic phase. As the tabulated data indicate, there is up to 50% excess of the intercalant present. Elimination of this excess improves thermal stability. Industrially Na-MMT undergoes ion exchange mainly with a quaternary ammonium chloride: R1R2R3R4N+ Cl-, to produce hydrophilic or hydrophobic intercalated MMT. The character and the interlayer spacing depend on the Ri- radicals. In several commercial organoclays R1 = R2 = -CH3 and R3 = R4 = hydrogenated tallow oil. Other radicals used to control interactions with the matrix are benzyl, hydroxyethyl, 2-ethylhexyl, etc. Intercalation involves a slow chemical reaction: MMT- Na+ + R4N+ Cl- ↔ MMT- R4N+ + Na+ ClOwing to low mobility of the MMT- R4N+ ionic pairs, the equilibrium is shifted to the right hand side (RHS). Nevertheless, intercalation in aqueous medium at T = 25-80 °C takes from t = 10 to 480 min, while the second stage intercalation in non-aqueous solvent may take up to 15 days. It results in a complex of MMT with two ammonium ions and two amines: MMT2- 2RN+H3 + 2RNH2. Typical characteristics of commercial organoclays are shown in Table 16 with abbreviations for intercalants listed in Table 17 and Appendix 6.3. The four types of Cloisite® organoclay produced by Southern Clay Products, Inc. are shown in Figure 27. Typically, their dry particle size is ca. 5 μm, with the sieve analysis: 10-vol% less than 2 μm; 50% less than 6 μm; and 90% less than 13 μm. 118
Intercalation of Clay
Table 16 Typical physical properties of commercial organoclays Organoclay
Intercalant
Weight loss on combustion (%)
H2O (%)
Spacing, d001 (nm)
Modifier conc. (meq/g & mmol/g*)
Somasif™ from CO-OP Chemical Co., Ltd. ME-100 none synthetic mica
1.25
MAE
2M2TA
3.0
MTE
M3O
2.5
MEE
M2EtOHC
2.3
MPE
M 2EPPOH
5.0
(CEC = 1.2 meq/g)
®
Cloisite from Southern Clay Products, Inc. Na-MMT
none
7
4
1.23
0
6A
2M2HTA
47
2
3.59
1.40 & 0.86
10A
2MBHTA
39
2
1.93
1.25
15A
2M2HTA
43
2
2.96
1.25 & 0.78
20A
2M2HTA
38
2
2.47
0.95 & 0.83
25A
2MHTL8
34
2
2.02
0.95 & 0.69
30B
MT2EtOH
32
2
1.86
0.90 & 0.85
93A
M2HTA
40
15). Regardless of the concentration, the amount of the intercalating composition should be > 4 times that of clay. The intercalant should be sorbed between and permanently bonded to the clay platelets, increasing the interlayer spacing of the phyllosilicate from 0.5-10 nm. Useful intercalants should be water-soluble and must have a functional group (e.g., carbonyl, hydroxyl, carboxyl, amine, amide, ether, ester, sulfate, sulfonate, sulfinate, sulfamate, phosphate, phosphonate, phosphinate functionalities) or aromatic rings that provide metal-cation complexing to the inner surfaces of clay platelets. Binding to the platelet surfaces is by metal cation either electrostatically bonding or chelating. Another mechanism involves bonding of the interlayer cations with intercalant aromatic rings. The intercalants should have sufficient affinity for the clay platelets to provide adequate interlayer spacing and to bind to the surfaces of the platelets, without additional coupling agents. An addition of metal cations (during intercalation and/or exfoliation) increases the suspension viscosity, most likely by complexing with the polar moieties of the intercalant molecules. The complexed metal salt-derived cations carry their dissociated anions along with the cations, in the interlayer space. Such a double intercalant complexing occurs on the opposed platelet surfaces, resulting in repulsion between closely spaced dissociated anions carried by the added cations, which results in increased interlayer spacing and more complete exfoliation. The concentration of metal salt ranges from 0.01 to 1 wt%, based on the dry weight of the phyllosilicate. The intercalated phyllosilicates can be exfoliated before or during mixing with solvents (e.g., alcohols, glycerols, glycols, aldehydes, ketones, carboxylic acids, amines, amides, etc.). Such a suspension may be used in a thixotropic composition, or for delivery of any active hydrophobic or hydrophilic organic compound, such as an active pharmaceutical, dissolved or dispersed in the carrier or solvent, in a thixotropic composition. Depending on the intercalation and exfoliation conditions (viz. T, pH, and ingredient concentration), the system viscosity can be adjusted in the range of 0.1-5,000 Pas. The compositions are thixotropic. The preferred water-soluble polymer intercalants are: P4VP, PVAl, and their mixtures. The weight ratio of intercalant to Na-MMT should be 1:5 to 1:3. The
124
Intercalation of Clay concentration of intercalant should be 20-90 wt%, increasing the interlayer spacing to d001 = 3.0-4.5 nm. The intercalating composition should include 25-50 wt% water. The water-soluble polymer can be added as a solid along with the clay when extruding or pug milling. Metal salt is dissolved in a suitable solvent (water or organic solvent) and added to the intercalating composition in the amount of 0.005 to 0.5 wt%. The intercalated clay was combined with various organic liquids (with and without water) to determine the effects of intercalate loading as well as temperature, pH and water content of the intercalating composition on viscosity. For example, mixing 10 wt% clay/P4VP intercalate into 84% glycerol, and 6% water resulted in a viscosity of 2-3 Pas. Heating the composition to gelation (100 °C) increased the viscosity to about 8 Pas, and when heated to 145 °C (then cooled to room temperature) it increased to 200-600 Pas. Addition of ethanol had a more dramatic effect. The composition: 20 wt% water, 70 wt% ethanol, and 10 wt% clay/P4VP complex showed viscosity in the range from 0.4-1 kPas, without heating. The development of intercalation technology as described by Beall et al. [1998] is a part of the technology developed over the years as evidenced by a series of patents [Beall et al., 1999a,b; Serrano et al., 1998a,b; Beall et al., 1996a,b]. In another invention, intercalates were prepared by contacting a clay either with water, or with an aqueous solution of a water-soluble polymer and/or a watermiscible organic solvent, e.g., alcohol, followed by contact with a monomeric organic pesticide or its solution. The latter would have a polar moiety, e.g., carboxylic acid, ester, amide, aldehyde, ketone, sulfur-oxygen or phosphorusoxygen moiety, cyano, or a nitro moiety. Best results were achieved using an aqueous solution of water-soluble polymer, to first intercalate the clay [Beall et al., 1999b]. Using > 10 wt% of an organic pesticide gave better sorption. Probably the organic pesticide displaces water and water-soluble polymer and bonds to the platelets by chelation with the exchangeable cation or via electrostatic or dipole/dipole interactions. Extrusion accelerates the intercalation of pesticide between activated clay platelets. The sorbed pesticide may increase the interlayer spacing from 0.5 to 10 nm. The intercalated clay containing a pesticide can be used directly as a pesticide. To intercalate Na-MMT with a herbicide (e.g., 2,4-dichlorophenoxyacetic acid butyl ester (2,4-D)) three processes were employed: 1. Mixing a dispersion of 2 wt% 2,4-D with a 2 wt% clay suspension in water for 4 h at room temperature. 2. Dry clay powder (about 8% by weight moisture) was gradually added to the 2 wt% aqueous dispersion of 2,4-D in water and agitated at room temperature for 4 hours. 3. 2,4-D was dry-blended with clay, the mixture hydrated with 35-38 wt% of water and extruded. All three methods resulted in clay-2,4-D intercalates. The spacing depended on the quantity of 2,4-D sorbed between clay platelets. Similar results were obtained with insecticides, e.g., chlorpyrifos, or the pesticide trifluralin. The intercalated Na-MMT showed d001 = 1.88 nm, an increase from 1.24 nm of untreated clay. When the wet sample was dried in a vacuum oven (10-3 torr at 60 °C for 48-60 h) causing the trifluralin to sublimate the spacing decreased to 1.237 nm. The experiment indicated that trifluralin might be fully released from the interlamellar space.
125
Clay-Containing Polymeric Nanocomposites
2.3.5 Intercalation by Monomers, Oligomers or Polymers Historically, the intercalation methods described in this part were developed later than those by organic cation, discussed before. Cation-exchange technology also profited from the addition of either: (i) an organosilane, (ii) a polar or (iii) an aromatic component. Thus, elements of this technology were used within the category of the cation-exchange methods in the two-step intercalation procedures. The present section will focus on the use of monomers, oligomers, cyclomers or macromolecules to enhance the intercalation and eventually will lead to total exfoliation of clay. The many variations of this method can be divided into two principal groups: 1. Intercalation of purified clay, e.g., Na-MMT. 2. Intercalation of organoclay (a two-step process). Each of these two can be further subdivided depending on whether monomer, oligomer, cyclomer or polymer was used, and whether the second step of intercalation was necessary. 2.3.5.1 Intercalation of Purified Clay by Hydrophobic Compounds Here intercalation of purified inorganic layered compound is of interest. Following the analysis of clay chemistry, two types of interactions can be explored: complex formation between an aromatic molecule and a clay cation that takes place in organic medium, and hydrogen bonding with water soluble monomers, oligomers or polymers. In one of the first patents on CPNC, a vinyl monomer (e.g., styrene) was used to intercalate the clay [Kamigaito et al., 1984]. Partly due to the high concentration of clay (85 wt% of MMT, vermiculite or Na-taeniolite) and partly to the small thermodynamic potential for styrene diffusion into interlamellar spacing, the effect was small. Thus, the cited values of the interlayer spacing were: before intercalation, d001 = 1.2-1.3 nm, and after expansion by sorption of styrene: d001 = 1.5 nm. These values indicate that initially the clay contained one to two monolayers of water. It is doubtful that styrene was able to expel this surfacebonded water, thus the expansion of d001 by about 0.2 nm may indicate the formation of a monolayer of styrene coordinated to the inorganic counterions within the interlayer space. According to the authors, the initial sorption process took ‘a very short period of time’. After the free radical polymerisation of sorbed monomer, the polymers showed a broad molecular weight distribution (MWD), viz. Mw/Mn ≥ 6. Some vinyl monomers may be converted into polymers with narrow MWD by anionic or group-transfer polymerisation, but these reactions are not applicable to condensation polymers (e.g., PA). The Kamigaito et al., patent [1984] also describes intercalation of MMT by ε-caprolactam in an aqueous medium. The process for the manufacture of CPNC is a bit convoluted, as it requires addition of dichlorodimethyl silane, filtering, drying, then addition of PA-66 dissolved in formic acid, heating the mixture to 200 °C (to effect copolymerisation of caprolactam with PA-66) and finally extrusion of a dry blend of concentrated CPNC with PA-66 pellets. The interlayer spacing was not given, but the final polymer mouldings showed significant improvement of modulus, strength and heat distortion temperature (HDT) in comparison to neat polymer. However, in the following patent the authors stated that the achieved degree of intercalation was low, dispersion of the silicate layers was not uniform, the 126
Intercalation of Clay clay aspect ratio was seriously reduced (thus reducing MMT effects on performances), the bonding between the clay and the matrix was not strong enough and the polyamide matrix had a broad molecular weight distribution, viz. Mw/Mn ≥ 6 [Okada et al., 1988]. Even if the monomer molecules diffuse into the interlamellar space they rarely bond strongly enough (by ionic or covalent bonds) to the clay surface, and the CPNC may be exfoliated, but not end-tethered. Several publications from the Toyota group indicate that even highly polar and relatively small monomers, such as ε-caprolactam, do not provide sufficient intercalation and bonding to the clay surface – Na-MMT must be pre-intercalated [Usuki et al., 1990; 1993a,b; 1995]. The latter process was carried out with 12-aminolauric acid (ADA) in aqueous medium. Obviously, instead of ADA another highly polar, water-soluble compound may be used, but as in any other cation-exchange intercalation process, the initial step of intercalation seems to require the presence of water. The intercalation with ADA results in nearly quantitative cation exchange; hence after polycondensation the PA-6 matrix is end-tethered and densely packed near the clay platelet surface. Prior to the emulsion or suspension polymerisation of styrene, the clay platelets dispersed in aqueous media (viz. Na-MMT) may be fully exfoliated. However, in the recovered polymer the interlayer spacing is quite small, d001 ≤ 1.7 nm, indicating re-aggregation. Thus, as a rule, to achieve a reasonable degree of intercalation in CPNC the intercalant must bind to the clay surface. The binding can be achieved by ionic bonding, covalent bonding (e.g., through ≡Si-OH groups) or at least through hydrogen bonding with hydrophilic compounds. 2.3.5.2 Intercalation of Purified Clay by Hydrophilic Compounds Owing to the hydrophilic nature of clay, intercalation by water-soluble monomers, oligomers or polymers is expected to be relatively simple. However, early attempts to intercalate MMT with PVAl, P4VP or PEG, were not successful. Greenland [1963] reported that PVAl containing 12% residual acetyl groups increased the d001 spacing only by ca. 1.0 nm. As the concentration of polymer increased from 0.25 to 4 wt%, the amount of polymer sorbed was reduced, indicating that sorption might only be effective at polymer concentrations of 1 wt% or less. Such a dilute process would be too costly, hence no further work was carried out toward commercialisation. Intercalation of Na-MMT by P4VP (MW = 40 kg/mol) was carried out in 1 wt% P4VP dissolved in ethanol/water solution. The clay was first repeatedly washed with ethanol. After dispersing it in the solution, filtering out and drying the basal spacing expanded to 2.6-3.2 nm [Levy and Francis, 1975]. It was concluded that: 1. The ethanol was needed to initially increase the basal spacing for the later sorption of P4VP to take place; 2. Water did not directly affect sorption of P4VP between the clay platelets; and 3. P4VP sorption was time consuming and difficult. Since 1995 Nanocor (a subsidiary of the AMCOL International Corp.; known since 1927 as American Colloid Co.) has patented several methods of chemical modification of clays. In an early patent, clay was intercalated using 30-80 wt% of a water-soluble monomer, oligomer (degree of polymerisation, DP = 2-15) or polymer (DP > 15) [Beall et al., 1996a]. The aim was to have the intercalant 127
Clay-Containing Polymeric Nanocomposites bonded to the clay platelets, thus increasing the interlayer spacing to d001 = 3-4.5 nm (the most desirable spacing!). To bond, the intercalant needed either a strong polar group (e.g., carbonyl, hydroxyl, carboxyl, amine, amide, ether, ester, sulfate, sulfonate, sulfinate, sulfamate, phosphate, phosphonate, phosphinate) or an aromatic ring able to form a metal-cation complex. Several types of interaction were envisaged, viz. by metal cation electrostatic bonding, chelation of the clay metal cations, bonding between clay inorganic cations and intercalant aromatic rings, by hydrogen bonding to the surface -OH groups, etc. The following water-soluble polymers or oligomers were reported useful as intercalants: •
P4VP (MW = 1-40 kg/mol). Its solubility can be adjusted by hydrolysis or by forming a sodium or potassium salt. Copolymers of vinylpyrrolidone and vinyl amide or γ-amine butyric acid, can be similarly treated. • Fully hydrolysed PVAl (MW = 2 to 10 kg/mol). • Fully or partially neutralised salts of a polyacrylic acid (PAA) or a polymethacrylic acid (PMAA, MW = 0.2 to 10 kg/mol). • Polymethacrylamide, poly(N,N-dimethyl acrylamide), poly(N-isopropyl acrylamide), poly(N-acetamido acrylamide), poly(N-acetamido methacrylamide), their copolymers or interpolymers containing PAA, or PMAA. • Polyvinyloxazolidone (PVO) and polyvinyl methyl-oxazolidone (PVMO), PEG, polypropylene glycol (PPG) and their copolymers. • Polymeric quaternary ammonium salts. • Sodium salts of acrylate/vinyl alcohol, olefin/maleic acid, polymethacrylate, polystyrene sulfonate, styrene/acrylate/PEG dimaleate, styrene/PEG maleate/ nonoxynol, etc. • Styrene copolymers with acrylamide, acrylate-ammonium methacrylate, maleic anhydride, PVO, etc. • Diverse copolymers, viz. cornstarch-acrylamide-sodium acrylate, diethylene glycol-amine-epichlorohydrin-piperazine; dodecanedioic acid-cetearyl alcoholglycol, ethylene-vinyl alcohol, hydroxy ethyl-polyethyleneimine, polyvinyl methacrylate-methacrylic acid, melamine-formaldehyde resin, phthalic anhydrideglycerin-glycidyl decanoate, metal salts of acrylic and polyacrylic acid, sucrose benzoate-sucrose acetate, isobutyrate-butyl benzyl phthalate, sucrose benzoatesucrose acetate isobutyrate-butyl benzylphthalate-methyl methacrylate, vinyl acetate-crotonic acid, and polysaccharide copolymers. • Prepolymers: urea-formaldehyde, urea-melamine-formaldehyde, etc. In the cited patent [Beall et al., 1996a] the preferred intercalants are P4VP, PVAl, and their mixtures at the weight ratio of intercalant-to-Na-MMT = 1:51:3. By contrast with the prior publication that indicated significant difficulties and achieved minor expansion of the interlayer space, the new intercalation in the presence of at least 10 wt% of water proceeded quite readily using any of the three methods. For example, to intercalate Na-MMT with P4VP (MW = 10 and 40 kg/mol) or PVAl (75-99% hydrolysed) the following processes were used: 1. Mix for 4 h aqueous 2 wt% solution of P4VP (MW = 10-40 kg/mol) with 2 wt% suspension of Na-MMT clay in water, at a ratio sufficient to vary the weight ratio of clay to P4VP from 4:1 to 1:4, based on dry mass. 128
Intercalation of Clay 2. Add the clay (ca. 8 wt% of moisture) to aqueous 2 wt% solution of P4VP in a ratio as in method (1) and mix for 4 h. 3. Dry blend P4VP with Na-MMT, and then add 35-38 wt% H2O (based on dry clay) and extrude the paste. When a dry blend of Na-MMT and powdered P4VP was mixed with 75 wt% water an exothermic reaction was observed. Apparently, the bonding reaction of a polymer to the internal face of the clay platelets is sufficient to engender an exothermic exfoliation. All of the three methods resulted in intercalation of MMT. The final spacing did not depend either on the method of preparation or MW of P4VP, but on the quantity of polymer sorbed between platelets. Exfoliation did not take place unless the clay contained at least 10 wt% water. The polymer-engendered expansion of the interlayer was demonstrated by drying the intercalated samples for 4 h at 120 °C – only a small change in the interlayer spacing was detected. The generality of the developed technology was demonstrated in later patents (see below), where Na-MMT was directly intercalated with insecticide or pesticide (e.g., 2,4-dichloro phenoxy acetic acid butyl ester), provided that their molecules were water-soluble and had strong polar groups. As shown in Figure 28 the interlayer spacing increased with the concentration of P4VP or PVAl in steps. When the intercalating composition contains < 16 wt% of intercalating polymer a monolayer is sorbed between the platelets, increasing the interlayer spacing by < 1 nm. When the concentration is in the range from 16 to 35 wt% the interlayer spacing increases by 1 to 1.6 nm. At loadings of 35 to 55 wt%, the interlayer distance is increased to about 2.0-2.5 nm, which corresponds to three layers of intercalant sorbed between adjacent clay platelets. At loadings of 55 to 80 wt% the interlayer distance is increased to 3.0-3.5 nm,
Figure 28 Interlayer spacing as a function of PVP or PVAl concentration (calculated on dry weight of Na-MMT). Data Beal et al. [1998].
129
Clay-Containing Polymeric Nanocomposites which corresponds to 4 and 5 layers of sorbed intercalant – this is considered the optimum interlayer distance for organoclays. When these are incorporated into polymeric matrix, the platelets with the intercalant molecules attached to the surface exfoliate. Such interlayer spacing has not been achieved by direct intercalation of a monomer, an oligomer or a polymer molecule, without prior sorption of an onium or silane coupling agent. The new method leads to easier and more complete exfoliation. The exfoliated compositions, especially the high viscosity gels, may be used for delivery of active compounds, such as an oxidising agent for hair lotions, drugs, cosmetics, oil well drilling fluids, paints, lubricants, food grade lubricants, etc. An important feature of the AMCOL invention is that the intercalate compositions can be manufactured in a concentrated form, e.g., as a master gel; having 20-80 wt% intercalated or exfoliated clay. Exfoliation should provide delamination of at least about 80 wt% of the intercalated material. For such relatively thorough exfoliation a shear rate > 10 sec-1 may be required. The preferred shear rate is 100-10,000 sec-1. The shearing may be imposed by mechanical means (in a Banbury, Brabender, continuous mixer or extruder), by thermal shocks (alternatively raising and lowering the temperature), by pressure alteration (0.05-6.0 MPa sudden pressure changes) or by ultrasonics (vibrations cause portions of the composition to be excited at different phases). Development of the intercalation technology described above has continued in the following patents [Beall et al., 1996a,b; 1999a; Serrano et al., 1998a,b; Ferraro et al., 1998]. In another document [Beall et al., 1999b], intercalates were prepared by contacting clay either with water, or with an aqueous solution of a water-soluble polymer and/or a water-miscible organic solvent, e.g., alcohol, followed by contact with a pesticide or its solution. The pesticide should have a polar group, e.g., carboxylic acid, ester, amide, aldehyde, ketone, sulfur-oxygen or phosphorus-oxygen moiety, cyano, or a nitro. As for P4VP, here also the results did not depend upon the method, but on the quantity of active ingredient sorbed between clay platelets. The intercalated Na-MMT showed d-spacing = 1.88 nm, an increase from 1.24 nm for untreated clay. In the following patent [Tsipursky et al., 1999] intercalation was also carried out in the presence of 25 to 50 wt% of H2O using a water-soluble monomer, an oligomer and/or a polymer. The shearing could be provided by mechanical means, thermal shock, pressure alteration, or by ultrasonics. The amount of intercalant in contact with Na-MMT was ca. 20 to 60 g per 100 g of clay. As before, the intercalant should have an aromatic ring and/or a polar group: carbonyl, carboxyl, hydroxyl, amine, etc. The preferred intercalants were P4VP, PVAl, polyacrylic or polymethacrylic acid polymers and copolymers. A wide variety of topically active compounds (e.g., cosmetics, industrial, medicinal) could be incorporated into a stable composition and evidently, so could the monomers. As before, the preferred interlayer spacing was stated as d001 = 3.0-4.5 nm. The viscosity, η = 0.02 to 5,000 Pas, could be adjusted by pH and addition of metal salts. A very long list of suitable salts was cited, from Al-acetate oxide to Zr-trifluoroacetyl-acetonate. For example, addition of 0.1 wt% of either AlCl2OH or Mg-acetate increased η by a factor of 5. Another disclosure [Lan et al., 1998; Lan et al., 2000] brings AMCOL’s intercalation methods closer to the mainstream of CPNC technology. Thus, clay was treated with 30 to 80 wt% of an organic intercalant, having paraffinic Cn130
Intercalation of Clay chain with n ≥ 6. The compound must also have a polar group, viz. hydroxyl, polyhydroxyl, carbonyl, polycarboxylic acids and salts, aldehydes, ketones, amines, amides, ethers, esters, lactams, lactones, anhydrides, nitrites, halides, pyridines and their mixtures. The intercalating composition should have a weight ratio of organic intercalant to clay ≈ 1:4. According to the authors, bonding by the intercalant polar ends causes the molecules to stretch in the direction normal to the platelet surface. This increases the interlayer spacing, while consuming little of the intercalant. As a result, there is sufficient expansion of d001 and sufficient concentration of the remaining free cations to ascertain sorption of polymerisable monomer, oligomer, and/or polymer molecules, e.g., of an epoxy resin. The process does not require either onium ions or silane coupling agents. It can be applied to all resins, but it has been specifically designed for the epoxy resins. Diluting the Na-MMT/intercalant concentrate with a monomer or oligomer, and then curing may lead to exfoliation. The presence of polymerisable monomer or oligomer in the clay galleries improves miscibility with the matrix polymer, hence a masterbatch can be mixed with additional polymer. Alternatively, the intercalant may be dispersed into a melt processable thermoplastic or thermosetting matrix oligomer or polymer, then clay and water added. The matrix polymer may have DP = 10-100, and a melt index MI = 0.01-12 g/10 min at the processing temperature. For example, Na-MMT (8 wt% water; CEC = 1.2 meq/g) was mixed at room temperature with epoxy resin and 1-dodecyl-2 pyrrolidone (DDP) in a 1:1 molar ratio to the Na+. Then, water was gradually added to the mixture. The solid-like mixture was extruded using a single screw extruder (SSE) and dried at 90-95 °C obtaining uniform powdered materials. XRD determined that incorporation of DDP and epoxy increased the interlayer spacings of Na-MMT, d001 = 1.23 to 3.4 nm. Similar spacing was obtained using different mixing sequences and methods, viz. by adding DDP and epoxy to clay slurry and then drying the mixture; by adding DDP/epoxy/water emulsion to clay, extruding using a twin screw extruder and drying. However, when DDP was eliminated from the formula the maximum interlayer spacing achieved by adding water and epoxy was d001 = 1.9 nm. In conclusion, water is the essential element for the co-intercalation by the epoxy and DDP molecules. Owing to the co-intercalation, the free liquid phase of DDP/epoxy/water disappears and the mixture becomes solid-like. DDP molecules bind to the interlayer Na+ and epoxy molecules diffuse into the interlayer spacing. The co-intercalate has the molar/weight ratio: DDP/epoxy/MMT = 1:1:0.75. Several types of polar organic compounds can be used as intercalants, including long chain ethers, esters or alcohols, having a polar end group that provides the molecule with a dipole moment greater than that of water. The listed examples include aliphatic; aromatic; aryl-substituted aliphatic; alkyl-substituted aromatic alcohols and phenols, e.g., manufactured from coconut, tallow and/or palm oils. Straight-chain acids can also be used, for example, stearic, oleic, linoleic, ricinoleic, canola, castor, tallow-based, soybean or coconut. Representative aldehydes include C6 to C28, phenyl acetaldehyde, etc. Amines (primary, secondary and tertiary) or amides are also suitable, viz. oleylamine, soya alkylamines, hydrogenated or not tallow or ditallow alkylamines, palmitamide, hexadecyl amide, stearamide, oleamide, linoleamide, etc. Suitable nitrites include hexanonitrile, n-nonadecanitrile and others. Lactones, lactams, pyridines and surface-active esters can also be used. 131
Clay-Containing Polymeric Nanocomposites These clay-intercalate complexes can be dispersed into thermosetting, thermoplastic matrix or elastomeric oligomers or polymers. For example, the indicated thermosets are: epoxides, polyamides, polyalkylamides, polyesters, polyurethanes, polycarbonates, and their mixtures. The cited thermoplastics are: polyamides (e.g., PA-4, PA-6, PA-66, PA-6,10), polyesters (e.g., PET, PBT, polycyclohexylene terephthalate (PCT)), and polyolefins (e.g., PE, PP) as well as many others, viz. PCL, PU, PSF, PPS, polyetherimide (PEI), polyketones, vinyls, acrylics, polyacrylonitrile, ionomers, and their blends. The elastomers may be selected from between chlorinated or brominated butyl rubber, TPU, polyester elastomers, PVC/NBR, polybutadiene (PBD), polyisoprene (IR), EPR, EPDM, chloroprene (CR), chlorosulfonated polyethylene (CSR), poly(2,3-dimethylbutadiene), fluoro-, silicone or polysulfide elastomers, etc. Exfoliation of the intercalated layered material should affect > 90 wt% of the layers, which for some systems may require a shear rate: 10 < γ˙ (sec-1) < 10,000, for others heating, or cycling the pressure between 0.05 and 6 MPa, with or without heating. The shearing can be either mechanical (e.g., during extrusion, injection moulding, compounding in a TSE or an internal mixer), or by thermal shock, pressure alteration, ultrasonics, etc. The amount of intercalated/exfoliated clay in a liquid matrix depends on the intended use and desired viscosity of the composition. A relatively high amount, i.e., from 10-30 wt%, may be used in solvent gels with viscosity η = 5-5,000 Pas. This may be achieved with 0.1-5 wt% organoclay, by adjusting the pH and/or by heating the composition to T = 75 to 100 °C. A masterbatch with 20-80% organoclay may be prepared, but the final composition should not exceed 10 wt% organoclay. The patent applications are quite broad, addressing: 1. Preparation of a wide variety of topically active compounds (e.g., cosmetic, medicinal, and industrial viz. antiperspirants, deodorants, antidandruff or antibacterial compounds, antifungal or anti-inflammatory compounds, anaesthetics, sunscreens, analgesics, antiseptics, antiparasitics, antibacterials, steroids, herpes treatment drugs, pruritic medications, etc.). 2. Thermoplastic or thermosetting moulding CPNC for the production of sheets and panels, formable by conventional processes, viz. vacuum or hot pressing, lamination of, for example, wood, glass, ceramic, metal or other plastics. The sheets and panels can also be produced by coextrusion. 3. Extrusion of 25-75 μm thick films and film laminates for food packaging, fabricated by conventional means, followed by biaxial stretching to orient the clay platelets parallel to the film surface. These films exhibit increased modulus, wet strength, and dimensional stability as well as decreased moisture adsorption, permeability to gases and liquids, etc. From the fundamental point of view, MMT being a weak silicilic acid has a negatively charged surface that should interact with a cationic monomer or polymer. For example, as the discussion above demonstrated, poly-4-vinylpyridine (PVP) may easily be protonated to form water soluble, positively charged polyelectrolyte, which intercalates the clay. Similarly, such monomers as protonated vinyl pyridines and their N-alkylated derivatives (4-VPH+X–) are expected to intercalate by ion exchange. The inserted monomer can be polymerised within the MMT galleys. Depending on the monomer concentration and acidity 4-VPH+X– may either form 1,2- or the quaternary 1,6-polyelectrolyte with the pyridinium units in the side or in the main chain, respectively. 132
Intercalation of Clay In polymer technology it is often advantageous to replace one component by a mixture of similar functional ingredients. For example, compatibilisation of two immiscible polymers is often more efficient by a mixture of compatibilisers than by either one. Similarly, blending one polymer with two fractions often results in better material. This philosophy has also been applied to clay intercalation [Lan et al., 2003]. Evidently, it is easier to mix known and available organic intercalants than to design and synthesise new ones. To illustrate the advantage of the mixing approach, consider the following situation. For the best performance of CPNC it is essential that the matrix is miscible with the organic tails of the intercalant attached to the clay surface. A good miscibility candidate is ethoxylated or polyalkoxylated ammonium salt (e.g., ETHOQUAD 18/25 or JEFFAMINE506). However, this compound has a high molecular weight, and if used alone it would reduce the clay content to less than 50 wt%. It was found that its 50:50 combination with ODA is easy to manufacture and use for CPNC. The new mixed-onium organoclays have a broader range of application in diverse polymeric matrices (e.g., PVP, PVAl, PEG, PTHF, PS, PCL, PA-6), but they have been especially designed for thermoplastic polyesters (PEST) such as PET. The CPNC may be used alone or as one component of the multi-layered structures with PEST, aliphatic or aromatic PA, EVAl, etc. For example, the patent describes preparation of CPNC by melt compounding, in a solvent or by reactive means, using an organoclay that was pre-intercalated with a mixture of at least two organic cations. In a given example, the PET-based CPNC (with 75 to 90% of exfoliated clay platelets) showed low permeability and good mechanical properties, thus it was suitable for bottle blow moulding. A similar approach was taken by Han et al. [2004]. The authors preintercalated Ca-MMT with trimethyl cetyl ammonium bromide (3MCA), aminoundecanoic acid (AUA), and a 1:1 mixture of these two. The organoclays were used to prepare CPNC with PU as the matrix. XRD and TEM showed that PU containing MMT-AUA was exfoliated, PU with MMT-3MCA was intercalated (d001 = 3.5 nm) and the one with mixed intercalant had an intermediate behaviour (small peak of MMT-3MCA remaining). The tensile properties of the CPNCs containing 5 wt% organoclay demonstrated an advantage of the mixed intercalant approach – the tensile modulus and strength were the best, with the elongation at break 35% higher than that of neat PU. Na-MMT (CEC = 0.83 meq/g) was intercalated in an aqueous solution of 2 wt% of poly-4-vinyl pyridinium bromide (at a level of 5×CEC) [Fournaris et al., 1999]. Polymerisation of 4-VPH+Br– in the clay galleries was fast and it resulted in the formation of quaternised ionene, independently of the polymerisation conditions. The quaternised ionene polymer and partially protonated poly-4vinylpyridine could also be dissolved in water and used to intercalate Na-MMT at a level of 4×CEC. The isotherms for the 1,2-PVPH+Br-, 1,6-PVP+Br–, and P4VP adsorption by Na-MMT were obtained by adding the appropriate amount of each intercalant to a suspension of Na-MMT and stirring it for 2 h. The products were isolated by centrifugation, washed, and then air-dried on glass plates. The amount adsorbed by the clay was calculated from analysis of the supernatant solution. Independently of the amount of polymer available for intercalation, only one macromolecular sheet of P4VP or the quaternised ionene was found to enter the interlayer space causing the interlayer space to expand by Δd001 = 0.4 to 0.54 nm. Thus, this amount of polymer suffices to coat the clay surfaces, making 133
Clay-Containing Polymeric Nanocomposites the clay anions not available for additional macromolecules. The surface saturation coverage increased in the order: partially protonated P4VP > quaternised ionene form > completely protonated P4VP. By contrast with neutral P4VP, the partially protonated P4VP could be adsorbed at different levels. XRD showed that the d001 spacing increases with P4VP loading (see Figure 29). At a concentration of P4VP > 70 wt% the XRD peak disappeared due to delamination. The FTIR spectra provided additional evidence of P4VP interaction with the clay surfaces. Fischer [1999] patented the use of block or graft copolymers for the preparation of CPNCs. By analogy to compatibiliser in immiscible polymer blends, the copolymer must have two types of structural units, one (A) to interact with clay the other (B) with polymer. The molecular weight of part A should be MW = 0.1 to 5 kg/mol (DP = 5 to 20), while for part B MW = 0.1 to 20 kg/mol. To interact with clay the structural units of part A should be hydrogen-bonding, e.g., vinyl pyrrolidone, vinyl alcohol, ethylene glycol, ethylene imine, vinyl pyridine, acrylic acid, acrylamide. To interact with the matrix the structural units of part B should be miscible with the matrix polymer and able to entangle. In the organoclay the inorganic clay content ranged from 5 to 600%, while in CPNC it constituted from 2 to 55 wt%. Judging by the examples, preferably the intercalant is a block copolymer, e.g., PS-b-PEG or PS-b-PVP. The intercalation is performed in THF, followed by solvent evaporation and melt compounding with the matrix polymer. Fischer et al. [1999; 2001] used this strategy to prepare CPNCs with either natural or synthetic clay and either PS or PMMA as matrix. The authors started directly with Na-MMT, Na-saponite, or Na- fluorohectorite (CEC = 0.85 to
Figure 29 XRD peak position(s) for Na-MMT intercalated with partially protonated poly-4-vinylpyridine. The peaks disappeared for loadings above 65 wt%. Data [Fournaris et al., 1999].
134
Intercalation of Clay 3.2 meq/g). Block or graft copolymers were used (viz. PS-b-PEG, PMMA-b-PEG, PMMA-co-polymethacrylic acid, PS-b-P2VP; MWblock = 1 to 27 kg/mol) having one part either identical to or miscible with the matrix polymer and the other capable of intercalating with the clay. The interaction between the compatibiliser and clay was preferentially ionic or hydrogen bonding. The intercalation was performed either in the melt (copolymers with PEG) or in solution. It resulted in individual clay platelets and/or short stacks containing 2-10 layers homogeneously distributed in the matrix. As expected, better mechanical performance was obtained for block copolymers having the miscible block with molecular weight above the entanglement molecular weight value: MW > Me. Several clays (e.g., sodium montmorillonite, hectorite, and laponite) were meltintercalated with PEG using microwave irradiation [Aranda et al., 2003]. The process was found to depend on irradiation time and power, amount and relative ratio of the reagents, and relative humidity. The process resulted in d001 ≅ 1.8 nm indicating formation of a hydrated double layer of PEG within the interlamellar galleries. Thus, 10 minutes of microwave irradiation of moderate power was able to accomplish a similar task as shear compounding in an internal mixer or TSE.
2.3.6 Two-Step Intercalation This process usually involves an organoclay that has been pre-intercalated with a suitable onium cation, then treated with low molecular weight polymerisable liquid (monomer, oligomer or cyclomer), which in turn may be polymerised. This reactive last step is essential as it leads to the ultimate level of intercalation – the exfoliation. An early example of this process has already been discussed in relation to the preparation of PA-6 nanocomposites [Okada et al., 1988]. The process comprised three steps: 1. Intercalation of Na-MMT (CEC = 1.8 meq/g) with ω-amino-dodecanoic acid (ADA) in the presence of inorganic cation, e.g., Cu2+ or Al3+; 2. Mixing the intercalate with ε-caprolactam, and 3. Polycondensation. As shown in Figure 25, each of the three steps contributed to the expansion of the interlayer spacing. The clay was first dispersed in aqueous, acidified intercalant solution, reacted with ADA and the precipitate was washed with water to remove excess ions. The organoclay paste was then mixed in a mortar with monomer and the mixture heated at 80 °C for 3 h to ensure dehydration and melting of the caprolactam. The homogeneous complex was then transferred to a closed stainless steel vessel and heated at 250 °C for 5 h. The resulting CPNC contained exfoliated clay platelets and showed significant improvement of the mechanical and thermal properties. Over the years several modifications of this basic process have been made. For example, it was found that the original process resulted in end tethering of all macromolecules, having the end amino groups of PA firmly attached to the clay surface. As a result, the CPNC had insufficient dyeability, as well as poor coating and printing properties. The modified method [Deguchi et al., 1992] comprised the following steps: 1. Dispersing >1 wt% of Na-MMT (CEC = 1.19 meq/g) in water solution of ADA and HCl, followed by stirring at 80°C for one hour. After washing, the mixture was filtered to obtain organoclay complex (abbreviated as 12MMT; d001 = 1.25 to 1.8 nm) containing 88 wt% H2O. 135
Clay-Containing Polymeric Nanocomposites 2. To the 12MMT-paste H2O and ε-caprolactam were added in a ratio: 12MMT/H2O/ε-caprolactam = 1:9:9, followed by mixing. 3. The produced organoclay/water/ε-caprolactam mixture was compounded with 72 wt% of either PA-6 (MW = 15 kg/mol) or PA-66 (MW = 20 kg/mol) in a TSE. The extruded strands were pelletised, and then caprolactam was extracted from the pellets using hot water. Excellent clay layer dispersion was observed under TEM in the moulded CPNC, with only single platelets or doublets visible. XRD and swelling tests indicated exfoliation. The new method resulted in CPNC in which layers of silicate are uniformly dispersed and not all amino end-groups are tethered. The material showed good mechanical properties, heat resistance, improved dye-affinity and whitening resistance during stretching. This process has been used for the commercial scale production of PA-6 nanocomposites. Two-step intercalation is particularly important for preparation of CPNC with non-polar polymers, viz. PO, regardless of whether the selected method is reactive or by melt compounding. Since these systems will be discussed in greater detail later, here only two examples are given. Heinemann et al. [1999] have prepared PO-based PNC starting with hectorite intercalated with dimethyl benzyl stearyl ammonium chloride (2MBSA), having d001 = 1.96 nm. When HDPE was polymerised in the presence of the intercalated clay, the XRD peak shifted to d001 = 1.40 nm. Best results were obtained for LLDPE-type ethylene-octene copolymers. The authors reported that performance of the reactively prepared CPNC was better than that prepared by melt compounding. However, the intercalated clay was found to interfere with metallocene catalysts. An early patent on the preparation of PO-based CPNC may serve as an illustration of the two-step melt intercalation method [Kato et al., 1997; Usuki et al., 1999]. Thus, MMT was intercalated with ODA and a guest molecule (MW = 0.100-100 kg/mol) having a polar group, e.g., a maleated PO. Alternatively, an oligomeric guest molecule with a polar group could be added to the oniumintercalated clay. Next, the complex was compounded with a non-polar PO matrix. The most important feature of the invention is the use of a polar guest molecule that can bond to the pre-intercalated clay layers and provide entangling possibility for the main PO macromolecules. The process is sensitive, as insufficient amounts of maleic anhydride (MAH) groups do not provide sufficient compatibilisation, and too high concentration may lead to phase separation of the guest molecules within the PO matrix, which in turn results in poor CPNC performance. Evidently, as in any other phase separation process, this also depends on other factors than MAH content, viz. molecular weight, temperature, pressure, stress field, etc. 2.3.6.1 Intercalation by Silylation In the patent applications of 1992 MacRae Maxfield et al. [1995] introduced traditional ‘sizing agents’ into nanotechnology. The aim was to obtain PA-based CPNC by melt compounding. It was found that cation-exchanged MMT could not be exfoliated during extrusion in a twin-screw extruder. However, the situation improved when the organoclay was treated with a silane coupling agent. In the intervening years better organoclays have been developed and today melt compounding PA with cation-exchanged MMT can result in exfoliated CPNC.
136
Intercalation of Clay However, the observed enhancement of dispersion and resulting CPNC performance by incorporation the sizing agents are valuable lessons for the many difficult systems that remain. The following procedure was described. A suspension of 5-15 wt% MMT in water was heated to 80 °C in a high-speed mixer with (NaPO3)6 then an ammonium salt (e.g., octadecyl-ammonium chloride) is added. The aminecomplexed clay was centrifuged, washed, dried, and ball milled to 100-mesh powder. The powder was re-dried at 100-160 °C in vacuum for 8-24 h in the presence of phosphorous pentoxide. Next, the intercalated clay was treated with amino-ethyl amino-propyl trimethoxy silane in an organic solvent (e.g., dioxane, DMSO, MEK). The modified clay (60 mmole of ammonium and 20 mmole of silane per 100 g of clay) was compounded into a masterbatch using a TSE. The masterbatch was finally diluted with PA-6 to mineral concentrations of 0.1, 0.05, and 0.01 wt%. The process was revised in the next patent [MacRae Maxfield et al., 1996]: 1. MMT (aspect ratio 15 ≤ p ≤ 300) was reacted with an organosilane, organotitanate or organozirconate capable of forming covalent bonds with the particles and either reacting with a polymer or at least miscible with the polymerisation product. For example, a suspension of 5-15 wt% MMT in an aqueous solution of sodium hexametaphosphate was heated at 50-90 °C. The dispersion was combined with a solution of caprolactam-blocked isocyanatopropyl triethoxy silane and 1-trimethoxysilyl-2-(m,p-di-chloro methyl)-phenylethane. The organosilane intercalated MMT was filtered, dried (at 60-160 °C for 8-24 h) and ground to about 100 mesh. 2. The intercalated MMT was combined with caprolactam and aminocaproic acid having cation-exchange capabilities. The mixture was polymerised. In the CPNC the amount of MMT varied from 0.1 to 10 wt% and the interlayer spacing d001 ≥ 5 nm. The platelets were uniformly dispersed. In 1998 Ogawa et al. [1998] described a general method of intercalation of layered silicate (Na-magadiite). The process involved silylation of interlayer silanol groups with organochlorosilanes with different functionality. The reaction created organically modified surfaces where organic groups are covalently attached to the clay surface. However, to start with, Na-magadiite was first ion exchanged with dodecyl-trimethyl-ammonium (3MDDA). The resulting intercalated clay (d 001 = 2.79 nm) was dispersed in a toluene solution of either octyldimethylchlorosilane (C 8H 17 (CH 3 ) 2 SiCl (1)) or octyltrichlorosilane (C8H17SiCl3 (2)), under a blanket of N2 for 48 h. The product was separated by centrifugation, then washed with an acetone/water mixture and dried. During the reaction de-intercalation of 3MDDA took place (evidenced by the absence of N in elemental analysis). The basal spacing decreased to d001 = 2.33 and 2.18 nm, for (1) and (2) respectively. The amounts of the attached organosilyl groups were determined to be 1.84 and 1.94 per 14SiO2 (molar ratio), respectively. The two differently modified magadiites showed different adsorption behaviour. The difference in the composition affected the reactivity, which in turn led to a different level of the basal spacing (e.g., after absorption of n-octyl alcohol: d001 = 3.20 and 3.18 nm, respectively). After silylation the organosilyl groups formed monolayers. When n-alkyl alcohols were introduced, they rearranged the microstructure, forming a bilayer and expanding the interlamellar spaces.
137
Clay-Containing Polymeric Nanocomposites Jeong et al. [1998] and Hudson [1999] described preparing PO-based PNC in three steps: 1. Functionalising MMT with an aminosilane. 2. Reacting the free amine with maleated PO (MW ≅ 20 kg/mol; ca. 1 wt% MAH groups) either in solution (reaction time ca. 30 min) or in melt. 3. Dispersing the organomodified clay in a semicrystalline PO. Cocrystallisation between the maleated and the semicrystalline PO is essential. For example, the coupling reaction between MMT and amino ethyl-dimethyl ethoxysilane was carried out for 2 to 4 h at T = 25-90 °C, in either ethanol and water, or dimethyl acetamide. The monoalkoxy silane may either react with the clay surface, self-condense to form dimer, or it may remain unreacted. Grafting may lead to precipitation. The reaction produces ethanol, which has been detected. Detection of the dimer by NMR is facile (resonance at 9.3 and 13.9 ppm). XRD of the silane treated MMT demonstrated increased interlayer separation with d001 = 2.4 to 7.3 nm. TEM showed the presence of many individual MMT-layers, but in CPNC containing 15% clay, most were found in stacks of 2-4 layers ca. 1 μm long. 2.3.6.2 Intercalation Utilising Epoxy Compounds Pinnavaia and his coworkers explored the use of epoxy compound as a second intercalant for layered materials. As early as in 1994 [Wang and Pinnavaia, 1994; 1998a,b; Wang et al., 1996], delamination of organoclays in epoxy resin was reported. The process involved heating an onium ion intercalated smectite with epoxy at T = 200-300 °C. The work resulted in several patents [Pinnavaia and Lan, 1998a,b,c; 2000a,b] and publications [Lan and Pinnavaia, 1994; Lan et al., 1995; Massam et al., 1998] from the group. While most of this work was directed towards production of thermoset CPNC, the possibility of using the two-step intercalation method (involving cation-exchange and epoxidation) for the manufacture of thermoplastic CPNC has also been suggested. Giannelis and Messersmith [1996] used a similar approach. An organically modified smectite (MMT-MT2EtOH; Cloisite® 30B) was dispersed in an epoxy resin together with diglycidyl ether of bisphenol-A (DGEBA). The system was cured in the presence of nadic methyl anhydride (NMA), and/or benzyl-dimethyl amine (BDMA), and/or boron trifluoride monoethylamine (BTFA) at T = 100-200 °C. Exfoliation of smectite was reported with interlayer spacings d001 ≥ 10 nm and good wetting of the silicate surface by the epoxy. The curing involved reaction between epoxy and the alkyl ammonium ions located in the galleries of the organically modified clay. In other words, the reaction resulted in direct attachment of the epoxy network to the silicate layers. The nanocomposite showed higher dynamic storage modulus in the glassy region and very much higher in the rubbery plateau region, when compared to the epoxy resin without clay. The patent focused on the preparation of clay-epoxy nanocomposites. In the year 2000, Ishida et al. [2000] proposed a more general approach to intercalation of smectite clays. It involves the use of either epoxy or PDMS as a ‘swelling agent’ that increases the interlayer spacing of the cation-exchanged smectites. Thus, Na-MMT was first intercalated by cation exchange reaction with either ADA or with HDA. The products were labelled B12 and B16, respectively. The epoxy derived from the DGEBA (Epon 825) is presented in 138
Intercalation of Clay Figure 5. Polymer (88 wt%), intercalated clay (10 wt%), and epoxy or PDMS (2 wt%) were manually mixed for 30 min above either the Tg or Tm of the matrix polymer. The XRD of B12 (no epoxy) showed the interlayer spacing, d001 = 1.6 nm. Upon addition of epoxy the peak became wider and located at a smaller diffraction angle, i.e., d001 increased to 2.3-1.9 nm. Thus, epoxy alone does not exfoliate the onium-intercalated clay. For the CPNCs the percentage of intercalation and/or exfoliation was calculated from XRD patterns as: Exfoliation (%) = 100×[1 - (clay peak area with epoxy)/(clay peak area without epoxy)] The extent of exfoliation was found to increase with the mixing time, but the time required for the completion of the process (from 4 to 120 min) depended on the type of system. The extent of exfoliation may be related to the polymer solubility parameter, δ, for epoxy and PDMS this is respectively, δ ≅ 9 and 8. Unfortunately, the authors do not indicate the mechanism involved in the clay ‘swelling’ by either of these two agents. The method was patented [Ishida, 2000]. 2.3.6.3 Intercalation Utilising Organic Anions As discussed in Section 2.2.4, there are three types of possible interactions that may be used when preparing CPNC: exchangeable cations in the interlayers, -OH groups on the silicate surface, and exchangeable anionic groups at the clay platelet edges. Only recently have the latter groups been targeted for the second step of smectite intercalation. Thus, natural or synthetic clay was first intercalated with a quaternary ammonium ion, and then treated with negatively charged organic molecules able to react with the edge cations. In recent patents from SCP 0.1 to 1.0 wt% (on dry weight of clay) of a high charge density anionic polymer such as a polyacrylate has been used [Powell, 2001a,b]. The clay edge treatment with polyacrylate was carried out before that with a quaternary ammonium ion. In a continuation of the SCP technology (see Section 4.3), polyacrylate was added to an aqueous slurry of MMT, which then was subjected to high shear in a MantonGaulin mill. After shearing the clay was treated with a branched chain quaternary ammonium ion. According to the author, the polymer became strongly attached to the clay edges, making them strongly anionic. In the subsequent treatment an alkyl quaternary ammonium cation not only reacted with the clay-surface anionic charges, but also with the edges, which resulted in uniform, hydrophobic coating, improved dispersability in the plastic matrix of the nanocomposite, and improved properties. In the patent example, PA-66 was melt compounded in a TSE with dry, doubly intercalated MMT. The latter was prepared as described above, with dimethyl hydrogenated tallow -2-ethyl hexyl ammonium methyl-sulfate (2MHTL8), described in a prior patent from SCP [Dennis, 1998]. Polyacrylate (e.g., Alcogum SL-76 or SL-78) was applied at a concentration of about 0.5 wt% by weight of the dry clay. The slurry was dewatered, dried and ground. The extrudates were subjected to wide angle XRD – total exfoliation was observed in CPNC with clay content of ≤5 wt%. 2.3.6.4 Intercalation Utilising Macrocyclic Oligomers (Cyclomers) The two-step intercalation process was used to produce CPNC with polycarbonates, polyesters or polyphenylene ethers as the matrix polymer. The
139
Clay-Containing Polymeric Nanocomposites first step was the customary onium ion exchange, the second incorporation of macrocyclic oligomers of these polymers [Takekoshi et al., 1996]. As in many prior patents, the onium ion could be either ammonium, phosphonium or sulfonium type. For example, MMT (CEC = 1.19 meq/g) was first intercalated with dodecylammonium chloride or N-hexadecyl pyridinium chloride. Next, the organoclay was mixed with a macrocyclic oligomer of terephthalic acid-coethylene-co-butylene glycols, dried under a vacuum and then placed in an oil bath at 190 °C. Finally, dioctyl tin dioctoxide was added to polymerise the oligomer. The melt became viscous in about 15 s and the resulting solid polymer composition comprised exfoliated MMT.
2.3.7 Intercalation by Inorganic Intercalants There is a large body of literature dedicated to the physicochemical modification of clays for use as catalysts or catalyst supports [see for example Pullukat and Shinomoto, 2001; Wypart et al., 2002; Sun et al., 2002]. During the last few years there has been an effort to use the insight gained into the chemistry of clays for the production of inorganically intercalated clays. It is expected that these new intercalated clays will have much greater thermal stability than the currently used ammonium-intercalated systems. During intercalation with inorganic intercalants pillared structures are formed between clay layers. Cations of Al, Zr, Be, Cr, Fe, Ni, Nb, Ta, Ho, and others have been used individually or in mixtures. Since the intercalated clays are subjected to calcination at T ≅ 600 °C and are mainly used in catalysis, high temperature stability has been essential [Mitchell, 1990]. However, even in catalysis this technology is new and (the starting material being a natural product) the reproducibility of created structures is poor. The use of inorganicallyintercalated clays for PNC is in statu nascendi. Hydrothermally stable porous structures were prepared by intercalating a smectite clay (e.g., MMT) with aluminium and rare earth salts, then calcination at T < 760 °C [McCauley, 1989]. The pillars were formed of Al and rare earth element oxides. They expanded the silica plates to an interlayer spacing of 1-5 nm. In 1979 Vaughan et al. [1979] dispersed smectite clay in an aqueous solution of aluminium chlorohydroxide (AlCl2OH) and mixed the suspension for about one hour at about 70 °C. The reacted clay was recovered and heated at T = 200-700 °C to form solid pillars between the clay layers of 0.9-1.8 nm height. Addition of ZrOCl2 increased the interlayer spacing to 2.2 nm. The pillars collapsed when heated to T > 650 °C. Katdare et al. [2000] reported that Na-MMT dispersed in water was mixed with AlCl2OH and subjected to ultrasonic agitation at 50 kHz for 20 min. After washing, the intercalated clay was calcined at 500 °C. Surprisingly, the XRD scan did not show any peak within the expected range of 2θ, which may indicate total exfoliation. This is promising news – hopefully, after some ‘compatibilisation’ step, these exfoliated individual clay platelets may be incorporated into a polymeric matrix. A Dow patent [Nichols and Chou, 1999] is the only one that professes to produce inorganically intercalated clays for the manufacture of CPNC. The intercalation involves organic and inorganic intercalants: 1. The clays of interest include MMT, nontronite, beidellite, hectorite, saponite, magadiite and vermiculite, double or mixed hydroxides, viz. Mg 6 Al 3.4 (OH) 18.8 (CO 3 )1.7H 2 O, chlorides, viz. ReCl 3 and FeOCl, 140
Intercalation of Clay chalcogenides: TiS2, MoS2, and MoS3; cyanides, e.g., Ni(CN)2; and oxides H2Si2O5, V5O13, HTiNbO5, Cr0.5V0.5S2, W0.2V2.807, Cr3O8,MoO3(OH)2, VOPO4-2H2O, CaPO4CH3-H2O, MnHAsO4-H2O, Ag6Mo10O33, etc. 2. The organic intercalant may be a water soluble polymer (e.g., PVAl, PEG, carboxymethyl cellulose, PAA, P4VP); a reactive organosilane compound, a cationic surface active agent such as quaternary ammonium salt having C12 to C18 moieties. The preferred quaternary ammonium cations may have such groups as: octadecyl trimethyl, dioctadecyl dimethyl, hexadecyl trimethyl, dihexadecyl dimethyl, tetradecyl trimethyl and ditetradecyl dimethyl. 3. The inorganic intercalant can be an inorganic polymer obtained by hydrolysing a metallic alcoholate: Si(OR)4, Al(OR)3, Ge(OR)4, Si(OC2H5)4, Si(OCH3)4, Ge(OC3H7), Ge(OC2H5)4, etc. The colloidal particles include the hydrolysed and dehydrated forms, e.g., Si(OH)4 or SiO2, Sb2O3, Fe2O3, Al2O3, TiO2, ZrO2 and SnO2, etc. The size of the colloidal particle should be about 12 nm. It is preferable to modify the multilayered material with a metallic alcoholate, e.g., Ti(OR)4, Zr(OR)4, PO(OR)3, B(OR)3 and the like alone or in combination, with Ti(OC3H7)4, Zr(OC3H7)4, PO(OCH3)3, PO(OC2H5)3, B(OCH3)3, B(OC2H5)3. Metallic chlorides such as TiCl4, metallic oxychlorides such as ZrCOCl2, and nitrate chloride can also be used. 4. The polymeric matrix can be thermoplastic, thermoset or elastomeric. The specifically named polymers include PO (e.g., PP, HDPE, LLDPE, ultra low density PE (ULDPE), EPR, EPDM, ethylene-acrylate acid copolymer (EAA), EVAc), PEST, PA, PC, acrylics (viz. PMMA, methylmethacrylate-butadienestyrene terpolymer (MBS)), styrenics (e.g., PS, styrene-butadiene rubber (SBR), high impact PS (HIPS), styrene-acrylonitrile copolymer (SAN), styrenebutadiene-styrene terpolymer (SBS)), TPU, etc. Prior to intercalating the clay is swollen in an aqueous (e.g., H2O with methanol or butanol) or an organic liquid (e.g., DMF, DMSO, hydrocarbons, halogenated hydrocarbons, benzene, xylene, cyclohexane, toluene, mineral liquids and oils). For example, a mixture of swollen clay and the polymerisable inorganic intercalant can be contacted with a hydrolysing agent for the polymerisable intercalant to form the inorganic polymer. In general, hydrolysis is conducted at T > 70 °C. Following intercalation, the clay is centrifuged and dried at 50-80 °C. The intercalant may optionally be calcined at T = 450-550 °C. Intercalation increases the original interlamellar gallery height of ≤ 0.4 to about 0.5 to 60 nm. Next, the intercalated, layered material can be dispersed in a monomer, which when polymerised would form the polymer matrix. Alternatively, it can be incorporated in the molten or dissolved polymer. Melt blending is the preferred method for preparing CPNC of a thermoplastic polymer. Typically, the polymer is melted and combined with the desired amount of the intercalated clay using an extruder, a Banbury or Brabender mixer, a continuous mixer, etc. Melt blending is carried out in the presence of an inert gas, such as argon, nitrogen or neon. Alternatively, the polymer may be dry mixed with the intercalant, then heated in a mixer and subjected to a shear sufficient to form the desired composite. Sufficient exfoliation is defined as when ≥ 80 wt% of the clay platelets (aspect ratio p = 10-2000) are individually and uniformly dispersed in the polymeric matrix. The patent does not provide any example of CPNC preparation with a specific polymer and there are no numerical values of performance. The claims are very 141
Clay-Containing Polymeric Nanocomposites broad, naming virtually any polymer, viz. a thermoset (e.g., phenolics, epoxy, urethane or urea resin), or thermoplastic polymer (including PO, PC, TPU, styrenics) or a vulcanisable or thermoplastic rubber (e.g., EPR, PU). Similarly, the description of the performance is quite generous. The unnamed CPNCs showed excellent balance of properties, viz. superior heat, chemical or ignition resistance, barrier to diffusion of polar liquids and gases, yield strength in the presence of polar solvents such as water, methanol, ethanol and the like, stiffness and dimensional stability. They are useful in diverse applications viz. in business equipment, computer housings, transport (automotive and aircraft), electronic, packaging, building and construction industry, etc. Bora et al. [2000] developed another method for the manufacture of intercalated, thermally stable organoclay. Since the metal complexes with organic ligands show good thermal stability, the authors expected that MMT intercalated with an appropriate metal complex might show superior thermal stability to that of onium cations. The aim of the work was to intercalate MMT with bulky, three-dimensional cationic metal complexes with ligand X. Two were selected: • Ni-CX1 ≡ [Ni{di-2-aminoethyl amine}2]Cl2 and • Ni-CX2 ≡ [Ni{2,2´:6´,2´´-ter-pyridine}2](ClO4)2. Na-MMT was converted to Ni-MMT by adding NiCl2 to an aqueous suspension of Na-MMT (CEC = 1.14 meq/g). The Ni-MMT could also be complexed with CX1 or CX2, but the structure was different from that obtained by ion exchange between Na-MMT and either Ni-CX1 or Ni-CX2. This reaction was carried out by adding Ni-complexes to an aqueous suspension of Na-MMT and allowing the reaction to proceed for 0.5 h, e.g.: Na-MMT + Ni-CX12+ → [Ni-CX1]-MMT + Na+ The product was separated, washed and then dried at 50 ± 5 °C in an air oven. Analysis of the dried samples indicated that complexes were adsorbed by MMT up to about 0.57 meq/g, i.e., up to stoichiometry. Oriented samples on glass slides were dried at room temperature, 100, 150, 200, 250, 300, 350 and 400 °C for about 1 h. XRD of [Ni-CX1]-MMT showed that on heating d001 decreased from 1.45 at 50 °C to 1.35 nm at 250 °C. However, XRD of [Ni-CX2]-MMT showed d001 = 1.94 at 50 °C, decreasing to 1.90 nm at about 350 °C. Accordingly, the FTIR spectra showed characteristic bands of these complexes up to about 250 and 350 °C, respectively. Thus, the thermal stability of MMT intercalated with Ni-CX1 and Ni-CX2 were found to be about 50 and 150 °C higher compared to that of their metal complex salts. The difference between Ni-CX1 and Ni-CX2 is noteworthy. In the former the ligand is aliphatic, in the other it is aromatic. Ligands with aromatic rings in the main chain show higher thermal stability than the aliphatic. This general observation was confirmed by results for the free complexes and their MMT salts. Furthermore, the interlayer spacing of [Ni-CX1]-MMT was significantly smaller than that for [Ni-CX2]-MMT. This may be due to π-interactions between the oxygen on MMT and the aromatic rings of Ni-CX2.
2.3.8 Melt Intercalation In a sense, melt intercalation is a misnomer – in reality the goal of the process is not so much intercalation as exfoliation of layered inorganic material in a molten
142
Intercalation of Clay polymer. However, since melt exfoliation is rarely achieved, there is some rationale for the term. Melt intercalation is the preferred method for the preparation of CPNC with thermoplastic matrix polymers. Usually, the polymer is melted and compounded with intercalated clay using an extruder, an internal mixer, a kinetic-energy mixer, etc. Compounding is carried out in the presence of an inert gas, e.g., N2. Alternatively, the polymer is first mixed with a compatibiliser (e.g., its functionalised homologue) then compounded with intercalated clay. Industrially, CPNC is considered to be exfoliated when ≥ 80 wt% of clay layers (aspect ratio p = 10-2000) are uniformly dispersed in the polymeric matrix in the form of stacks comprising not more than two platelets. It is remarkable that macromolecules with significantly larger radius of gyration than the interlamellar gallery height: rg2
1 /2
> d001 − 0.96
are able to diffuse into the galleries. Evidently, the process will take place only if it leads to a decrease of the free energy (ΔG), i.e.:
ΔGintercalation = ΔH - TΔS < 0
(15)
The enthalpic (ΔH) contribution usually comes from the chemical interaction between clay and the intercalating compound, whereas the entropic one (ΔS) usually comes from the ‘randomisation’ of the macromolecular segments placement, e.g., diffusion into spaces devoid of macromolecular presence. However, the conformational energy loss caused by chain stretching from the random coil configuration into elongated structures inside the galleys, and associated with this the topographical and energetic constraints (e.g., caused by adsorption of the macromolecules on the clay surface) hinders the diffusion. Thus, it is expected that for successful melt intercalation there must be an enthalpic driving force for the functionalised macromolecules. Furthermore, the interlayer spacing should be large, at least the same order of magnitude as the diameter of the macromolecular Gaussian coil, and sufficient time must be provided for the diffusion process to reach the centre of the stacked platelets. Two relations can be used as guides: 1. The self-diffusion coefficient, Ds ∝ 1/Mn2. 2. The distance travelled by the diffusing macromolecules is proportional to the square root of the diffusion time, viz. ldiff ∝ t1/2. In practice numerous factors influence the process, affecting the validity of these simple relations. The melt intercalation process can readily be divided into static and dynamic. Usually, the static process takes place under vacuum at temperatures at least 50 °C above the transition temperature (Tg or Tm) in the absence of mixing, thus often it is also called melt annealing. Dynamic melt intercalation is more-or-less a standard compounding operation, performed in an extruder, internal mixer and similar processing equipment. 2.3.8.1 Quiescent (or Static) Melt Intercalation There is no commercially viable process that would start with dry, unaltered clay powder and thermoplastics melt and produce exfoliated CPNC. For example, 143
Clay-Containing Polymeric Nanocomposites attempts at a solventless intercalation of Na-MMT or Li-MMT with PEG or PS failed [Vaia et al., 1993]. However, PS was intercalated with an organoclay (Na-MMT ion-exchanged with alkylammonium chloride) by mixing the two powders, pressing the mixture into a pellet, and heating the pellet in a vacuum at 165 °C. XRD showed that PS increased the interlayer spacing by about 0.7 nm, which corresponds to a monolayer of stretched PS macromolecule. Interestingly, the intercalated PS had no Tg within the temperature range from 50 to 150 °C. The lack of chemical bonding was demonstrated by dissolving/dispersing the CPNC in toluene and quantitatively extracting the PS. For PEG the static intercalation (heating the polymer/Na-MMT mixture at 80 °C) was slow, requiring from 2 to 6 h. The achieved final interlayer spacing was modest, d001 = 1.77 nm [Vaia et al., 1995a]. The method failed for PS – it was found that prior to melt intercalation the MMT had to be modified by treatment with dimethyl tallow organosilicate that expanded the interlamellar gallery to at least 0.7 nm. Furthermore, the process led to intercalation, but not to exfoliation. PDMS-base CPNC was prepared by ultrasonication of organoclay (MMT-2M2HTA) with silanol-terminated-PDMS. The system was crosslinked using a mixture of tetraethyl orthosilicate (TEOS) and tin 2-ethyl hexanoate. Exfoliation was facilitated by the presence of water at a concentration corresponding to about a monolayer on the clay surface [Burnside and Giannelis, 1995; 2000]. In the latter publication exfoliation in the crosslinked PDMS/dimethylditallow-MMT system was confirmed, but only intercalation (d001 increased from 2.5 to 3.9 nm) was obtained when a copolymer, poly(dimethyl siloxane-co-diphenyl siloxane), with 14-18 mol% of diphenyl, was used instead of PDMS. Furthermore, not even intercalation was achieved for either polymer with Na-MMT. In another study, organoclay was prepared by reacting, e.g., Na-MMT with 2M2ODA [Vaia et al., 1994] or Li fluorohectorite (FH) with ODA [Vaia et al., 1995b]. In the latter study, 25 mg of organoclay was mixed with 75 mg of PS and formed into pellet, which in turn was heated at T > Tg(PS) ≅ 100 °C. Melt intercalation took place in a quiescent system as a function of the annealing temperature (Ta = 155-180 °C), annealing time (ta ≤ 400 min), and molecular weight of the intercalating PS (MW = 30-300 kg/mol). The extent of intercalation was determined from XRD. Thus the interlayer spacing of organoclay (d001 = 2.13 nm) during annealing with PS (having MW = 30 kg/mol) progressively increased to d001 = 3.13 nm. The amount of polymer (Q) that diffused into the interlamellar galleries over time (t) followed the dependence: Q(t ) / Q∞ = 1 −
∑ (4 / α ) exp{−( D / a )α ∞
2
m
m −1
mt
}
(16)
where: αm is the m-th positive root of the Bessel zeroth-order function, and D/a2 is the effective diffusion rate. Its temperature dependence was found to follow the Arrhenius equation with the activation energy ∂ln(D/a2)/∂(1/T) = Ea ≅ 166±12 kJ/mol. The molecular weight dependence was found to follow the dependence: (D/a2) ∝ 1/MW1.6. The authors observed that intercalation by PS of commercial interest (MW = 300 kg/mol) at T = 250 °C should take about 10 min. The melt annealing would not result in exfoliation. There is no information about how the terminal interlayer spacing depends on the molecular weight of the intercalating polymer. In the following contribution by the same authors [Vaia et al., 1996] the quiescent melt intercalation was conducted not only with PS (PS30 and PS400 with MW = 30 and 400 kg/mol, respectively) but also with poly(3-bromo styrene) 144
Intercalation of Clay (PS3Br; MW = 55 kg/mol). The microstructure of the organoclay/polymer systems was examined by XRD and high-resolution transmission electron microscopy (HRTEM). In contrast to the averaging of XRD, HRTEM provides information on local structure, spatial distribution of structural elements and defects. When PS30 was used, the same interlayer expansion was observed as that reported earlier, viz. d001 = 2.13 increasing to 3.13 nm. The HRTEM showed the presence of ordered clay microstructures similar to that observed for nonintercalated (by PS) organoclays, but slightly expanded. The publication in part answers the earlier question as to whether the final interlayer spacing depends on the polymer molecular weight. Partial intercalation by PS400 resulted in similarly ordered structure as that observed for PS30, but the intercalated regions expanded more, viz. the measured interlamellar spacing of 2-3 nm, compared to 2.17 determined for PS30. Since the unperturbed radius of gyration for PS30 and PS400 coils is: 1/2 ≅ 5 and 18 nm, respectively, in neither case would the interlayer spacing accommodate the Gaussian coil in 3D. Intercalation by PS400 was only partial. The HRTEM micrographs showed domains of intercalated and of non-intercalated (by PS) clay platelets with diameter ranging from 0.05 to 0.5 μm formed by stacks of 50 to 100 parallel platelets. In the intercalated and non-intercalated domains the number of layers in the stack was similar – intercalation by PS just pushed the layers further apart. On the basis of the HRTEM information it was concluded that polymer intercalation occurs as a front that diffuses into organoclay interlamellar galleries from the external edges. The non-intercalated (by PS) domains were more prevalent towards the centre of the stack. Structural defects in the clay crystalline structure and irregularities in the layer stacking have been observed. In contrast with the PS-intercalated layered stacks, intercalation with PS3Br resulted in exfoliation, evidenced by the lack of a well-defined XRD peak as well as by the HRTEM micrographs. Evidently, the presence of -Br brings about stronger interaction between the intercalated clay and the polymer than that existing between clay and PS. Since the clay was intercalated with the primary octadecyl ammonium, interaction between -Br and either C18-NH3+ or with -OH groups at the platelets edges is possible. As observed under HRTEM, individual platelets were intermixed with small stacks of 10 to 20 platelets with the interlayer spacing ranging from 2.1 to 6.0 nm. The heterogeneous microstructure indicates that the formation of these systems occurs by a more complex process than simple sequential intercalation of individual layers starting from the surface of the primary particle. The clay crystalline structure may contain defects or chemical inhomogeneities that facilitate polymer transport or result in enhanced polymersilicate interaction. Long-range electrostatic forces tend to maintain layer stacks. However, exfoliation most likely takes place layer-by-layer from the limits of the original organoclay stack. The earlier studies of PS/organoclay [Vaia et al., 1993] led to the conclusion that PS macromolecules form monolayers inside the organoclay galleries. This significantly changes the chain dynamics, e.g., resulting in absence of Tg for PS within the expected range of 50-150 °C. Evidently, the presence of the onium compound in the galleries must have influenced the macromolecular behaviour, but the effects of 2D steric restriction must have also affected the chain dynamics. Krishnamoorti et al. [1996] melt intercalated lightly deuterated PEG (Mw = 180 kg/mol, Mw/Mn = 1.2) into Li-FH. The system was probed using thermally stimulated current (TSC), XRD, DSC, SANS and solid-state NMR. 145
Clay-Containing Polymeric Nanocomposites PEG chains were confined to 0.8 nm galleries. The 2H-NMR spectra were obtained at the same temperatures for the bulk and the intercalated d-PEG. As the temperature increased from 220 toward 250-270 K, a central peak of the intercalated d-PEG developed earlier and became narrow. Thus, the enhanced local chain dynamics of the intercalated d-PEG at low T may have resulted from the absence of chain entanglements. However, within the high T-range (T = 320-340 K), the intercalated d-PEG showed a broad base structure, whereas the bulk d-PEG had a single narrow signal, indicating that the silicate layers restrict macromolecular chain motion. Intercalation into the narrow interlamellar space with height of 0.8 nm must have engendered interactions (hydrogen bonding?) between the PEG segments and the platelet surface. Furthermore, the narrow space between the clay layers topologically restricted the d-PEG chains. The DSC experiments on an intercalated CPNC indicated the absence of any thermal transitions of PEG; Tg or Tm. However, TSC suggested that the transition, which in bulk PEG takes place at about Tg = -55 °C, upon intercalation increased to Tg ≅ 60 °C. The TSC peak for the intercalated system was broad and shallow, indicating a low level of molecular cooperativity near Tg. The most comprehensive studies of the quiescent intercalation of organoclay by molten polymers included: several intercalating organic cations, synthetic and mineral clay, polymers (different chemically and with different molecular weight) and annealing temperature [Vaia, 1995; Vaia and Giannelis, 1997a,b]. The material characteristics are given in Table 18.
Table 18 Material characteristics for quiescent melt intercalation [Vaia and Giannelis, 1997b] Component
Code
Characteristics
Clays
F S M
Li+-fluorohectorite; CEC = 1.5, a = 0.39, Corning Inc. Li+-saponite; CEC = 1.0, a = 0.58, Southern Clay Products Na+-MMT; CEC = 0.8, a = 0.72, Swy-1, U. Missouri, Source Clay Repository
Ammonium salts
2C18 Q18 Cn
Dimethyl dioctadecyl ammonium bromide (2M2ODA) Tri-methyl octadecyl ammonium bromide (3MODA) n-Alkyl-ammonium chlorides (CnH2n+1NH3+Cl-, with n = 6, 9-16, and 18)
Polymers
PS30 PS90 PS400 PS3Br PVCH PVP
Polystyrene; Mw = 30, Mw/Mn = 1.06, Tg = 96 °C Polystyrene; Mw = 90, Mw/Mn = 1.06, Tg = 100 °C Polystyrene; Mw = 400, Mw/Mn = 1.06, Tg = 100 °C Poly-3-bromo styrene; Mw = 55, Mw/Mn = 2, Tg = 113 °C Poly(vinyl cyclohexane); Mw = 97, Tg = 120 °C Poly(2-vinylpyridine); Mw = 150, Mw/Mn = 2.1, Tg = 104 °C
Notes: CEC = cation exchange capacity (meq/g) a = clay surface area per cation (nm2) Mw = weight-average molecular weight (kg/mol) Tg = glass transition temperature (°C)
146
Intercalation of Clay As before, dry organosilicate (25 mg) and polymer powder (75 mg) were mechanically mixed and formed into a pellet using a hydraulic press at a pressure of 70 MPa. Melt intercalation was carried out by annealing the pellet in vacuum at T > Tg. Usually the samples were annealed to equilibrium, indicated by no further changes in the X-ray diffraction pattern. The work brought out several important observations: 1. Of the three clays the largest value of d001 was systematically observed for F and the smallest for M, having, respectively, the highest and the lowest CEC value. Note also that for the three clays there is a difference in the distribution of the anionic charge on the silicate surface. The increased localisation of surface charge for tetrahedrally-substituted S does not appear to affect CPNC formation when compared to the more dispersed surface charge of octahedrally-substituted M. 2. Li+-fluorohectorite (FH, clay F) was intercalated with the primary ODA (F18) and subsequently annealed with each of the three PS. XRD showed intercalation with d001 = 3.11 nm. However, while the intercalation for PS30 was achieved after annealing for less than 6 h at 160 °C, to intercalate with PS90 took 24 h, and PS400 more than 48 h. Since the XRD peak was quite similar for the three polymers it was concluded that the kinetics of intercalation are strongly affected by the polymer molecular weight, but that the structure of CPNC prepared under quiescent conditions is independent of it. 3. The ammonium salt used in the cation exchange may have an important effect on intercalation by molten polymer. Two aspects are worth stressing: a. At T = 160 °C PS30 was unable to diffuse into the interlamellar galleries of the F-clay intercalated with alkyl-ammonium chlorides having n ≤ 12. Intercalation with alkyl ammonium with n = 6 to 12 engendered the same initial expansion of the interlayer spacing, viz. d001 ≅ 1.8 nm (see also Figure 23), but apparently alkyls with n < 12 do not sufficiently shield the hydrophilic clay surface to assure PS interpenetration. Annealing with PS30 at T = 180 °C expanded even F8 to the maximum spacing d001 ≅ 3.1 nm, i.e., increasing the gallery height by about 1.1 nm. This maximum value was found to be independent of the alkyl cation, of the annealing temperature, and of the PS molecular weight. b. The difference between primary and quaternary ammonium cations can be evaluated by comparing the intercalation of PS30 into FC18 (primary) with that into FQ18 (secondary) and F2C18 (quaternary). For the three compositions the interlayer spacing was found to be, respectively, d001 ≅ 3.11, 3.64 and 3.80 nm. Thus, all three cations facilitated the secondary intercalation by PS, but different types of ammonium cation did influence the final gallery height. The difference could originate in the difference in the ionic interaction strength for the primary, secondary and quaternary onium ions as well as in the size of the quaternary group. 4. Melt intercalation also depends on the chemical nature of the intercalating macromolecules. Thus, in this series of styrene-derivative polymers with M2C18 the interlayer spacing increased with the relative acid/base character and the side-group polarisability in the order: poly(vinyl cyclohexane) (PVCH) 4.4 nm. The presence of short stacks in both products indicated that exfoliation was not complete. Improved mechanical properties were reported, e.g., addition of 3.16 wt% of inorganic clay resulted in 38% higher modulus and 30% higher strength (than PA-6) keeping the Izod impact strength about the same. Intercalated and exfoliated CPNC of PA-6 with organoclay were prepared in a TSE at T = 250 °C [Varlot et al., 2001]. The organoclay was Na-MMT cationexchanged with methyl octadecyl bis-2-hydroxyethyl ammonium methyl-sulfate (MODA2EtOH) and dimethyl dioctadecyl ammonium chloride (2M2ODA), which resulted in the interlayer spacing of d001 = 1.96 and 3.68 nm, respectively. After compounding, the CPNC specimens were found, respectively, exfoliated and intercalated into short stacks containing 2-4 MMT platelets. The nanocomposites were anisotropic. The preferential orientation of the MMT platelets as determined by SAXS was parallel to the flow direction during injection moulding. The orientation of the MMT plates and the PA-6 lamellae is expected to play a major role in the mechanical properties. In the CPNC the matrix crystallised preferentially in the γ-form. 2.3.8.2.5 Melt Intercalation in PEG Matrix In his doctoral thesis of 1995 Vaia [1995] described the preparation of polymer electrolyte nanocomposites for batteries. Thus, PEG was melt intercalated into Na-MMT or Li-MMT with CEC = 0.8 meq/g. The process was carried out by grinding a mixture of, e.g., 28 wt% of PEG (MW = 100 kg/mol) with 72 wt% Na-MMT, then compressing it into a sheet at 80 °C for up to 6 h. The final d001 spacing was 1.77 nm, the same as that obtained in the solution intercalation method. Considering the high solid loading larger interlayer spacing could not be expected. Judging by the cited patents from AMCOL, the intercalation process is facilitated by the presence of water (from 6 to 10 wt%), and increasing the PEG content may increase the interlayer spacing. 2.3.8.2.6 Melt Intercalation in PO Matrix Development of CPNC with a PA as a matrix has been carried out since the late 1960s. Owing to the polar nature of these polymers and advances in the chemistry of the ammonium ion, at present melt exfoliation in the PA-matrix is conducted on an industrial scale. By contrast, the melt intercalation of clays in a PO-matrix is at a relatively early stage of development. The information from the patent literature will be discussed later. Here the focus will be on the more basic aspects of PO-melt intercalation technology. By contrast with PA or PEST, POs are nonpolar, nonreactive and sensitive to thermomechanical oxidation on the tertiary carbon by the free radical mechanism. The fundamental question remains whether it is at all possible for PO macromolecules to diffuse into the hydrophilic, repulsive environment of the clay interlamellar galleries. As with PA, here also most of the work involves MMT modified with organic compounds. Because of the difficulties encountered during polymerisation in the presence of clay [Heinemann et al., 1999], melt intercalation is the preferred 165
Clay-Containing Polymeric Nanocomposites technique. The large resin manufacturers are pursuing the development of the polymerisation-intercalation method with moderate success. During the last decade, PP technology underwent a dramatic evolution that resulted in a rapid increase of PP application in various industries, including transportation. Thus, it is of no surprise that PP-based CPNCs have been of interest to researchers around the globe. The early work on PO-based CPNC was published by the Toyota group. Its strategy was based on a multistep intercalation: 1. Primary intercalation of clay by cation exchange with onium salt; 2. Secondary intercalation with a compatibiliser, e.g., PO macromolecules grafted with a polar compound (such as MAH); and 3. Melt compounding with the matrix PO. The two critical steps are the selection of the organic cation (able to react with clay anions and interact with the compatibiliser), and selection of the polar compatibiliser, which will react with organoclay and remain miscible with the basic resin. Some success was also obtained by combining steps 2 and 3 into one, by modifying the PO only slightly, and compounding it directly with organoclay as a polar matrix. In studies by Usuki et al. [1997] (which provided the basis for the first patents in this field [Usuki et al., 1996; 1999]) Na-MMT (CEC = 1.19 meq/g) was intercalated with 45.8 wt% of dimethyl distearyl ammonium chloride (2M2ODA), which increased the MMT interlayer spacing to d001 ≅ 3.28 nm. Next, the organoclay was suspended in toluene and blended at a weight ratio of 1:1 with the compatibilising ‘main guest molecule’ (hydroxylated oligo-olefin: OO-OH; e.g., hydrogenated polybutadiene with telechelic -OH groups, Mw ≅ 3 kg/mol, Polytail H from Mitsubishi) then dried. The interlayer spacing increased to d001 ≅ 3.87 nm, but when the ratio of OO-OH-to-organoclay increased to 10 the XRD peak disappeared altogether, indicating that d001 ≥ 8.8 nm. Still better results were obtained by either reducing the molecular weight of the OO-OH compound, or by replacing the -OH groups by either -COOH, acid anhydride or epoxy (Polytail EP from Mitsubishi Kagaku). Next, PP (Mw = 30 kg/mol) was melt-compounded in an internal mixer at 220 °C with either: Na-MMT, the MMT-2M2ODA organoclay or the exfoliated OO-OH/organoclay material. The TEM images taken of these three compounds showed, respectively, a micron-level dispersion, the presence of large stacks, and exfoliation with only short stacks. Thus, good dispersion of clay platelets in molten PP was achieved when prior to compounding the clay platelets were exfoliated by the combined effects of onium ion and the OO-OH compatibiliser. The authors stressed good miscibility of the MMT-2M2ODA organoclay with PO, e.g., PP, butyl rubber (BR), polyisoprene (IR), polybutadiene, etc. In another publication [Kato et al., 1997] two-step exfoliation was carried out. Thus, Na-MMT was intercalated with octadecyl ammonium chloride (ODA), which increased the MMT interlayer spacing to d001 = 2.17 nm. Melt blending it with either maleated PP (PP-MA; Mw = 30 kg/mol, acid value = 52 mg KOH/g) or hydroxylated PP (PP-OH; Mw = 20 kg/mol, OH value = 54 mg KOH/g) was carried out in an internal mixer at 200 °C for 15 min, with the ingredient ratio changing from 1:3 to 3:1. The interlayer spacing of CPNC containing 50 wt% of organoclay was d001 = 3.82 and 4.40 nm for the PP-MA and PP-OH matrices, respectively. However, for an organoclay concentration of 25 wt% d001 = 7.22 nm was 166
Intercalation of Clay obtained. Note, that the PP-MA matrix had lower molecular weight than commercially desirable, but still above the entanglement molecular weight: (Mw > Me = 2.8 kg/mol [Porter, 1995]). Furthermore, under the same conditions the same resin with lower acid value (PP-MA; Mw = 12 kg/mol, acid value = 7 mg KOH/g) was unable to expand the organoclay spacing. The authors concluded that to obtain intercalation one polar group per 25 PP mers is required. Unfortunately, there is no explanation of the MAH group reaction(s) with the organoclay – is it with the primary ammonium ion, with clay surface -OH groups, or the surface cations at the platelet edges? In the following publications from Toyota [Kawasumi et al., 1997; Hasegawa et al., 1998] again a three-step melt intercalation was used. Thus, Na-MMT was first intercalated with ca. 32 wt% ODA, which increased the interlayer spacing from d001 = 1.2 to 2.2 nm. The same procedure was repeated with synthetic fluoromica (FM) (Somasif ME100 from CO-OP Chem.). As the ‘main guest molecule’ two grades of PP-MA were used (Yumex 1001 and 1010 from Sanyo, with Mw = 40 and 30 kg/mol, acid number = 26 and 52 mg KOH/g, Tm = 154 and 145 °C, respectively). The powders of organoclay (7.3 wt%), PP (melt flow rate MI = 16 g/min; 70.8 wt%) and Yumex (21.9 wt%) were dry-blended, then compounded in a TSE at 210 °C. About 5 wt% of the clay (MMT or FM) was incorporated. The miscibility of PP with PP-MA was examined at 200 °C by means of an optical microscope – Yumex 1001 was found miscible while Yumex 1010 was immiscible. The degree of clay dispersion in injection-moulded specimens was determined by XRD and TEM. It was found that compounding organoclay with PP caused a reduction of the interlayer spacing, while that with either PP-MA increased it to d001 = 5.9 and 6.4 nm for MMT and FM, respectively. In three-component systems (organoclay/PP-MA/PP) a significant difference in the degree of dispersion was observed for systems with Yumex 1001 and 1010. In the first case, the miscibility of the PP/PP-MA blend resulted in a high degree of dispersion (displayed in TEM micrographs), but with residual short stacks detected as a shoulder on the XRD (d001 ≅ 3.3 nm). In the case of the immiscible PP/Yumex 1010 system the XRD peaks indicated reduced interlayer spacing to d001 ≅ 5.5 and 5.9 nm for MMT and FM, respectively. The authors stated that there is ‘strong hydrogen bonding between the maleic anhydride groups and the oxygen groups of the silicates’, suggesting that MAH groups react with the clay surface -OH functionalities. Evidently, preparation of CPNC with such non-polar polymers as PP requires thermodynamic compatibilisation. The ‘compatibilising’ molecules (e.g., PP-MA) must be selected considering several aspects: 1. Interaction with pre-intercalated clay; 2. Miscibility with the matrix polymer; 3. Ability to entangle with the matrix, and 4. The effects on the crystallinity of the matrix. As Kawasumi et al. have shown [1997] the miscibility of PP-MA with PP is of great importance. As expected, when the MAH content exceeds a critical acid value (≤ 52 mg KOH/g) the immiscibility with PP results in a micron-size dispersion of PP-MA/organoclay phase in PP. Thus, a high degree of maleation leads to immiscibility and poor results. Similarly [Kato et al., 1997], when the degree of maleation is too small, there is insufficient interaction with the organoclay and 167
Clay-Containing Polymeric Nanocomposites again platelet dispersion is poor. However, miscibility depends on concentration, temperature and pressure [Utracki and Kamal, 2002b]. Furthermore, the structure of the copolymer is important. When maleation is conducted in the presence of a relatively high concentration of free radicals, random grafting of individual MAH groups onto PP short chains takes place. Grafting PP with a small amount of free radicals in the presence of styrene may lead to high molecular weight PP with few SMA-chains attached to it. Significant differences in the compatibilising capabilities of these two types of PP-MAH compounds are to be expected. It is noteworthy that the commercially available PP-MA compounds widely differ in purity, polydispersity and type of grafting hence their performance in CPNCs vary widely. Hasegawa et al. [2000a] explored the earlier observation from the Toyota laboratory that melt compounding of lightly maleated PO (compatibiliser not needed) with pre-intercalated clay may lead to a high degree of dispersion. Thus, ca. 5 wt% of organoclay (MMT-ODA) was melt-incorporated into each of the three, slightly maleated POs: PP (Mw = 210 kg/mol; MI = 150 g/10 min; MAH = 0.2 wt%), PE (MI = 2 g/10 min; MAH = 0.62 wt%), and EPR (Mw = 270 kg/mol; MI = 0.06 g/10 min; MAH = 0.55 wt%). The blending was carried out at 200 °C in a TSE (D = 30 mm, L/D = 45.5). Excellent dispersion of clay platelets was shown by XRD and TEM (in decreasing order) for EPR, PP and PE. In the latter system residual short stacks with d001 ≅ 3 nm were detected. The authors also showed that under the same compounding conditions, non-maleated PO gave a coarse, ca. 1 μm in size, dispersion of organoclay aggregates with d001 ≅ 2.2 nm. Next, the effects of organoclay concentration on melt intercalation were studied [Hasegawa et al., 2000b]. Again, Na-MMT (CEC = 1.19 meq/g) was intercalated with ODA, which increased the interlayer spacing from d001 = 1.2 to 2.2 nm. The organoclay (2.1 to 5.3 wt%) was melt compounded with PP-MA (Mw = 209 kg/mol; acid number = 2.1 mg KOH/g; MAH = 0.2 wt%) in a TSE at 200 °C. Good dispersion was obtained – the XRD spectra were free of d001 peaks, but a small shoulder for the highest clay content indicated the presence of short stacks. The tensile modulus increased linearly with clay content (by nearly a factor of two), whereas the increase of tensile strength was more modest (from 21 to 25 MPa), stabilising at about 4 wt% clay loading. Compounding of organoclay with non-maleated PP gave a coarse dispersion of organoclay aggregates with unchanged interlayer spacing (d001 = 2.2 nm), higher modulus, but lower elongation at break and strength than the neat resin. A three-step procedure was successfully used for melt intercalation of PP (MI = 1.5 g/10 min) with PP-MA (MAH = 1 wt%, MI = 1.2 g/min) and MMT intercalated with 2M2ODA [Zhang et al., 2000b]. Mixing PP/PP-MA (ratio 50:1) with organoclay was carried out in a TSE at 200 °C. As a result, the interlayer spacing of the organoclay (d001 = 1.74 to 3.6 nm) progressed towards exfoliation, leaving only a small shoulder in the XRD spectrum (short stacks with d001 ≅ 3.4 nm). Tensile and impact strength at T = 5 °C were determined (see Figure 33). Thus, incorporation of organoclay increased the tensile strength from about 29 to 31 MPa. The effect of organoclay on impact strength is more difficult to judge, as addition of 0.1 wt% organoclay increased it from 9 to 26 kJ/m2, whereas further addition of organoclay erratically reduced it to 6 kJ/m2. One may suspect that PP-crystallinity and/or PP/PP-MA miscibility plays a role. More recently, the effects of additional shearing were investigated [Zhang et al., 2003]. During the packing stage of injection moulding an oscillatory flow through the mould cavity 168
Intercalation of Clay
Figure 33 Tensile and impact strength (at 5 °C) versus organoclay content for PP/PP-MA/organoclay [Zhang et al., 2000b].
was generated. The CPNC was partly intercalated – shearing improved the degree and homogeneity of dispersion. However, this improvement was at a cost of platelet orientation in the flow direction. Some bent MMT platelets were observed under TEM. As a result the improvement of tensile strength was modest, decreasing with clay loading from positive 19% to negative 19% for 1 and 10 wt% of organoclay, respectively. Several modifications of these procedures have been proposed. For example, the interlayer spacing of an ammonium-intercalated MMT was expanded by swelling with an organic solvent with boiling point: BP = 100-200 °C, viz. ethylene glycol, naphtha or heptane [Wolf et al., 1999]. The swollen organoclay was then compounded with PP in a TSE at 250 °C, replacing the solvent molecules by PP segments and evacuating the solvent. The extent of dispersion is unknown, but the authors reported absence of an XRD peak in the 2.0-4.0 nm range. The aspect of the thermal stability of the organic cation used to intercalate Na-MMT was also analysed [Lee et al., 2000a]. PP (Mw =278, Mn = 72 kg/mol), PP-MA (Mn = 23 kg/mol, 2 wt% MAH) and organoclay (Cloisite® 20A or 30B, i.e., MMT with 2M2HTA or MT2EtOH, respectively) were melt-mixed for 10 min at 210 °C in an internal mixer at the weight ratio of about 70:22:8, respectively. The XRD of the mixtures with C20A showed that the interlayer spacing increased from that of organoclay (d001 = 2.47 nm) to 3.4 nm for C20A with PP-MA, and to full exfoliation for the three-component CPNC. The opposite trend was seen for mixtures with C30B. After compounding with PP-MA and/or with PP/PP-MA the C30B interlayer spacing (d001 = 1.86 nm) was reduced to about 1.4 nm. Furthermore, it was found that compounding under a blanket of N2 or in air gave different results. The effect was also observed for binary PP mixtures with organoclay (without PP-MA) where a change of the interlayer spacing was not 169
Clay-Containing Polymeric Nanocomposites expected. The FTIR measurements indicated thermal desorption and decomposition of organic ions in C30B. By contrast, C20A heated with PP retained its relatively large spacing, which in the presence of PP-MA expanded all the way to exfoliation. The C30B spacing is small to start with and upon heating it readily collapses, preventing PP-MA diffusion into interlamellar space. This observation is particularly interesting since C30B melt-exfoliated during melt compounding with PA-6 in a TSE at T = 200-250 °C [Dennis et al., 2000]. Evidently PA chain ends were able to react with the -OH groups of the intercalant either preventing their decomposition or forming their own PA-clay bonds through amine or amide groups. Cloisite® C6A, C20A, C25A, and C30A (see Table 16) were also used with three PP resins of different melt index (PP-L, MI = 820; PP-M, MI = 200; and PP-H, MI = 60) and four PP-MA resins (unknown MW) containing 0.55, 1.1, 3 and 5 wt% MAH [Kim et al., 2000a]. First, PP-L was mixed with 5 wt% of organoclay in an internal mixer at 200 °C and 20 rpm for 10 to 30 min. In systems containing C20A (MMT-2M2HTA) the interlayer spacing increased by 0.73 nm, while for the three other organoclays it decreased by 0.1 to 0.2 nm. However, since the magnitude of the XRD peak decreased on mixing, some exfoliation may occur. Next, three component mixtures: PP:PP-MA: Cloisite® = 80:15:5 wt%, were prepared. The best results were obtained using PP-L and PP-MA containing the lowest concentration (0.55 wt%) of MAH. Apparently, the phase separation at higher MAH content interfered with exfoliation. In another report [Ryul et al., 2001] PP-M with Cloisite ® C10A (2MBHTA) or C20A (2M2HTA) was compounded in an internal mixer (T = 180 °C, 75 rpm, t = 20 min) equipped with an ultrasonic wave generator. In this series of experiments the PP-MA was replaced by styrene. The role of ultrasonics was to generate free radicals as well as to help in breaking the organoclay stacks, thus facilitating intercalation. A high degree of dispersion was obtained in the case of PP-M:styrene:C10A = 80:15:5 wt%; for C20A the d001 peak shifted from 2.39 to 3.65 nm. Ultrasonics caused chain scission of PP followed by copolymerisation with styrene. The superiority of C20A (2M2HTA) over the other organoclays indicates that the most favourable conditions for achieving a high degree of dispersion are: 1. A relatively large initial interlayer spacing of organoclay that makes diffusion of the compatibilising molecules possible, 2. A relatively small area of the clay surface blocked by the intercalating ion, 3. Availability of sufficient concentration of -OH groups on the clay surface, 4. Chemical bonding between the compatibiliser reactive group (e.g., MAH) and ≡Si-OH group, 5. Miscibility of the compatibiliser with the base resin, and 6. High enough molecular weight of the compatibiliser to obtain entanglement with the matrix molecules. The above observations are also valid for CPNC prepared with synthetic hectorite (FM, Somasif ME100 from CO-OP Chem.) [Reichert et al., 2000]. First, the synthetic clay (FM) was treated with protonated alkyl amines, viz. butyl (C4), hexyl (C6), octyl (C8), dodecyl (C12), hexadecyl (C16) and octadecyl (ODA). The primary ammonium ion intercalation resulted in expansion of the interlayer spacing from 0.95 to 1.98 nm (see Figure 34). PP (Tm = 163 °C; MI = 3.2 g/10 min)
170
Intercalation of Clay was compounded with 20 wt% of PP-MA (type E: Mw = 22.9 kg/mol, Mw/Mn = 2.6, MAH = 2.9 wt% or type H: Mw = 32 kg/mol, Mw/Mn = 8, MAH = 4.2 wt%) and 0, 5 or 10 wt% of the organoclay in a CORI at 300 rpm and T = 190 to 230 °C. Only C12, C16 and C18 amines produced organoclays (d001 ≥ 1.68 nm) that could be dispersed in the compatibilised PO matrix. The interlayer spacing was found to increase with MAH content, viz. 10 wt% of FM intercalated with either C16or C18-amine, when blended with 20 wt% PP-g-MA (type-H) and 70 wt% PP gave the best performance, viz., increase of Young’s modulus by 129%, yield stress increase by 32%, notched Izod impact strength decrease by 18% and elongation at break decrease by 99%. XRD and TEM indicated that full exfoliation was not achieved. The miscibility of PP-MA with PP was not examined. Manias et al. [2001] reacted Na-MMT with either 2M2ODA or ODA obtaining two organoclays. PP was grafted with ca. 0.5 wt% of p-methylstyrene (PP-MS; Mw = 200 kg/mol), then the p-methyl group was converted into either -(CH2)3-OH (PP-OH) or -CH2-maleic anhydride (PP-MA). A block copolymer of PP with PMMA was also prepared (PP-b-PMMA; Mw = 15 kg/mol). CPNCs were: (i) Extruded at 180 °C using a SSE with residence time up to 10 min, (ii) Compounded in an internal mixer at 170 to 180 °C for up to 20 min, or (iii) Ultrasonicated in trichlorobenzene.
Figure 34 Synthetic clay (Somasif ME10) intercalated with salts of primary amines, CnH2n+1NH2. The interlayer spacing (d001) and amine content (w) are plotted as functions of the number of carbon atoms in the amine (n). The amount of adsorbed amine and the stepwise increase of d001 may be approximated with the linear dependence with the correlation coefficient, r = 0.980 and 0.996, respectively. Data [Reichert et al., 2000].
171
Clay-Containing Polymeric Nanocomposites According to XRD and TEM all three methods resulted in similarly structured CPNC: intercalated tactoids and exfoliated platelets. Thus, reacting MMT2M2ODA with PP-MA resulted in expanded interlayer spacing from d001 = 2.4 to 3.0 nm, whereas mixing it with either PP-MS or PP-OH gave d001 = 3.4 nm. The MMT-ODA was found to expand when blended with PP-b-PMMA from d001 = 2.2 to 3.2 nm. The short note does not contain either the postulated mechanism of reaction between the functionalised PP and the organoclay or information about how the chemical structure and blending time affected the exfoliation. It seems that in the series of MMT-2M2ODA with grafted PP the interlayer expansion was controlled by the presence of the styrenic group – the presence of MAH gave lower expansion than that obtained from the non-polar -CH3 group. MMT-ODA with PP-b-PMMA also showed a similarly modest expansion of the interlayer space by Δd001 ≅ 1 nm. Oya et al. [2000] examined the role of clay as well as a secondary intercalant on the degree of dispersion and mechanical properties of CPNC with PP. Thus, synthetic hectorite (FM), MMT and mica (MC) were intercalated with a quaternary ammonium ion (2MHDODA) of the formula: (CH3)2N+(C16H33)(C18H37). Next, the organoclay was dispersed in a toluene solution of diacetone acrylamide (CH2=CHCONHC(CH3)2CH2COCH3; DAAM) with a free radical AIBN initiator. After polymerisation of DAAM a solution of PP-MA was added, the product was precipitated, washed, dried, and finally melt compounded with PP. In parallel, CPNC was similarly prepared without DAAM. In all cases only intercalation was obtained – the largest interlayer spacing (and the overall best dispersion) was obtained for doubly intercalated FM prepared with poly-DAAM. However, the best mechanical properties at 3 wt% of clay loading were these of CPNC with MC. The authors speculated that the difference originates in the relatively high modulus of MC pristine platelets. In conclusion, mica may be a better nanofiller than FM, but the performance of CPNCs with MMT was nearly as good. The additional polymerisation step did not result in the expected improvement of dispersion and/or mechanical performance. Intercalated nanocomposites of PP-MA with 2, 4 and 7.5 wt% clay were prepared via melt extrusion at 200 °C in a TSE [Nam et al., 2001]. The organoclay was synthesised by an ion exchange reaction between Na-MMT (CEC = 1.10 meq/g; d001 = 2.31 nm after ion exchange) and ODA. The PP was modified by grafting with 0.2% MAH (PP-MA; Mw = 195 kg/mol and Mw/Mn = 2.98). It was found that the interlayer spacing (as measured by XRD) decreased with the clay concentration (from d001 = 3.24 to 2.89 nm), while the thickness of the short clay stacks increased (see Figure 35). The XRD and TEM data indicated that the extent of exfoliation and/or stacking is controlled by the amount of clay. The grafted MAH-groups in the PP-MA chains promoted interactions with the clay particles by diffusion of PP chain into the space between the silicate galleries. The authors focused on the effect of the organoclay on PP-MA crystallinity and hierarchical structure formation. The hierarchical structure starts with the intercalated clay spacing of 2¯3 nm, to crystalline lamellae thickness of 7-15 nm and spherulitic texture of 10 μm diameter. Two important observations were made. After crystallisation at 80 °C, the PP-MA formed a rod-like crystalline texture of about 10 μm length consisting of an interfibrillar structure. The total volume fraction of the crystalline phase was about constant, but with the increasing clay content the γ-phase crystalline content increased from zero to 10.5%. The formation of the γ-phase originates in the reduction of the PP-MA chain mobility 172
Intercalation of Clay
Figure 35 Interlayer spacing, d001, stack thickness, t, and its aspect ratio, p, in PP-MA/stearyl ammonium-MMT. Data [Nam et al., 2001].
affected by the intercalation of PP chains into the interlamellar gallery space. The orthorhombic unit cell of the γ-phase consisted of two parallel helices with an inclination of 40° to the lamellar surface. The diffraction peak of the 130-plane of the γ-phase appears at 2θ = 19.3°. The clay particles are located inbetween the crystalline fibrils made of assembly of lamellae. Thus, even when the lamellar parameters (thickness and spacing) remain about constant, the crystalline fibril size and interstack correlation distance decrease with clay concentration. The CPNC showed enhanced moduli (compared to the matrix) related to the clay content and aspect ratio by the Halpin-Tsai relation. The system: PP (Mw = 280 kg/mol) with PP-MA (Mw = 47 kg/mol; 2.5 wt% MAH) and organoclay (ODA and n-decyl amine, treated with silane; d001 = 2.3 nm) was melt-mixed in an internal mixer at 180 °C and 150 rpm [Marchand and Jayaraman, 2001]. The aim was to maximise the degree of clay dispersion by varying the PP-MA content (0 to 14 vol%) at a constant organoclay content of 1.8 vol%. The optimum concentration of PP-MA was found to be ca. 9-10 vol%. Furthermore, it was reported that a higher degree of exfoliation was obtained using ODA than its silane-treated homologue. In a recent publication from NCL in Pune, a SSE (T = 160 to 200 °C, at 45 rpm) was used to melt-blend PP/PP-MA into organoclays [Kodgire et al., 2001; Hambir et al., 2001; 2002]. The organoclays used in this study were Cloisite® 6A (2M2HTA; see Table 16), and Nanoclay (N1; Na-MMT with CEC = 1.35 meq/g ion exchanged with ODA) from Nanocor Inc. The material characteristics are shown in Table 26. The composition of the CPNC was: 84 wt% PP, 12 wt% PP-MA, and 4 wt% organoclay. XRD of the Na-MMT gave the interlayer spacing d001 = 1.2 nm. For the intercalated clays, Cloisite® 6A (C6A) and Nanoclay® ODA (N1), respectively, three peaks: d001 = 1.2, 1.8 and 3.3 nm, and a single: d001 = 2.45 nm, 173
Clay-Containing Polymeric Nanocomposites
Table 26 Material characteristics for PP, PP-MA and organoclays [Hambir et al., 2001] Sample
Sample Grade code
MW Supplier (kg/mole)
PP
PP1
Vestolen 2000
164
Hüles, Germany
PP
PP2
Koylene M0030
241
IPCL, India
PP
PP3
Vestolen 7000
36 2
Hüles, Germany
Polybond
PB
PB 3150; 1 wt% grafted MAH
188
BP Chem., USA
Vinbond
VB
VB 100; 0.65 wt% grafted MAH
151
Vin Enterprises, India
Cloisite®
6A
47 wt% of 2M2HTA
-
Southern Clay Products
Nanoclay
N1
ODA
-
Nanocor
were observed. In the presence of C6A the XRD showed the presence of multiple peaks, indicating non-uniform structure. Melt-blending the two organoclays with PP and PP-MA (Polybond 3150 from Uniroyal Chemical) resulted in a reduction of the XRD peaks’ intensity. This was particularly evident with C6A; for N1 the interlayer spacing shifted to a higher value (d001 = 2.7 nm), but the reduction of intensity was not as pronounced. Thus, both systems did not exfoliate, but the degree of dispersion was better in the presence of C6A. However, as data in Figure 36 and Table 26 indicate, it is N1 that yields higher rigidity. The effect may originate in the different type and/different content of the crystalline phase or the platelets’ aspect ratio. The performance of PP was enhanced by incorporation of PP-MA and organoclay. About a 35% increase in the tensile modulus and about a 10% increase in the tensile strength were observed. The thermal decomposition temperature increased from 270 to about 400 °C (for C6A). The most interesting observation is the dramatically different large-scale morphology for PP and CPNC based on it; while PP3 crystallises in well-known spherulites, the presence of PP-MA/organoclay (PP3/PB/6A) changed it into a fibrillar structure. After prolonged crystallisation the fibres grow in length and diameter, but do not revert to the spherulites. Furthermore, the PP/clay system crystallises at temperatures higher by ca. 10 to 20 °C than neat PP, suggesting strong nucleating capabilities of the organoclays. In another study from the same laboratory [Hambir et al., 2002], three grades of PP and two of PP-MA were used (see Table 26). Judging by XRD spectra, melt intercalation with PB was superior to that of VB – evidently the acidity of the latter graft copolymer was insufficient. The data also indicated that the use of lower MW PP resulted in a higher degree of dispersion (judged by the reduction of the peak areas). Tethered PP/organoclay CPNCs were prepared via melt compounding in a TSE at T = 180, 190, 200 and 190 °C (from hopper to die), at a screw speed of 174
Intercalation of Clay
Figure 36 Temperature dependence of the tensile storage modulus, E´, for PP and PP/PP-MA/organoclay for: Cloisite® 6A and Nanocor N1. Data [Kodgire et al., 2001].
180 rpm. For testing, dried pellets of the nanocomposite were injection-moulded at 200 °C with a mould temperature of 30 °C [Liu and Wu, 2001]. The organoclay was prepared from Na-MMT (CEC = 0.80 meq/g) by ion exchange reaction with trimethyl hexadecyl ammonium bromide (3MHDA). In the presence of dibenzoyl peroxide the product was compounded in an internal mixer for 60 min with glycidyl methacrylate, at a weight ratio of 13:2. This epoxidised product (E-MMT) was then compounded with PP. XRD of Na-MMT, 3MHDA and E-MMT gave the following interlayer spacings: d001 = 1.24, 1.96 and 2.98 nm, respectively. XRD of PP with 5 wt% of 3MHDA indicates that the d001 peak of the organoclay remains at the same position, i.e., PP does not intercalate into the organoclay. By contrast, XRD of PP with different loadings of E-MMT (1, 3, 5, and 7 wt%) showed expansion of the interlayer spacing, viz. 4.92, 5.09, 4.61 and 4.75 nm, respectively. Thus, E-MMT can lead to intercalation but not to exfoliation. The mechanical properties of the CPNC improved with addition of organoclay, increasing the storage modulus (by 40% at 7 wt% loading), not affecting the impact strength and slightly reducing Tg. The addition of clay did not change the crystal structure of PP, however silicate layers acted as nucleating agents, increasing the crystallisation peak temperature of PP from Tc = 110.5 °C to about 120 °C. Nanocor recently recommended CORI screw design for the manufacture of CPNC with PO as a matrix [Qian et al., 2001]. A Leistritz TSE with D = 27 mm and L/D = 36 was used at 300 to 500 rpm and T = 170 to 190 °C. Two vents at 13D and 24D were used. The recommended configuration is given in Table 23. The process involves two-steps: 1. Preparation of masterbatch containing about 50 wt% of organoclay, a compatibiliser (ca. 25 wt%), and a matrix polymer (e.g., PP-MA and PP). 2. Dilution to the desired clay content (e.g., 2 to 7 wt%). 175
Clay-Containing Polymeric Nanocomposites The latter process may be carried out using either a TSE or a SSE [Cho et al., 2002]. For example, first, organoclay/PP/PP-MA masterbatch was prepared. It contained 50 wt% Nanomer® I.30P (MMT-ODA) with 25 wt% PP-MA and 25 wt% PP. Next, dilution of the masterbatch with PP (to 6 wt% organoclay in the final CPNC) was carried out in either a SSE (containing different mixing screws or add-on mixers) or in a CORI. For comparison, PP nanocomposites were prepared following a single-step procedure. Thus, PP, PP-MA and 6 wt% I.30P were compounded in a CORI. The interlayer spacing of all six CPNC batches prepared by different methods was small, viz. d001 = 2.7 to 2.8 nm. Thus, intercalation not exfoliation was obtained. The mechanical properties of the neat resin, CPNC prepared by dilution of the masterbatch either in a SSE or in a TSE, as well as of CPNC prepared by direct compounding are presented in Table 27. CPNC prepared by the two-step method showed slightly better performance than that prepared by direct compounding in a TSE, but dilution in a TSE results in a marginally better performance than that in any SSE configuration (independent of which mixing unit was installed). The difference between the two-step and single-step is most likely related to the difference in d001. The origin of the increased interlayer spacing may be related to the total residence time of the CPNC subjected to melt compounding. Since the stated aim of the process is exfoliation, more detailed discussion can be found in Section 2.4 and Part 4. By contrast with the customary use of ammonium-intercalated clays, Pozsgay et al. [2001] treated Na-MMT with N-cetyl pyridinium chloride. Melt compounding of this organoclay with PP in an internal mixer produced rather disappointing results as the tensile strength of PP was found to decrease with the organoclay loading (from 1 up to 10 wt% clay). The decrease superimposed on data obtained for a PP mixture with neat Na-MMT, indicated a lack of miscibility between PP and the cetyl C16H33-paraffin group of the intercalant. As reported by Wang et al. [2001a], organoclay may also act as a compatibiliser in polymer blends. The authors dispersed 10 wt% of Nanomer I.30 TC (MMT-ODA). Comparing the binary blend PA-6/PP = 9/1 with the three-component system PA-6/PP/I30 = 9/1/1 it was found that addition of organoclay significantly improved the tensile properties, viz. modulus increased from 2.5 to 4.7 GPa, tensile strength
Table 27 Mechanical properties of PP nanocomposites [Cho et al., 2002] Material Process
Nanomer® Flex I.30P (%) strength, σb (MPa)
Flex modulus, Ef (GPa)
HDT (°C)
d001 (nm)
PP 6523 (neat resin)
0
33.7
1.08
85
-
CPNC
SSE dilution
6
44.5 ± 0.5
1.71 ± 0.03
102
2.79 ± 0.01
CPNC
TS E dilution
6
45.9
1.81
104
2.8
CPNC
Direct
6
43.8
1.57
100
2.7
176
Intercalation of Clay from 65 to 73 MPa, but at a cost of impact strength, which decreased from 70 to 13 J/m. Compatibilisation of the blend by addition of PP-MA reduced the modulus to 3.4 GPa, but slightly improved the strength and impact strength (to 75 MPa and 17 J/m, respectively). Evidently, in the four-component system PP-MA migrated to the clay-PA-6 interface, thus its role as the blend compatibiliser was reduced. The fundamental principle of additive incorporation into a multiphase polymeric system is that it must be inserted into a specific phase – if in the blend one phase is rigid and the other ductile, increasing the volume of each of these with the same additive will have opposite effects. For example, it is known that incorporation of PS into the rubbery phase of HIPS increases the ductility of the system, while incorporation of glass fibre (GF) into PA-66 in its blend with ABS increases the modulus [Utracki, 2002]. Obviously, the same principles are valid for CPNC with polymer blend as a matrix. Recently two blends, PC/ABS and PA-6/ABS were prepared by melt compounding with MMT-3MHDA [Wang et al., 2004]. The TEM micrographs showed that in the former CPNC clay platelets dispersed in the ABS phase, whereas in the latter system, they were dispersed mainly in the PA-6 phase. The dynamic self-organisation of organoclay was studied, but unfortunately there is no information regarding the effect on performance. Extensive studies on melt intercalation of PP were carried out by Ton-That et al. [2002]. The authors focused on CPNC containing PP melt compounded in a TSE at 200 °C with 2 wt% of Cloisite® 15A and 4 wt% of a compatibiliser. Several commercial PP-MA compounds, a PP-AA and an onium-terminated PP have been used as compatibilisers. XRD spectra have shown that the interlayer spacing d001 ≤ 3.9 nm was achieved. TEM confirmed this by displaying a number of short stacks along with individual platelets dispersed in the matrix. The highest values of tensile modulus were obtained for PP-MA with 0.5 wt% MAH and with PP-AA with 6 wt% acrylic acid. To close this discussion on the melt intercalation of PP, it is necessary to stress that in many reports all the observed differences in the performance of CPNC and neat PP matrix resin are assigned to the more-or-less exfoliated clay. However, the behaviour of this semicrystalline polymer strongly depends on crystallinity: crystallographic form, type and size of crystal, total crystallinity, etc. Rare are the publications that even consider these effects. In 2001 Saujanya and Radhakrishnan published interesting work on the effects of calcium phosphate (CaPO3) particles on the crystallinity of PP. The authors prepared nanosized CaPO3 using an in situ deposition technique in the presence of PEG or PEG/PVAc. The composite was prepared by grinding together 2, 5 or 10 wt% of CaPO3 with PP in an agate mortar and pestle. A small quantity of the powder was isothermally crystallised on the microscope hot stage. The growth of spherulites as well as the intensity (grey scale) of transmitted light in the crosspolar mode was recorded. The results were compared with those for PP containing conventionally prepared CaPO3. Both filled PP as well as neat PP crystallised in the monoclinic α-crystalline phase. However, as the size of CaPO3 particles (d) decreased, the isothermal crystallisation rate increased dramatically – this is illustrated in Figure 37 as a decrease of the crystallisation half-time (t1/2) versus 1/d. The overall crystallisation rate (G) was found to follow the dependence: log G ∝ (1/d). Similarly, the nucleation efficiency of CaPO3 particles was found to increase with their inverse diameter, resulting in a decrease of the ultimate spherulite size and enhanced optical transparency of the PP/nanoparticle composites. These results were further confirmed by DSC analysis. 177
Clay-Containing Polymeric Nanocomposites
Figure 37 Crystallisation half-time versus reciprocal diameter of calcium phosphate powder in PP. Data [Saujanya and Radhakrishnan, 2001].
The dependence of t1/2 versus 1/d presented in Figure 37 confirms that at the same volume fraction the efficiency of a nucleating agent depends on its surface area. The dependence is not linear, suggesting that as the particle size is reduced to nanosize, additional factors start to play a role, viz. particle shape, aspect ratio, curvature, increased proportion of the high energy surface atoms at a cost of the shielded atoms in the interior, etc. The smallest CaPO3 particle with d = 7 nm (i.e., specific area, Asp = 69 m2/g) has a surface area one order of magnitude smaller than MMT (Asp = 700-800 m2/g). However, it is organoclay not neat mineral that gets incorporated into the PP matrix hence the shielding effect of the organic intercalant plays an important role – often the clay surface is fully shielded from the matrix. Another point that often escapes attention is the actual location of the nanoreinforcing platelets in the semicrystalline polymer matrix. The thermodynamics of crystallisation require that any foreign material should be expelled from the growing crystal. Hence, the organoclay must be located mainly in the non-crystalline part of the CPNC, in the liquid phase of PP at T > Tg ≅ 0 °C or in the glassy part at lower temperatures. Since the crystallinity of PP in CPNC may be as high as 68%, the clay concentration in the non-crystalline phase may be three times higher than average. Owing to the crowding effect, the interlayer spacing in this case may be reduced by a factor of up to 3. Thus, on the one hand large effects of organoclay addition into a semi-crystalline matrix may not be as significant as those in an amorphous one and on the other the mechanism of the nanoclay nucleating activity should be taken into account. The key information from the relatively large number of reports on melt intercalation of PP/organoclay systems is summarised in Table 28. As far as the tensile modulus is concerned, the data indicate that at low clay loading it increases linearly with clay content. For exfoliated CPNC with PA-6 as the matrix the dependence (up to 10 wt% of clay) may be expressed as:
E R ≡ E / Ematrix
1 + a0 w ; a0 = 0.2
where w is clay (inorganic part only) content. 178
(22)
Intercalant
2M2ODA
ODA
ODA
Aminosilane H2NR1 -SiR2R3R4
Ammonium intercalated
ODA
Clay
Na-MMT
Na-MMT
Na-MMT or FM
MMT
MMT
Na-MMT
PP 30 kg/mol
PP + 0.2 wt% MAH; Mw = 210 kg/mol
Solvent with BP = 100 – 200 °C
PP-MA, 1-5 wt% MAH; Mn = 20 kg/mol
PP-MA (Yumex 1001, acid # 26)
none
PP
PP
PP
Telechelic diol or PP-MA should have good chemical affinity to 2M2ODA
Performance
Maxfield et al., 1995, Usuki et al., 1997
Ref.
2.4 to 7.3
No XRD peak in the range 2.0-4.0 nm Exfoliation + shoulder d001 ≅ 3.4
In xylene at 120 °C, or in melt TSE at 250 °C (solvent evaporation) TSE at 200 °C
Kawasumi et al., 1997; Hasegawa eet al., 1998
Tensile modulus up by a factor of 1.9 (at 5% loading), tensile strength up by 19%
Hasegawa et al., 2000a
Wolf et al., 1999
Hudson, 1999 The key is bonding through co-crystallisation between the grafted PP and the main PP
Miscibility of the PP/PPMA blends is of critical importance
7.22 for 33 wt% Optimum when one polar Kato et al., 1997 of organoclay group per 25 PP-mers is present
> 3.87 to exfoliation
d001 (nm)
TSE at 210 °C; 5.9 & 6.4 for and injection MMT & FM moulded
Internal mixer at 200 °C
Internal mixer at 220 °C
Polymer Process
PP-MA; acid = 52; none or PP-OH; OH = 54
Hydroxylated polybutadiene or PP-MA
Compatibiliser
Table 28 Summary of melt intercalation in PP matrix
Intercalation of Clay
179
180
2M2ODA
2M2HT
MT2EtOH
2M2HT
2MHTEtOH PP-MA, 0.55, 1.1, 3, 5 % MAH
n-alkyl PP-MA, MAH = amines 2.9 wt% C4 – C18 5 & 10 wt%
2M2ODA
Na-MMT
Cloisite® C20A
Cloisite® C30B
Cloisite® C20A
Cloisite® C25A
FM
Na-MMT
PP-OH or PP-MA
PP-MA, 0.55, 1.1, 3, 5 % MAH
PP-MA, 2 wt% MAH
PP-MA, 2 wt% MAH
PP-MA; 1 wt% MAH
PP + 0.2 wt% MAH; Mw = 210 kg/mol
ODA
Na-MMT
Compatibiliser
Intercalant
Clay
-
PP
PP
PP
PP
PP
PP
none
Exfoliation?
1.8 at 5 wt% C25A
3.2 at 5 wt% C20A
1.4
SSE at 180 °C; Stacks (3.4 nm) ultrasonics + platelets (US) in solvent
CORI, 300 rpm, 190 to 230 °C
Internal mixer at 200 °C
Internal mixer at 200 °C
Internal mixer at 210 °C
Exfoliation at 8% C20A
Exfoliation + shoulder 3.4
TSE at 200 °C Internal mixer at 210 °C
Exfoliation + short stacks at 5.3 wt% clay
d001 (nm)
TSE at 200 °C
Polymer Process
Table 28 Continued... Ref.
Improved mechanical properties and reduced flammability
Young's up by 129%, yield stress up by 32%, Izod impact down by 18%, elongation down by 99%.
Worst case
Best for low MW PP & the lowest MAH content
Bad performance – composite
Good performance
Tensile strength from 29 to 31 MPa
Manias et al., 2001
Reichert et al., 2000
Kim et al., 2000
Kim et al., 2000
Lee et al., 2000a
Lee et al., 2000a
Zhang et al., 2000b
Relative tensile vs. T: E´ = Hasegawa et al., 1.5 to 2.5 with max at T = 2000b 50 °C. Note that Young’s modulus E(PP) = 780, E(PP-MA) = 429 MPa
Performance
Clay-Containing Polymeric Nanocomposites
2MHDODA
2MBHT
2M2HT
ODA
ODA
2M2HT; 4 wt%
ODA; 4 wt% PP-MA Polybond 3150; 1 wt% MAH
MMT, FM, or mica
Cloisite® C10A
Cloisite® C20A
Na-MMT
MMT
Cloisite® 6A
Nanoclay
PP-MA Polybond 3150; 1 wt% MAH
PP-MA, 2.5 wt% MAH
PP with 0.2% MAH
Styrene
Styrene
Polyacrylamide + PP-MA
PP-OH or PP-MA
ODA
Na-MMT
Compatibiliser
Intercalant
Clay
PP
PP
PP
PP
PP
PP
PP
-
d001 (nm)
SSE at 160 to 200 °C
SSE at 160 to 200 °C
2.7
1.2 to 3.3
Intercalation
Intercalation; 2.9 to 3.2
TSE at 200 °C Internal mixer at 180 °C
3.65
Exfoliation?
Internal mixer at T = 180 °C + US
Internal mixer at T = 180 °C + US
2.3 to 3.8
SSE at 180 °C; Stacks at 3.2 + US in solvent platelets
Polymer Process
Table 28 Continued...
Nam et al., 2001
Ryul et al., 2001
Ryul et al., 2001
Oya et al., 2000
Manias et al., 2001
Ref.
Kodgire et al., 2001; Hambir et al., 2001; Hambir et al., 2002
et al., 2002
Higher modulus than with Kodgire et al., 2001; Hambir et 6A (increased by 42%); al., 2001; Hambir TS increased by 10%
Better dispersion than for Nanoclay, but poorer performance
Best results with ODA and Marchand and Jayaraman, 2001 10 wt% PP-MA
PP-MA formed rod-like crystals; higher modulus
Ultrasonics caused copolymerisation of styrene
Ultrasonics caused copolymerisation of styrene
At 3 wt% clay modulus increased by 67%, TS by 26%, Izod by 95%
Improved mechanical properties and reduced flammability
Performance
Intercalation of Clay
181
182
3MHDA; 5 wt%
Na-MMT
N-cetyl pyridinium
ODA
OD A
2M2HT; 2 wt%
2M2HT; 2 wt%
Na-MMT
Nanomer I.30 TC
Nanomer
Cloisite 15A
Cloisite 15A
Nanomer® ODA; I.30P 6 wt%
Intercalant
Clay
MA-g-PP (Polybond 3150), 4 wt%
PP-MA; 0.2 wt% of MAH AA-g-PP (Polybond 1001), 4 wt%
PP-MA; 1.5 to 5.8 wt% MAH
None
Organoclay grafted with glycidyl methacrylate PP-MA
Compatibiliser 5.09 with epoxidised organoclay
CORI for master 2.7 to 2.8 batch, then either CORI or SSE for diluting to 6 wt% Internal mixer ?
TSE at T = 180, 190, 200 °C, 180 rpm; IM
d001 (nm) Storage modulus up by 40% (7 wt% loading), same impact strength, Tg. Down Tc up by 10 °C Flex modulus up by 67%, flex strength up by 36%, HDT up by 19 °C
Performance
Tensile strength of PP down with organoclay loading PP/PA-6 TSE at T = 180- 3.7; organoclay Modulus increased by 230 °C, 90 rpm as compatibiliser 36%, tensile strength by 15%, impact strength, down 70 to 13 J/m. TSE at 200 °C 3.03 at 4 % organoclay PP 6100 TSE at 200 °C, 3.5 Improved: Flex strength: 200 rpm 13%, modulus: 22%; tensile strength: 1%; modulus: 15%, Izod: 45% PP 6100 TSE at 200 °C, 3.5 Improved: Flex strength 200 rpm 16%, modulus 32%; tensile strength 3%; modulus 9%, Izod 25%
PP
PP
PP
Polymer Process
Table 28 Continued...
Ton-That et al., 2002
Ton-That et al., 2002
Okamoto et al., 2001
Wang et al., 2001a
Pozsgay et al., 2001
Sakai, 2002; Cho et al., 2002; Lan and Quiqan, 2000
Lui and Wu, 2001
Ref.
Clay-Containing Polymeric Nanocomposites
Intercalation of Clay As discussed in Section 3.5.1 (Micromechanics of CPNC), well-dispersed CPNC with PP as the matrix follow the same dependence. This is quite remarkable since the parameter ao is a measure of the hydrodynamic volume which depends on the aspect ratio – for hard spheres the Einstein prediction is ao ≈ 0.005 (for poorly dispersed organoclays the constant ao ≈ 0.07). Thus, the observed dependence indicates that in these solid CPNCs the aspect ratio of clay platelets is p ≈ 277. In short, at least as far as tensile modulus is concerned, one can get similarly high performance form the PP/clay system as that of the renowned PA-6/clay. As evident from this discussion, PP is the principal polyolefin of interest. Numerous patents and research publications on CPNC with PO as the matrix usually mention other polymers, but the cited examples invariably focus on PP. One of the exceptions is a publication by Wang et al. [2001b], who prepared LLDPE-based CPNC using MAH grafted LLDPE (PE-MA) and one of four organoclays, viz. Cloisite® 20A (2M2HTA), and synthesised alkyl ammoniumMMT (CEC = 1.19 meq/g) with alkyls: dodecyl, hexadecyl or octadecyl (respectively C12M, C16M, ODAM). The interlayer spacings of C20A, C12M, C16M, and ODAM were determined by XRD as d001 = 2.47, 1.36, 1.79 and 1.85 nm, respectively. Two types of PE-MA were used, one a direct product of MAH grafting on LLDPE, and the second obtained by mixing neat LLDPE with PE-MA (0.85 wt% MAH). The CPNC were prepared by melt compounding in an internal mixer at 140 °C and 60 rpm. The XRD scans at 2θ = 2 to 10°, for CPNCs with 5 wt% of either C20A or ODAM did not show a d001 peak indicating exfoliation – confirmed by TEM. For the two other organoclays there was a significant reduction of the peak area – only a trace of a peak was observed for C16M and a weak one for C12M. Thus, for exfoliation of alkyl ammonium-MMT in an LLDPE matrix at least C16-alkyl chain in the organoclay is required. The authors also determined that the critical level of MAH grafted onto LLDPE is 0.1 wt% – below this level even C20A or ODAM could not be exfoliated. Several publications report that full exfoliation in PE/clay CPNC has been achieved. However, these were prepared by polymerisation of ethylene in the presence of organoclay. The encountered problem is that the system is not thermodynamically stable, and during the subsequent processing the platelets reaggregate, reducing the interlayer spacing and the performance.
2.3.9 Temperature and Pressure Effects on Interlamellar Spacing The ‘standard’ intercalation procedure (e.g., by ammonium salts) is carried out under mild conditions, viz. ambient pressure, P = 0.1 MPa, and T = 20-80 °C. However, melt exfoliation of CPNC may involve much more severe treatment. There is little information in the literature as to how T and P affect the interlayer spacing. What is available, seems to be system-specific. For example, the temperature effects on the interlayer spacing were found to depend on the onium ion and (to a lesser extent) on the type of clay. In Figure 38 variation of the interlayer spacing, d001, as a function of T is presented for ammonium salt intercalated magadiite (a) and beidellite (b); circles represent tetradecyl ammonium-tetradecyl amine complex (C14H29N+H3× C14H29NH2), while triangles represent dimethyl ditetradecyl ammonium [(CH3)2N+(C14H29)2]. Magadiite is a layered sodium silicate: Na2Si14O29.9H2O. Evidently, the interlayer spacing variations with temperature depend on the ammonium radical, and to a 183
Clay-Containing Polymeric Nanocomposites
Figure 38 Interlayer spacing, d001, as a function of temperature for (a) magadiite and (b) beidellite intercalated with salts of tetradecyl ammoniumtetradecyl amine (C14; circles) or dimethyl ditetradecyl ammonium (2(C1C14); squares; open squares show values after cooling). Data [de Siquira et al., 1999].
184
Intercalation of Clay lesser extend on the type of clay [de Siquira et al., 1999]. The authors explained the observed changes in the interlayer spacing as caused by molecular rearrangement of the ammonium radicals, melting of the n-paraffin groups, or changes of hydration. Information on the effects of pressure (P) is even more difficult to find. However, there are indications that with increasing P the interlamellar gallery height decreases. High-resolution data were obtained using time-of-flight neutron diffraction of hydrated Ca- and Na-smectite and vermiculite [de Siquira et al., 1999]. The measurements were conducted in a special Ti/Zr cell that could withstand simultaneously T ≤ 350 °C and P ≤ 200 MPa. The data are presented in Figure 39. One of the more interesting conclusions from that work was that the interlayer water is denser than that in the bulk: ρinterlayer = 1.06 versus ρ bulk = 0.874 g/ml. In other words, there is a reduction of free volume in the vicinity of the clay particle surface. If so, the compressibility should also be reduced. The transitionless reduction of the interlayer spacing of Ca-smectite gives the isothermal volumetric compressibility of the hydrated clay: κ = 4.8-10-5 MPa-1, to be compared with bulk water compressibility of κ = 4.5-10-4 MPa-1 – a reduction by a factor of 9.
2.3.10 Layered Nanofillers, other than Montmorillonite Excepting the patents that list a plethora of diverse layered nanofillers, the majority of published research on CPNC and all the industrial activities focus on swellable smectite clays, such as natural montmorillonite or synthetic hectorite (quite similar to MMT). However, the properties of polymer/clay nanocomposites do depend on the kind of layered material that has been used. Thus, it is advisable to consider the possibilities of CPNC formation with clays other than MMT.
Figure 39 Pressure dependence of d-spacing for calcium- and sodium-smectite and vermiculite. Data [de Sequira et al., 1999].
185
Clay-Containing Polymeric Nanocomposites 2.3.10.1 Kaolinite The structural formula for kaolinite is A14Si4Ol0(OH)8, and it is a 1:1 type clay, hence different from the commonly used 2:1 type smectites. Its lattice consists of one sheet of tetrahedrally coordinated SiO4 and one sheet of octahedrally coordinated AlO2(OH)4. Kaolinite has a low value of CEC = 0.02-0.04 meq/g and large aspect ratio. In neat kaolinite the adjacent cells are spaced about 0.71 nm across the (001) plane. A layer of -OH covers the octahedral sheet forming strong hydrogen bonds between the layers, hence only a limited number of polar guest species, viz. N-methyl formamide (NMF) or dimethyl sulfoxide (DMSO), can be introduced. However, once the layers are separated, the -OH functionality may be used for hydrogen bonding with some polymers. Thus, kaolinite has quite a different structure than MMT, viz. one side of the interlayer space is covered with hydroxyl groups of the AlO2(OH)4 octahedral sheets and the other side is covered by oxygen atoms of the SiO4 tetrahedra. As a result, CPNC with kaolinite are expected to exhibit different behaviour from those with MMT. In 1996 Tunney and Detellier provided an overview of the earlier work. The first incorporation of a polymer into kaolinite was reported in the early 1990s. In a series of publications Sugahara and his colleagues [Sugahara et al., 1992] intercalated kaolinite with monomeric acrylonitrile, acrylamide or vinylpyrrolidone then induced thermal polymerisation. However, this method did not provide a means for controlling the molecular weight and its distribution. The first direct intercalation of kaolinite with macromolecules took place more recently [Tunney and Detellier, 1996]. The motivation for this work was a search for anisotropic ionic conductivity. In other layered materials two-dimensional confinement of PEG provided the desirable performance. In other words, here total exfoliation was neither expected nor desired. The applied method consisted of two steps: 1. Intercalation of kaolinite with a solvent, and 2. Compounding the intercalated kaolinite with a polymer, which progressively displaced the solvent molecules in the interlamellar galleries. The authors have shown that it is possible to prepare two distinct ethylene glycol phases of kaolinite, with d001 = 0.94 and 1.08 nm. The former intercalate had the ethylene glycol unit covalently bound to the interlayer aluminol surface, while the other had a more weakly bound intercalated phase. The logical next step was to intercalate larger oxyethylene-based molecules into kaolinite. Thus, first DMSO or NMF was used to intercalate kaolinite, and then PEG-3400 or PEG-1000 displaced the low molecular weight intercalant. This was accomplished by heating PEG with ca. 17 wt% DMSO-intercalated kaolinite in a round-bottom flask for 9 days at T = 155 °C. The product was purified and dried in an oven at 100 °C for 3 days. Elemental analysis: carbon 7.90%; hydrogen 2.63%. XRD gave d001 = 1.112 ± 0.004 nm. The calcination weight loss was 29.1%. Several samples were prepared from combinations of clay, intercalating solvents and melt-intercalating PEG resins. Their interlayer expansion was comparable, ranging from d001 = 1.085 to 1.119 nm, i.e., providing an interlamellar space of Δd001 ≅ 0.37-0.41 nm. This indicates that the intercalated oxyethylene units are arranged in a flattened monolayer conformation. It is noteworthy that intercalation, using PEG-1000 dissolved in either water or 1,4-dioxane, was
186
Intercalation of Clay unsuccessful. The success of the intercalation by the melt method could be due to a strong concentration effect of the polymer and lower stabilisation energy of polymer segments. Larger interlayer expansion was obtained by reacting kaolinite with phenylphosphonic acid (PPA) in a water/acetone (1:1) solution at 95 ± 5 °C for up to 19 days [Guimarães et al., 1998]. The topotactic reaction was stoichiometric, viz.: Al2Si2O5(OH)4 + 4 H2O3PPH → Al2Si2O5(OH)(HO3PPH)3×2H2O + 3H2O As a result of reaction the interlayer spacing increased from that of kaolinite (d001 = 0.716 nm) to the spacing of the kaolinite phenyl-phosphonate (KPP): d001 = 1.502 to 1.645 nm, i.e., providing an interlamellar space of Δd001 ≅ 0.7860.929 nm. The analysis suggested that phenyl-phosphonate groups were grafted to the kaolinite platelets. The materials were stable up to ca. 450 °C. In another report [Guimarães et al., 1999] hydrated kaolinite phenylphosphonate (KPP-hyd) was reacted with hexylamine. The reaction produced a stable light-yellow compound (KPP-hex). XRD gave d001 = 1.636 nm, consistent with the size of the intercalating hexylamine. The thermal analyses showed that at T ≅ 230 °C hexylamine molecules decomposed, but the resulting KPP remained stable up to 498 °C, at which temperature the grafted phenyl phosphonate started to decompose and dehydroxylation of kaolinite takes place. In 1999 Gardolinski et al. [1999] intercalated kaolinite in two steps, first reacting it at 60 °C with an aqueous solution of DMSO. After washing and characterisation, the resulting complex: Al 2 Si 2 O 5 (OH) 4 (DMSO) 0.4 was stoichiometrically reacted at RT with N-methyl-2-pyrrolidone (NMP). The product was Al2Si2O5(OH)4(NMP)n (where n = 0.39 ± 0.02). XRD gave d001 = 1.231 nm, i.e., an expansion of 0.11 nm over the DSMO complex. The presence of NMP molecules in the interlamellar space resulted in a notable enhancement of the thermal stability – while the DMSO-complex decomposes at 175 °C, the NMP complex remains stable up to 431 °C. In the next publication from this group, first kaolinite was reacted with DMSO, producing a kaolinite-DMSO complex. The complex was subsequently reacted with either PEG or bacterial polyhydroxybutyrate (PHB) in the molten state at 130 or 180 °C, respectively [Gardolinski et al., 2000]. The displacement of DMSO molecules took several days – highly ordered polymer/kaolinite CPNC was obtained. In the interlamellar galleries of kaolinite, PEG macromolecules formed a polymeric monolayer. The characteristics of the kaolinite-complexes are listed in Table 29. As mentioned above, Sugahara and his colleagues synthesised several kaolinitepolymer compounds (viz. kaolinite-PAN, kaolinite-PAAM, or kaolinite-P4VP) by first intercalating kaolinite with the appropriate monomers and then polymerised by thermal treatment. The serious drawback of this procedure is the lack of control of the degree of polymerisation of the resulting polymers. Recognising the problem, the authors developed an alternative intercalation method [Komori et al., 1999b]. The publication is particularly valuable as that the authors also reported the unsuccessful attempts. The process resembles the ‘common solvent method’ discussed in the next Chapter. Thus, pre-intercalated kaolinite at room temperature is dispersed in a polymer solution. For example when kaolinite-ammonium acetate was used as the intermediate for the intercalation of P4VP (Mw = 10 kg/mol) dissolved in water, the displacement was
187
Clay-Containing Polymeric Nanocomposites
Table 29 Characteristics of intercalated kaolinite [Gardolinski et al., 2000] Content (wt%)
d001 (nm)
Δd001 (nm)
Decomposition-T (°C)
(nil)
(nil)
0.716
0
(450)
DMSO
8.8
1.121
0.404
175
PEG
28.9
1.116
0.399
349
PHB
15.5
1.170
0.453
309
Intercalant
not successful, leading to de-intercalation of ammonium acetate. However, when kaolinite-methanol (K-MeOH) was used as the intermediate and methanol as the solvent for P4VP, the 24 h long polymer infusion increased the interlayer spacing. The product was centrifuged and yellowish white powders were obtained without washing. K-MeOH was also used to intercalate such organic species as alkylamines, p-nitroaniline, and ε-caprolactam. XRD of kaolinite, K-MeOH and kaolinite-P4VP determined that d001 = 0.72, 1.11 and 1.24 nm, respectively. Elementary analysis showed that the number of P4VP mers per kaolinite unit was 1.9, hence the formula: [Al 2Si 2O5(OH) 4]×(C6H9NO)1.9. The complex kaolinite-P4VP can easily be destroyed by washing with ethanol or water – P4VP could be removed after a few minutes of washing. The authors determined that out of the 1.9 mers per unit cell, about 1.1 are adsorbed on the stack surface with only 0.8 mers in the interlayer space, i.e., twice as much as is obtained by in situ polymerisation (0.4 P4VP mers per kaolinite-unit cell). Komori et al. [2000] reported a partial methoxylation of -OH groups in a kaolinite. Kaolinite-NMF complex (d001 = 1.08 nm) was dispersed in methanol and mixed for a day. After centrifugation, the complex was redispersed and the procedure repeated seven times. The final, wet K-MeOH complex (d001 = 1.11 nm) was separated and dried in air to convert it into methoxylated compound (K-OMe; d001 = 0.86 nm). The amount of methoxy groups was determined by analysis. The results can be written as Al2Si2O5(OH)4-x(OCH3)x with x ≅ 0.36. The authors stressed that even when methanol can readily be removed from the K-MeOH complex, methoxy groups are more stable, e.g., they and water molecules remain in the interlayer space of kaolinite after air drying. The next publication from the group focused on the usefulness of K-OMe as an intermediate for the preparation of CPNC with PA-6 as the matrix [Itagaki et al., 2001]. A two-stage process was used, first preparing kaolinite-PA-6 intercalation compound, using a kaolinite-6-amino-hexanoic acid (AHA) as a precursor, then kaolinite-PA-6 was melt compounded with PA-6 (commercial 1015B grade from Ube Industries Co.). The compounding was carried out using a TSE at 240 °C and 300 rpm. XRD gave d001 = 1.15 nm indicating that the layered structure of kaolinite was not exfoliated. For comparison, PA-6 was also compounded without clay (PA-6(extruded)) and with unmodified kaolinite (PA-6/kaolinite). The PA-6/MMT commercial 1015C2 grade from Ube was used as a reference. The properties of these CPNCs are listed in Table 30. The relative properties of the last two 188
Intercalation of Clay
Table 30 Mechanical properties of kaolinite-PA-6 type nanocomposites [Itagaki et al., 2001] Material
Clay content (wt%)
Tensile strength, σb (MPa)
Tensile modulus, E (GPa)
Notched Izod impact strength, NIRT (J/m)
PA-6 (neat)
0
74.0
1.13
27.8
PA-6 (extruded)
0
72.7
1.16
34.5
Kaolinite-PA-6
1.42
79.2
1.29
27.7
PA-6/kaolinite-AHA
1.48
77.0
1.25
21.6
PA-6/kaolinite-PA-6
1.38
80.4
1.33
27.7
PA-6/MMT
1.80
89.1
1.36
25.4
compositions in the table are of interest. Taking into account different clay loadings, the performance ratio of PA-6/kaolinite-PA-6 to that of PA-6/MMT (per 1 wt% of clay) is: 1.18, 1.28, and 1.2, for the tensile strength, tensile modulus and impact strength, respectively. Evidently, the properties are not proportional to the clay content, thus the performance ratios showing about 20% better performance for CPNC with kaolinite than with MMT may not be correct. However, it is intriguing that kaolinite with only slightly expanded interlayer spacing offers at least comparable performance to that of commercial-grade CPNC with exfoliated MMT. Since kaolinite in the PA-6/kaolinite-PA-6 material is not exfoliated, the enhancement of properties must be related to the surface effects. The surface of kaolinite particles is covered by -OH groups, which interact with PA-6 via hydrogen bonding. The particle size of kaolinite is larger than that of MMT, which might also contribute to the enhanced properties. 2.3.10.2 Micas and Synthetic Micas The composition and polymorphism of micas vary considerably (see Section 2.2.2). There are 34 phyllosilicate minerals with a layered or platy texture that are classified as micas. The commercially important micas are muscovite and phlogopite. Mica sheets are transparent to opaque, resilient, reflective, refractive, dielectric, chemically inert, insulating, lightweight and hydrophilic. Mica is also stable when exposed to electricity, light, moisture and extreme temperatures. It has been fabricated into parts for electronic and electrical equipment. Micas have a 2:1 sheet structure, similar to MMT, but the maximum charge deficit is in the tetrahedral layers and contains K+ that is held tenaciously in the interlayer space. The XRD basal spacing d001 ≅ 1.0 nm is broad and skewed toward wider spacings. The illite’s CEC = 0.2-0.3 meq/g of dry clay. Since the natural mica sheet diameter can be almost as large as one metre (e.g., in Calcutta – nowadays Kolkata – museum), the numerical value of the aspect ratio is ridiculously large. The intercalation strategies for micas and smectites are the same. Exfoliated mica sheets in polymeric matrix with aspect ratio of p > 1,200 have been reported [Yano et al., 1993; 1997]. 189
Clay-Containing Polymeric Nanocomposites The early work on mica/polymer systems was aimed at improving the performance of mica for electrical applications. For example, in 1981 Ishizaka and Fujii patented composites comprising an organosiloxane resin, mica and acid phosphates, suitable for use in the manufacture of mica products with superior mechanical strength, electrical characteristics, water resistance and heat resistance. Synthetic micas are also available, e.g., Barasym SMM-100, a muscovitetype; mica-montmorillonite with CEC = 0.7 meq/g or Somasif ME100, a fluoromica (or fluorohectorite) with CEC = 0.7-0.8 meq/g and d001 = 0.95 nm. By contrast with natural micas the synthetic ones have been reported as having low aspect ratio, thus inefficient for improving the barrier or mechanical properties. However, it may be that the low aspect ratio is not inherent, but rather caused by mechanical attrition during compounding in molten polymer. Micas: ‘DM clean A’ from Topy Ind. Co. have CEC = 1.19 meq/g and a high aspect ratio of p = 1230. The chemistry of mica was discussed in Section 3.4. Yano et al. [1993] prepared CPNC with polyimide (PI) as matrix in which 2 wt% of one of the four clays (hectorite, saponite, MMT and synthetic mica) was dispersed. The clays had CEC = 0.55, 1.00, 1.19 and 1.19 meq/g, and aspect ratio: p ≅ 46, 165, 218, and 1,230, respectively. They were intercalated with DDA, filtered, washed and freeze-dried. The interlayer spacing was: d001 ≅ 1.5 (hectorite), and 1.8 nm for the three other clays. To prepare PI/clay films, 2.49 wt % of organoclay was dispersed in dimethyl-acetamide and vigorously stirred for 3 h at 90 °C, diamino diphenyl ether was added and the mixture was stirred at 30 °C for 30 min, then pyromellitic dianhydride was introduced and the mixture stirred for an additional 6 h. The resulting solution was spread on a glass plate, solvent was evaporated off for 2 days then the film was heated at 100 °C for 1 h, at 150 °C for 1 h, and at 300 °C for 2 h under N2. A 60 μm thick film of polyimide with 2 wt% of clay was obtained. The XRD spectra showed that CPNC with MMT and synthetic mica had no peaks indicating exfoliation, this was confirmed by TEM. XRD of PI with either hectorite or saponite showed peaks at d001 = 1.5 nm, but smaller in intensity than those in the respective organoclays. This implies that roughly one layer of organic molecules exists in the interlamellar space. TEM of CPNC with saponite showed dispersion of individual platelets with a small number of tactoids. In the PI/hectorite system most clay formed aggregates. This difference in dispersability goes in the reverse direction to the aspect ratio, hence it is either the CEC or, more probably, the chemical reactivity of these clays that leads to these effects. It is to be expected that the higher the aspect ratio and degree of dispersion, the higher the PI properties. This indeed is the case as far as water vapour permeability is concerned (see Figure 40). Two elements account for the reduction of permeability (see also Part 5.3): dispersion of high aspect ratio of oriented platelets and reduction of the free volume. The relative permeability coefficient (PR) (see Figure 41) is given by: PR = P/P0 = d/d´ = [1 + pφ/2]-1
(23)
where φ is the volume fraction of the platelets, P is the permeability coefficient at platelet content φ, P0 is that at φ = 0 and p = L/W is the aspect ratio. The relation was derived assuming perfect alignment of individual clay platelets – if for a given system this assumption is not obeyed, the aspect ratio necessary to fit such data will be a projection of the true aspect ratio on a plane perpendicular to the
190
Intercalation of Clay flux direction. As shown in Figure 40, the permeability reduction (at constant concentration of clay in PI matrix) is fully accounted for by the clay platelets’ aspect ratio. Furthermore, addition of 2 wt% of synthetic mica reduced the PI thermal expansion coefficient at 100 °C by 40%. Smaller factors were obtained for the other clays with smaller aspect ratio. The same intercalation method was successfully used for Na-MMT and synthetic fluoromica (FM, Somasif ME-110 from CO-OP Chem. Co.) [Kawasumi et al., 1997; Hasegawa et al., 1998]. First the clays were intercalated with ODA, then organoclay powder (7.3 wt%), PP (melt flow rate = 16 g/min; 70.8 wt%)
Figure 40 Relative permeability of polyimide containing 2 wt% of clays versus aspect ratio [Yano et al., 1997]. The line was calculated from Equation 23.
Figure 41 Tortuosity model: d = film thickness; d´ = tortuous path of diffusing molecules; L and W are platelet diameter and thickness, respectively, hence the aspect ratio p = L/W.
191
Clay-Containing Polymeric Nanocomposites and PP-MA (21.9 wt%) were dry-blended, then compounded in a TSE at 210 °C, incorporating about 5 wt% of clay (MMT or fluoromica). It was shown that compounding organoclay with PP caused a reduction of the interlayer spacing, while in the presence of PP-MA it increased to d001 = 5.9 and 6.4 nm for MMT and FM, respectively. In other words, neither one of these two clays was exfoliated, but the degree of dispersion was higher for the system with FM. The authors presented a strong case for the critical importance of PP/PP-MA blend miscibility – well dispersed clay platelets were evident in TEM micrographs for miscible systems, with a shoulder on the XRD spectrum indicating the presence of tightly spaced short stacks. In 1998 Katahira et al. [1998a,b,c,d] published a series of articles on the use of mica for the production of PA-6-based PNC. Thus, purified Na-mica flakes were dispersed in hexanoic acid-ω-ammonium phosphate, prepared by treating ε-caprolactam with phosphoric acid. The intercalation was rapid even at 20 °C, but the exchange of Na+ was accelerated by heating to T > 60 °C. The monomerintercalated mica showed d001 = 1.47 nm. During polymerisation (under low or high pressure) the mica exfoliated. Owing to the high dispersion of high aspect ratio clay platelets excellent amelioration of the bending strength and modulus were obtained for the CPNC. Thus, the formation of CPNC consisted of three steps: 1. Protonation of ε-caprolactam by phosphoric acid, 2. Intercalation of mica by ion exchange of Na+ with protonated lactam, 3. Polymerisation at T ≥ 260 °C which expanded the interlamellar spacing all the way to exfoliation. Badesha et al. [1998] prepared CPNC with fluoroelastomer matrix and dispersed in it mica-type clay (SCPX-984 from SCP). The aspect ratio of mica platelets ranged from p = 50 to 1000. The clay was first intercalated with 2M2ODA. The fluoroelastomers of interest were copolymers and terpolymers of vinylidene fluoride, hexafluoropropylene, and tetrafluoroethylene, known commercially as Viton, Fluorel, Aflas, Tecnoflon, etc. These may be cured with, e.g., a bisphenol and organophosphonium salt (accelerator). Other additives, such as colouring agents, processing aids, conductive fillers, initiators and accelerators may also be added. The intercalated mica could be dispersed in the matrix by compounding, e.g., milling prior to curing. During compounding the fluoroelastomer chains penetrate the organophilic clay, causing each platelet to be surrounded by polymer, hence exfoliating. The exfoliated nanocomposite may be formed with or without going through the intermediate stage of intercalation. An example describes addition of 10 phr of organoclay to 100 parts of Fluorel, and then milling the compound for 15 min on a two-roll rubber mill with a tight nip at 27 to 38 °C. XRD indicated that no intercalation had taken place. However, when the milling temperature was increased to 50 °C, the interlayer spacing increased to d001 = 3.3 nm – hence intercalation. However, at still higher milling temperature, T = 66 °C, fully exfoliated CPNC was produced. Zilg et al. [1999a] investigated the correlations between the polymer composition, type of clay, its concentration and mechanical properties. The matrix was based on diglycidyl ether of bisphenol-A (DGEBA) cured with hexahydrophthalic anhydride. Three types of clay were used: 1. Synthetic fluoromica (FM; Somasif ME 100 from CO-OP Chem.),
192
Intercalation of Clay 2. Purified Na-MMT (particle diameter 15 μm, BET surface area 25 m2/g from Südchemie AG) and 3. Synthetic hectorite (FH; Optigel SH from Südchemie AG). The clays were made organophilic by ion exchange with various alkyl ammonium ions: dodecylamine (DDA), N,N-dimethyl didodecyl amine (2M2DDA), N,N,N-trimethyl dodecyl ammonium chloride (3MDDA), N-methyl dodecyl N,N-bis(2-hydroxyethyl)-ammonium chloride (MD2EtOH; AKZO Ethoquad C12), 12-aminododecanoic acid (ADA), a,x-bis(aminopropyl)-terminated oligo(propylene oxide), known as Jeffamine of the D series (JAD 130, 300, 500, 800), N,N,N,N-dimethyl dioctadecyl ammonium chloride (2M2ODA) and N,N,N,N-dimethyl benzyl-octadecyl ammonium chloride (2MBODA). Upon intercalation the interlayer spacing of FM increased in the expected sequence, e.g., after ion exchange with MD2EtOH the neat clay spacing of d001 = 0.94 increased to 1.74 nm, and then upon matrix polymerisation to d001 = 6.79 nm. The authors reported that (at 5 wt% of clay content) FM was difficult to disperse beyond short stacks. Easier to disperse (but still in form of short, intercalated stacks) was MMT, while FH exfoliated quite readily. Enhanced toughness was associated with the formation of dispersed anisotropic laminated nanoparticles consisting of intercalated silicates. The authors confirmed the earlier report from Pinnavaia’s laboratory that d001 spacing increases with the onium salt alkyl chain length, and that primary amines increase the interlayer spacing more than the quaternary ones. During curing, the former most likely reacted with epoxy which the latter were unable to do. With 10 wt% of the ammonium salt used on the silicates, not all were properly intercalated, i.e., XRD showed a diffraction peak indicating intercalation. On the other hand (e.g., for FH intercalated with 2MBODA) TEM micrographs showed randomly dispersed individual platelets, suggesting exfoliation. An interesting observation came from comparing the spacings calculated from XRD and AFM micrographs, viz. 6.8 and 11 nm, respectively. The explanation offered is that the individual clay platelets are quite flexible and bend under the force of the AFM test tip. The mechanical properties of four series of epoxy-based CPNCs are summarised in Figure 42. Surprisingly, there is little improvement of properties by 2MBODA intercalation – the tensile strength of neat MMT or FH is higher (mid-concentrations) than that after intercalation. The best fracture toughness was reported for FM and MMT, intercalated or not. In another publication from the same laboratory [Zilg et al., 1999b], CPNCs based on a PU matrix were prepared by dispersing synthetic fluoromica (FM; Somasif ME100 from CO-OP Chem. Co.). The clay was manufactured by heating Na2SiF6 with talc. The FM lamellae consist of a sheet of octahedral alumina sandwiched in between two sheets of tetrahedral fluorosilica. Because in the octahedral sheet Mg2+ ions substituted for some Al3+ ones, the lamellae are anionically charged. Similarly, like MMT, FM behaves as a weak silicilic acid, with Na+ as counterions in the interlamellar gallery. During dispersion, the fairly large FM particles of average diameter around 5,000 nm (aspect ratio p ≤ 5,000) were broken down into anisotropic nanoparticles resembling short fibres of 500 nm length and 20 to 50 nm diameter, thus the effective aspect ratio (peff) for the clay was only peff ≅ 20 to 50. Two series of CPNC samples were prepared, with nonintercalated and with intercalated FM. In the latter case, FM was intercalated with 2MBODA (Ethoquad C12) in aqueous medium. Dry, intercalated FM was dispersed 193
Clay-Containing Polymeric Nanocomposites
Figure 42 Mechanical properties of four series of epoxy-based CPNC: MMT - not intercalated purified bentonite; MMT/B - MMT intercalated with N,N,N,N-dimethyl benzyl octadecyl ammonium chloride (2MBODA); FM/B - fluoromica intercalated with 2MBODA; H - not intercalated hectorite; H/B - hectorite intercalated with 2MBODA. Data [Zilg et al., 1999a].
in trihydroxy-terminated oligo(propylene glycol) with Mn = 3.8 kg/mol, which, after high shear mixing was cured with diisocyanatophenyl methane, accelerated with 0.6 wt% N,N-dimethyl-benzylamine at 80 °C. According to XRD, the interlayer spacing of neat FM (d001 = 0.9 nm) increased in the CPNC to 8.8 nm. 194
Intercalation of Clay The clay lamellae were well dispersed within the matrix, but without full exfoliation. The mechanical behaviour of the two series of PU/FM nanocomposites is presented in Figure 43. Intercalation significantly improved the tensile strength (by 60-240%) and elongation at break (by 130-400%), but it had a negative effect on modulus. The CPNC seems to have clay platelets bonded to PU. In conclusion, dispersion of intercalated synthetic fluoromica in a polyurethane matrix results in the formation of intercalated (not exfoliated) structures that offer significant improvement of mechanical properties (in comparison to neat PU). An interesting use of PA-6-based CPNC was reported by Cheng et al. [2000]. The authors found significant differences in the suitability of PA-6 and its CPNC for the preparation of microporous membranes. While neat PA-6 generated asymmetric membranes with tight skin and a porous sublayer, the intercalated mica/PA-6 system (PA-6/mica nanocomposites, M1030D from Unitika; Mn = 15 kg/mol, MI = 19) produced skinless microporous membrane with an open, bicontinuous structure that could easily be adjusted. Hence, while PA-6 produced membrane useless for microfiltration, the mica-filled CPNC engendered perfectly adjustable, useful products. To close this discussion on CPNCs with non-MMT layered nanofillers, the use of layered double hydroxides (LDH) should be mentioned. A recent review described the synthesis and characterisation of these materials [Leroux and Besse, 2001]. The LDH ideal structural formula is: [MIIxMIII1-x(OH)2]intra[Am-x/m·nH2O]inter where MII and MIII are metal cations, A is the anion, and intra and inter denote the intralayer domain and the interlayer space, respectively. The structure consists of edge-sharing M(OH)6 octahedra. Partial MII to MIII substitution induces a positive charge for the layers, balanced with the presence of the interlayer anions. LDH are
Figure 43 Mechanical behaviour of PU/FM and PU/FM-2MBODA nanocomposites. (Full symbols – PU/FM, i.e., with non-intercalated clay; open symbols – PU with FM-2MBODA (Ethoquad C12)). Data [Zilg et al., 1999b].
195
Clay-Containing Polymeric Nanocomposites often prepared via coprecipitation using MII and MIII salts at constant pH, mostly basic conditions. Their charge density is significantly higher than that of Na-MMT, viz. 0.25 to 0.40 nm2/charge versus 0.70 nm2/charge for the latter. Correspondingly, the CEC for the LDH material ranges from 0.6 to 4.8 meq/g, whereas that for Na-MMT is about 1 meq/g. A lower CEC makes exfoliation easier. There are three principal methods of LDH intercalation: (a) Using such monomers as aniline, pyrrole, ω-aminoacid, methyl methacrylate, vinylbenzene sulfonate, vinylpyrrolidone, vinyl acetate, etc.; (b) Direct intercalation of extended polymer chains in the host lattice, for example, PEG/MoO3, poly(p-phenylene)/molybdenum bronze, PANI/MoS2, or PEG/NbSe2. Sometimes, a two-step intercalation (as for MMT) may have to be used; (c) Transformation of the host material into a colloid and precipitation in the presence of the polymer. Thus, numerous methods for the preparation of LDH/polymer systems have been used, viz. coprecipitation, exchange, in situ polymerisation, two-step surfactantmediated incorporation, hydrothermal treatment, reconstruction, or restacking. The latter method, effective via the exfoliation of the LDH layers, appears to be more appropriate for the capture of monomers. Examples of polymer/LDH systems are given in Table 31. These multicomponent systems are thermally more stable than the pristine inorganic compounds, leading, for example, to potential applications in flame retardant composites. A large variety of LDH/polymer systems may be tailored considering the highly tunable interlayer composition coupled to the choice of the organic moiety. DNA may be stabilised in the interlayer space of (Mg2AlNO3) as a gene reservoir. Some of the intercalated polymers present excellent physical properties such as conductive properties (PANI), insulation (PS), or ion-gate properties (polypyrrole (PPY)), but are difficult to process because of a lack of mechanical strength. Numerous studies have focused on the use of conductive polymers as capacitor, rechargeable battery materials or in electrochromic windows. Hsueh and Chen [2003] prepared LDH (by co-precipitation in NaOH solution of Mg(NO3)2·with Al(NO3)3), with polyimide (PI) and epoxy matrix, respectively. According to TEM and XRD, in both nanocomposites the LDH were exfoliated. In parallel, the mechanical performance of these systems was enhanced. In CPNC with a PI matrix the elongation at break (εb) and maximum tensile strength (σy) were obtained at ca. 4 and 5 wt% organoclay loading, respectively. Similarly, for CPNC with epoxy matrix εb reached maximum at about 3 wt%, while σy continuously increased up to the highest organoclay concentration of 7 wt%. For both systems the tensile modulus and thermal decomposition temperature also increased, while the thermal expansion coefficient decreased with LDH loading. The fine dispersion of the inorganic component was illustrated by photographs, which showed that in the whole range of LDH concentration the epoxy nanocomposites remain transparent. While exfoliation of any layered nanofiller is easier in a matrix with polar groups, dispersion of clay or LDH in non-polar polymers is difficult. For these reason two publication from B. Qu laboratories are of particular interest [Chen and Qu, 2003; Chen et al., 2004]. The first describes exfoliation of Mg3Al(OH)8(C12H25SO4), or MgAl-LDH for short, by intercalation with PE-g-MAH in a solution of xylene under reflux. 196
Intercalation of Clay
Table 31 Examples of polymeric nanocomposites based on layered double hydroxide (LDH) [Leroux and Besse, 2001] Polymer
LDH
Method*
d (nm)
PANI
Cu0.66Cr0.33(OH)2(terephthalate)0.17·nH2O
a
1.33
PANI
Cu0.66Al0.33(OH)2(hexacyanoferrate)0.17·nH2O
a
1.35
PVAl
Ca0.66Al0.33(OH)2(OH)-0.33·H2O
a
1.8
Poly(α, β-aspartate)
Mg0.74Al0.26(OH)2(CO3)2-0.13(NO3)-0.006·0.32H2O
a, b
0.90
PSS
Mg0.66Al0.33(OH)2(CO3)2-0.17·nH2O Zn0.75Al0.25(OH)2(CO3)2-0.13·nH2O
cop
2.08
PSS
Zn0.75Al0.25(OH)2(CO3)2-0.13·nH2O
co p
2.16
a
2.32
2-
PS
Mg0.66Al0.33(OH)2(terephthalate)
PAA, PVS
Mg0.66Al0.33(OH)2(CO3)2-0.17·0.17·nH2O
cop
1.20, 1.31
PAA, PVS
Zn0.75Al0.25(OH)2(CO3)2-0.13·nH2O
co p
1.24, 1.33
PVS
Co0.75Al0.25(OH)2(OH)-0.13·nH2O
cop
1.33
cop
1.24, 1.31, 1.96
·0.17·nH2O
0.17
PAA, PVS, PSS Ca0.66Al0.33(OH)2(CO3)2-0.17·nH2O
Polyacrylate
Mg0.66Al0.33(OH)2(NO3)-0.33·nH2O
a
1.34
Polyacrylate
Ni0.7 Fe0.3(OH)2·nH2O
b
1.26
PSS
Zn0.66Al0.33(OH)2(Cl)-0.33·0.63H2O
a-c
1.56, 2.12, 1.98
PEG
Cu0.66Cr0.33(OH)2(Cl)-0.33·1.14H2O
b
3.01
PEG/alkenyl sulfonic acid
Cu0.66Cr0.33(OH)2(Cl)-0.33·1.14H2O
b
3.74
a
1.42
PANI sulfonate Cu0.66Cr0.33(OH)2(dodecylsulfate)-0.33·nH2O
Notes: PANI = polyaniline, PVAl = poly(vinyl alcohol), PSS = poly(styrene sulfonate), PS = Polystyrene, PAA = poly(acrylic acid), PVS = poly(vinyl sulfonate), and PEG = poly(ethylene glycol) *Methods a, b, c are described in the text, while cop indicates a templating reaction, during polymer coprecipitation with a [Mg2Al] or [Zn3Al] LDH material [Oriakhi et al., 1996]
197
Clay-Containing Polymeric Nanocomposites The platelets of LDH were about 0.48 nm thick and ca. 70 nm in diameter. The nanocomposites containing 5 wt% LDH had a higher decomposition temperature (by ca. 60 °C), and greater thermal stability than PE-g-MA. In the second publication Zn3Al(OH)8(C12H25SO4), or ZnAl-LDH for short, was prepared by spontaneous self-assembly in an aqueous solution of Zn(NO3)2, Al(NO3)3, and C 12H 25SO 4Na at pH ≅ 10. As before, the exfoliation of ZnAl-LDH was accomplished by refluxing its suspension in a xylene solution of (unmodified!) LLDPE for 24 h. The exfoliated nanocomposites contained up to 20 wt% of ZnAl-LDH. Better thermal stability was observed. The method is expected to be applicable to other polymers, viz. PP, PS, rubbers, and the polar ones. Another interesting layered material is Zr(HPO 4) 2·H 2O, α-zirconium phosphate (α-ZrP). Clearfield et al., synthesised it in 1964 and for years studied its ion exchange ability [Clearfield et al., 1969; 1972]. In comparison to MMT the α-ZrP has higher charge (CEC = 2), smaller interlayer spacing (d001 = 0.76 nm), and it can be prepared with a desired aspect ratio (at least 100) and particle size distribution. Crystalline α-ZrP has the layers formed by Zr and O atoms of the phosphate groups, with one -OH group pointing into the interlamellar galleries. More recently, to study the interrelations between CPNC structure and performance, the team prepared the first nanocomposites with exfoliated α-ZrP in epoxy [Sue et al., 2004]. Thus, α-ZrP was synthesised, intercalated with monoamine-terminated polyether (Jeffamine M715), and then 1.9 vol% of the intercalated α-ZrP was incorporated into DGEBA. According to TEM, XRD and the optical transparency of the cured nanocomposites, full exfoliation was achieved. The influence of hydrogen-bonded α-ZrP intercalate on the mechanical properties of epoxy was investigated. The volume percentages of the specimen components were: α-ZrP = 1.9, Jeffamine = 18.7 and DGEBA + curing agent = 79.4 vol%. Thus the cured specimen contained a large amount of the plastifying intercalant: CH3O(CH2CH2O)14CH2CH2NH2. The rubbery plateau modulus of the nanocomposite was about 4.5 times higher than that of the matrix. Upon the addition of 1.9 vol% α-ZrP, the tensile modulus increased by 50%, and the yield strength improved by 10%, but the elongation at break was drastically reduced. However, the mode-I critical stress intensity factor, KIC, indicated that the fracture toughness is not significantly affected by the addition of the intercalated nanofiller – the epoxy resins are inherently brittle and/or notch-sensitive. Nanocomposites of PA with up to 65 wt% of hydroxyapatite have been prepared for use as a load-bearing bioactive material for bone repair or substitution [Jie et al., 2003].
2.3.11 Summary of the Intercalation Methods Clay intercalation for use as rheological additives, catalyst supports or nanoreinforcements has been practiced for nearly 70 years. The initial applications were for aqueous systems, and then for organic ones (e.g., thickening of oils or greases). Intercalation for the preparation of CPNC is more recent, but the old technology (developed for the greases) dominates the market. The developments are summarised in Table 32.
198
Intercalation of Clay
Table 32 Summary of intercalation methods Clay
Intercalant
d-spacing (nm) Comments
MMT
None; (CEC = 1.0 to 1.2 meq/g)
0.96
Dry clay
MMT
Water
1.3
Ambient humidity
MMT
Ethylene glycol or sorbitol 1.7
Thickener
Clays with CEC CH3(CH2)n NH4+ cations = 0.8 to 1.5 meq/g
1.3-2.2
Spacing depends on CEC and on n (stepwise)
Clays
Phosphonium or sulfonium salts with aliphatic or aromatic radicals
(as above)
Alternatives to ammonium; in aq. or organic solvent
Clays (CEC ≥ 0.75)
Onium salt + organosilane ?
Na-MMT
Aq. solution of MB2HTA-chloride
⇒ exfoliation
Organic solvent thickener
Na-MMT
Aq. solution of hydroxy- ⇒ exfoliation polyoxy-ethylene-tri-alkylammonium chloride
Thickening of highly polar organic solvents
MMT
Aq. solution of PVAl
Na-MMT
Aq. solution of ω-amino- ⇒ exfoliation C12-18 acid onium ion, viz. H3N+C12 H24 COOH and H3N+C12 H25 with caprolactam
MMT
ODA-, 2MODA, 2M2ODA cation, and aminoethyl aminopropyl trimethoxysilane
⇒ exfoliation
For PA-based CPNC (AlliedSignal)
Na-MMT + aqueous HCl
Liquid polybutadiene (PBD)
⇒ exfoliation
For rubber-based CPNC (Toyota)
Na-MMT (CEC Aq. 2M2ODA chloride, = 1.2 meq/g) then toluene solution of end-hydroxylated PBD
⇒ exfoliation
For rubber-based CPNC (Toyota)
Na-MMT (CEC Aq. chloride melt= 1.2 meq/g) compounding with lowMW PP-MA
⇒ exfoliation
For rubber-based CPNC (Toyota)
≈ 2.0
Rheological additive
PVAc 88% hydrolysed For PA-based CPNC (Toyota)
199
Clay-Containing Polymeric Nanocomposites
Table 32 Continued Clay
Intercalant
d-spacing (nm) Comments
A layered material, e.g., MMT
Hydrolysed alcoholates: Si(OR)4, Al(OR)3, Ge(OR)4, Si(OC2H5)4, Si(OCH3)4, Ge(OC3H7)4, Ge(OC2H5)4
⇒ exfoliation
For PO-based CPNC (Dow)
Clay filler
Amino-functional silane + ⇒ exfoliation carboxylated or maleated semicrystalline PO
For PO-based CPNC
Acidified clay; H+MMT
Lewis bases, esp. primary ⇒ exfoliation amines
Tested for epoxyand PU-based CPNCs
Synthetic smectite
Aq. solution of Intercalation ethylenediamine + polyol, then oleophilic solution of triethyl phosphate (TEP)
2 weeks intercalation for fire-resistant clays
MMT
Aq. solution of PVP (at least 10% of water is needed)
1.5-11.0; 3-4.5 Thickener or for is optimum drug-delivery
MMT
Polar, water soluble compounds
⇒ exfoliation
Smectite
Aq. solution of AlCl2(OH) 2-6 (rigid) + ZrOCl2
MMT
Waterless PEG in 2-6 h
1.77
Organoclay
Monomer or polymer
⇒ exfoliation
Inorganic Inorganic polymer having 1.5-6.1 layered materials colloidal particles, e.g., hydrolysable metallic alcoholates
200
For CPNC as well For catalyst support
For PA-type For TS, TP or rubber-based CPNC
Exfoliation of Clays
2.4
Exfoliation of Clays
The high surface-to-volume ratio of nanoparticles leads to a high reinforcement efficiency. Thus, CPNCs with well dispersed platelets at a low clay loading of 2 to 5 wt% show highly increased modulus, yield strength, DTUL as well as reduced flame propagation and permeability. In the crystallisable polymer matrix, the clay platelets (if not totally covered by the intercalant organic tails) promote faster crystallisation and higher levels of crystallinity, which results in improved solvent and moisture resistance, but reduced impact strength. The presence of clay may result in modification of the crystalline structure of the matrix polymer (e.g., α to γ transformation in PA-6), which may promote enhancement of the performance characteristics. Owing to the nature of the nanoscale reinforcement, the CPNC may be treated as an improved grade of a homopolymer, hence it may be used as a replacement for its matrix polymer in diverse multicomponent polymeric systems, viz. blends, composites or foams. For example, Akkapeddi in 2000 reported using CPNC for making either short or long glass fibre (GF) reinforced composites, getting good processability (e.g., fast moulding cycle), low density, further improvements of modulus, strength, moisture resistance, permeation barrier, etc. The materials were aimed at automotive parts (viz. fuel system components, fuel tanks, door and rear quarter panels, consoles, door panels, pillars, under hood components), packaging (containers, films for food and electronics packaging), appliances, building & construction, electrical & electronic, lawn & garden, power tool applications, etc. There are four basic structures for polymer/clay mixtures: 1. Conventional clay-filled composite with micron-sized aggregates of clay particles. 2. Nanocomposites with intercalated clay. 3. Exfoliated nanocomposites with locally ordered structure, where the ordering is imposed by flow and concentration, φ > φmax. 4. Exfoliated nanocomposites with disordered structure, φ < φmax. Not all performance characteristics depend to the same degree on exfoliation. However, as Kojima et al. [1993] have shown (see Table 33), the benefits increase with the degree of dispersion. Thus, the goal of CPNC technology is to achieve the highest possible degree of clay exfoliation, i.e., the best dispersion and distribution in a polymeric matrix. Exfoliation is the last step in the preparation of CPNC. The methods for achieving it can be discussed under three titles:
201
Clay-Containing Polymeric Nanocomposites
Table 33 Effect of the degree of dispersion on performance. Data [Kojima et al., 1993] Clay (wt%)
E (GPa)
σ (MPa)
NIRT (kJ/m2)
HDT (°C)
PA-6
0
1.1
69
2.3
65
Intercalated
5
1.0
61
2.2
89
4.2
2.1
107
2.8
14 5
Material
Exfoliated
Notes: E is tensile modulus; σ is yield stress; NIRT is notched Izod impact strength at room temperature; HDT is heat deflection temperature
1. Polymerisation in the presence of organoclay. 2. Melt compounding a polymer with a suitable organoclay complex. 3. Other exfoliation methods: • Combining the organoclays with latex. • Ultrasonic exfoliating of organoclay particles in a low MW polar liquid. • Others, viz. sol¯gel templating, co-precipitation, etc. In Part 4 of this book the patent literature on CPNC is summarised, first with thermoplastic polymer matrices, then with those of thermosets and elastomers. Since the maximum performance is achieved when clay is exfoliated, the exfoliation methods are discussed there in depth. For this reason, this chapter on the basic elements of CPNC technology will focus on the general aspects of exfoliation technology.
2.4.1 Principles The layered inorganic materials that are to be used as nanofillers, e.g., natural or synthetic clays, have strong ionic and van der Waals interactions and small interlamellar spacing, much too small for allowing organic molecules, monomers, oligomers or polymers, to diffuse into. Expansion of the interlayer spacing to total exfoliation is accomplished in several steps. Since clays are strongly hydrophilic, the first step is suspending them in an aqueous medium. Few types of hydrophilic, highly polar, low MW polymers (viz. PEG or P2VP) may be used to exfoliate directly such a swollen clay. For the great majority of CPNCs with hydrophobic polymeric matrices the swollen clay must be intercalated with one or two intercalants (e.g., onium salt and silane or an epoxy compound). The hydrophobic monomers and polymers require that the nanofiller is pre-intercalated, and often compatibilised. The following four pathways have been distinguished [Usuki, 2001]: 1. Hydrophilic matrix with strong polar groups, e.g., P2VP or PEG:
clay clay 202
water
water
hydrophilic polymer
CPNC
swollen clay polar organic
swollen clay
CPNC
Exfoliation of Clays 2a. Hydrophobic matrix with strong polar groups, e.g., PA or PEST:
clay
monomer
intercalant(s)
intercalated clay
expanded clay
polymerisation
CPNC
2b. Hydrophobic matrix with strong polar groups, e.g., PA:
clay
intercalant(s)
+ polar polymer
intercalated clay compounding
CPNC
3. Hydrophobic non-polar matrix, e.g., PP:
clay
intercalant(s)
+ non-polar polymer
intercalated clay + compatibiliser compounding
CPNC
The difficulty in achieving exfoliation increases in the sequence from pathways 1 to 3. The key to good performance is on the one hand obtaining initial expansion of the interlayer galleries for penetration by monomer(s) or polymer, and on the other good thermodynamic miscibility between the preintercalated clay platelets and the polymeric matrix. In the case of the reactive exfoliation processes (pathways 1 and 2a) an important additional condition is that the polymerisation inside the gallery is at least as fast as that outside it. When this condition is not met, the polymer formed outside the interlamellar galleries may hinder expansion of the interlayer space, leading to intercalation but not exfoliation. It is noteworthy that the clay and the intercalating onium salt often affect the polymerisation rate – one must ascertain that these effects are positive, i.e., that these catalyse the reaction. Another often forgotten requirement is the stability of the clay-matrix bond. The CPNC must survive the forming stage! There are very few polymers that can be processed below 200 °C. Above this temperature the ammonium intercalants (especially quaternary) are not thermally stable (especially in the presence of oxygen and shear field). Reaggregation of clay platelets (that were exfoliated during in situ polymerisation) during melt processing of CPNC has been reported. As the listed pathways indicate, either a single one- or a two-step sequential intercalation-cum-compatibilisation may be required. Thus, the pathways focus on the chemical aspects of intercalation. Unspecified in these are the contributions of physics: the thermodynamics that controls the phase behaviour as well as P and T effects on the interlayer spacing, molecular diffusion, flow which introduces dispersive forces, radiation-absorption which may accelerate the exfoliation process under the influence of, e.g., ultrasonics, etc. Pathway 1 has been presented in sufficient detail in Section 2.3.5.2, hence there is no need to discuss it again. Pathway 2a is pertinent to virtually any polymeric nanocomposite, with thermoplastic, thermoset or elastomeric matrix hence a brief description of these systems will be given. Finally, pathways 2b and 3 belong to the most important mechanical exfoliation methods and need more detailed analysis. 203
Clay-Containing Polymeric Nanocomposites
2.4.2 Polymerisation in the Presence of Organoclay Several different methods have been explored to prepare CPNC by chemical reactions. Following Usuki’s suggestions [Usuki, 2001], one may categorise them as: 1. Monomer intercalation – viz. preparation of PA-6 nanocomposites. 2. Monomer modification – viz. preparation of acrylic-based nanocomposites. 3. Non-reactive intercalated clays – viz. preparation of styrenic-type CPNC. 4. Co-vulcanisation – viz. preparation of NBR-based nanocomposites. 5. Common solvent method, frequently used to prepare PI-based CPNC. 6. Others. In the methods 1 and 2 the clay is intercalated with a compound that subsequently enters the polymerisation reaction – polycondensation in method 1, and radical polymerisation in method 2. Thus, as a result of either of these two processes end-tethered CPNC is obtained. 2.4.2.1 Monomer Intercalation – PA-6 Nanocomposites Preparation of CPNC via in situ polymerisation was used at the Toyota Institute for the first nanocomposites with PA-6 as matrix. The process involved three steps [Deguchi et al., 1992]: 1. Intercalation of Na-MMT with ω-amino dodecanoic acid chloride in water – the interlayer spacing increased from d001 = 0.96 (dry) to 1.3 (wet) to 1.8 nm (intercalated). 2. Mixing the intercalated clay with ε-caprolactam and water at the ratio of 1:9:9 - the interlayer distance further increased to d001 = 3.87 nm. At this stage, a catalyst and an activator were added. 3. Polymerisation at 100 and 250 °C for 48 h followed by TSE extrusion. A high level of exfoliation was obtained, with rare short stacks containing twoto-three clay platelets evident in TEM micrographs. The following year, Usuki et al. [1993a] found that clay, intercalated with ω-amino dodecanoic acid can be swollen by molten ε-caprolactam (30 to 98 wt%). Polymerisation under nitrogen at 250 to 270 °C for 48 h resulted in well-dispersed silicate layers in a PA-6 matrix. The TEM images showed single platelets even in CPNC containing 31 wt% MMT. At 2 wt% of organoclay loading, the modulus increased to 1.5 times that of PA-6, the heat distortion temperature increased from 75 to 140 °C, and the moisture permeability was reduced by 50%. At the same time the density increased from 1140 to 1150 kg/m3, i.e., by a theoretically predictable 0.88% [Ube Industries, Ltd., 2000]. The method was reported economic, efficiently producing a PNC with any PA matrix, having a wide viscosity range for all processing operations. The material had good mechanical properties, heat resistance, improved dye affinity and whitening resistance during stretching. When discussing exfoliation it is important to pay attention to clay concentration and the aspect ratio, p. During shear flow platelets rotate and the maximum packing volume fraction of such ‘encompassed volume’ spheres is φmax = 0.62. It can be shown that for such platelets:
φmax = 0.93/p
(24)
When: φ > φmax the platelets cannot rotate and have to locally align. In this case the distance between them can be calculated from geometry as: 204
Exfoliation of Clays
d001 = (hclay ρclay / ρ polymer )(100 / wlocal − 1) + hclay ≈ + a0 + a1 / w
(25)
where hclay is the platelet thickness, ρclay and ρpolymer are densities, and wlocal is the local clay concentration (in wt%), a 0 and a 1 are corresponding equation parameters. Owing to the tendency of the clay platelets to align parallel to each other in local stacks, the clay concentration within the stack is not known, hence the parameters in Equation 25 cannot be calculated from first principles. However, since it can be assumed that the concentration within stacks is proportional to the average, global concentration: wlocal ∝ w (wt%), d 001 should be inversely proportional to the clay content. Thus, for organoclays with large aspect ratio the formation of local stacks takes place at low concentration, and the interlayer spacing may be small, not because of a lack of exfoliation, but due to the crowding effects. It is worth noting that at φ > φmax the interlayer thickness is independent of the aspect ratio – p only dictates the concentration at which the local stacking starts. Okada and Usuki [1995] published data on the interlayer spacing in PA-6/clay, measured by XRD and TEM. Using Equation 25 the maximum packing for free platelet rotation is predicted for organoclay content below 1.13 wt%. From the relation, at a clay content of 2 wt% the interlayer spacing d001 is 44.2 nm (see Figure 44). Evidently, starting with ω-aminoacid end-tethered CPNC were produced. Emulsion polymerisation was used to prepare CPNCs with PMMA as a matrix [Choo and Jang, 1996]. Thus, MMA was polymerised in the presence of Na-
Figure 44 Basal spacing in CPNC versus organoclay content as measured by XRD and TEM. The organoclay was MMT intercalated with ω-dodecyl acid: NH2(CH2)11COOH and ε-caprolactam . Data [Okada and Usuki, 1995]. The line indicates Equation 25 dependence with a0 = 1.37; a1 = 0.856; and the correlation coefficient r = 0.991.
205
Clay-Containing Polymeric Nanocomposites MMT. The intercalating PMMA macromolecules were found oriented parallel to the clay lamellae. The interlayer distance decreased with MMT loading from d001 =1.73 nm at 10 phr to 1.28 at 50 phr (33 wt%). Both the thermal stability (char formation on decomposition) and tensile properties were enhanced. The system was not end-tethered, but the ion-dipole interactions were strong enough to be the driving force for the introduction of MMA monomer into the interlamellar galleries and bonding of the PMMA molecules to clay surfaces. This work was extended to the emulsion copolymerisation of ABS in the presence of clay [Jang et al., 2001]. The clay content varied from 10 to 50 phr and correspondingly the interlayer spacing changed from d001 = 1.75 to 1.56 nm. Thus, it seems that the process results in expansion of Na-MMT particles (ca. 500 nm diameter) by monomer intercalation. In either case clay was not exfoliated. Similar results were obtained by adding an aqueous dispersion of Na-MMT to SBR latex, then coagulating the emulsion, washing, drying and curing [Zhang et al., 2000b]. Clay ‘bundles’ 4 to 10 nm thick and about 200 to 300 nm long were observed, thus intercalation not exfoliation was obtained. Better results for MMT exfoliation in PMMA matrix were obtained using the monomer modification method, described in Section 4.2.2.2. Exfoliation was also obtained using a polar intercalant-compatibiliser, e.g., PCL [Kim et al., 2001a]. The other approach is based on clay intercalation with a compound that chemically participates in the subsequent polymerisation. Full exfoliation was reported for a PE/MMT system prepared by Ziegler-Natta polymerisation of ethylene in the presence of organoclay [Jin et al., 2002]. The latter was Cloisite® 30B (MMT with MT2EtOH) treated with TiCl4 to affix the catalyst to the intercalant –OH group. As was expected, polymerisation in the presence of Na-MMT had little effect on the clay interlayer spacing. The authors processed the exfoliated polymerisation product by compression moulding or TSE extrusion. The processing resulted in a partial re-aggregation of MMT platelets with d001 ≅ 1.4 nm. The extent of re-aggregation depended on the processing conditions – it was slight for compression moulding, but quite pronounced for extrusion. The polymerisation route was also selected by Alexandre et al. [2002a,b]. Thus PE was polymerised in the presence of either MMT or hectorite. The clays were first treated with trimethyl aluminium-depleted methyl aluminoxane, and then a Ti-based constrained geometry catalyst and monomer were incorporated. The tensile properties of the resulting CPNC were poor and essentially independent of the nature and content of the silicate. When hydrogen was used to control the molecular weight of PE the tensile modulus was significantly increased (from 0.69 to 2.48 GPa for PE and PE with 11.4 wt% clay), but at a cost of a dramatic reduction of the strain at break (from 244 to 3%, respectively). Clay exfoliation in the reaction products was confirmed by XRD and TEM. However, as it has been reported before, during melt processing of the CPNC the interlayer spacing partially collapsed. Evidently, the system is thermodynamically unstable and the strong solid-solid interaction between clay platelets drives the phase separation. 2.4.2.2 Monomer Modification – Acrylic-Based Nanocomposites Usuki et al. [1995] introduced the term ‘monomer modification’ to indicate that the organoclay was intercalated with a reactive compound, which subsequently participates in polymerisation. One may presuppose that preparation of PA-6/clay CPNC (which starts with ω-amino dodecanoic acid) also belongs to this category. 206
Exfoliation of Clays However, while PA polycondensation results in a single attachment of a macromolecule to a clay platelet, the monomer modification strategy results in a copolymerisation of the reactive monomers, one type of these being ionically bonded to clay. Thus, the resulting copolymer may have numerous molecular units bonded to the clay surface. Monomer modification will usually start with intercalation of clay using molecules that have at least one group capable of reacting with other monomer(s), forming a multitethered, exfoliated CPNC. This approach has been used for the preparation of PMMA/organoclay [Biasci et al., 1994]. The organoclay was MMT intercalated with quaternary ammonium: either 2-(N-methyl-N,N-diethyl ammonium iodide)-ethylacrylate (QD1) or with 2-(N-butyl-N,N-diethyl ammonium bromide)-ethylacrylate (QD4) – owing to the bulkiness of the intercalating molecules only up to 58% of the clay active sites were exchanged. The radical copolymerisation of MMA with the intercalated clay was carried out either in bulk or in acetonitrile at 60 °C. As a result of intercalation and subsequent polymerisation the interlayer spacing increased to d001 = 1.45 to 1.78 nm, respectively. The length of MMA sequence in the formed copolymers was 3.5 and 1.1 to 8.2, after bulk and solution reaction, respectively. Apparently, bridging the interlamellar gallery by the macromolecular chain prevented exfoliation. When the sequence was reversed (intercalating Na-MMT with MMA/QD1 or MMA/QD4 copolymers) the interlayer spacing increased to d001 = 2.96 nm and thermal stability was enhanced. Usuki et al. [1995] prepared CPNC by first copolymerising ethyl acrylate (90 mol% of EA: CH 2 =CHCOOC 2 H 5 ), acrylic acid (10 mol% of AA: CH2=CHCOOH), and a quaternary ammonium salt of dimethyl aminopropyl acrylamide: Q = (CH2=CHCOOCH2CH2CH2N+(CH3)2C3H7)Cl– (0.12, 36, 56, and 0.89 mol%). Next, the copolymer was mixed with an aqueous suspension of Na-MMT to induce intercalation. Four CPNCs were prepared with a clay content of 1, 3, 5, and 8 wt%. CPNCs with MMT ≥ 3 wt% showed high viscosity and yield stress. In comparison to the neat copolymer, incorporation of the end-tethered clay increased viscosity by 4 to 6 orders of magnitude. The CPNCs were used for the preparation of transparent films (crosslinked with melamine). Gas permeability through the film followed Equation 23, with p ≅ 100 (see Figure 45). Heinemann et al. [1999] prepared PO-based PNCs starting with hectorite intercalated with dimethyl benzyl stearyl ammonium chloride (2MBODA), having d001 = 1.96 nm. When HDPE was polymerised in the presence of the intercalated clay, the XRD peak shifted to d001 = 1.40 nm. Best results were obtained for LLDPE-type ethylene-octene copolymers. The authors reported superior performance of the reactively prepared CPNCs over that prepared by melt compounding. However, the intercalated clay was found to interfere with the metallocene catalysts and the processing of these CPNCs resulted in re-aggregation of the clay platelets. A bulk polymerisation method was used to prepare PS-based PNC [Fu and Qutubuddin, 2000; 2001], starting with MMT intercalated by dimethyl vinylbenzyl dodecyl ammonium chloride (2MVBDDA). Owing to good miscibility of 2MVBDDA with styrene, a uniform dispersion of platelets was obtained. Polymerisation for 48 h at 60 °C resulted in exfoliation. At 7.6 wt% of clay, the modulus of the PNC was about 70% higher than that of PS. MMT was intercalated with either oligo (n = 25)-ethylene-glycol diethyl methyl ammonium chloride (SPN) or with methyl-trioctyl ammonium chloride (STN). The organoclays were dispersed in either MMA or styrene by ultrasonication for 207
Clay-Containing Polymeric Nanocomposites
Figure 45 Oxygen permeability through crosslinked acrylic film versus clay content. The line was calculated from Equation 23 with p = 100 . Data [Usuki et al., 1995].
7 h then radically polymerised. For comparison, PMMA and PS were also synthesised in the presence of the quaternary amines [Okamoto et al., 2000; 2001b,c]. In the earlier publication, polymerisation of the CPNCs systematically resulted in decreased interlayer spacing, viz.: 1. MMA/STN; before polymerisation d001 = 2.96 - after 2.66 nm. 2. MMA/SPN; before polymerisation fully exfoliated - after a shoulder at 4.55 nm. 3. Styrene/SPN; before polymerisation nearly totally exfoliated with only a small shoulder at d001 = 3.65 - after a large peak at 3.65 nm. This behaviour suggests a thermodynamically driven phase separation. The dynamic mechanical temperature scans indicated only a minor change in the storage modulus. In the latter publication only the SPN organoclay was used. Polymerisation of the MMA/SPN system was carried out as in the earlier publication, but in the presence of small amount of polar comonomer, viz. N,N-dimethyl aminopropyl acrylamide (PAA), N,N-dimethyl aminoethyl acrylamide (AEA) or acrylamide (AA). The radical polymerisation resulted in exfoliation. The storage modulus versus T scan indicated a significant increase, viz. from 1 to 3 GPa. Similarly good exfoliation was reported for PS/organoclay systems [Zhang et al., 2003]. The authors used one non-reactive intercalant (3MHDA) and three reactive ones, each containing one methacryloyloxy and two methyl groups, and in addition either a benzyl, octyl or hexadecyl group. Na-MMT was preintercalated with each of these, and then dispersed in styrene. The suspension was transformed into oil-phase for γ-ray initiated suspension polymerisation. At 5 wt% organoclay loading, exfoliation was obtained for all three reactive organoclays, while only intercalation occurred with the MMT-3MHDA. 208
Exfoliation of Clays Soap-free emulsion polymerisation of MMA was used to prepare exfoliated CPNC with PMMA as a matrix [Choi et al., 2001]. Thus, the reaction was carried out in two steps. In the initial step, Na-MMT was dispersed in water with 25% (of the total amount) of MMA and 2-acryl-amido-2-methyl-1-propane sulfonic acid (AMPS). The ratio of MMT/AMPS varied from 8 to 67. AMPS played a triple role as a surfactant, intercalating agent and coreactant. In the second step, the remaining monomer was used to complete the reaction. XRD and TEM showed that the PNCs containing up to 10 wt% of Na-MMT were exfoliated. Surprisingly at 10 wt% of Na-MMT the exfoliation occurred after 10 min of reaction. The molecular weight of PMMA in PNC slightly decreased with the AMPS content, while the glass transition temperature (Tg) and storage modulus (E´) significantly increased. 2.4.2.3 Non-Reactive Intercalated Clays Translucent acrylic nanocomposites were recently described [Dietsche et al., 1999; 2000]. The CPNC was prepared by polymerisation of methyl methacrylatedodecyl methacrylate copolymer in the presence of 2-10 wt% bentonite intercalated with N,N,N,N-dimethyl dioctadecyl ammonium (2M2ODA). Addition of n-dodecyl methacrylate improved interactions and accounted for the improved stiffness-to-toughness balance, higher Tg and thermal stability, in comparison to the corresponding copolymer. Evidently, this polymerisation strategy does not create covalent bonding between clay and the acrylic matrix. The only bonding that could take place is by cocrystallisation of the C12 paraffinic chains. Recent patents from Eastman [Barbee and Matayabas, 2000; Matayabas et al., 2000] described PET-based CPNC with enhanced barrier properties. Thus, MMT was intercalated (with a quaternary onium salt) and then treated with a secondary intercalant, e.g., PEG, PCL or vitamin E. The organoclay (d001 = 2.2 to 4.2 nm) was incorporated into PET either by polycondensation or melt compounding, followed by solid-state polymerisation. TEM showed mostly individual platelets with only few tactoids and aggregates. PC-based CPNC was prepared starting with PC-cyclomer and MMT intercalated with dimethyl ditallow ammonium cation (2M2TA; B34 from Rheox, Inc.; d001 = 2.47 nm) well washed with water/ethanol to remove excess quaternary ammonium salt [Huang et al., 2000]. Cyclic oligomers of PC have lower solution and melt viscosities (compared to the corresponding polymer), thus the intercalation process is significantly easier. For example, the organoclay was dispersed in CH2Cl2 and mixed with either PC or PC-cyclomer. After 5 min of mixing with PC, the basal reflection was identical to that of B34. By contrast, intercalation with PC-cyclomer was quick, increasing d001 to 3.62 nm. The enhanced intercalation rate may be related not only to the lower viscosity, hence enhanced diffusivity of the cyclomer, but possibly to the difference in molecular architecture as well, the absence of end groups and the intermolecular interactions between the cyclomer and the clay surface. Similar differences between the intercalation processes of PC and PC-cyclomer were observed during melt processing. After 24 h of melt annealing with PC d001 = 3.27 nm was obtained – the same as obtained by intercalation from solution. By contrast, melt annealing with PC-cyclomer readily resulted in d001 = 3.8 nm. Furthermore, when a mixture was compounded in an internal mixer for 1 h at 180 °C, an exfoliation was achieved. Subsequently, raising the temperature to 209
Clay-Containing Polymeric Nanocomposites 240 °C for 10 min, caused ring-opening polymerisation, converting the cyclomer to linear PC (MW = 40 kg/mol). TEM showed the presence of individual layers as well as tactoids consisting of 3 to 5 platelets. Thus, again partially exfoliated CPNC was obtained without covalent bonding between the clay and the matrix. 2.4.2.4 Co-Vulcanisation Toyota was the first to develop a process for producing polyolefin (PO)-based CPNC [Usuki et al., 1989]. However, the patent claims extend to ‘vinyl-based polymeric compound, a thermosetting resin and a rubber’. Ammonium salt having a terminal vinyl group was used to intercalate MMT. The product was mixed with a vinyl-based monomer and/or oligomer, e.g., ethylene, propylene, butadiene, methylmethacrylate, styrene, etc. The mixture could be polymerised either in bulk, suspension or in solution reaction, by a radical, cationic, anionic, coordination or condensation mechanism. Polymerisation took place within the interlayer space, expanding the interlayer distance to d001 > 3.0 nm. However, to prepare exfoliated CPNC with a rubber, first MMT was reacted with low molecular weight liquid rubber. For example, in an aqueous solution of DMSO a liquid acrylic rubber (amino-terminated butadiene-acrylonitrile rubber (ATBN), MW = 3400, 16.5% AN), HCl, and Na-MMT (clay platelet thickness = 1.0 nm, p = 1000, CEC = 1.19 meq/g) were dispersed. The reaction product was filtered and dried. Pulse NMR indicated that a strong bond was formed between MMT and ATBN, with ca. 20% of the rubber molecules restricted near the interface. XRD indicated that the (001) peak of MMT had disappeared, d001 > 8.8 nm, and the MMT platelets were uniformly dispersed in the matrix. Next, the complex (containing 32.5 wt% of MMT) was cooled with liquid N2, crushed by a hammer mill and mixed with NBR (AN = 33%). The blends (containing 5 or 10 wt% MMT) were vulcanised with sulfur. Superior performance in tensile, dynamic viscoelastic, and swelling tests were observed. A similar method has been used to prepare ATBN-intercalated MMT, subsequently mixed with NBR and vulcanised [Kojima et al., 1993a]. TEM of the resulting CPNC showed intercalated, well-dispersed short stacks of MMT dispersed in the matrix. The permeability reduction for H2 and H2O through the CPNC (containing 3.6 wt% of MMT) when compared to a standard rubber with 10 vol% of carbon black was reduced by 37 and 26%, respectively. 2.4.2.5 Common Solvent Method – Polyimide Based Nanocomposites Yano et al. [1993; 1997] discovered that organoclay intercalated by ion exchange with dodecyl ammonium chloride (DDA), could be homogeneously dispersed in N,N-dimethyl acetamide (DMAc). Thus, to prepare PI/clay films, organoclay was dispersed in DMAc, then diamino diphenyl ether, pyromellitic dianhydride was added and the solution stirred for 6 h. The film was cast from a homogeneous mixture of organoclay and polyamic acid, and was heated at 300 °C to polymerise. With 2 wt% of clay the water and CO2 permeability decreased to 50%. XRD and TEM showed that the CPNC was exfoliated [Lan and Pinnavaia, 1994]. As an extension of this procedure, nematic liquid crystal (LC) with (up to 2 wt%) MMT was prepared [Kawasumi et al., 1998]. Thus, MMT was intercalated with a variety of ammonium cations, including 4-cyano-(4´-biphenyloxy)-undecyl ammonium salt, which showed enhanced miscibility to the LC. 210
Exfoliation of Clays The intercalated clay was than dispersed in dimethyl formamide (DMF) and the LC was added. The solvent was slowly evaporated at 50 °C under vacuum, while stirring. When the clay had good affinity for LC, the system was homogeneously dispersed. The CPNC exhibited a bi-stable and reversible electrooptical effect between a light scattering state and transparent state, which could be selected by changing the frequency and voltage of the applied electric field. In 1999 Yang et al., further examined the common solvent method for the preparation of CPNC with PI as matrix. The work focused on the influence of the intercalation agents upon dispersability of organoclay in a selected solvent (DMAc) as well as that in PI. First, Na-MMT was intercalated with either amino acids, primary aliphatic amines or quaternary ammonium salt, viz.: (1) p-aminobenzoic acid (ArNCO), (2) ethanolamine (HONH), (3) N,N-dimethyl aminoethyl methacrylate (DMAEM), (4) dodecylamine (12CNH), (5) 1-hexadecyl amine (16CNH), (6) hexadecyl-trimethyl amine (3MHDA) and (7) 6-aminohexanoic acid (6NCO). The interlayer spacing of the organoclays increased from d001 = 1.26 nm (Na-MMT) to, respectively, 1.27, 1.29, 2.16, 2.87, 3.70, 4.05, and 4.96 nm. However, the dispersability did not follow the same order – the best dispersability in DMAc and PI was observed for the intercalants 1, 2, 5 and 7, the worst for 3 and 6. Evidently, the dispersability depends on the type of functional groups present in the solvent or PI and the bulky group of the intercalation agent. Thus, one can disperse clay aggregates (or tactoids) well without exfoliation, but the CPNC properties depend on the homogeneous dispersion of MMT platelets. The organosoluble PI was based on pyromellitic dianhydride (PMDA) and 4,4´-diamino-3,3´-dimethyl-diphenyl methane (MMDA). The organoclay was added to DMAc and agitated for 3 h at 90 °C before adding it to MMDA solution in DMA, which was followed by addition of PMDA. The mixture reacted at room temperature for 6 h then it was cast on a glass and heated at 100 °C for 6 h, 150 °C for 4 h and 270 °C for 2 h under N2 to obtain MMT/PI hybrids. The properties of the final product greatly depended on the degree of dispersion. Only when the MMT was well dispersed did the CPNC show good performance, e.g., simultaneous high strength and toughness, improved thermal stability, decreased thermal expansion coefficient, retention of the solubility of the polyimide matrix and high optical transparency. The chemical structure of an intercalation agent imposes great influence on the dispersion of MMT. MMT modified with 1-hexadecyl amine (HDA) showed the best dispersion behaviour and the best set of properties. Tyan et al. [1999a,b] observed that imidisation of polyamic acid (PAmA) is accelerated by the presence of organoclay. The organoclay was prepared by intercalating Na-MMT (CEC = 0.764 meq/g) with p-phenylene diamine. During the intercalation only one amine group was converted into a cation (–NH3+). The other -NH2 group was able to react with the dianhydride end group of PAA when these molecules diffused into interlayer galleries. PAA was synthesised by dissolving 4,4´-oxydianiline (ODAn) and pyromellitic dianhydride (PMDA), in DMAc at 25 °C under N2. To a viscous PAmA solution a suspension of organoclay in DMAc was added then mixed to obtain PAA/organoclay/DMAc. During the reaction, the intercalant became an integral part of the PI molecules, making these nanocomposite more thermally stable and mechanically stronger. The spin-coated films containing 0, 2, 5 or 7 phr of organoclay were dried then heated at 150, 200, 230 and 250 °C. XRD indicated that the interlayer spacing 211
Clay-Containing Polymeric Nanocomposites of organoclay was d001 = 1.549 nm, but all PAA/organoclay systems were exfoliated. In the presence of organoclay the imidisation temperature and the imidisation time were reduced. For example, the imidisation temperature was reduced from 300 to 250 °C, while at 250 °C the reaction time was reduced to 15 min (for 7 phr of organoclay). Initially, the reaction followed the first-order kinetics: ln(1-p) = -kt
(26)
where p is the extent of reaction at a reaction temperature T, k is the rate constant and t is the reaction time. The temperature effects were well-described by the Arrhenius dependence: k = A exp{-Ea/RT}
(27)
where A is the pre-exponential factor, R is the gas constant, and Ea is the activation energy (see Figure 46). In the next paper 4,4´-oxydianiline (ODAn) was used as an intercalant [Tyan et al., 2000]. After drying in a vacuum oven at 80 °C TGA indicated the presence of 5.41 wt% structural water and 29.0% of the intercalant. The interlayer spacing of the organoclay was d001 = 1.5 nm. PAmA was prepared by dissolving ODAn in DMAc under N2 then adding 3,3´-4,4´-benzophenone tetracarboxylic dianhydride (BTDA). The system was stirred for 1 h, which produced a viscous PAmA solution to which different quantities of organoclay suspension in DMAc were added (to yield 0, 1, 2, 3, 5, and 7 wt% of organoclay in the product), and then mixed for 12 h. The organoclay-PAmA suspension was cast, and then solvent was removed under vacuum at 30 °C over 48 h. Imidisation was carried out in an air circulation oven at 100, 150, 200, and 300 °C for 1 h and then at 400 °C for 5 min. In the final product full exfoliation was obtained. The
Figure 46 Arrhenius plot of the imidisation rate constant for 0, 2, 5 and 7 phr of organoclay in the CPNC. Parameters of the Arrhenius equation are listed. Data [Tyan et al., 1999a,b].
212
Exfoliation of Clays modulus, the maximum stress and the elongation at break of these CPNCs are increased with organoclay content – see Figure 47. CPNC with PI as a matrix has also been produced starting with MMT (CEC = 1.15 meq/g) intercalated with N-hexadecane pyridinium chloride (HP-MMT) [Gu and Chang, 2001]. PAA was prepared in a THF/MeOH solution, dissolving in it at room temperature first ODAn then pyromellitic dianhydride (PMDA) and mixing the resulting solution for 3 h. Then a solution of triethylamine (Et3A) was added and the process took another 3 h of stirring to complete. Next, to a stirred viscous pale-yellow solution a suspension of HP-MMT was added at 30 °C. After 6 h of stirring, the solution was coated on a glass plate to a thickness of 250 μm. The film was dried at 25 °C for 30 min, 40 °C for 30 min, and 80 °C for 2 h, and then cured at 150, 200, and 300 °C under N2 to produce a transparent PI-based CPNC film with a thickness of 26 ± 2 μm. Results showed that the organoclay was fully exfoliated in the polymer matrix, had high modulus (2.38 GPa), tensile strength (110 MPa), elongation at break (11%), and low coefficient of thermal expansion (35%, compared to 100% for neat PI). In the next paper from the group, the same HP-MMT was used to prepare CPNC with PI as a matrix using two methods [Gu et al., 2001]: 1. Blending a DMAc solution of ODAn with the organoclay dispersion in DMAc before adding PMDA; and 2. Blending a DMAc solution of PAmA with the organoclay dispersion in DMAc. Compositions containing 1, 3, 5 and 10 wt% of organoclay were prepared – the interlayer spacing was: d001 = 2.2 to 1.4 nm, respectively. Tensile, thermal, dielectric and water absorption properties of these CPNCs were studied. The properties depended on the clay type, clay concentration, and the method of preparation. The best performance was obtained using 3 wt% HP-MMT prepared by the
Figure 47 Mechanical properties (tensile modulus, E; tensile strength, σ; and elongation at break, ε) of PI with MMT-ODA. Data [Tyan et al., 2000]. The three functions are well approximated by straight lines – their parameters are given.
213
Clay-Containing Polymeric Nanocomposites second method: tensile strength = 110 MPa, tensile modulus = 2.35 GPa, elongation at break = 14.5%. Delozier et al. [2002] used several methods to prepare PI-based CPNC, with MMT that was intercalated with a long chain aliphatic quaternary ammonium cation. The methods included: 1. Mixing the organoclay into a high molecular weight poly(amide acid) solution; 2. Mixing as in (1), followed by sonication; and 3. In situ preparation, starting with dispersion of organoclay in N-methyl-2pyrrolidinone (NMP), then addition of 4,4´-oxydianiline (ODAn) and 3,3´,4,4´-benzophenone tetracarboxylic dianhydride (BTDA) to prepare high molecular weight poly(amide acid). The imidisation step involved a hightemperature treatment (i.e., 1 h each at 100, 200 and 300 °C in air or N2). The best results were obtained using the latter approach - the specimen obtained using the other methods had aggregated clay particles that caused film failure in tensile tests (stress concentration). However, high shear mixing using a homogeniser improved intercalation and exfoliation, but the films were extremely brittle making mechanical properties impossible to measure. The CPNCs that contained 3-8 wt% of organoclay, were characterised by DSC, TGA, TEM, XRD and thin film tensile properties. After thermal treatment of amide acid films that caused imidisation, the films with clay were darker than those without and the interlayer spacing was reduced. During imidisation at T = 200 to 300 °C thermal degradation of the quaternary ammonium took place. The effect was less pronounced when imidisation was carried out under N2. The degree of dispersion was assessed from TEM micrographs. There is evidence that in the in situ prepared CPNC exfoliation dominates. The exfoliated particles were 200¯700 nm long and 1¯10 nm thick. However, XRD analysis indicated that the interlayer spacing was about constant, d001 = 1.34 nm - the peak intensity increased with clay content. It is noteworthy that this spacing is closer to that known for Na-MMT than for the organoclay (d001 = 2.37 nm). The polyimide/organoclay hybrid films exhibited higher room temperature tensile moduli, but lower strength and elongation to break than the control films. The Tg for the neat ODAn¯BTDA system was 283 °C and remained constant within 4 °C for the hybrid films. At 5 to 8 wt% of clay the temperature for 5% weight loss was ca. 517 °C. Chang et al. [2001b,c] studied the preparation of CPNC in polybenzoxazole (PBO) matrix. The common solvent method was chosen with DMAc as the solvent. MMT (CEC = 1.19 meq/g) was intercalated with primary hexadecyl ammonium salt (MMT-HDA). Polyamic acid (PAmA) was synthesised in DMAc, then the suspension of MMT-HDA in DMAc was added. The solution was cast, solvent was evaporated at 50 °C and the films, 10-15 μm thick, were thermally treated. The conversion to PI was carried out at 300 °C for 1 h under N2. The final conversion to PBO was accomplished at 550 °C. CPNCs with 0, 1, 2, 4 and 8 wt% of organoclay were prepared. Good exfoliation in systems containing up to 4 wt% MMT-HDA was found, although short stacks were visible in TEM micrographs. The specimens were thermally stable up to 611 °C, retaining about 75% of weight at 900 °C – clay presence hardly affected these properties. However, the tensile strength was doubled for 4 wt% loading and modulus increased by 37%. A one-step method for the solution preparation of PI nanocomposites was also proposed [Huang et al., 2001a,b]. First, MMT was intercalated with primary (dodecyl or hexadecyl, DDA or HDA, respectively) or quaternary (trimethyl 214
Exfoliation of Clays hexadecyl, 3MHDA) ammonium ions. The organoclay was then dispersed in m-cresol by mixing and ultrasonication for 1 h at 100 °C. To the suspension the reactants (diphenyl ether-tetracarboxylic anhydride and diamino dimethyl diphenyl methane) were added. The reaction was conducted at room temperature for 2 h, and at 180 °C for 3 h. The solution was cast, then heated stepwise at 70, 120, 200 and 270 °C for a total of 22 h. The interlayer spacing of MMT (d001 = 1.3 nm) increased upon intercalation to 1.75, 3.37 and 4.01, for DDA, HDA and 3MHDA, respectively. Nanocomposites comprising: 0, 2, 3.2, 5, 10 and 20 wt% of MMT were prepared with MMT-3MHDA. Fully exfoliated CPNCs were obtained only for 2 and 3.2 wt% of clay. The mechanical properties linearly increased with clay content up to 5 wt%. In the full range of investigated compositions the decomposition temperature increased from 518 to 524 °C, while Tg(PI) = 286 °C increased by 4 °C and the coefficient of thermal expansion (CTE) hyperbolically decreased following the dependence: CTE = a0 + a1/(a2 + w)
(28)
The PI data for MMT content (w) versus CTE followed the dependence with a0 = 30.0 ± 4.0; a1 = 203.9 ± 78.5; a2 = 5.19 ± 1.59; and standard deviation σ = 1.85, correlation coefficient squared r 2 = 0.9995 and coefficient of determination (CD) = 0.988 (see Figure 48). Delozier et al. [2003] investigated the influence of the cation exchange capacity (CEC = 0.63 to 1.11 meq/g) on the degree of exfoliation in a PI matrix. To adjust the CEC the Na-MMT was ion-exchanged with 1N LiCl, then centrifuged and placed in an air circulating oven at T = 120 to 170 °C for 24 h – the higher the temperature, the lower the resulting CEC. Next, the treated clay was intercalated with an aromatic primary di-ammonium compound: 1, 5-di(3-amino-phenoxy)-3-oxapentane (BAOD). CPNC with polyamic acid was prepared in NMP (16% solids) at room temperature, under N2 in 24 h. The solution was cast, dried, then cured at T ≤ 300 °C. The first series of specimens was prepared with clays having different CEC, but at a constant
Figure 48 Coefficient of thermal dispersion of PI containing 0-20 wt% of MMT. Data [Huang et al., 2001a,b]. The broken line follows Equation 28.
215
Clay-Containing Polymeric Nanocomposites clay loading of 3 wt%. XRD showed the presence of a peak for all specimen with corresponding values of d001 = 1.20 to 1.31 nm, with the largest value for CEC ≅ 0.7 meq/g. However, TEM showed a random dispersion of individual clay platelets and short stacks. Using these selected clays, the second series of specimens was prepared with clay loading of up to 8 wt%. A small enhancement of the tensile modulus was accompanied by reduction of strength and elongation at break. TGA showed a significant increase of thermal stability. Better dispersion was obtained by dispersing MMT-DDA in a NPM solution of photosensitive polyamic acid (PAA) [Hsu et al., 2003]. The latter was prepared by polymerising pyromellitic dianhydride, oxydiphthalic anhydride, and oxydianiline. The photosensitive formulations contained 2,3,4-tris(1-oxo-2-diazonaphthoquinone5-sulfonyloxy)-benzophenone as the photosensitiser and 3 wt% organoclay. The films were transparent and tough. XRD and TEM indicated exfoliation before and after thermal imidization. The thermal expansion coefficient was 23% lower than that of film that did not contain the organoclay. The tensile modulus increased by 14%, while the strength and elongation at break decreased by 15 and 33%, respectively. An intrinsically photosensitive PI/MMT system was prepared by solution polymerisation at 180 °C of 4,40-diamino-3,30-dimethyldiphenylmethane (MMDA) with benzophenone-3,30,4,40-tetracarboxylic dianhydride (BTDA) and isoquinoline in the presence of MMT-HDA [Liang et al., 2004]. The solution was either cast or spin coated, devolatilised and imidized at T ≤ 280 °C. XRD and TEM indicated exfoliation in the full range of MMT concentrations. Excellent mechanical properties were obtained – in the full range of MMT content (0 to 3 wt%) the modulus, tensile strength and elongation at break increased by 211, 48 and 11%, respectively. At the same time the thermal expansion coefficient was reduced by up to ca. 32%, and Tg increased by 6 °C. While the addition of organoclay did not affect the inherent PI solubility, the presence of MMT significantly reduced the rate of solvent absorption. Finally, the photolithographic properties of PI remained unaffected by the clay presence, but only for concentrations up to 2 wt%. Polybenzoxazines are thermoset phenolic resins developed to overcome the shortcomings of the novolacs and resoles. They show excellent properties, viz. heat resistance, good electrical properties, flame retardance, dimensional stability, toughness, stable dielectric constant, and low moisture absorption - all these at relatively low cost. They can be synthesised from inexpensive raw materials, cured without strong acid or base catalyst, and do not release by-products during polymerisation, thus they are attractive candidates for many applications [Riess et al., 1985]. Agag and Takeichi [2000] prepared CPNC of polybenzoxazine (PBOa) by first intercalating Na-MMT with octyl (OA), dodecyl (DDA) or stearyl (ODA) ammonium chloride, then mechanically mixing with different amounts of a PBOa precursor, bifunctional bis(3-phenyl-3,4-dihydro-2H-1,3-benzoxazinyl) isopropane (B-a). The ring opening polymerisation of pristine B-a started at 223 °C, while in the presence of organoclay at T = 177 to 190 °C. According to XRD, the MMT intercalated with either DDA or ODA became totally exfoliated in the product. Viscoelastic measurements showed that the Tgs of the CPNCs were higher than that of neat PBOa. Increased storage modulus, decomposition temperatures and thermal stability were also noted. However, as the authors
216
Exfoliation of Clays remarked, the method was difficult because of the high viscosity of the monomer and the small difference between its gelation temperature and the melting point. More recently, several types of CPNCs with PBOa as a matrix have been prepared [Takeichi et al., 2002]. The authors described the effects of the preparation method and that of type and content of organoclay on the CPNC properties. Two monomers were used: B-a and monofunctional 3-phenyl-3, 4-dihydro-2H-1,3-benzoxazine (P-a). These produced crosslinked PB-a and linear PP-a matrix, respectively. Na-MMT (CEC = 1.19 meq/g) was intercalated with ammonium salts of such amines as tyramine, 2-phenyl-ethyl amine, aminolauric acid, and DDA. The nanocomposites were prepared either by melt or solution method. The melt method involved incorporating organoclay into B-a or P-a above its melting point (Tm = 100 and 60 °C, respectively). The solution method involved dispersing the organoclay in a solvent NMP or tetrahydrofuran (THF) at 80 °C for 1¯2 h and then blending with either B-a or P-a at 80 °C for 3 h. The blends were cast on glass plate, dried at 60 °C for 16 h, then cured at 100, 150, 200 and 240 °C for 1 h each. Good dispersion was obtained in NMP, but poor in THF. The CPNCs were transparent, red-wine coloured with thickness ranging from 0.2 to 0.4 mm. The intercalation increased the interlayer spacing, from d001 = 1.24 nm of Na-MMT only to ≤ 1.83 nm. Thus, it is remarkable that using either method a high degree of exfoliation in the cured nanocomposites was obtained. Evidently, changing the solvent, concentration and procedure affected the result, viz. when THF was used, a small XRD peak remained indicating intercalated clay, but when NMP was used total exfoliation was achieved. As before, increasing the organoclay concentration reduced the level of platelet dispersion. The effect of clay addition on P-a was much larger than on B-a. The smaller P-a molecules penetrated into the clay galleries more easily and the number in contact with the catalytic clay surface should be larger than with the bigger B-a molecules. The melt method also resulted in exfoliated structures [Agag and Takeichi, 2000]. DSC showed that the incorporation of any type of organoclay lowered the curing temperature. The effect was particularly large for a small amount of organoclay - addition of 5 wt% reduced the onset curing temperature by ca. 50 °C, suggesting a catalytic effect of the organoclay on ring opening polymerisation. A plot of Tg versus clay content showed a maximum at about 4 wt%, with all the CPNC values higher than that of neat resins. As evidenced by TGA, an addition of organoclay improved the thermal stability – the best results were obtained for MMT intercalated with tyramine (p-ethyl-amine phenol). In summary, the work on PBOa-type CPNCs illustrates the importance of miscibility and chemistry on the exfoliation process. Even when the organoclay had small expansion of the interlamellar galleries (Δd001 ≤ 0.59 nm), but was well dispersed in a solvent and/or monomer, good exfoliation was achieved after curing. As in thermoplastic systems, here also the best results were obtained when the matrix polymer was chemically bound to clay by reactive intercalant (tyramine). High energy shearing may help, but it does not replace good thermodynamics and chemistry of the system. The common solvent method has also been used to prepare POSS-type nanocomposites in PI matrix [Tsai and Whang, 2001a]. The polyimide/ polysilsesquioxane-like (PI/PSSQ-like) films had 3D structure with linear PI blocks
217
Clay-Containing Polymeric Nanocomposites and a crosslinked PSSQ-like structure. The system showed higher thermal stability and char yields than pure PI from 4,4´-diamino-diphenyl ether and 3,3´-oxydiphthalic anhydride (ODPA). In a series of X-PIS films (PI modified with p-amino-phenyl tri-methoxysilane (APTS), where X is the molecular weight of each PI block), decreasing the PI block length enhanced the storage modulus, tensile modulus, and Tg but reduced the α-relaxation damping peak intensity, density and elongation. The change in the moduli and Tg may be caused by an increase in the crosslinking density and rigidity of the network. The changes in the peak intensity and density may be caused by increased free volume or the PI interblock separation and decreases in the interblock PI chain interaction. The decrease in elongation is related to increased rigidity. The activation energy of the α-transition depended on the length of the PI block. A maximum value was reached for the chain length of 10 kg/mol because of two opposing factors: crosslinking density and free volume. In a series of X-PIS-y-PTS (X-PIS modified with phenyl trimethoxysilane (PTS), where y is the weight-ratio percentage of PTS to APTS-polyamic acid) films with a constant PI block length, the storage modulus, tensile modulus, Tg, density, and α-relaxation damping peak intensity decrease with the PTS content. This could be because of the increase of the PSSQ-like domain size with the PTS content, which leads to the introduction of more free volume or interblock separation and to a decrease in the interblock chain interaction force. 2.4.2.6 Other Methods – Epoxy-Based Nanocomposites CPNCs with thermoset matrix were already considered in the preceding section where the common solvent methods were discussed. However, owing to the specificity of the epoxy systems it seems desirable to summarise specifically the work carried out on their synthesis. A short review on these systems was recently published [Wang et al., 2000a]. As these authors remarked, the strategy for the preparation of epoxy-based CPNC depends on whether the cured system is glassy (with Epon 826) or elastomeric (with Epon 828 – see Figure 5). In the first case, it was advantageous to disperse organoclay in epoxy overnight at 50 °C before adding the curing agent. In the second case, organoclay should be added directly to a mixture of epoxy with a curing agent. Kornmann, in his doctoral thesis of 2001 offers an overview of nanocomposites technology with a special emphasis on the preparation of CPNC with epoxy resin matrix. Pinnavaia et al., explored the use of epoxy compounds as the second intercalant of layered materials. As early as in 1994 [Wang and Pinnavaia, 1994; 1998a,b], exfoliation of organoclays in a glassy epoxy resin was reported. The process involved heating preintercalated smectite with epoxy resin at T = 200-300 °C. Synthetic Na+-magadiite was intercalated with a series of ammonium ions, viz. CH3(CH2)17NH3-n+(CH3)n, where n = 0-3, by ion-exchange reaction in the presence of neutral amine [Wang et al., 1996]. For primary amine (ODA or C18; n = 0) the interlayer spacing, d001 = 3.82 nm, was found consistent with the structure of onium ions and neutral amine inclined at an angle of about 65°. As the number of methyl groups increased from n = 1 to 2 and 3, d001 decreased to 3.74, 3.20 and 3.41 nm, respectively. Epon 828 was used along with Jeffamine (PPG-bis(2-aminopropyl ether)) as the curing agent. The desired amount of organomagadiite was added to the mixture and stirred for 60 min, degassed and cured at 75 °C for 3 h and at 125 °C for a different length of time. When that time exceeded 5 min, even for systems with 15 wt% organoclay, full exfoliation was recorded. In the cured epoxy matrix good exfoliation was obtained 218
Exfoliation of Clays for magadiite intercalated with either the primary or secondary onium ions (n = 0 or 1). In the case of higher amines (n = 2 or 3) XRD indicated an interlayer spacing of d001 = 7.82 and 4.10 nm, respectively. The observed strong effects of the intercalant structure was interpreted as due to the high acidity of ODA, which is reduced by the presence of methyl groups, especially when n = 2 or 3. The acidity is responsible for the intercalant’s catalytic effects on the curing reaction within the interlamellar galleries, which lead to exfoliation. Good enhancement of the tensile strength was reported. The enhancement was shown to depend on the degree of exfoliation, e.g., at 10 wt% of clay the tensile strength increased by a factor of 4.8, 3.5, and 2.3 for the exfoliated, ordered exfoliated and intercalated systems, respectively (n = 1-3). The transparency and high barrier properties makes these CPNCs attractive as packaging materials and protective films. The early work resulted in several patents [Pinnavaia and Lan, 1998a,b]. The authors observed that the extent of epoxy resin diffusion into the interlamellar galleries depends on the hydrophobicity controlled by the chain length of the alkylammonium cations. Thus, to start with Na-MMT was cation exchanged with alkylammonium cations with different alkyl chain length (CnH2n+1, where n = 4, 8, 10, 12, 16 or 18). Equivalent amounts of the epoxy resin (Epon 828 27.5 wt %) and the polyether-amine (72.5 wt %) were mixed at 75 °C for 30 min, then 10 wt% of the organoclay was added, mixed and cured. The XRD results indicated that the clays with alkyl chain length n ≥ 10 were exfoliated. The mechanical performance was significantly enhanced by exfoliation. In later patents [Pinnavaia and Lan, 1998c; 2000a,b], Na-MMT was treated with HCl to prepare protonated clay, H-MMT (d001 = 1.05 nm). Next, polyetheramine was added to the aqueous suspension and stirred at room temperature for 6 h (d001 = 4.6 nm). After evaporation of the suspending medium the gel of intercalated clay was mixed with Epon 828. The XRD pattern showed absence of the clay diffraction peaks, suggesting exfoliation. The tensile strength and modulus were superior in comparison to CPNC prepared with alkylammonium cation intercalated clay (see Figure 49). However, it is worth stressing that the presence of ≥ 5 wt% MMT reduced the heat of reaction [Butzloff et al., 2001]. Exfoliation was observed only at a MMT concentration below 2.5 wt% – above this limit the system was progressively less exfoliated and more intercalated. Intercalated CPNCs in glassy epoxy resin were obtained by dispersing Nanomer I.28E (MMT-3MODA) into Epon 825 at 65 °C, degassing and curing with a stoichiometric amount of Jeffamine D400 at 65 °C for 15 min [Zerda and Lesser, 2001]. The XRD data indicated that the interlayer spacing of organoclay, d001 = 2.28 nm, increased during pre-swelling in Epon to 3.42 nm, then shrank during curing to 3.21 nm. The tensile modulus increased from 2.8 GPa for neat resin to 3.2 GPa at 12.5 wt% of clay. However, the tensile stress was reduced from 70 to 46 MPa. For the middle range of organoclay concentration (2.5 to 10 wt%) the fracture energy nearly doubled. The results are understandable considering the presence of large scale (10-30 μm) aggregates, evident in SEM micrographs. The measured fracture energy release rate (G1C) was shown to depend on the generated surface area, as measured by AFM. Lü et al. [2001] studied the dispersability of MMT intercalated with either CH3(CH2)17NH3+- (ODA) or CH3(CH2)17N(CH3)3+ (3MODA) in epoxy resin, which was a reaction product of the diglycidyl ether of bisphenol-A (DGEBA; MW = 380 g/mol) with p,p´diaminodiphenylmethane (DDM) or methyltetrahydrophthalic anhydride (MeTHPA). The CPNC was prepared in two steps: 219
Clay-Containing Polymeric Nanocomposites 1. Dispersing organoclay in DGEBA (either directly at 70-80 °C for 20-60 min, or in chloroform at room temperature for 60 min), and 2. Incorporation of the curing agent followed by degassing, casting and curing. The products have been studied using XRD and DSC. Dispersing the organoclays into epoxy increased the interlayer spacing from d001 = 2.4 to 3.7 nm, indicating enhanced intercalation. The two mixing procedures gave equivalent results. Using DDM as a curing agent, full exfoliation was rapidly achieved for MMT intercalated with the primary amine (ODA), but not when MMT was intercalated with the quaternary onium (3MODA) - evidently, the quaternary ammonium ion does not have the catalytic activity observed for ODA. However, anhydride curing with MeTHPA resulted in exfoliation of either organoclay. As the authors remarked, it is important that the exfoliation time is shorter than the gel time – in the studied systems both these processes required nearly the same time, viz. at 80, 100, 120 and 150 °C the exfoliation time (as observed by XRD) was 60, 15, 8 and 5 min, respectively. Furthermore, the exfoliation must take place before the extragallery epoxy matrix reaches its gel point. The factors facilitating the curing inside the interlamellar gallery space enhance the exfoliation. Nanomer 1.30E (MMT-ODA) was dispersed in a matrix obtained by curing Epon 828 with m-phenylene diamine (MPDA) by stirring and sonication for about 30 min and degassing under vacuum [Chin et al., 2001]. To the mixture 5, 14.5 or 25 phr of MPDA was added at 60 °C, the mixture was heated for 2 h at 80 °C and for 2 h at 135 °C. The structural evolution in the CPNC was examined by time-dependent small-angle X-ray scattering (SAXS) using synchrotron radiation. A conventional XRD, AFM and DSC were used to probe the static structure and the reaction kinetics. The interlayer spacing of the Nanomer 1.30E
Figure 49 Tensile strength and tensile modulus of epoxy-based CPNC. The subscript ‘H’ indicates a new method of intercalation that starts with protonated H-MMT; ‘18’ indicates systems where clay was intercalated with C18 H37 NH3+ cation. Data from [Pinnavaia and Lan, 1998c; 2000a,b].
220
Exfoliation of Clays was d001 = 2.3 nm, increasing upon dispersion in Epon to 3.9 nm. Addition of the curing agent and heating for 2 h at 80 °C did not change the latter spacing. Heating the mixture at 135 °C for additional time caused progressive expansion of the interlamellar space toward exfoliation: d001 ≥ 20 nm. The latter spacing was observed for less than an equimolar amount of the curing agent and clay concentration up to 5 wt% (for 20 wt% clay d001 = 8.3 nm). When Epon was cured with more than equimolar MPDA, exfoliation was suppressed – high curing agent concentrations seem to favour extragallery crosslinking. Exfoliation was obtained for the amount of curing agent ranging from zero to less than its stoichiometric amount. The time-dependent SAXS measurements indicated that exfoliation proceeded by progressive delamination of the stacks. While most authors prepare CPNC with epoxy (DGEBA cured with acid anhydride) matrix using either MMT-ODA or MMT-3MODA, recently MMT pre-intercalated with an ‘alkyl ammonium salt with hydroxyethyl groups’ was used [Zhang et al., 2004]. According to TEM, at 3 wt% of organoclay loading only intercalated stacks with d001 ≅ 5.5 nm were obtained. However, in spite of a lack of exfoliation, excellent performance was obtained: impact strength increased by around 85%, tensile strength by 22%, storage tensile modulus (at T < Tg) by 33% and Tg increased by ca. 14 °C. A new approach to clay exfoliation in epoxy matrix involves the use of aromatic amine hardener as the clay intercalant, dispersing such organoclay in epoxy, and then curing the system [Ma et al., 2004]. For example, m-xylylenediamine (m-C6H4(CH2-NH2)2 or MXD for short) was acidified and used to pre-intercalate Na-MMT. The epoxy (DGEBA) was added to the desired quantity of purified organoclay suspension, and after stirring the mixture was dried. Finally, CPNC was obtained by curing the mixture with a stoichiometric amount of 4-aminophenyl sulfone (DDS). Within 2θ = 1 to 9o XRD did not show any peak that may have indicated exfoliation. However, TEM showed the presence of randomly dispersed, ca. 3 nm thick clay platelets, possibly bridged by acidified MXD. The authors speculated that the disorderly exfoliated structure originated from the large crosslinked epoxy molecules connected to the clay platelet via the MXD hardener-intercalant. No information on performance of this CPNC was given. Kornmann et al. [2001] studied the effect of the cation-exchange capacities (CEC) on the formation of CPNC in epoxy matrix. Two MMTs with CEC = 0.94 and 1.4 meq/g (MMT09 and MMT14, respectively) were intercalated with ODA. Epon 828 and a polyoxyalkylene diamine curing agent (Jeffamine D-230) were used. First, the epoxy was mixed with the desired amount of organoclay at 75 °C for several hours, and then a stoichiometric amount of Jeffamine was added. The mixture was outgassed under vacuum, poured into a mould, then cured for 3 h at 75 °C and post-cured for 12 h at 110 °C. The dispersion was studied using XRD, SEM and TEM. The interlamellar spacing of the two clays increased from d001 = 0.97 for MMT09 to 1.72 nm for organophilic MMT09-ODA, and from d001 = 1.21 for MMT14 to 2.14 nm for the MMT14-ODA. Surprisingly, swelling the former organoclay in Epon resulted in exfoliation (d001 > 8.8 nm), whereas the latter slowly intercalated to reach d001 = 3.47 nm after 24 h. The high surface energy of MMT is known to attract polar species such as epoxy molecules hence they diffuse into the interlamellar galleries. If these are able to polymerise, the next molecules are able to diffuse into the interlamellar galleries causing exfoliation, hence self-polymerisation of epoxy in MMT09-ODA may be postulated. Within 221
Clay-Containing Polymeric Nanocomposites MMT14-ODA there is a significantly larger amount of the intercalant present, leaving less space for the Epon molecules to diffuse into, and probably less opportunity to contact the ammonium group. It is expected that curing the already exfoliated mixtures of Epon with MMT09-ODA will result in exfoliated CPNC. However, exfoliation (as defined by the absence of a 001 peak in XRD spectra) was also achieved in systems comprising MMT14-ODA. Although XRD showed exfoliation, TEM indicated that clay platelets formed stacks with interlamellar spacing of 11 and 9 nm for MMT09 and MMT14, respectively. Furthermore, a XRD peak at 2 θ = 20° that corresponds to crystallographic planes 110 and 020 of the clay, was observed and its intensity increased with the clay content. Its presence demonstrates that the XRD analysis is sufficiently sensitive to detect the presence of the clay in the nanocomposite and to quantify it. SEM was also used to investigate dispersion of MMT14 in the nano and micro-composites. The latter system was prepared by dispersing the non-intercalated clay in the matrix. Here XRD showed two peaks: a broad one at d001 = 1.4 nm and a sharp one at 0.97 nm. For both systems clay aggregates were observed, albeit significantly smaller in the (supposedly) exfoliated nano than in the microcomposite (ca. 10 μm particles). Recently, high performance epoxy-nanocomposites were prepared by dispersing synthetic sodium fluorohectorite (FH; Somasif ME-100; CEC = 1.0 meq/g) in a matrix composed of tetraglycidyl 4,4´-diaminodiphenyl methane (TGDDM) cured with 4,4´-diamino diphenyl sulfone (DDS) [Kornmann et al., 2002]. The FH was intercalated using one of the following compounds: ODA, 1-methyl-2-norstearyl3-stearinoacid-amidoethyl-dihydro-imidazolinium metho-sulfate (W75), hydroxyethyldihydro-imidazolinium chloride (HEODI), and ricinyl-dihydroimidazolinium chloride (RDI). ODA is known to produce organoclays that can be exfoliated during curing of epoxy systems, while the dihydro-imidazolines are known to impart good thermal stability. The matrix was N,N,N´,N´-tetraglycidyl 4,4´diamino-diphenyl methane (TGDDM) cured with 4,4'-diamino-diphenyl sulfone (DDS). The influence of the intercalants on curing reactions, the morphology, matrix Tg and the mechanical properties was investigated. TGDDM epoxy resin was heated to 100 °C under vacuum in a high shear mixer then an organoclay was added and the mixture was mixed for 3.5 h. Next, the temperature was increased to 140 °C and a stoichiometric amount of the DDS was added, mixed for 30 min and then poured into a mould. Curing was performed in several steps, e.g., 2 h at 140 °C + 2 h at 177 °C + 7 h at 200 °C. DSC studies of the curing reaction indicate that FH has no influence on TGDDM polymerisation – the reaction with or without the clay had the exothermic peak at Tpeak = 316 °C. However, 10 wt% of FH-W75, FH-HEODI or FH-RDI reduced the exothermal peak by 6 ºC and with 10 wt% of FH-ODA, by 11 °C. Thus, the intercalants do catalyse curing, but the dihydro imidazolines are not as effective as the ODA ions. This suggests that FH-ODA would produce the best exfoliation, if not for the thermal decomposition of ammonium ions which starts near 200 °C. The interlayer spacing of intercalated FH and that of cured epoxy nanocomposites is presented in Table 34. Intercalation expands the interlamellar gallery by Δd001 = 0.96 to 2.26 nm. Once the organoclays are dispersed in the epoxy matrix, the gallery increases for FH-ODA, FH-HEODI and FH-RDI whereas it decreases for FH-W75. In the latter case the TGDDM and the DDS were not able to diffuse between the layers, the polymerisation took place outside 222
Exfoliation of Clays
Table 34 Organic content and interlamellar spacing of cured epoxy nanocomposites. Data [Kornmann et al., 2002] Materials
Intercalant as % of clay CEC
Interlamellar spacing d001 (nm) In organoclay
In epoxy
-
0.94
-
FH-ODA
99
1.9
4.0
FH-W75
67
3.2
2.5
FH-HEODI
90
2.2
4.1
b
FH-RDI
110
2.5
7.5
c
FH
a
Notes: At 5 wt% clay, cured as indicated in the text b In ordered places c Irregular distribution a
the interlamellar galleries, compressing them by Δd001 = 0.7 nm. Apparently, FHW75 is immiscible with some components of the epoxy matrix. At least part of the reason for immiscibility is that only partial ion exchange took place during intercalation. None of the nanocomposites produced contained fully exfoliated clay. TEM confirmed the interlayer spacing deduced from the XRD data – indeed, the largest average spacing was observed for FH-RDI, but some nonintercalated platelets were also visible. The SEM micrographs also indicated the presence of clay aggregates, particularly bad (some as large as 20 to 30 μm) for FH-W75 and FH-ODA. This is an important observation! Without the evidence of SEM one would concentrate the interpretation of data on the nanometre-scale morphology, whereas for a number of performance characteristics (for example the impact properties) the micron-scale aggregates may be more important. The tensile properties of the nanocomposites were examined. The clay concentration effect on the tensile modulus is shown in Figure 50. The results for FH-W75 are clearly the worst, falling even below the microcomposites with nonintercalated FH. Of the remaining three systems, the best is the one with FHRDI and the worst with FH-ODA – here at last the expected correlation with the degree of platelet dispersion is observed. The fracture energy for all the intercalated systems was found to be higher than that for the microcomposite with FH, but again the data for FH-W75 were the lowest and these of FH-RDI the highest. At 5 vol% clay loading the fracture energy was nearly twice as large as that for the neat matrix resin. Epoxy resins are inherently brittle and several methods have been used to improve toughness without loss of stiffness. Since CPNC with an epoxy matrix frequently show improved modulus and reduced ductility, a growing number of publications have appeared, trying to solve this problem. Thus, for example, Ratna et al., 2003 used a combination of linear (DGEBA) and 11-arm star branched epoxy to prepare CPNC with 5 wt% of MMT-ODA. A high degree of intercalation was obtained. In spite of the phase separation between the two epoxies, good performance was reported, viz. enhancement (in comparison to 223
Clay-Containing Polymeric Nanocomposites
Figure 50 Tensile modulus versus fluorohectorite (ME-100) content in high temperature epoxy systems. Data [Kornmann et al., 2002]. For explanation of the Figure legend see text.
cured DGEBA) of flexural strength and modulus by 21 and 26%, respectively, and impact strength by 114%. Fröhlich et al. [2003] toughened their epoxy (hexahydrophthalic acid anhydride-cured DGEBA) with PPG-PEG liquid rubber treated with methyl stearate. Up to 15 wt% organoclay (synthetic fluorohectorite pre-intercalated with MDD2EtOH) was dispersed as short stacks with d001 = 2.94 nm. The thermal, and mechanical properties were determined. Addition of the additives reduced the modulus and improved the toughness. In parallel, the yield strength, and Tg also decreased. At a later date the same epoxy system was reinforced by a diversity of layered silicates intercalated with phenolic alkyl imidazoline amide cations [Fröhlich et al., 2004]. Natural and synthetic clays (bentonite and fluorohectorite, respectively) were pre-intercalated and used at loadings of up to 10 wt%. For comparison, the same epoxy was prepared with talc as well. Intercalation expanded the interlayer spacing to about d001 = 3.3 to 3.7 nm, but dispersing it in epoxy slightly reduced the gallery height (by ca. 0.3 nm). The mechanical performance of the samples with organoclay was quite similar to that of the talcfilled composite. The poor performance of organoclay-filled CPNC seems to have been caused by immiscibility between the organoclay tails and the matrix. 2.4.2.7 Other Methods – PU-Based Nanocomposites Polyurethanes (PU) have been mixed with nanosize particles for a number of decades. Carbon black, metal and silica particles have been used. Three examples selected from between many will be mentioned. 2.4.2.7.1 Metal Particles PU was dissolved in dimethyl acetamide and a solution of metal ions (iron, cobalt, nickel and copper) was added [Chen et al., 1996]. The metal ions were reduced 224
Exfoliation of Clays by sodium borohydride under mild conditions into amorphous fine powders. Polar polymer and low metal concentration favour smaller particles. The polymer chains prohibited excessive aggregation of the metal atoms and had a protective effect on the fine metal powders. An energy disperse X-ray spectrometer connected with TEM proved that the dispersed particles were metal clusters with the sizes ranging from 10 to 150 nm. 2.4.2.7.2 Silica A series of PU nanocomposites was prepared with 0-50 wt% of silica, having a particle size of about 12 nm [Havni and Petrovic, 1998]. SEM showed regular distribution and small spacing between neighbouring particles at all concentrations. For all samples XRD showed a single broad maximum at about 6° and a shoulder at 20°. The Tg of PU increased with the silica concentration. A parallel series of PU filled with micron size silica had higher density, hardness and modulus than nanosilica filled systems, but the tensile strength and elongation at break were dramatically better for the latter systems. Composites with nanosilica were clear and transparent while those with micron size silica were opaque. Petroviç et al. [2000] prepared PU nanocomposites by dispersing amorphous SiO2 particles (average diameter 12 nm) in MEK. Next, PPG was added, MEK was removed, MDI added and the system was cured at 100 °C for 16 h. The final concentration of SiO2 was up to 50 wt%. Properties of these nanocomposites were compared to those of microcomposites containing crystalline SiO2 with an average particle diameter of 1.4 μm. The nanocomposites showed 3-fold higher tensile strength and elongation at break, but lower modulus and hardness. The sol-gel process has also been used to generate nanosized silica particles within a PU matrix [Goda and Frank, 1997]. At lower concentration the silica penetrated the hard segment domains of PU, at higher it disrupted its ordered structure. 2.4.2.7.3 Cadmium Sulfide Particles (CdS) Nano-CdS particles were prepared by reverse micellisation [Hirai et al., 1999]. They were dispersed into PU, by surface-modification with 4-hydroxythiophenol or 2-mercaptoethanol, followed by polyaddition of ethylene glycol with toluene2,4-diisocyanate. The resulting CdS-PU powder could be dissolved in organic solvents such as DMF to make nanoCdS/PU transparent films that showed quantum-size effects. These were used for photocatalytic generation of hydrogen. 2.4.2.7.4 Organoclays Several already cited broad patents on CPNCs have stated that the invented technology is also applicable to PU [Usuki et al., 1989; Nichols and Chou, 1999, Beall et al., 1996a; 1999b]. One of these general patents was granted to Pinnavaia and Lan [2000a,b]. However, the cited examples are only for epoxy-based CPNC. The patent information will be discussed later on in this book. PUs are prepared by the reaction of isocyanate and polyols: HO-R-OH + OCN-R´-NCO → -(-CO-NH-R´-NH-CO-O-R-O-)nThe polyol component -R- usually contains a base, such as a secondary or tertiary amine. Therefore, these may react with the protons within the clay interlamellar gallery enhancing the intercalation. 225
Clay-Containing Polymeric Nanocomposites Wang and Pinnavaia [1998b] prepared thermoset PU-based CPNC by curing a PU network in the presence of MMT intercalated with C12H25NH3+ or C18H37NH3+ (Nanocor C12A and C18A, respectively). The organoclays were solvated in one of several polyols commonly used in PU synthesis, viz. ethylene glycol (EG), polyethylene glycol (PEG), polypropylene glycol (PPG), and glycerol propoxylates (Voranols). The solvation could be carried out at room temperature, but it was more rapid at higher temperatures, say T = 50 °C for 12 h. The d001 -spacing depended mainly on the chain length of the onium ion. As shown in Table 35, the observed values of interlayer spacing agree with calculations (dcalc). Initially, the onium ion paraffin tails in the non-solvated clay are oriented parallel to the clay platelets, as judged by the values d001 = 2.22 and 2.30 nm, then they reorient from this position to optimise solvation by polyol. PU-based PNCs were formed by adding a methylene diphenyl diisocyanate prepolymer (MDDP; MW = 1050; functionality = 2.0) to a mixture of organoclay and Voranol V230-238. Before use, the polyol was degassed under vacuum at 100 °C for 6 h. The diisocyanate was added and the mixture was stirred at 70 °C for 15 min before degassing at 95 °C under vacuum. The mixture that contained ≤ 10 wt% of organoclay was pourable. The bubble-free mixture was poured into a mould for curing at 95 °C for 10 h under N2. The alkylammonium-exchanged ions of the intercalated clay were considered to react with isocyanate and were counted as contributors to the polymerisation stoichiometry; hence the amount of V230-238 was reduced accordingly. No catalyst was used. During curing at 95 °C the interlayer spacing increased with time from d001 = 3.74 to 5.08 nm. The presence of the Braggs reflections indicated that the clay layers formed large tactoids. However, the distance between the nanolayers was in the range where matrix reinforcement could be expected. Inorganic fillers are commonly used in PU to reduce formulation cost and to increase stiffness, but the improvements in modulus are compromised by a loss
Table 35 Basal spacing of MMT intercalated with alkylammonium salt then solvated with polyols. Data [Wang and Pinnavaia, 1998] Organoclay Cloisite® type
Intercalated/solvated Air- Glycol PEG300 PPG2000 V230dried (62)b (300)b (2000)b 238a (700)b
Calculated V230- V230112a 056a b (1500) (3000)b
dcalc
d001 (nm) C12A
2.22
3.39
3.20
3.31
3.23
3.29
3.29
3.10
C18A
2.30
3.67
3.66
3.78
3.74
3.71
3.80
3.87
Notes: a Glycerol propoxylate (Voranol from Dow); b Molecular weight of polyols; c Calculated basal spacing for a vertical orientation of the onium paraffin tails
226
Exfoliation of Clays of elastic properties. By contrast, PU-based CPNC exhibits improvement of toughness, as well as of modulus. An illustration of the mechanical behaviour is shown in Figure 51. The tensile strength, tensile modulus, and strain-at-break versus organoclay concentration are plotted for PU prepared from V230-238, MDDP and MMT intercalated with ODA (C18A). Evidently, incorporation of nanoclay both strengthened and toughened the elastomeric matrix in comparison with the neat polymer. Even when the clay platelets are in the form of intercalated tactoids, they strengthen, stiffen and toughen the matrix. At an organoclay loading of 10 wt% the values of all three functions at least doubled. It is noteworthy that the PU-based nanocomposites retained high optical transparency. Chen et al. [1999b] prepared CPNC with PU as the matrix going through a stage of polycaprolactone-based nanocomposite as an intermediary. Thus, first Na-MMT in aqueous solution was intercalated with 12-aminolauric acid. The adduct was collected, washed, dried, ground and screened through a 325-mesh sieve. Next, the organoclay was dispersed in ε-caprolactone which subsequently was polymerised for 3 h at 170 °C while stirring. PU was prepared in two steps: 1. Mixing 4,4´-diphenyl-methane diisocyanate (MDI), polycaprolactone diol and DMF for 2 h at 70 °C, then reacting the mixture with 1,4-butanediol for 30 min. 2. The PCL-based PNC was added and the combined mixture was allowed to react for 30 min. The prepolymer solution was spread, dried under vacuum and cured at 80 °C for 10 h. XRD showed that the clay was exfoliated and that the reaction product comprised crystalline PCL. Thus, the prepared CPNC was a complex system with segmented PU forming the matrix chemically bonded to the randomly dispersed PCL-based exfoliated nanocomposites. Size exclusion chromatography (SEC) or gel permeation chromatography (GPC) indicated that the Mn of PCL was only 3.62 kg/mol, and
Figure 51 Mechanical behaviour versus organoclay content of PU-based CPNC. Data [Wang and Pinnavaia, 1998b].
227
Clay-Containing Polymeric Nanocomposites that in PU, Mn decreased with the content of PCL-based nanocomposite, from Mn = 52 to 13 kg/mol. The miscibility of the PU/PCL-clay blend was not studied. The mechanical properties versus clay content plot showed a surprising behaviour (see Figure 52). For example, in tensile tests the tensile strength increased linearly with clay content, but the elongation at break went through a sharp maximum at 0.74 wt% clay. In the lap-joint shear strength tests (ASTM D1002) the modulus increased linearly with clay content while the strength went through a local maximum at a clay content of 1.3 wt%. CPNC were prepared by dispersing synthetic fluoromica (FM; Somasif ME100; aspect ratio p ≤ 5,000) in a PU matrix [Zilg et al., 1999b]. The ME-100 behaves as a weak silicilic acid, with Na+ as counterions in the interlamellar gallery. FM was intercalated in aqueous solution with methyl-dodecyl bis(hydroxyethyl)ammonium chloride (MDD2EtOH; Ethoquad C12). After drying it was dispersed in trihydroxy-terminated oligo(propylene glycol) having Mn = 3.8 kg/mol, then reacted with diisocyanato-phenyl methane catalysed with N,N-dimethylbenzylamine (NNDB) at 80 °C. The reaction rate was slightly slowed by the presence of organoclay. At 10 wt% clay loading the CPNC was not exfoliated, but d001 increased from 0.9 (dry ME-100) to 8.8 nm. Evidently, the intercalant and polyol diffused into ME-100 interlamellar galleries. TEM showed that the intercalated clay formed stacks, 500 nm in length and 20 to 50 nm in thickness. The stacks were well dispersed within the PU matrix. As shown in Figure 53, the tensile strength and elongation at break increased with the organoclay content. However, a less satisfactory correlation was reported for the tensile modulus and the Shore-A hardness – the data were scattered. PU-based CPNC were prepared starting with Na-MMT (CEC = 0.76 meq/g) which was intercalated (3 h stirring in aqueous suspension at 60 °C) with either 12-aminolauric acid (ADA) or benzidine (BZD), forming the complexes: MMT-
Figure 52 Tensile (ASTM D412) and lap-joint strength (ASTM D1002) test data for PU containing dispersed PCL-based CPNC. Data [Chen et al., 1999b].
228
Exfoliation of Clays
Figure 53 PU-based CPNC – tensile strength and elongation at break versus fluoromica clay content. Data [Zilg et al., 1999b].
ADA and MMT-BZD, respectively [Chen et al., 2000]. The intercalated clay was washed, dried under vacuum at 80 °C for 12 h, then ground and screened through a 325-mesh sieve. In a parallel operation, MDI and polytetramethylene glycol (PTMG) at a molar ratio of 2:1 were dissolved in DMF and heated to 90 °C for 2.5 h to form a prepolymer. To the prepolymer 1,4-butanediol was added with rapid mixing at 90 °C for 10 min. Next, an appropriate amount of organoclay (to give 1, 3 or 5 wt% of organoclay in CPNC) was dispersed in 10 ml of DMF, added to the prepolymer mixture and the reaction was allowed to proceed to completion in 3 h at room temperature under mixing. The final concentration of PU in DMF was 30 wt%. After removing DMF at 80 °C and degassing the mixture an elastic film was obtained. XRD of MMT-ADA and MMT-BZD gave d001 = 1.7 and 1.54 nm, respectively. The difference was attributed to the size of the intercalant molecule. The PU-based PNCs prepared with 1, 3 and 5 wt% MMT-ADA, as well as with 1 and 3 wt% MMT-BZD were all exfoliated. This conclusion was supported by TEM micrographs – randomly placed, about 1 nm thick layers of organoclay were observed. However, the PNC with 5 wt% MMT-BZD had a broad peak of d001 = 2.47 nm indicating intercalation. The molecular weight of PU in neat resin and in CPNC prepared in the presence of MMT-ADA or MMT-BZD was Mn ≅ 11 kg/mol. Similarly, for all seven samples the glass transition temperature was -58 ≤ Tg ≤ -59 °C. Also, as expected, the water absorption was found to vary from 1.3 to 1.7 wt%. Surprisingly, TGA analysis of PU-based CPNC containing MMT-ADA showed that it started to degrade faster than neat PU. This may be due to the presence of extra ADA unattached either to clay or to PU. However, at higher temperatures (T > 350 °C) the PU/clay nanocomposite displayed higher thermal resistance than that of PU. The stability of the MMT-BZD/PU system was better than that with MMT-ADA. 229
Clay-Containing Polymeric Nanocomposites The mechanical properties of the PU-based PNC are shown in Figure 54. Addition of MMT-ADA only slightly improved the tensile strength and elongation at break over that of neat PU. However, the effect of MMT-BZD was quite significant – improvement of the tensile strength by a factor of 2.0 and the elongation at break by a factor of 2.8. The large contrast between the effects of MMT-ADA and MMTBZD was explained by the interactions between the intercalant, clay and PU matrix. ADA has one terminal -NH2 group that might form a complex with Na-MMT and still be able to react with a -NCO to form urea. BZD had two terminal -NH2 groups that can participate in these interactions/reactions. Therefore, ADA can form linear chains while BZD a crosslinked one. Schematics of these interactions/reactions are presented in Figure 55, which illustrates the molecular structure in PU-based PNC formed in the presence of (a) MMT-ADA and (b) MMT-BZD. Ma et al. [2001] prepared elastomeric PU based nanocomposite starting with MMT (CEC = 0.9-1.0 meq/g, with particle size of about 40-70 μm) intercalated with trimethyl hexadecyl-ammonium chloride (3MHDA). Three types of polyol of different MW and functionalities were used: PPG1, and PPG2 having Mn =1 and 2 kg/mol, respectively, and glycerol propoxylate (GPO3, with Mn = 3 kg/ mol). Dry organoclay was dispersed in GPO3 for 2 h. The degassed dispersion was mixed with the PU prepolymer, prepared by reacting PPG1 and PPG2 with toluene diisocyanate (TDI) at 80 °C. The mixture was cured at 80-90 °C for 2 h. CPNCs containing either PPG1 or PPG2 were also prepared. The XRD data indicated that MMT was intercalated with the interlayer spacing d001 ≤ 4.6 nm. The tensile strength and elongation at break reached maximum values at ca. 8 wt% clay. The maximum improvement of the tensile strength and elongation at break was by a factor of 2 and 5, respectively (based on neat PU). TGA showed that the temperature for the maximum thermal decomposition increased from Td = 377 (PU) to 385 °C. The best properties were observed for CPNC comprising GPO3. Nanocomposites prepared by first dispersing the organoclay in low MW polyol had better mechanical properties than those prepared by dispersing organoclay in high MW polyol. In 2001 Hu et al., reported a similar degree of dispersion. The authors first expanded the interlayer spacing of MMT-3MHDA in PEG and then added TDI and caused polymerisation. The reported interlayer spacing was: for MMT d001 = 1.26; for organoclay 1.96; and for CPNC with (10 wt% organoclay) d001 ≤ 4.85 nm. In accord with these observations, only intercalation could be expected in systems containing clay intercalated with quaternary ammonium ions. An interesting strategy has been used by Yao et al. [2002]. Different amounts of Na-MMT (from Nanocor) were mixed with modified polyether polyol (MPP) at 50 °C for 72 h. The mixture was then blended with a known amount of modified 4,4´-diphenylmethylate diisocyanate (M-MDI) and moulded, curing it at 78 °C for 168 h. Thus, MPP was used as an intercalating agent that expanded the interlamellar galleries from Δd001 = 0.14 to 0.64 nm – evidently, only a moderate level of intercalation was achieved. It was observed that with an increasing amount of clay the heat of reaction was reduced. The reduction was ascribed to immobilisation of MPP that intercalated the clay, being strongly bonded to the solid surface. In spite of the low degree of dispersion, the CPNC showed significant improvement of the tensile strength and strain at break, increasing with the amount of clay. At a clay loading of 21.5 wt%, tensile strength increased by ca. 44% and the strain at break by 20%. The storage modulus at T < Tg of the soft segments increased by 350%. 230
Exfoliation of Clays
Figure 54 Mechanical behaviour of PU-based CPNC containing MMT intercalated either with 12-aminolauric acid (ADA) or with benzidine (BZD) [Chen et al., 2000].
Figure 55 Schematic representation of the molecular structure in PU-based CPNC formed in the presence of (a) ADA-MMT and (b) BZD-MMT. Reproduced from Chen et al., 2000, with permission from Elsevier.
231
Clay-Containing Polymeric Nanocomposites The thermal conductivity initially decreased with clay loading (by ca. 10% for 17% clay) then returned to the original level. MMT is known to be hygroscopic and, as TGA experiments indicate, it may contain a significant amount of adsorbed and chemically bound water [Ogawa et al., 1992; Menéndez et al., 1993]. The -NCO groups readily react with water, for example, and 1.0 g of water will consume 18.0 g of toluene-diisocyanate (TDI). Cloisite® 10A (MMT-2MBHT) was de-watered by dispersing it in toluene followed by azeotropic distillation for 6 h [Zhang et al., 2003]. Next, the dehydrated organoclay was dispersed in PEG, reacted with TDI at 80 °C for 4 h, and then cured at 90 °C for 15 h under vacuum – the clay concentration ranged from 0 to 7 wt%. The authors reported that only about 4.1% of the organoclay -OH groups were able to react with -NCO groups of TDI, hence grafting the PU matrix directly to the clay surface. The reaction was detected by FTIR. XRD of Cloisite® 10A and the CPNC gave d001 values of about 2.3 and 3.5 nm, respectively. In spite of the lack of exfoliation the mechanical properties were found to be greatly improved, viz. the dynamic tensile modulus, E, increased by a factor of 10, the tensile strength and the elongation at break by about 160%. These improvements in the intercalated systems are clear evidence of the benefits of direct matrix-filler bonding.
2.4.3 Melt Exfoliation Melt blending is the preferred method for preparing CPNC with a thermoplastic polymer matrix. Typically, the polymer is melted and combined with the desired amount of the intercalated clay in an extruder, internal, kinetic energy or a continuous mixer. Melt blending is carried out in the presence of an inert gas, such as argon, neon or nitrogen. Alternatively, the polymer may be dry mixed with the intercalant, then heated in a mixer and subjected to a shear sufficient to form the desired degree of dispersion. While total exfoliation can be achieved for some diluted systems, the industrial standard requires that ≥ 80 wt% of the clay platelets (aspect ratio p = 10 to 1500) are individually and uniformly dispersed in the polymeric matrix, with the rest forming short stacks. There are several advantages of the melt versus the reactive exfoliation method. The process is more economic, it is closer to the ultimate product manufacturer, hence it is better suited for rapid changes of formulation, it is more versatile offering ready adaptability, it involves more participants with the technical knowhow (hence it offers more rapid penetration of the industry), and finally it does not require that a big polymer production line is dedicated to developing a material with uncertain application and hesitant market. As discussed in Section 2.3.8, the beginning of melt exfoliation started with the quiescent, diffusion-controlled melt intercalation process. For example, Vaia et al. [Vaia et al., 1993; Vaia and Giannelis, 1997b] prepared PS-based CPNC by such a static melt processing method. Melt exfoliation of clay in PDMS matrix was pioneered by Burnside and Giannelis [1995; 2000]. Na-MMT was first intercalated with dimethyl ditallow ammonium bromide (2M2TA), and then dispersed in the matrix by sonication. Similarly, silicone rubber-type CPNCs were prepared by melt intercalation [Wang et al., 1998]. In this work Na-MMT was intercalated with trimethyl hexadecylammonium bromide (3MHDA). The CPNCs with ≤ 20 phr of organoclay (≤ 9 wt% of clay) were prepared by mechanical mixing. The resulting nanocomposites,
232
Exfoliation of Clays characterised by XRD, TEM and TGA, were reported intercalated with d001 = 3.71 nm. Thus, the ordered organosilicate structure of clay was still present in the CPNC, forming stacks that were ca. 50 nm thick and uniformly dispersed in the silicone rubber matrix. The mechanical properties and thermal stability of the hybrids were very close to those of aerosilica-filled silicone rubber. Melt intercalation/exfoliation of hydrophilic polymers, such as P4VP, PEG or PVAl is relatively easy. The strategies used have been described in Section 2.3.5.2. More recently, Zanetti et al. [2000a,b] described the melt exfoliation of fluorohectorite (fluoromica, FM) in a PVAc matrix. The melt-produced CPNCs show a similar set of properties as those obtained by the reactive route, viz. superior heat, chemical or ignition resistance, barrier to diffusion of gases and polar liquids such as water, methanol, ethanol etc., yield strength, stiffness and dimensional stability. These CPNCs may be useful in diverse applications viz. in business equipment, computer housings, transport (automotive and aircraft, viz. in under the hood and exterior body trim applications), electronic, packaging, building and construction industry, etc. However, this method is at an early stage of development with numerous unanswered questions still remaining. A consensus emerges that the clay platelets, e.g., individual lamellae of MMT, should be treated as any other structural plates, with their own value of mechanical performance characteristics, e.g., modulus, maximum strain at break, etc. There is TEM evidence that in shear the clay platelets tend to bend. The work on FM showed extensive reduction of the aspect ratio, but it is not known whether the phenomenon is general and which mechanism controls the attrition. 2.4.3.1 PA-Based CPNCs Maxfield et al., from AlliedSignal (now Honeywell) [1995; 1996; Christiani and Maxfield, 1998] were the first to propose melt exfoliation in a PA-6 matrix. The aim was to produce PA-based CPNC with advantageous performance, without infringing on the Toyota reactive exfoliation method. The method comprises three steps: 1. Treating an aqueous suspension of clay (the preferred platelet diameter is 15 ≤ D ≤ 300 nm) with peptising Na6P6O18 at 50-90 °C, and then with organosilanes (e.g., aminoethyl aminopropyl trimethoxysilane), organotitanates or organozirconates. The process increases the interlayer spacing to d001 ≥ 5 nm. The authors suggested that to facilitate intercalation the size of clay particles may be reduced, or the mixture may be exposed to heat, ultrasonic cavitation, microwaves, etc. The complexes of organosilanes, organotitanates and/or organozirconates (with or without onium salt) improved the thermal stability at T > 300 °C. 2. Saturating the dried clay (interlayer spacing d001 ≥ 1.5 nm) with a monomer, e.g., ε-caprolactam, (the patent claims are broad, stretching to ‘one or more thermosetting and/or thermoplastic polymers or rubbers’) then polymerising it, which increases the interlayer spacing to d001 ≥ 5 nm. 3. Compounding the modified clay at T ≥ Tm with matrix polymer until the desired level of exfoliation is reached. The authors emphasised the formation of γ-phase crystalline PA-6 by compounding with clay.
233
Clay-Containing Polymeric Nanocomposites Melt compounding was also used by Liu et al. [1999] to prepare PA-6/organoclay PNC. First, Na-MMT was intercalated with octadecyl ammonium chloride (ODA), which increased d001 from 0.98 to 1.55 nm. Next, PA-6 was dry blended with ≤ 17 wt% of the organoclay, then melt-compounded in a TSE and injection moulded. Melt blending increased the interlayer spacing to d001 = 3.68 nm and reduced XRD peak intensity, indicating a partial exfoliation. Similar results were claimed by the RTP Co. for melt compounded PA-6 [Dahman, 2000]. In the latter study various organoclays were used at different concentrations. Melt exfoliation was carried out in a TSE. As described in Section 2.3.8.2.4, Dennis et al. [2000; 2001] reported that melt blending of PA-6 with Cloisite® 30B (2M2EtOH) resulted in exfoliation, thus the fundamental studies on the intercalation processes were conducted with PA-6/Cloisite® 15A (2M2HTA) as a model system. The work has shown that at least for this organoclay/polymer system, intercalation depends solely on the residence time and the residence time distribution inside the processing equipment. The observed difference in the ability of Cloisite® 30B and Cloisite® 15A organoclays from SCP to exfoliate demonstrated the importance of the chemical treatment of the clay, which controls the clay/polymer interactions. As shown in Table 16 the exfoliating Cloisite®-30B (methyl-tallow-bis-2-hydroxyethyl) differs from the intercalating Cloisite® 15A (bis-methyl-bis-hydrogenated tallow) on three accounts: 1. It contains one (non-hydrogenated) tallow group versus two hydrogenated in C15A, 2. It contains two hydroxyethyl groups versus two methyl groups of C15A and 3. It has lower substitution of organic ions (0.9 for C30B versus 1.25 meq/g for C15A). In short, C30B has a lower density of the organic intercalant within the interlamellar galleries than C15A and at the same time it has higher polarity, which originates from two -OH groups and unsaturation of the tallow group. Thus, Cloisite® 30B has stronger attraction for PA molecules causing them to diffuse into galleries, and more space for them to do so. In another publication from the same laboratory an experimental organoclay SCPX-2004 was used [Cho et al., 2000; Cho and Paul, 2001]. This new organoclay is a close cousin of Cloisite® 30B – in SCPX-2004 the tallow is replaced with rapeseed, which contains ca. 95% of C22 alkyl chain with single unsaturation at C13 (erucyl). In both cases the same Na-MMT was used with CEC = 0.91 meq/g. SCPX-2004 has been successfully used for exfoliation via direct melt compounding of PA-6 (Mn = 29.3 kg/mol) with 0-20 wt% of the organoclay, using either a TSE or a SSE at 240 °C. The extruded pellets were injection moulded into standard test mould. For the SCPX organoclay, XRD gave the interlayer spacing d001 = 1.8 nm. CPNC (with 5 wt% clay) was prepared either in a TSE or a SSE; the former process engendered full exfoliation, while the latter only a partial exfoliation, but large domains of unaltered organoclay remained. TEM confirmed these observations. The individual layers were found aligned along the flow axis. The average clay platelet thickness and length were determined as approximately 3 and 120 nm, respectively. At a clay content of 3.16 wt% it produced a respectable modulus (3.7 GPa versus 2.7 GPa for neat PA-6 and 4.3 GPa for PA-6/C30B) and excellent elongation at break (38% versus 40% for neat PA-6 and 10% for PA-6/C30B). 234
Exfoliation of Clays TGA showed that due to the degradation of the quaternary alkylammonium, the nanocomposites had lower stability than neat PA-6. A series of CPNC with 5 wt% clay were prepared varying temperature (T = 230-280 °C) and screw speed (N = 80-280 rpm) in the TSE. Their mechanical properties were also given. To detect tendencies, the published data were fitted to a linear relation: Y = a 0 + a1 T + a 2 N
(29)
where Y is any of the four variables listed in Table 36, N is screw speed, and a1 and a2 are equation parameters. Evidently, there is no correlation between the two compounding variables and impact strength as well as elongation at break. These properties may be more related to the crystallinity of the specimens than the process variables. However, there is strong correlation for the modulus (r2 = 0.9999) and the yield stress (r2 = 0.9998). Increasing the compounding temperature and screw speed results in higher modulus. One may speculate that the degree of platelet dispersion also increases. For yield strength the coefficient a1 < 0, hence an increase of T reduces the yield strength. Most likely, thermal decomposition of the SCPX quaternary alkylammonium weakens the bonding between dispersed clay platelets and the matrix (see also Figure 32). Cho and Paul also plotted the relative strength, modulus and elongation at break versus the mineral clay content using data from four sources: their own laboratory and from the literature [Christiani and Maxfield, 1998; Liu et al., 1999; Okada et al., 1988; Okada and Usuki, 1995]. Three of the CPNCs were prepared by melt compounding, whereas the fourth one was prepared by the in situ reactive exfoliation method. Cloisite® 30B gave slightly higher modulus and strength than SCPX, but the latter was far superior as far as the elongation at break is concerned. The plots indicated a good linear increase of the relative
Table 36 Statistical analysis of CPNC performance versus compounding conditions. Data [Cho and Paul, 2001] Variables
Izod impact strength (J/m)
Tensile modulus, E (GPa)
Yield strength, σY (MPa)
Elongation at break, εb (%)
Standard deviation
4.2889
0.0656
2.0258
5.4628
R-squared, r2
0.9961
0.9999
0.9998
0.9901
Coefficient of determination
0.3287
0.6832
0.6792
0.0556
Range of values
38 - 47
3.66 - 3.85
81.1 - 87.6
29 - 38
Parameter value a0 =
37.2 ± 27.7
3.39 ± 0.42
91.0 ± 13.1
40.6 ± 32.2
Parameter value a1 =
0.003 ± 0.110
0.00059 ± 0.0016
–0.054 ± 0.052
–0.03 ± 0.14
Parameter value a2 =
0.030 ± 0.030
0.00095 ± 0.00046
0.026 ± 0.014
0.010 ± 0.039
235
Clay-Containing Polymeric Nanocomposites modulus with clay content, a similar increase of the tensile strength and poorly correlating decrease of the elongation at break. The authors also reported a synergistic effect on tensile strength and modulus when the exfoliated nanocomposite was used as the matrix for a glass fibre reinforced composite. Work from the same laboratory by Fornes et al. [2001] indicated that during melt intercalation two processes simultaneously take place: thermodynamicallydriven diffusion of macromolecules into the interlamellar galleries, and mechanical peeling off of platelets or short stacks. If during the former process the diffusioncontrolled intercalation sufficiently reduced the solid-solid interactions between the adjacent platelets, the latter process may lead to exfoliation. It is noteworthy that the minimum interlayer thickness of the organoclay needed for the PA-6 macromolecules to diffuse into is relatively small, d001 = 1.8 nm. However, the chemical character of the intercalated onium cation, hence miscibility with PA-6, is important as the diffusion of PA-6 molecules expands the interlamellar gallery height at least by a factor of two. The stress field present during the compounding also seems essential for efficient exfoliation. The next contribution from this group was analysis of the behaviour of a series of CPNCs with PA-6 as the matrix and containing up to 4.5 wt% of organoclay [Fornes et al., 2004]. Three types of organoclays with quaternary ammonium ions were tested. These were MMT-type, intercalated with: tetramethyl (4MA), trimethyl hydrogenated-tallow (3MHTA), or dimethyl di(hydrogenatedtallow) (2M2HTA). The nanocomposites were prepared by melt compounding in a TSE. XRD and TEM demonstrated that the degree of dispersion increased from 4MA (the worst) to 3MHTA. The modulus and yield strength increased in the same order, while the elongation at break showed the opposite trend. These results indicate that for melt intercalation by PA-6, a minimum interlayer spacing is required – the minimum being located between d001 = 1.36 and 1.80 nm, measured for MMT-4MA and MMT-3MHTA, respectively. The second important information concerns the interaction between the PA-6 macromolecules and the organoclay – in spite of larger initial interlayer spacing (d001 = 2.42 nm) MMT2M2HT did not disperse as well as MMT-3MHTA. Evidently, a more thorough covering of clay platelet in MMT-2M2HTA by immiscible (with PA-6) long alkyl tails than that produced by MMT-3MHTA is to blame. Partial access of the clay surface to PA-6 chain-ends provides compatibilisation, similar to that provided by PO-MAH in CPNC with PO as the matrix. Exfoliated CPNC was obtained by melt compounding PA-6 with doublyintercalated MMT [Liu et al., 2003]. Thus, MMT was first intercalated with 3MHDA, and then with the diglycidyl ether of bisphenol A. A high level of clay dispersion was achieved up to 5.6 wt% of inorganic content. At this loading the mechanical properties (with reference to neat PA-6) increased by: tensile modulus 112%, tensile strength at yield 71%, flexural modulus 89%, flexural strength 54%, impact strength 34%, etc. In parallel, the HDT of PA-6 of 62 °C increased to 157 °C. Hasegawa et al. [2003] described quite a revolutionary method for the production of PA-based nanocomposites. Thus, PA-6 was compounded in a TSE with an aqueous suspension of Na-MMT (2 wt% solids). During compounding the water was removed by vacuum-aided devolatilisation. The PA-matrix was not hydrolysed. According to OM and TEM the clay platelets were exfoliated and dispersed homogeneously. The tensile and flexural moduli of the CPNC were 28 and 14% higher than those of neat PA-6, respectively. Similarly, the tensile 236
Exfoliation of Clays and flexural strengths were higher by 28 and 12%, respectively. The Izod impact strength of the new nanocomposites was 12% lower compared to PA-6 and HDT increased from 75 °C for the neat resin to 102 °C. Gas permeability was 31% lower than that of neat PA-6. In short, the properties of the new PA-6/MMT nanocomposites were comparable to those of commercial CPNC from Ube (see Table 62), prepared by compounding PA-6 with 2 wt% MMT-ADA organoclay. CPNCs with PA-66 as a matrix were prepared in a high strain rate TSE [Nair et al., 2002]. The final composition contained 1 to 15 wt% of unidentified organoclay from SCP. While no details on the compounding protocol were given, the strategy adopted by the group involves: 1. The use of organoclay suitable for melt compounding. 2. Preparation of a concentrate by compounding the organoclay with the matrix resin in a TSE. 3. Using a continuous or batch mixer, dilute the concentrate to the final composition and pelletise. 4. Adjust the MW of PA by solid-state polycondensation. 5. Form final object by injection moulding, extrusion, etc. The processing conditions are not critical, but moderation is recommended – the process must provide sufficient residence time at sufficient shear stress at a temperature near to the Tm of the matrix resin. According to the authors, the variables that mostly affect the dispersion are: screw design, rotational speed, throughput rate and the feeding method. The cited publication focused on the fracture mechanics study of PA-66 with different types of layered silicate having nanoscale (fully dispersed) or multiscale (mixed nanoscale/microscale) structure. Independently of the structure and concentration, the fracture toughness was proportional to the size of the crack-tip plastic zone at fracture. At low clay concentration, the toughening effects in CPNC increased with exfoliation and strong clay/matrix interactions. However, at high clay concentrations, where the initiation toughness was low, no significant effect of these microstructural factors was observed – the composites with mixed nanoscale/microscale particulates exhibited the best toughness. In the latter systems high initiation toughening from the nanoclay was combined with high propagation resistance from the micron sized particles. The organoclays seem to influence shear flow at the crack tip, thus playing an important role in the toughening mechanism. Incarnato et al. [2003] melt compounded a commercial ADS copolyamide with up to 9 wt% Cloisite® 30B. The specimens showed mixed intercalated/exfoliated clay dispersion with altered matrix crystalline morphology, e.g., high content of the γ-crystalline phase, and reduced crystallisation temperature. 2.4.3.2 PO-Based CPNCs Plastics comprise over 15 wt% of an automobile weight. Owing to its low price and good performance the share of PP is expanding – it already constitutes 68% in an Opel Astra. Incorporation of 4 wt% of exfoliated clay should improve the flexural strength of PP by ≥ 50%, the modulus by at least 35%, and HDT by ≥ 32 °C, while maintaining the ductility to 10 °C. This makes the development of PP-based nanocomposites commercially important. Serious efforts toward melt 237
Clay-Containing Polymeric Nanocomposites exfoliation of layered materials in PP have been carried out in Japan, North America and Europe. Since most of these efforts lead to intercalation, in part they have already been discussed in Section 2.3.8.2. Usuki et al. [1997] described an early, simple route to PO-based CPNC. First, Na-MMT was cation-exchanged with dimethyl distearyl ammonium chloride (2M2ODA, clay content = 52.8 wt%, d001 ≅ 3.3 nm). Next, an olefin oligomer with telechelic -OH groups (a diol, carbon number = 150 to 200) was dissolved in toluene, and different quantities of 2M2ODA were added to it. After evaporation of toluene, the d001 of the doubly intercalated MMT-2M2ODA/PO was found to depend on the diol-to-2M2ODA ratio: for the ratios 1:1, 3:1 and 5:1 the XRD peak position did not change, giving d001 ≅ 4.4 nm, indicative of the presence of short stacks, but when the ratio was 10:1 a full exfoliation was obtained. Finally, 5 wt% of MMT2M2ODA/PO (diol-to-2M2ODA = 1:1) was compounded with PP in an internal mixer at 220 °C until the torque reached a plateau (ca. 5 min). XRD and TEM showed that in the resulting CPNC the clay was exfoliated and uniformly dispersed. It is noteworthy that in the absence of diol, organoclay dispersed only into micronsize aggregates. The authors concluded that the oligomeric or telechelic diol has good chemical affinity to 2M2ODA, so it can be inserted into the interlamellar space in the presence of toluene. After evaporation of toluene, the -OH groups hydrogen bond with MMT. The presence of oligomer increased the interlayer spacing enough for the PP macromolecule to diffuse into the predominantly hydrocarbon filled interlamellar galleries, eventually exfoliating the organoclay. The short note does not report on the performance of these systems, but since the hydrophobic organoclay is dispersed in PP without covalent bonding between these two or efficient entanglement formation, one would not expect strength to be improved. Another report from the Toyota laboratory also describes melt-intercalation of doubly intercalated clay in PP [Usuki et al., 1999]. MMT was intercalated with a primary ammonium salt (ODA), and then dispersed in a toluene solution of endhydroxylated PBD or maleated-PP. The complex was well-dried, added to molten PP at 200 °C and mixed for 30 min. The interlayer spacing of the product was d001 = 3.82 nm. However, as reported by Kato et al. [1997], adequate exfoliation can only be obtained by melt compounding when the acid value of the MA-PP is reasonably high. Thus, MMT-ODA (d001 = 2.17 nm) was melt blended with PPMA in an internal mixer at 200 °C for 15 min. For the PP-to-organoclay ratio = 3:1, d001 = 7.22 nm was found, indicating advanced intercalation. The authors concluded that one polar group per 25 PP-mers is required to achieve exfoliation. In 1997 Kawasumi et al. [1997] described another version of the process. Thus, PP-based CPNC was prepared by compounding PP, PP-MA, and either MMT (CEC = 1.19 meq/g) or fluoromica (FM) intercalated with stearyl ammonium chloride (ODA). Two types of PP-MA were used: Yumex 1001 and Yumex 1010 from Sanyo Chem., with Mw = 40 and 30 kg/mol, acid number 26 and 52 mg KOH/g, Tm = 154 and 145 °C, respectively. Intercalation of MMT increased the interlayer spacing from d001 = 1.2 to 2.2 nm (inorganic content 67.4 to 68.4 wt%). Similarly, nearly quantitative intercalation of ODA in FM was achieved (d001 = 2.2 nm; inorganic content 68.8 to 70.8 wt%. Exfoliation was carried out by first dry blending the three ingredients, and then compounding them in a TSE at T = 210 °C. The dried pellets were injection-moulded. Miscibility of the PP/PP-MA blends was evaluated using an optical microscope. Phase separation was observed in PP/ Yumex 1010 blends, while PP/ Yumex 1001 mixtures were miscible. Evidently, miscibility of PP/PP-MA depends on the relative 238
Exfoliation of Clays concentration of ingredients and temperature, but mainly on the concentration of MAH in PP-MA. Yumex 1010 contains 52 mmole of KOH/g of polar MAH groups that tend to form their own phase. In consequence, a relatively strong XRD peak was observed for the PP/Yumex 1010 systems (d001 = 5.9 nm), indicating lack of exfoliation. By contrast, the XRD of PP/Yumex 1001 exhibited a relatively small and highly reduced peak indicating a high degree of exfoliation with only few short stacks present. TEM confirmed these results. In compositions containing PP with either MMT-ODA- or FM-ODA, but without PP-MA compatibiliser, the clay formed particles several hundred micrometres in size. Thus, these materials should be considered to be conventional mineral-filled composites. Evidently, addition of compatibilising PP-MA is essential if exfoliation is to take place. Thus, the polar molecules, viz. telechelic diol or PP-MA, must first intercalate into the interlamellar space. The driving force of this secondary intercalation originates from the hydrogen bonding between the polar groups of the compatibiliser (-OH, MAH or -COOH) and the clay surface =Si=O, ≡Si-OH or the intercalant =NH groups. As the interlayer spacing of the clay increases, the solid-solid interactions between the layers are reduced. Under the influence of a strong shear field the doubly intercalated clay disperses in the PP matrix. However, when the doubly intercalated clay is immiscible with the matrix, phase separation occurs and the clay concentrates in domains where exfoliation is unable to take place. Another modification of these methods was published by Hasegawa et al. [2000a]. Thus, MMT-ODA was compounded with three slightly maleated polyolefins, viz. PP (Mw = 210 kg/mol; MI = 150 g/10 min; MAH = 0.2 wt%), PE (MI = 2 g/10 min; MAH = 0.62 wt%), and EPR (M w = 270 kg/mol; MI = 0.06 g/10 min; MAH = 0.55 wt%). The blending was carried out at 200 °C in a TSE (D = 30 mm, L/D = 45.5). Excellent dispersion of clay platelets was shown by XRD and TEM (in decreasing order) for EPR, PP and PE. In the latter system short stacks with d001 ≅ 3 nm were visible. Similarly, Okamoto et al. [2001a] prepared CPNC using organoclay (MMTODA from Nanocor Inc.) and PP-MA (from Exxon Chem.; Mw = 195 kg/mol, Mw/Mn = 2.98; 0.2 wt% of MAH; Tm = 141 °C). Introduction of such a small amount of MAH was enough to obtain sufficient interaction between the PP matrix and the organoclay. Three CPNCs with MMT contents of 2, 4 and 7.5 wt% were prepared at 200 °C in a TSE (TEX30α-45.5BW from JSW). The extruded and pelletised strands were moulded into ca. 2 mm sheets. The interlayer spacing of the organoclay, d001 = 2.31 nm increased to 3.24, 3.03 and 2.89 at the three clay concentrations, respectively. At the same time TEM microscopy showed a fine dispersion of the clay platelets, with about 150 nm in length and about 5 nm in thickness. The XRD diffraction peaks most likely resulted from short stacks comprising 2-3 platelets. The effects of organoclay concentration on melt intercalation were also investigated [2000b]. Thus, MMT-ODA was melt compounded with PP-MA. Good dispersion was obtained - XRD spectra were free of d001 peaks, but a small shoulder for the highest clay content indicated the presence of short stacks. Apparently, the chemical similarity of the paraffinic intercalant to PP matrix and the PP-MA compatibiliser improved the degree of exfoliation, albeit by a small amount. Another approach to the double intercalation strategy for the preparation of PO-based CPNC was proposed by Hudson [1999] and by Jeong et al. [1998]. The three steps are: 239
Clay-Containing Polymeric Nanocomposites 1. Functionalising MMT with an aminosilane. 2. Reacting the free amine with maleated PO (MW ≅ 20 kg/mol; about 1 wt% MAH) either in solution or in melt. 3. Dispersing the organomodified clay in a semicrystalline PO. XRD of the silane treated MMT demonstrated an increase in interlayer separation with d001 = 2.4 to 7.3 nm. TEM showed the presence of many individual MMTplatelets, but in CPNC containing 15% clay most were in stacks of 2-4 layers, ca. 1 μm long. Co-crystallisation between the maleated and the semi-crystalline PO is essential to provide a mechanism of stress transfer between the matrix and clay particles. Serrano et al. [1998] patented CPNC that comprised 40-99.95 wt% EVAl and more than 2 wt% of exfoliated MMT. No onium ion or silane coupling agent was required, provided that water-soluble intercalants were used, e.g., P2VP or PVAl. These hydrogen-bonded to the platelet surface, covering the Na+ and thus shielding the EVAl from its degrading influence. For example, MMT was compounded in a TSE with an intercalant at T = Tm + 50 °C. The CPNCs could be used as external, heat-resistant body parts for the automotive industry; for tyre cords; food wrapping with enhanced gas barrier properties; electrical components; food-grade drink containers, etc. The process for preparation of exfoliated PO nanocomposites developed in Nanocor laboratories involves the use of proprietary organoclays, I.30P and I.31PS (containing a silane coupling agent) [Lan and Quian, 2000; Nanocor’s Technical Service Group, 2001, Nanocor Technical data sheets, 2001]. The process involves addition of 2-5 wt% of PP-MA which acts as a compatibiliser between 2-6 wt% of organoclay and PP. The compounding was carried out in a Leistritz CORI TSE with L/D = 36. The screw configuration is shown in Figure 30. The compounding temperature (including the devolatilisation zone) was 180 °C, increasing progressively to 190 °C at the die. Two approaches were examined. In the first a masterbatch that containing 50-60 wt% of organoclay was prepared, then diluted to the final composition of 2-6 wt% organoclay. In the second, the final composition was compounded directly. In both cases a high degree of exfoliation was obtained, although some short stacks were evident on the TEM micrographs. Good properties were reported, viz. tensile strength increased by 8 or 15%, tensile modulus increased by 29 or 28%, flexural strength increased by 18 or 22%, etc., for direct (single pass) compounding, and the two-step process, respectively. These values were calculated from the authors’ data for CPNC with 4 wt% organoclay. They seem to be short of the expected benchmark set by the automotive industry. There is quite a bit of data scattering. Within the experimental error the single pass provides as good CPNC as the one involving masterbatch. One of the reasons may be matrix degradation – since PP is prone to thermomechanical chain scission, two passes through an extruder are expected to engender more damage. Polypropylene-based CPNCs were also prepared by melt intercalation with two organoclays, three PP-MA containing 1.5 to 5.8 wt% MAH and a PA-6 resin [Wang et al., 2001a]. The two organoclays were MMT-ODA (Nanomer I.30 TC) and Cloisite ® 20A (C20A; MMT-2M2HTA from SCP). The compounding was carried out in an intermeshing, CORI TSE (Leistritz ZSE-27; L/D = 40) operating at T = 180 to 230 °C and at screw speed N = 100 to 200 rpm. To achieve better dispersion, PP-MA was mixed with C20A, then PP was added and compounded to make PP/PP-MA/C20A nanocomposites. 240
Exfoliation of Clays Since PA-6 has Tm ≅ 219 °C, I.30 with better thermal stability than C20A was chosen for the preparation of CPNC with PP/PA-6 blend. The compounding sequence was altered, first blending PP with PP-MA, and then adding I.30 at 100 rpm. PA-6 was incorporated using two procedures: 1. Two-step process where first PA-6, PP-MA, and I.30 were mixed, then PP was added; 2. Three-step process, where first PP-MA was mixed with I.30, then with PA-6, and finally with PP. As summarised in Table 37 the compounding resulted in intercalation. The presence of PA-6 did not increase the d001 value, but reduced the intensity of the XRD diffraction peak, indicating a higher degree of exfoliation. The molecular weight and MAH content of the compatibilising PP-MA strongly affect the nanostructure and the properties of PP nanocomposites. PP-MA with a lower molecular weight and a high MA content leads to better dispersion, but reduces the mechanical performance. Similarly, addition of PA-6 increases the interlayer spacing, but reduces the tensile strength. Addition of a high molecular weight PP-MA with lower MAH content may improve both stiffness and toughness, but improvements are limited viz. 14% for tensile strength, 49% in
Table 37 Interlayer spacing in PP/PP-MA/clay and PP/PP-MA/ PA-6/clay nanocomposites. Data [Wang et al., 2001a] Material
d001 (nm)
Cloisite® 20A
2.3
PP + 5 wt% Cloisite® 20A
2.48
PP-MA (1.5 wt% MAH) + 25 wt% Cloisite® 20A
2.9
PP-MA (2.5 wt% MAH) + 25 wt% Cloisite® 20A
3.0
®
PP-MA (5.8 wt% MAH) + 25 wt% Cloisite 20A
3.7
PP + 15 wt% PP-MA (1.5 wt% MAH) + 5 wt% Cloisite® 20A
3.0
PP + 15 wt% PP-MA (2.5 wt% MAH) + 5 wt% Cloisite® 20A
2.9
PP + 15 wt% PP-MA (5.8 wt% MAH) + 5 wt% Cloisite® 20A
3.6
Nanomer I.30 TC
2.3
PA-6 + 5 wt% Nanomer I.30 TC
3.71
PP + 15 wt% PP-MA (1.5 wt% MAH) + 5 wt% Nanomer I.30 TC
2.85
PP + 5 wt% PP-MA (1.5 wt% MAH) + 5 wt% Nanomer I.30 TC + 10 wt% PA-6
3.7
241
Clay-Containing Polymeric Nanocomposites tensile modulus and 30% in Izod impact strength. The best overall performance was obtained from the system: PP + 15 wt% PP-MA (1.5 wt% MAH) + 5 wt% Cloisite® 20A. Following the successful direct melt exfoliation of Na-MMT in a PA-6 matrix by Hasegawa et al. [2003] the method was extended to a Na-MMT/PP system [Kato et al., 2003]. A TSE with D = 32 mm, and L/D = 77 was used at T = 200 to 220 °C and screw speed of 300 rpm, with residence time of ca. 6 min. The barrel was divided into three functional sections: (1). mixing, (2). exfoliation, and (3). devolatilisation. Into section (1), first PP and PP-MAH were fed from the main hopper to be melt compounded with Na-MMT (fed a bit downstream). Into section (2) ca. 10% of water was injected (at the processing temperatures the saturated vapour pressure is 1.5 to 2.3 MPa) to expand the interlayer spacing of Na-MMT, and facilitate reactions between MAH and the clay surface. In section (3) a two-stage devolatilisation was carried out. As a result, the clay was uniformly dispersed as either exfoliated platelets or intercalated short stacks. The mechanical properties of these nanocomposites were nearly the same as those of conventional CPNC prepared by compounding with organoclay, for example, the flexural modulus of the new CPNC was 2.76 versus 2.88 GPa. 2.4.3.3 PCL-Based CPNCs With the growing concern for recycling industrial and municipal waste, biodegradable polymers and their nanocomposites are becoming increasingly attractive [Akovali et al., 1998]. As discussed in Section 5.4.2, polylactic acid (PLA) has been selected by Toyota for progressive replacement of PP composites in automotive applications. Poly-ε-caprolactam (PCL) is another biodegradable aliphatic polyester with good potential for the consumer market. However, performance of both these resins, PLA and PCL, should be enhanced by incorporation of exfoliated clay platelets. In a series of papers from Mons-Hainaut the focus has been placed on the development of a melt compounding process for the preparation of CPNC with PCL as the matrix, which would show significant improvement of mechanical properties at low clay loading [Pantoustier et al., 2001; Lepoittevin et al., 2002a]. Thus, commercial PCL (Mn = 49, Mw = 69 kg/mol) was mixed with several clays: (C1) Na+-MMT; (C2) MMT¯DDA; (C3) MMT-ODA; (C4) MMT-2M2ODA (Cloisite®25A); and (C5) MMT-MHT2EtOH (Cloisite®30B). The mixing was carried out in a two-roll mill at 130 °C for 10 min, followed by compression moulding into 3 mm thick specimens. Compositions containing 1, 3, 5 and 10 wt% of MMT were prepared. The interlayer spacing of the neat organoclays (after dispersing 3 wt% in PLC) was: d001 = 1.21/1.23; 1.38/1.37; 1.87/2.56; 2.9/3.6; and 1.84/3.1 nm, for C1 to C5, respectively. Evidently, Na+-MMT and MMT¯DDA did not intercalate, whereas modest intercalation was obtained for the remaining organoclays, where the diffusion of PCL macromolecules increased the interlamellar galleries by Δd001 = 0.69, 0.7 and 1.26 nm, respectively. A summary of the mechanical performance is displayed in Figure 56. Except for the modulus, the best performance from between the six materials is that of neat matrix, PCL. 242
Exfoliation of Clays
Figure 56 Tensile properties of polycaprolactam (PCL) and its mixtures containing 3 wt% of MMT in the form of organoclays C1 to C5 – see text. The vertical broken lines are drawn to help judge the magnitude of the property change in maximum stress at break, tensile modulus and the maximum strain at break. Data [Pantoustier et al., 2001].
A partial explanation for the poor mechanical performance can be found in the reduction of crystallinity. To compensate for it a higher clay loading is required. Thus, CPNC with up to 10 wt% MMT were prepared with C4 and C5 organoclays [Lepoittevin et al., 2002a,b]. The highest tensile modulus was obtained for 10 wt% C5 (increase by a factor of 1.85), but the material became brittle (7% elongation at break). A better balance of properties was obtained for a 5 wt% C5 loading, viz. modulus increased by 45% and the elongation at break, εb = 560%. The CPNC also showed improved thermal stability and flame resistance. Lepoittevin et al. [2002b] also prepared CPNC with PCL as the matrix via in situ intercalative ring-opening polymerisation of ε-caprolactone catalysed by dibutyl-tin dimethoxide, Bu2Sn(MeO)2. For comparison, 1, 3, 5, and 10 wt % of C1, C4 and C5 organoclays were incorporated. The C1 and C4 content had no effect on the PCL molecular weight (Mn = 14 to 24 kg/mol), but as C5 content increased, the Mn was found to decrease, probably due to chain transfer by the intercalant’s –OH groups. The polymerisation resulted in a modest intercalation of C1 (d001 increased from 1.2 to 1.63 nm for the 3 wt% in PCL matrix); in intercalation of C4 (d001 increased from 1.86 to 2.68 nm); and in nearly full exfoliation of C5, confirmed by TEM. TGA showed better thermal stability for the exfoliated than for the intercalated CPNC, both exceeding the degradation temperature of neat PCL. Tortora et al. [2002] also used the polycondensation route, but starting with MMT pre-intercalated with ω-amino dodecyl acid, 0 to 44 wt% of MMT-ADA. According to XRD exfoliation was achieved for CPNC with up to 16 wt% clay. 243
Clay-Containing Polymeric Nanocomposites Incorporation of clay slightly reduced the Tm (from 62 to 57 °C), but significantly reduced the matrix crystallinity as well as the onset of degradation temperature: Td = 382 (for PCL) to 188 °C (for CPNC with 44 wt% clay). The reduction could in part be explained by the decrease of the matrix molecular weight with increasing organoclay content (from Mn = 11.6 to 6.5 kg/mol). The permeability of dichloromethane or water vapours through these CPNCs was also significantly reduced. The same method was used to prepare CPNCs of PCL polymerised in the presence of up to 44 wt% of MMT-ADA [Pucciariello et al., 2004]. As the organoclay content increased the molecular weight decreased, viz. Mn from 11.6 to 6.5 and Mw from 14.9 to 7.8 kg/mol. According to XRD the nanocomposites with low organoclay content (6 and 18 wt%) were exfoliated, whereas those containing 22, 30 or 44 wt% were only intercalated. In parallel with the interlayer spacing there was a significant difference in thermal behaviour – the CPNC with exfoliated structures showed an increase of Tc, Tm and the heats of transition, whereas the intercalated ones showed the reverse trend – these thermal properties decreased with organoclay content. Surprisingly, as the clay content increased the crystallisation rate decreased, but the PCL morphology remained spherulitic. Since clay platelets act as nucleating agents, the slower crystallisation most likely resulted from hindrance of molecular motion in the progressively more and more crowded interlamellar galleries. In related work, PCL was polymerised in the presence of 3 wt% of co-intercalated MMT [Gorrasi et al., 2004]. The co-intercalants were: (CH3)2(CH2CH2OH)(C16H33)N+ Cl– or 2MEtOHHDA, and (CH3)3(C16H33)N+ Cl– or 3MHDA. The co-intercalation was carried out with the ratio of these two ammonium salts ranging from 0 to 25, 50, 75, and 100%. Similarly, as in the preceding communication, the PCL molecular weight decreased with 2MEtOHHDA (i.e., -OH group) content (the clay loading being constant) from Mn = 56 to 28 kg/mol. According to XRD, only the CPNC with more than 50% 2MEtOHHDA were exfoliated, confirming the importance of good bonding between clay and matrix. The tensile modulus increased with the organoclay hydroxyl group content to about 9-fold of the value for neat PCL. Furthermore, as the –OH group content increased, so did the grafting density of PCL to clay, and as a consequence the gas and vapour diffusion (e.g., dichloromethane or water) were reduced. Considering on the one hand the relative ease of producing exfoliated CPNC with PCL, and on the other hand its miscibility with several commercially important resins (e.g., SAN, ABS, PVC, CPE, PEST, PEMA, etc.), it is only natural that this CPNC would be used as a vehicle for nanoreinforcing PCL-miscible polymer blends [Choe et al., 2002]. To ascertain system miscibility, 20 to 40 wt% of PCL (MW = 10 to 100 kg/mol) with well dispersed organoclay has been used. The preferred clay is MMT pre-intercalated with CH 3 (CH 2 ) 17 N + (HOCH 2 CH 2 ) 2 CH 3 Cl - , acidified CH 3 (CH 2 ) n -NH 2 or CH3(CH2)nNHR ( where n = 8 to 18, and R is a hydrocarbon). As the cited examples show, melt compounding at T ≤ 140 °C for the required time (e.g., 20 min!) results in exfoliation, whereas melt compounding at T > 140 °C results in intercalation with d001 < 4 nm. Compounding in a single step involves simultaneously mixing organoclay with PCL and a miscible resin. Compounding in two steps involves first, the preparation of organoclay concentrate (≤ 30 wt%) in PCL, and subsequently diluting it with the selected blend component (e.g., 244
Exfoliation of Clays PVC, SAN or ABS). It was found that incorporation of organoclay improved mechanical properties (tensile modulus increased by 350%) and heat deflection temperature (by 50 °C). With these advantages, CPNC may replace the conventional ABS resin and its blends with PVC, for example, in electronic housing applications. To prepare CPNC of SAN, Lee and Kim [2004] dispersed Na-MMT in caprolactam, polymerised the suspension, and then melt compounded the product with SAN. Surprisingly, XRD indicated full exfoliation within the PCL domains, while Tg showed good miscibility between PCL and SAN. The stress-strain curves from the tensile tests at room temperature indicated significant improvement of the stress and elongation at break. 2.4.3.4 Other Systems Melt compounding is becoming a method of choice for CPNCs with a variety of polymeric matrices. The method has not been successful in the case of a PS matrix [Tanoue et al., 2003]. The authors compounded three grades of PS with Cloisite® 10A under diverse processing conditions. The interlayer spacing versus residence time in the TSE initially slightly increased to decrease at longer times toward the value of non-intercalated MMT. The reason for this behaviour was a rapid thermomechanical decomposition of organoclay by the Hofmann elimination mechanism, followed by chain scission of PS. Better results could be obtained using MMT pre-intercalated with either imidazolium cations [Wang et al., 2003], or with ammonium-ion terminated oligostyrene [Su et al., 2004]. Owing to the higher decomposition temperature, and good miscibility with the matrix, a satisfactory dispersion of clay platelets has been obtained. Intercalated CPNC with PMMA as the matrix were prepared by melt mixing with 10 wt% of commercial organoclays, viz. Cloisite® -10A, -20A, -25A and –30B [Kumar et al., 2003]. The interlayer spacing of the organoclays was found to increase by 0.7 to 1.4 nm. The glass-transition temperature of these CPNC was lower than that of neat PMMA by ca. 10 °C. Melt compounding was also used to disperse 1 to 10 wt% of Cloisite® 30B in an epoxy matrix [Yasmin et al., 2003]. Thus, a three-roll mill was used at 60 °C to exfoliate the clay. The tensile stress-strain curves indicated that as the organoclay loading increased the modulus increased as well (by 67% at 10% loading), but the strain at break decreased (by a factor of ca. 5.7). Incomplete degassing, leaving voids in the specimen was most likely responsible for these modest improvements. However, since the method does not require solvent it is environmentally-friendly and worth pursuing.
2.4.4 Functional CPNC The previous examples described CPNC prepared as structural materials, with improved strength, modulus, and/or high barrier properties. However, the methods are quite general, applicable to other materials, designed to perform specific functions. A few examples will be mentioned. 2.4.4.1 Liquid Crystal/Clay Composite (LCC) Kawasumi et al. [1998] prepared nematic liquid crystal (LC) with up to 2 wt% of MMT. First, the clay was intercalated with a variety of ammonium cations, 245
Clay-Containing Polymeric Nanocomposites including primary and quaternary alkyl (C8, C12, C18, etc.) as well as primary and tertiary amines with biphenyl groups, e.g., 4-cyano-(4´-biphenyl-oxy)-undecyl ammonium salt, that showed good miscibility with LC. MMT intercalation was conducted in aqueous medium at 50 °C stirring the suspension for 3 h, washing the precipitate with ethanol and freeze-drying it. The interlayer spacing varied with the intercalant from d001 = 1.37 to 3.28 nm. Next, the organoclay (0.2 to 2 wt% in the final composition) was dispersed in DMF or dimethyl acetamide (DMAc) and the LC was added. The solution was spin-coated on slide glass, DMF slowly evaporated at 50 °C, and then under vacuum. When the clay had good affinity for LC, the system was homogeneous. The CPNC exhibited bistable and reversible electrooptical effects between a light scattering and transparent state that could be selected by changing the frequency and voltage of applied electric fields. To activate transparency a 50 ms electrical field (60 Hz, 100 V) application was required. To induce opacity a 50 ms high frequency field (1.5 kHz, 100 V) was needed. When the field was switched off the transparency (in the first case) or opacity (in the second) remained. The memory effect is related to the orientation of clay particles caused by application of the low frequency field and randomisation by the high frequency one. This new CPNC is a potential candidate for light controlling glasses, high information display devices (that do not require active addressing devices), erasable optical storage devices, etc. 2.4.4.2 Biodegradable CPNC with Polylactic Acid (PLA) Lactic acid (LA), a colourless liquid with chemical formula CH3CHOHCOOH was identified by Carl W. Scheele in 1780. It exists in two optically active forms, dextro- and laevo-, often designated as D-lactic acid and L-lactic acid. Ordinary (or racemic) lactic acid is optically inactive. Pharmaceutical and food industries prefer the L-lactic acid and lactates since the human body does not metabolise the D-form. Lactic acid has been produced by fermentation controlled by specific bacteria, using sugar as the starting raw material. While LA can be synthesised, its commercial production is based on fermentation. The starting material is usually starch from any plant (e.g., corn, potatoes or sweet potatoes). In the first step, starch is recovered from plant material and enzymatically converted to glucose, which in turn is fermented by microorganisms (e.g., Lactobacillus amylovorus) to PLA [Yan et al., 2001]. For high productivity the fermentation requires immobilisation of the bacteria and a complex mixture of nutrients. Carothers in 1932 produced a low molecular weight polylactic acid (PLA) by heating lactic acid under vacuum. Following this discovery DuPont manufactured medical grade sutures, implants and packaging for controlled drug release applications. Today the two principal routes to PLA are: 1. Direct condensation followed by chain coupling, and 2. Ring-opening polymerisation of lactide(s). In the first process the polycondensation water is removed by azeotropic distillation and high vacuum – the approach used by Carothers (nowadays by Mitsui Toatsu Chem.) to produce a low to intermediate MW polymer. This product may be used directly, or if high MW is required, it can be chemically coupled using isocyanates, epoxies, anhydrides, carbodiimides, ortho-esters, etc. The second process also starts with polycondensation of LA under mild conditions, 246
Exfoliation of Clays to produce a low MW intermediary PLA, which in turn can be catalytically converted to cyclic dimer – the lactide. The latter has three forms that can be separated by distillation: L-, D-, and DL- or meso- [Kõhn et al., 2003]. The ring opening polymerisation of the lactide leads to a wide range of molecular weights. Several manufacturers supply PLA, viz. Galactic, Mitsui Toatsu Chem., Mitsubishi Plastics, Inc. (‘Ecoloju’ Biodegradable Plastic Films and Sheets), Shimadzu, Trespaphan-Celanese AG (biaxially oriented PLA, BOPLA), and Cargill Dow Polymers. The latter company announced a major expansion aimed primarily at the fibre and nonwovens market. Typical properties of PLA are listed in Table 38. L-PLA shows better mechanical properties than meso-PLA. The former properties are comparable to these of typical commodity resins, viz. PP, PS or PVC. The main problem is the relatively low Tg and crystallinity. In consequence, PLA has a tendency to permanently deform under stress at T > 50 °C. It is expected that incorporating nanoclays may solve both problems. Solution crystallised PLA was shown to reach over 80% crystallinity [Fischer et al., 1973]. Another problem – the relatively high density of PLA – may be eliminated by microfoaming a properly formulated CPNC.
Table 38 Physical properties of polylactic acid (PLA) Property Molecular weight (kg/mol) Glass transition temperature, Tg (ºC)
Typical values 100 to 300 55-70
Melting temperature, Tm (ºC)
130-215
Heat of melting, ΔHm (J/g)
8.1-93.1
Crystallinity, X (%) Surface energy (dynes) Solubility parameter, δ ( J / mL )
10-4 0 38 19-20.5
Density, ρ (kg/m3)
1.25
Melt flow rate, MFR (g/10 min)
2-2 0
Permeability of O2 & CO2 (f mol/m s Pa) Tensile modulus, E (GPa) (orientational effects)
4.25 & 23.2 1.9-4.1
Yield strength, L-PLA/meso-PLA, σy (MPa)
70/53
Strength at break, L-PLA/meso-PLA, σB (MPa)
66/44
Flexural strength, L-PLA/meso-PLA, I (MPa)
119/88
Elongation at break, εB (%)
100-180
Notched Izod impact strength, L-PLA/meso-PLA, (J/m)
66/18
Vicat, L-PLA/meso-PLA, (°C)
165/52
247
Clay-Containing Polymeric Nanocomposites In 1998 Rhone-Poulenc obtained a US patent for nanocomposites prepared by dissolving polyethylene glycol- and/or polypropylene glycol-polylactic acid copolymers in an organic solvent followed by adding it to an aqueous suspension of clay [Spenleuhauer et al., 1998]. CPNC could be recovered by precipitation (without additional colloidal protective agents) or by microfluidisation and solvent evaporation. Also in 1998, Toyota established the Biotechnology & Afforestation Business Department. The group, elevated to the level of an independent business division, actively promotes business activities in forestation, agriculture and other fields contributing to the solution of such problems as food shortages and environmental degradation. In 2001, Toyota jointly with Mitsui & Co., Ltd started a joint venture in Indonesia. The aim is to produce ca. 100 kton/year of sweet potatoes from 6,000 ha dedicated farmland for the production of PLA and animal feed. Production will start in early 2004. The PLA plant will be built next to the animal feed processing plant. In the meantime there is serious ongoing research and engineering design effort for the manufacture of lactic acid, its polymerisation into PLA, and upgrading the performance of the latter by means of CPNC technology. The main concerns are heat resistance, mechanical performance and durability. Toyota believes that upgraded PLA can be used to replace PP fabrics and mouldings in automobile interior parts, such as seat mouldings, interior door finish, pillar garnish, etc. Some of these parts were already introduced in a concept car presented at the 2002 motor shows. CPNCs with PLA as a matrix can be prepared either by polymerisation in the presence of organoclay (e.g., by the monomer intercalation method) or by melt compounding. Ray et al. [2002a] selected the latter route. PLA (D-enantiomer content of 1.1-1.7 wt%) was vacuum dried, and dry blended with MMT-ODA. The inorganic content varied from 2 to 4.8 wt%. The effect of a compatibiliser, a low molecular weight o-PCL (α,ω-hydroxy-terminated with Mw = 2 kg/mol), was also studied. The mixtures were melt-compounded in a TSE at 190 °C. XRD and TEM provided evidence of intercalation but not exfoliation. As a result of melt-compounding the interlayer spacing of MMT-ODA, d001 = 2.31 nm, increased to higher values, dependent on the clay content, viz. d001 = 3.10 (for 2 wt%) and 2.89 nm (for 4.8 wt%). Incorporation of the compatibiliser did not change the XRD peak position, but it increased its intensity. The number of clay platelets per stack was calculated as 10 to 13. Good bonding between the clay platelets and the matrix was evidenced by melt rheology – the low frequency storage modulus strongly increased with the organoclay content. The degree of crystallinity of PLA was determined as 36% – incorporation of either organoclay or PCL increased it to ca. 50%. However, the latter additive simultaneously decreased the Tg by about 5 °C. The two additives also had opposite effects on the storage modulus – the clay increased it, the compatibiliser decreased its value. Interesting and unexpected results were obtained by burning the CPNC. The resulting cinder was a highly porous, ‘house of cards’, a sort of ceramic foam, with density, ρ = 0.187 g/ml. Since the BET surface area of MMT is 780 m2/g and that of the ceramic foam is 31 m2/g, the reduction is about 25-fold [Ray et al., 2002b]. XRD indicated that the sintered material had amorphous structure. A more detailed study of the effect of clay on PLA crystallinity was published separately [Ray et al., 2002c]. In this case, instead of MMT-ODA + PCL compatibiliser, a fluoromica (FM; CEC = 1.2 meq/g; p ≈ 250) pre-intercalated with methyl coco-alkyl bis(2-hydroxyethyl)-ammonium ion (MC2EtOH) was 248
Exfoliation of Clays used. Melt compounding of PLA with 4 wt% of this organoclay was carried out in a TSE at 210 °C. Compounding increased d001 from 2.1 to 3.1 nm, but it reduced the PLA Mw from 177 to 150 kg/mol. The effect of the organoclay on crystallinity was quite modest, viz. the degree of crystallinity increased by ca. 4%, Tg decreased by 4 °C, Tm was unchanged and Tc increased by 28 °C. However, at the same time the improvement of the mechanical properties of PLA was significant – flex modulus, storage modulus and flex strength at 25 °C all increased by: 26, 25 and 9%, respectively. Surprisingly, incorporation of the organoclay accelerated the rate of PLA biodegradability by ca. 60%. In the fourth paper of this series [Ray et al., 2003], PLA was melt compounded in a TSE at 210 °C with 4, 5 or 7 wt% of organoclay. The latter was MMT (CEC = 0.9 meq/g; p ≈ 100), pre-intercalated with trimethyl octadecyl ammonium (3MODA). Melt compounding only slightly reduced the Mw of the PLA matrix (from 177 to 165 kg/mol), and virtually did not affect Tg and Tm. The strongest increase of the PLA degree of crystallinity was for 4 wt% clay loading – an increase from 36 to 65%. At the higher organoclay contents the crystallinity decreased to 56 and 54%, respectively. Again, these nanocomposites were only intercalated. While the interlayer spacing of organoclay was d001 = 1.93 nm, incorporation into PLA melt at the level of 4, 5 and 7 wt% organoclay increased this value to 3.05, 2.7 and 2.8 nm, respectively. The data also indicate that here the number of platelets in a short stack was about 4. Nevertheless, the improvement of the mechanical properties (see Figure 57) was quite impressive – note that the tensile strength and the elongation at break go through a local maximum at about 4 wt%. On the other hand, the Young’s modulus and the heat distortion temperature linearly increase with the clay content in the whole range of compositions. In Figure 58 the relative oxygen permeability is shown versus organoclay content. The experimental data
Figure 57 Mechanical properties of a PLA/MMT-3MODA system as functions of the organoclay content. The data points are experimental [Ray et al., 2003], the lines have been included to guide the eye.
249
Clay-Containing Polymeric Nanocomposites
Figure 58 Relative oxygen permeability of a PLA/MMT-3MODA system as a function of organoclay content. The calculated dependence suggests low a clay platelet aspect ratio of p = 24 or their misalignment. Data [Ray et al., 2003].
follow Equation 23 with the aspect ratio, p = 24. Since the average number of platelets in the stack is about 4, and the interlayer spacing is about 3 nm, the effective length of the stack is 288 nm – to be compared with the estimate by the authors of approximate platelet diameter, D ≈ 200 nm. As has been noted before, incorporation of MMT enhanced the biodegradability of PLA. It is interesting that the enhancement is not observed during an initial period (30 < ti < 50 days). Afterwards, without the organoclay the biodegradation proceeds at a slow rate, whereas in its presence it is dramatically increased – all specimens decomposed in 60 days. Since the biodegradation proceeds by hydrolysis of the ester linkages, one may speculate that the difference is related to the size of the PLA crystals, as well as the presence of ≡Si-OH hydroxyl groups. The melt flow behaviour of these systems strongly suggests that these groups react with PLA, causing end tethering. However, such a non-catalysed reaction is slow and numerous ≡Si-OH groups are most likely left behind. Paper 5 of this series examines the influence of different organoclays on the performance of PLA nanocomposites [Ray et al., 2003b]. At a constant loading of 4 wt%, the best performance was obtained using synthetic fluoromica (FM) pre-intercalated with methyl di-polyethylene glycol coco ammonium ion: the flexural modulus increased by 26%, strength by 9%, HDT increase by 17 °C, and O2 permeability was reduced by a factor of 2.8. Furthermore, biodegradability of this system was twice as rapid as that of neat PLA. The final contribution in this series [Ray and Okamoto, 2003] focused on the flow and foamability of PLA containing up to 7 wt% of MMT-ODA. Non-linear viscoelastic behaviour was observed in dynamic, steady state and extensional flow fields. High quality foams with bubble size of 2.59 ± 0.55 μm were produced. 250
Exfoliation of Clays Recently, fundamental studies on the crystalline and supermolecular structures of CPNC with PLA as matrix were carried out [Pluta et al., 2002]. The specimens were prepared by melt compounding in an internal mixer (at 60 rpm and 180 °C for 10 min). Thus PLA (MW = 166 kg/mol; 4.1 mol% D-lactide) was mixed with 0.3 wt % Ultranox 626 stabiliser and 3 wt% (inorganic content) of a nanofiller. As the latter, either Na-MMT (CEC = 0.92 meq/g; p = 250 to 500; d001 = 1.22 nm) or MMT intercalated with MHT2EtOH from SCP were used. The compounds were compression moulded at 185 °C into 0.5 mm thick sheets. Three solidification procedures were used: quenching, annealing of the quenched sample at 120 °C, and crystallisation from the melt at 120 °C for 3 h. XRD showed that meltcompounding MHT2EtOH with PLA resulted in intercalated CPNC – the interlayer spacing increased from 2.1 to 3.14 nm. By contrast, Na-MMT only formed a microdispersion in PLA matrix. Microscopic observations of the specimen structures are summarised in Table 39. The authors noted that in the nanocomposites the clay short stacks were incorporated into the interlamellar and/or intralamellar regions. The data are a bit disappointing as the nanofiller had negligible effects on the Tg as well as on crystallinity. However, it was found that during melt compounding with clay or organoclay PLA degraded less than without it, viz.: decrease of Mn for PLA, NC and MC was 41.2, 19.6 and 22.1%,
Table 39 Morphology of PLA and its micro- and nano-composites. Data [Pluta et al., 2002] Tg (°C)
T1 (°C)
Tm (°C)
Δ Hm (J/g)
PLA(Q) Amorphous
58.4
108.3
129.6
3.6
PLA(A)
Thin spherulites
58.4
125.4
153.7
42.3
PLA(C)
Spherulites D ≈ 330 μm
52.9
124.0
154.1
52.6
NC(Q)
Amorphous
58.0
113.7
130.9
1.0
NC(A)
Thin spherulites
58.2
121.9
152.7
38.7
NC(C)
Spherulites D ≈ 200 μm
54.9
124.0
154.3
49.0
MC(Q)
Amorphous
58.5
110.6
130.0
1.6
MC(A)
Thin spherulites
58.8
124.3
153.1
39.5
MC(C)
Spherulites D ≈ 150 μm
54.1
124.0
154.3
51.6
Material
Code*
PLA
PLA + 3 wt% MHT2EtOH
PLA + 3 wt% Na-MMT
Crystallinity**
Notes: * Q = quenched, A = sample Q annealed at 120 °C, C = crystallized from melt at 120 °C for 3 h ** D = spherulite diameter, Tg is the glass-transition temperature, T1 is the onset temperature for cold crystallisation and Tm is the melting temperature
251
Clay-Containing Polymeric Nanocomposites respectively. The processability of these compositions was comparable to that of the neat PLA. An improvement of the CPNC thermal stability under oxidative conditions was also noted. CPNC based on polycaprolactone (PCL) were prepared by melt mixing [Di et al., 2003]. The exfoliation of Cloisite® 30B was achieved for an organoclay loading below 10 wt%. In parallel with the degree of dispersion was the increase of the PCL crystallisation temperature. A significant enhancement in mechanical properties and thermal stability was reported. For packaging applications, Paul et al. [Paul et al., 2003] studied the preparation and properties of CPNC with plasticised PLA. To make these systems competitive with the standard packaging resins (viz. PE or PP) such PLA properties as thermal stability and gas barrier properties should be enhanced. Thus PLA (Mw = 155 kg/mol) was plasticised with PEG (Mw = 1 kg/mol; 5, 10 or 20 wt%) and melt-compounded with a nanofiller (1 to 10 wt% of clay) in the presence of 0.3 wt% of Ultranox® 626 stabiliser. Four Cloisite‚s from SCP were used: Na-MMT, C25A, C20A and C30B (see Table 16). The compounding was done in an internal mixer at 180 °C and 20 rpm for 4 min, then at 60 rpm for 3 min. The compound was pressed into 3 mm thick sheets. According to XRD, at a loading of 3 wt% of these four nanofillers, only intercalation was obtained. The interlayer spacings increased from d001(Na+) = 1.21 to 1.77 nm, d001(C25A) = 2.04 to 3.24 nm, d001(C20A) = 2.36 to 3.65 nm, and d001(C30B) = 1.84 to 3.80 nm. The intercalation process was complicated by two secondary intercalants, PLA and PEG. Intercalation of Na-MMT was most likely done by PEG, but in the case of organoclays both partake in the process. Plasticisation reduced the Tg from 55 to 15 ± 1 °C, and slightly increased the Tm from 167 to 170 ± 1 °C, i.e., the nanofillers have been found unable to affect the PLA transition temperatures. On the other hand, the decomposition temperature was significantly increased, e.g., addition of 3 wt% of C30B increased it by 40 °C. The flammability was also suppressed. The mechanical properties were not reported. Bionolle, a commercial biodegradable polybutylenesuccinate (PBS) was used as CPNC matrix for a series of organoclays [Someya et al., 2004]. The latter were based on MMT pre-intercalated with primary and tertiary ammonium cations (e.g., DDA, ODA, ADA, N-lauryldiethanolamine, and 1-[N,N-bis(2-hydroxyethyl) amino]-2-propanol). MMT-ODA provided CPNC with the largest interlayer spacing, d001 = 3.27 nm, but even in this case the improvement of mechanical performance was relatively minor, for example, at 3 wt% clay loading tensile modulus increased by 40% while strength decreased by 9%. A similar enhancement of rigidity and reduction of strength was reported by Okamoto et al. [2003] for CPNC with PBS as matrix – only intercalation was achieved. Several other water-soluble polymers have been proposed for biodegradable applications of CPNC, viz. PVAl, EVAl, PEG or PCL. While melt blending resulted in intercalation, the reactive route leads to exfoliated CPNC [Pramanik, et al., 2001; 2002; Kwiatkowski and Whittaker, 2001; Pantoustier et al., 2002]. However, none of these materials has been as intensely promoted as the CPNC with MMT and PLA. 2.4.4.3 Poly(N-Vinyl Carbazole)/MMT Polymerisation of purified N-vinyl carbazole (NVC) was conducted in the presence of dehydrated MMT either in bulk, just above Tm = 64 °C, or in benzene solution at
252
Exfoliation of Clays 50 °C [Biswas and Ray, 1998]. Both methods resulted in a similar conversion versus time plot. The cationic polymerisation at 50 °C was catalysed by MMT. First, the NVC was adsorbed by the MMT surface, forming a complex either with the counterion or with metal oxide impurities in MMT (e.g., Fe2O3), which resulted in electron transfer that initiated polymerisation. Up to 10 wt% of poly-N-vinyl carbazole (same as poly-9-vinyl carbazole, PVK) was retained by MMT after extraction by benzene, indicating the presence of end-tethered macromolecules. XRD showed that as a result of polymerisation the characteristic MMT diffraction peak (d001 = 0.99 nm) was significantly reduced in intensity and a new one (d001 = 1.46 nm) appeared. The latter indicated intercalation, but judging by its low intensity, probably some exfoliation took place. Considering the high concentration of MMT in these systems (≤ 43 wt%) exfoliation was not to be expected. Incorporation of MMT significantly improved the thermal stability; whereas PVK totally decomposes at 700 °C, the CPNC lost only 12.2 wt% at 683 °C. However, the most astonishing result of the process was the increase of DC conductivity by 10 orders of magnitude: from 10-16 to 10-6 S/cm, and doping may increase it even further. In the subsequent paper the authors discussed polymerisation of NVC in the presence of FeCl3-impregnated MMT [Ray and Biswas, 1999]. XRD displayed only one d001 = 0.98 nm peak, thus the MMT-FeCl3 complex was not intercalated during polymerisation. TEM showed the clay particles to be in the range 3040 nm. However, the DC conductivity of the CPNC was in the range 3.1 x 10-5 S/cm at 5.9 wt% of FeCl3, dependent on the FeCl3 loading. In this context it may be interesting to note that polymerisation on the surface of nanoparticles can be used for the preparation of self-assembled monolayers. For example, brush-type core-shell macromolecular structures were polymerised on the surface of Au particles with diameter d = 2-5 nm by means of surface-initiated living cationic polymerisation of 2-oxazolines [Jordan et al., 2001]. In 2002 Suh and Park prepared CPNC with MMT-ODA dispersed in a PVK (Mw = 100 kg/mol) matrix. Three organoclays were prepared, containing 0, 50, 75 and 100% of intercalant per 100% of the clay’s CEC. Dissolving PVK in chlorobenzene, and then adding a suspension of 1 wt% organoclay resulted in CPNC. XRD of the organoclay indicated that the interlayer spacing increased with ODA content from d001 = 1.25, 1.72, 1.77 and 1.89 nm, for 0, 50, 75 and 100% CEC, respectively. The spacing in dry, cast films of PVK/MMT-ODA also depended on the ODA content. Evidently, for 0% (neat MMT) a non-intercalated, simple, filled system was obtained. Intercalated CPNC were obtained for compositions containing 75 and 100% CEC of ODA. However, the system containing a low density of the intercalant in organoclay, PVK/MMT-50%ODA was found exfoliated. This, seemingly surprising observation is in excellent agreement with thermodynamic theories discussed in Section 3.15. Yu et al. [2004] prepared a series of CPNCs by dispersing an organoclay in NVC, and then UV-photoinitiating in situ polymerisation with triarylsulfonium salt as the initiator. The organoclay was Na-MMT pre-intercalated with acidified diallyl amine: (CH2=CH-CH2)2NH2+ Cl–. The final CPNC contained 0 to 1 wt% of clay. Within this range of composition, Mn decreased from 2273 to 1438 g/mol, while Tg increased from ca. 48 to 78 °C. This Tg increase is intriguing, considering the low clay loading, poor exfoliation, and the low and decreasing (with clay content) molecular weight of the matrix (note that Tg decreases with Mn as: Tg = Tg∞ - A/Mn).
253
Clay-Containing Polymeric Nanocomposites 2.4.4.4 Polydiacetylene An attempt was made to intercalate MMT, vermiculite or mica with diaminediacetylene (DADA), monoamine-diacetylene (MADA), or ω-aminoacid-diacetylene (AADA), then polymerise these using 60Co irradiation [Srikhirin et al., 1998]. For this purpose, 14-amino-10,12-tetradiynoic acid and 10,12-docosadiyndiamine (diacetylenic diamine), were synthesised. The approach was to intercalate the clays with a cationically terminated diacetylene monomer by cation exchange. The intercalation was carried out by mixing the clay with diacetylene ammonium salt solution in 10% EtOH/water for up to seven days at 70 °C. XRD and FTIR confirmed the intercalation. The interlayer spacing of the intercalate decreased with the number of washings, e.g., from d001 = 4.5 to 3.7 nm. The interlayer spacing and polymerisability of the intercalated diacetylene were found to depend on the molecular length, the CEC of the clay, the type of diacetylene molecule, and the solvent treatment. Polymerisation of a diacetylene may occur when the monomer has proper packing in the solid state. The polymerisation results in a conjugated backbone, providing π-electron delocalisation along the backbone which gives polydiacetylenes their unique electrical and optical properties. The polymerisation of the intercalated diacetylene should result in growth of the polydiacetylene chain along the interlayer gallery. However, packing of the intercalated monomers controls the reactivity of the diacetylene/clay complex. Since the intercalated DADA/MMT, AADA/MMT and AADA/vermiculite lie flat on the clay surface they lack the proper geometry to be polymerised. The intercalated MADA/vermiculite and MADA/MMT are tilted with respect to the clay surface, but polymerisation was only observed in MADA/vermiculite where the intercalated diacetylene has proper packing geometry and monomer density (smaller area per intercalated molecule). The conductivity of I2-doped polyaniline (PANI) approaches that of Cu, but the polymer is brittle and sensitive to humidity and oxygen. Recently MMT was intercalated with aniline hydrochloride, and then aniline was electropolymerised, increasing the interlayer spacing to d 001 = 1.47 nm [Feng et al., 2001]. Unfortunately, the note did not report data on the stability and conductivity of this potentially interesting material. 2.4.4.5 Clay-Functional Organic Molecules The intercalation of indigo blue into clay is credited for the preservation of vivid colours in Maya’s frescos. Several patents from AMCOL (e.g., [Beall et al., 1999b]) describe the direct intercalation of clay with organic molecules having functional properties, e.g., fungicides, pesticides, insecticides, acaricides and other agricultural or medically active compounds. These systems were prepared by dispersing clay in an aqueous medium, followed by adding a solution of a functional compound. The latter must have a polar moiety, e.g., carboxylic acid, ester, amide, aldehyde, ketone, sulfur-oxygen or phosphorus-oxygen moiety, cyano, or a nitro moiety. Judging by XRD diffraction the compounds intercalated into the interlayer galleries. In some cases strong ionic bonding was obtained, in others (viz. the pesticide trifluralin) the bonding was of the physical adsorption type, allowing the compound to be fully released from the interlamellar space. There is another group of compounds of industrial interest: molecules with specific functionalities, e.g., catalysis, photoluminescence, photochromism, optical 254
Exfoliation of Clays nonlinearity, etc. Using such molecules as intercalants may impose molecular ordering that enhances the performance. Furthermore, intercalation often results in improved thermal and oxidative stability caused by reduced thermal motion and/or reduced oxygen diffusivity. Recently [Kim et al., 2001] nanocomposites of organic laser dye-hectorite complexes were prepared. Thus, nanocomposites of hectorite (CEC = 0.44 meq/g) with 7-diethylamino-4-methyl coumarin dye were produced. By alternate adsorption of positively charged polyelectrolyte and a negatively charged hectorite/ coumarin complex, a multilayered film was fabricated. The coumarin was acidified and then intercalated into hectorite by the ion exchange reaction. The original interlayer spacing of hectorite (d001 = 0.98 nm) slowly increased with coumarin content up to the total exchange of inorganic cations (from 1.34 to 1.39 nm), then more rapidly up to 2.22 nm at a coumarin loading of 20.4 wt% (252% of the CEC). This suggested different packing of coumarin molecules within the interlayer galleries – a flat monolayer below 100% CEC and a tilted monolayer above. As a consequence, CPNC films with high molecular order and thermal stability were prepared for photonic applications. A multilayered composite film showed a linear increase in the absorption and in the fluorescence intensity with the number of deposited layers. Other photofunctional chromophores and fluorophores with a high degree of molecular orientation order and enhanced electrooptical properties can be prepared. By adjusting the intercalant-clay interactions the quantum efficiency of fluorescence can be adjusted. Molecular dynamic simulation of the adsorption of methylene blue by MMT, beidellite and muscovite has been carried out, predicting d001 = 1.23 to 1.57 nm [Yu et al., 2000b]. CPNC have been prepared by polymerising 2-ethynyl pyridine within MMT [Liu et al., 2001]. Intercalation was carried out in benzene, by first dispersing in it dehydrated Ca2+-MMT (CEC = 1.20 meq/g), adding 2-ethynylpyridine (2-EPy) and then stirring the mixture at 65 ± 3 °C for 24 h. Initially, the suspension was red. The colour darkened gradually to reach dark brown colour when the reaction was terminated. Thus, 2-EPy polymerised spontaneously within the interlayer galleries. The complex was centrifuged and washed several times with benzene then dried. Up to 21.3 wt% of 2-EPy was adsorbed, increasing the interlayer spacing up to d001 = 1.64 nm, consistent with the van der Waals dimension of a pyridine ring (0.66 nm). Thus, the polyacetylene macromolecules are flat on the MMT surface with pyridine rings tilted with respect to the surface. 2.4.4.6 Super-Absorbent CPNC MMT was dispersed in an aqueous solution of Na-acrylic acid, N,N´-methylenebis-acrylamide (crosslinker) and K2S2O8 (free radical initiator) [Lin et al., 2001]. The reaction (4 h at 60-70 °C) resulted in a slightly yellow powder. The water absorbency (Q) strongly depended on the crosslinker concentration (e.g., 1.1 at 0.05% to 0.7 kg/g at 0.13%), MMT content (maximum Q was found at 30 wt% MMT), and the degree of neutralisation. Interlayer spacing has not been measured. 2.4.4.7 Emulsion Polymerisation of CPNC The preferred methods for the preparation of PS-based CPNCs are emulsion, solution or bulk (co-) polymerisation that start with organoclay dispersed in a monomer phase. Since clay intercalation is usually in water, emulsion 255
Clay-Containing Polymeric Nanocomposites polymerisation is an obvious choice. However, when the hydrophobicity induced by intercalation is insufficient, the clay will not be transferred to the monomer phase, leading to a limited improvement of the resulting polymer properties. For example, SBR-based PNC was prepared by emulsion polymerisation in the presence of onium-intercalated MMT. The PNCs showed superior barrier and mechanical properties [Elspass et al., 1997; 1999]. SAN-based PNCs [Noh and Lee, 1999] were prepared by emulsion or solution polymerisation. Thermogravimetric analysis showed that at 500 °C the amount of remaining residue was: 5 wt% of SAN, 20 wt% of solution-type PNC and 70 wt% of emulsion-type. There is a detailed discussion on the dispersion of organoclays during emulsion or suspension polymerisation in Section 4.1.4.1.
256
Introduction
Part 3 Fundamental Aspects
Clay-Containing Polymeric Nanocomposites
Thermodynamics
3.1
Thermodynamics
Nanostructures are intermediate in size between molecular and micron-size structures. They contain a countable number of atoms and thus resemble molecules. Their small size and structure result in strong interactions and high surface-to-volume ratio. Nanostructures may show specific electronic and magnetic characteristics, often dominated by the quantum effects. Their properties depend upon the size, the shape and arrangement of the atoms; for example, substitution of Mg for Al in the octahedral layer of smectites is essential for intercalation and exfoliation in a polymeric matrix. Thus, there is a need to develop new fundamental tools capable of describing and predicting properties of the nanoscale structures.
3.1.1 Glass Transition in Thin Films As the size of the particles decreases, the transition temperatures change as well. For example, the melting point of bulk PE is about Tm ≅ 400 K, but the Tm of its nanosized particles with diameter of 5 and 13 nm is 218 and 266 K, respectively. Similar reductions were reported for the glass transition temperatures (Tg) of thin film. In the latter case, three types of behaviour were distinguished: (1) for thin films grafted to a solid substrate, (2) for supported films, and (3) for unsupported films [Forrest and Dalnoki-Veress, 2001]. The two latter cases are under active investigation at present. 1. For type (1) the Tg depends on the degree of molecular bonding and it may either decrease or increase with decrease of film thickness. 2. For type (2) there is a reduction of Tg with reduction of film thickness that seems to be independent of molecular weight. An empirical relation was derived: (30) Tg = Tgb ⎡1 − ( a / h) ⎤ ⎢⎣ ⎥⎦ where Tgb is the glass transition of the bulk phase, h is the film thickness, and a and δ are empirical parameters which depend on the system, viz. for PS a = 3.2 nm and δ = 1.8; for PMMA a = 0.35 nm and δ = 0.8. It was reported that the behaviour was independent (or nearly so) of the substrate. A mechanism that may explain such Tg behaviour is the presence of a thin surface layer having high molecular mobility and higher free volume fraction than that in the bulk [Forrest and Dalnoki-Veress, 2001]. The thermodynamic analysis of the surface/interface energy indicated that there are two basic mechanisms responsible for the minimisation of the surface energy: high δ
257
Clay-Containing Polymeric Nanocomposites concentration of highly mobile chain ends and migration toward the surface of low molecular weight components (low MW fractions and additives) [Helfand and Tagami, 1971; 1972]. Recently, Monte-Carlo (MC) simulation was used to study the behaviour of thin polymeric films. As shown in Figure 59, the computations gave results closely resembling the experimental results. Furthermore, it was found that the thin films have a fluid-like interfacial region where mobility is considerably higher than in the bulk film [Jain and de Pablo, 2002]. The computed results indicated that it is not an effect of different skin composition, but rather the chain conformation. In agreement with Helfand and Tagami calculations, near the surface MC simulations predicted high concentration of chain ends and reduced density. In this region the linear macromolecules have nonGaussian coil configuration. 3. For type (3) the Tg depends on both film thickness and weight-average molecular weight (Mw). The reduction of Tg is significantly more pronounced than in type (2). For high Mw the following dependence was found to describe the behaviour: (31) Tg − Tg* = b ln(Mw / Mw* ) × (h − h* ) where symbols with asterisk (*) are reference variables. For PS: b = 0.70 ± 0.02 K/nm, Mw* = 69 ± 4 kg/mol, h* = 10.3 ± 0.1 nm, and Tg* = 423 ± 2 K. The dependence is illustrated in Figure 60. Chow [2002] proposed the glass transition theory for unsupported, nanoscale polymeric films with thickness < 100 nm. The theory is based on the Gibbs and DiMarzio time-independent theory that Tg is related to the loss of configurational entropy of a liquid during the cooling experiment. The entropy in turn can be related to the free volume fluctuations, which depend on the temperature and film thickness. The Langevin equation was used to derive the following dependence: 2θ ⎫ ⎧ (32) ln Tg (h, N ) / Tg∞ = − k B N / 2 N o ΔC p ⋅ exp ⎨− (h / ho ) ⋅ N o / N ⎬ ⎩ ⎭ where Tg and Tg∞ are the polymer glass transition temperature in the film and in the bulk, respectively; h and ho are the film thickness and its critical value equal to the radius of gyration; kB is the Boltzmann constant; N and No are the number of statistical segments in a polymer and in a reference state; ΔCp is heat capacity. The exponent θ = 0.7 was determined for polystyrene. The relation predicts that the Tg of low molecular weight polymer thin films has higher value than that of high molecular weight. This unexpected result is confirmed by experiments and the empirical dependence is illustrated in Figure 60. The author also derived an expression predicting variation of Tg with the strength of the polymer-substrate interactions. For low (or absent) interactions the dependence for free film was recovered; hence Tg(film) < Tg(bulk). For strongly interacting systems the Tg in thin film was predicted to have a higher value than that observed for the bulk polymer: Tg(film) > Tg(bulk). Another type of nanosized particles with unexpected Tg behaviour was recently described [Mi et al., 2002]. The authors prepared single-macromolecular coil size globules of polyacrylamide (PAAM; Mw = 5,000 to 6,000 kg/mol) by atomisation spraying of a dilute polymer solution, followed by drying. Thermal analysis showed that the globules had a higher Tg than that of the same bulk
[
258
] (
)
[
]
Thermodynamics
Figure 59 Glass transition temperature for thin, supported PS films. Line represents experimental data, points computed as Tg* ≡ k B Tg ε −1 (where kB is the Boltzmann constant and e is the nearest-neighbour interaction energy – see text).
Figure 60 Glass transition temperature for thin, freestanding PS films. Lines computed from Equation 31 for the indicated molecular weights.
259
Clay-Containing Polymeric Nanocomposites polymer, viz. 205 versus 193 °C, respectively. The explanation offered is based on the configurational entropy of a single macromolecule in the bulk and in the globular state. The analysis indicated that polymer chains in the latter state should have Tg higher by a few percent (absolute T-scale), e.g., by ca. 10 °C, which indeed was observed. However, this analysis neglects the surface energy difference (which acts in the opposite direction), as well as the interaction of the globule with the substrate (which tends to increase Tg). In conclusion, three factors have been identified as affecting the glass transition: molecular weight (MW), film thickness (h), and the interactions between film and the substrate (χ). The theory predicts that high molecular weight polymer diffusing into narrow interlamellar galleries will have high mobility under the condition of poor interaction with the clay, but poor interactions means absence of the driving force for diffusion. On the other hand, strong interactions means good ‘motivation’ for the macromolecules to diffuse, bind with the clay substrate and ‘solidify’ into a layer of adsorbed macromolecules with low segmental mobility.
3.1.2 Nanothermodynamics In the early 1960s Hill noted that the equilibrium thermodynamic properties of sufficiently small systems differ from those predicted by the classical thermodynamics of large systems [Hill, 1994]. More recently, the field was renamed nanothermodynamics [Hill, 2001a; 2001b]. Originally, the treatment considered homogeneous, small systems, such as individual macromolecules or colloidal particles, each composed of n-elements. For n → ∞ the macroscopic properties are recovered. However, when n is small there are at least two significant contributions to the energetic state of such a system, the first depends on the surface energies, and the second on the entropy related to the dynamic state of the particles. For example, a single platelet of MMT, owing to its thickness (0.96 nm) and large aspect ratio, has a specific surface area of ca. 800 m2/g, and about 36% of its atoms are on the surface. Considering the high surface energy of solids (e.g., surface energy of freshly cleaved mica is γ = 4,500 mN/m) the surface effects are expected to dominate the thermodynamic state of nanoparticles. Since in most systems the miscibility and phase equilibria are determined by a fine balance of small energetic and configurational contributions, the surface effects are expected to influence them, and thus to be critical for the performance of CPNC. For the principle of nanothermodynamics to apply to a macroscopic system, it must be visualised as an ensemble of small subsystems (or aggregates) independent of each other. The number of elements n in each aggregate does not have to be the same. A good example is a semicrystalline substance having numerous small crystals of various sizes that contain n molecules. The substance may show size-related effects, e.g., of the melting point, width of the melting transition, heat capacity, glass transition temperature, etc. Other examples are provided in recent publications on statistical thermodynamics of metastable droplets [Hill and Chamberlin, 1998], on ferromagnetism [Chamberlin, 2000] or on phonon scattering by the boundary of nanocrystals [Shrivastava, 2002]. In the latter paper it was shown that the specific heat of crystalline materials depends on particle size. One of Gibbs contributions to the equilibrium thermodynamics was addition of the chemical potential to the basic energy-heat-work equation. This 260
Thermodynamics concentration-dependent contribution is essential for understanding and computing diverse thermodynamic properties, viz. chemical reactions, phase separation, gas solubility, osmotic pressure, etc. In a sense, incorporation of the nanoscale effects into the thermodynamics is the next step in generalisation of the thermodynamic behaviour. Thus, Hill formally wrote the total (subscript ‘t’) energy of a macrosystem as: dEt = TdSt − PdVt +
∑ μ dN i
i, t
+ edn
(33)
i
e ≡ (∂Et / ∂n) St, Vt, Ni ,t
where T is temperature, P is pressure, V is volume, S is the entropy, μi is the chemical potential and Ni is the number of molecules of species ‘i’. The last term in this dependence, edn originates from the extra contribution of the small system. In Hill’s concept, any macroscopic system may be described in terms of contribution from all elements, viz.: Xt º nX, where X = E, S, V, Ni, with n being an extensive variable. The variable e is a sort of chemical potential of the whole ensemble, i.e., a ‘subdivision potential’ that varies with T, P and μi: de = − SdT + VdP −
∑ N dμ i
i
(34)
i
− S = (∂e / ∂T ) P, μ ; V = (∂e / ∂P)T, μ ; Ni = (∂e / ∂μ i )T, P, μ i
i
j
The nanothermodynamic properties usually differ in different ‘environments’. For example, a rigid incompressible aggregate of n = 102 or 103 spherical molecules in a constant temperature bath may be considered as having two ‘environmental variables’, n and T. The aggregate entropy can be computed considering the translational, rotational and vibrational motions of the molecules. However, one may ‘construct’ such an aggregate considering that the constant-temperature bath contains individual molecules with chemical potential, μ. Depending on the value of μ the molecules will form an aggregate containing n´ molecules, hence now the ‘environmental variables’ are μ and T. One may choose a value of μ that will result in n = n´. However, the calculated entropy for the first case (n, T) will be smaller than that for the second (μ, T). The reason for the discrepancy is the presence of fluctuations in the second case that cannot be directly incorporated in the first case. However, as the aggregate approaches macroscopic size, i.e., when n → ∞, this entropy difference will disappear or in other words, when an ensemble contains only one aggregate, the ‘classical’ macroscopic equilibrium thermodynamics is recovered. Hill considered four sets of ‘environments’: (N, V, T); (N, P, T); (μ, V, T); and (μ, P, T). For example, the ‘subdivision potential – e’ of a one-component small system in the four sets of variables was computed. 1. Environmental Variables N, V, T: From the partition function: Q( N , V , T ) =
∑ exp{− E ( N , V ) / kT} i
i
(35)
the following relationships are derived: − kT ln Q = E − TS = − PV + μN + e − d ( kT ln Q) = − SdT − PdV + μdN
(36)
261
Clay-Containing Polymeric Nanocomposites 2. Environmental Variables N, P, T: From the partition function:
Δ( N , P, T ) =
∑
j, V
{
}
exp − E j ( N , V ) / kT exp{− PV / kT }
(37)
the following relationships are derived:
− kT ln Δ = E − TS + PV = μN + e
− d ( kT ln Δ ) = − SdT − VdP + μdN 3. Environmental Variables μ, V, T: From the partition function:
Ξ (μ, V , T ) =
∑
j, N
{
}
exp − E j ( N , V ) / kT exp{− Nμ / kT }
(38)
(39)
the following relationships are derived: − kT ln Ξ = E − TS − μN = − PV + e − d ( kT ln Ξ ) = − SdT − PdV − Ndμ
(40)
4. Environmental Variables μ, P, T: From the partition function: ϒ ( μ , P, T) =
∑
j, N, V
{
}
exp − E j ( N , V ) / kT exp{Nμ / kT } exp{− PV / kT }
(41)
the following relationships are derived: − kT ln ϒ = E − TS + PV − μN = e
− d ( kT ln ϒ ) = − SdT + VdP − Ndμ = de
(42)
The latter case is particularly important since here the system is ‘completely open’ and all environmental variables are intensive. Note that the three variables: μ, P and T cannot all be independent in a macroscopic system but they can in a small system. In nanosystems these intensive variables determine the mean size N (an extensive variable). This type of system is found in biology, in bulk magnetic materials, a Gibbs surface excess resulting from an adsorbent molecule at the end of a one-dimensional lattice gas, or metastable supersaturated gaseous states near the gas-liquid transition point. Recently, this treatment was extended to one-dimensional adsorption of gas [Hill, 2001a; 2002b]. Consider a surface of B adsorbent, with molecules independent of each other, from each of which a chain of M binding sites extends vertically. B and M are both very large. The B molecules, at T and μ, go on and off the sites. There is an interaction energy (w) between any two adsorbed molecules on nearest-neighbour sites of a chain. Also, there is an interaction energy w′ between an adsorbent molecule and an adsorbed molecule that occupies the first site of the chain. The mean number of adsorbed molecules per chain is denoted N ex and typically its magnitude is of the order of 1 to 10. For B surface excesses (small system) free energy is: E = TS + μN ex B + eB
(
)
dE = TdS + μd N ex B + edB
So far, nanothermodynamics have not been applied to PNC. 262
(43)
Thermodynamics
3.1.3 Vaia’s Lattice Model for Organoclay Intercalation by Molten Polymer 3.1.3.1 Introduction In his PhD thesis of 1995 Vaia used a lattice model to describe the thermodynamic behaviour of CPNC comprising organically modified layered silicates, e.g., mica or MMT (the work was published two years later [Vaia and Giannelis, 1997a; 1997b]). Initially, the interlayer galleries are occupied by hydrated alkali metal cations, which during intercalation are exchanged for organic ones – most often quaternary alkylammonium ones. Because of the negative charge on the silicate surface and crowding of intercalant groups, the alkyl tail of the alkylammonium molecule projects away from the surface. However, as the data in Figure 23 demonstrate, the expansion of the interlamellar gallery proceeds step-wise, at lower efficiency than would be expected from fully stretched hydrocarbon chains. At the processing temperature of CPNC, Tprocess ≥ 150 °C, the interlayer structure is expected to be disordered with a density comparable to that of liquid alkanes. Preparation of CPNC involves the formation of nanostructures where the matrix molecules (monomers, macromers or cyclomers introduced during the reactive process or macromolecules incorporated during melt compounding) diffuse into the interlayer galleries, either as a result of static annealing, dynamic mixing or both. Polymer diffusion into galleries expands the silicate layers. Depending on the degree of penetration and resulting interlayer spacing, either intercalated or exfoliated nanocomposites are obtained. Their structure depends on the length, density and type of intercalant, on the type, molecular weight and structure of the polymer, on the concentration of the pre-intercalated clay, as well as the polymer-clay and polymer-intercalant interactions. Vaia modelled CPNC as two parallel platelets pre-intercalated with short chains tethered to the clay surface, and then dispersed in an infinite sea of molten polymer (infinitely diluted system). During the second stage of the intercalation the macromolecules diffuse into the interlamellar galleries. Within these galleries the polymer forms a dilute solution with intercalant molecules. The basic assumptions are that the system is incompressible, has constant polymer density, and the intercalant is end-tethered. Furthermore, it was assumed that the interactions that account for the miscibility/immiscibility of polymer/organoclay systems are polymer-intercalant and polymer-clay type – the clay-clay interactions and their variation with the gallery height were neglected. The final state of the intercalation is given by the equilibrium thermodynamics. The author assumed that the interlamellar gallery height does not exceed the full extension of the intercalant chains on both gallery surfaces, thus the model accounts only for the initial intercalation of organoclay by molten polymer. A thermodynamic description takes into account the entropy-related configurations of the constituents and the interactions between them. The external forces (e.g., shearing) are not considered. Derivation starts with the assumption that the macroscopic thermodynamic description of the system is obtained considering a single sandwich of the organoclay clay (having infinite breadth and width and the initial separation h0) embedded in molten polymer. As the macromolecules diffuse into the gallery, the interlamellar spacing increases to h. The Helmholtz free energy change, ΔF, associated with the interlayer expansion from ho to h, is given by the internal energy change (associated 263
Clay-Containing Polymeric Nanocomposites with intermolecular interactions), ΔE, and the combinatorial entropy change (associated with configurational changes of the constituents), ΔS:
ΔF ≡ F(h) − F(ho ) = ΔE − TΔS
(44)
where T is the absolute temperature. Accordingly, the negative value, ΔF < 0, would indicate that, from the equilibrium thermodynamics point of view, the interlayer expansion is favourable. The average interlayer volume fraction, ϕˆ i , of the end-tethered intercalant chains (ϕˆ 2 = ho / h) and the intercalated polymer (ϕˆ 1 = 1 − ϕˆ 2 ) are expressed in terms of gallery height, h. The end-tethered intercalant molecules can only reach the distance h∞ of fully extended alkyl chain length, thus uniform mixing will not be possible when the interlayer height exceeds its double size, h > 2h∞. Therefore, the mean-field approximation is only valid for ϕˆ 1 ≤ 1 − (ho / 2h∞ ) . 3.1.3.2 Entropic Contributions There are three contributions to the configurational entropy, ΔS: (1) from changes associated with the intercalated clay, (2) from the diffusing polymer and (3) from the intercalant molecules. During polymer diffusion into the interlamellar galleries the clay platelets are pushed apart, but their translational entropy is relatively small and Vaia considered this contribution negligible. For the polymer, the configurational entropy arises from changes of the macromolecular confinement, from its unperturbed random coil structure in the melt to a solution with the intercalant molecules within the gallery; hence there is a loss of the macromolecular configurational entropy. By contrast, the tethered intercalant chains gain configurational freedom (thus entropy) as the gallery height increases. Thus, ΔS has an entropy loss associated with confining the macromolecules and the entropy gain of the tethered intercalant during the gallery height increase from h0 to h:
ΔS = ΔS polymer + ΔSintercalant
(45)
For thermodynamically favourable intercalation of organoclay by polymer the Helmholtz free energy should be negative hence the favourable entropic contribution, ΔS, should be positive. The entropic contribution per interlayer volume (superscript ‘v’) of a macromolecule diffusing into the interlayer gallery was expressed in terms of the Dolan and Edward’s theory for confined random-flight polymer chains with excluded volume:
(
v ΔS polymer / R = −(ϕˆ 1a1 / ν1 h) π 2 a1 / 6h + u 3 / m1
)
(46)
where v1, al, and m1 are the molar volume per segment, the segment length, and degree of polymerisation of the polymer, respectively. Excluded volume is included in the dimensionless parameter, u. Since all the parameters in Equation 46 are positive, the entropic contribution of the polymer diffusing into clay gallery is negative hence it hinders the intercalation process. Using estimated numerical values of parameters (see Table 40) the entropic effect per surface area, ΔSA/R, was computed. To express the second contribution, ΔSintercalant, the authors used the HugginsFlory lattice model for polymer solutions. The original relations were modified to account for the restricted freedom of the end-tethered alkyls and the influence of the silicate surface on their conformation. The entropy change per interlayer volume was expressed in terms of the increased layer separation from h0 to h: 264
Thermodynamics
Table 40 Parameters used for computing the thermodynamic properties. Data [Vaia, 1995] Parameter
Symbol
Value
Polymer segment length
a1
2.5a2
Polymer molar volume
v1
3v2
-1/2
Excluded volume
um1
0.8
C18H37-NH3+
m2 = 9.5
Intercalant segment length (nm)
a2
0.25
Intercalant molar volume (ml)
v2
33
Initial gallery height (nm)
h0
1.3
Final gallery height (nm)
2h∞ = 2a2m2
4.75
Intercalant (ODA)
ΔSivntercalant / R = (ϕˆ 2 / v2 )(ψ s − ψ so ) ln c − (ϕˆ 1 / m1 v1 ) ln ϕˆ 1 ≅ (ϕˆ 2 / v2 )(ψ s − ψ so ) ln c (47) with ψ ≅ (a / h) cos 2 (πh / 2h ) s
2
∞
where νi, mi, and ϕi are the molar volume per segment, the number of segments per chain, and the interlayer volume fraction, respectively. The statistical surface fractions of interlayer sites near the surface that are occupied by the tethered chains are expressed as: ψs ≡ ψs (h); ψso ≡ ψsh0 ≥ ψs h ≥ h0). The statistical surface factor, c, accounts for the inaccessibility of sites on the far side of the surface (c = 0.75 for z = 4, decreasing to 0.5 for z → ∞, where z is the lattice coordination number). Finally, a2 is the intercalant segment length of the tethered chain and R is the gas constant. Since ln c < 0, Equation 47 predicts that the entropic contribution of the intercalant is positive, thus favourable for intercalation. The parameters of the configurational entropy that originates from the polymer, intercalant and their sum are listed in Table 40; the computed dependencies are presented in Figure 61. Evidently, the loss of polymer entropy during its diffusion into a gallery is partially compensated by the configurational gains of the intercalant molecules. While initially there is a possibility of a small increase of the interlayer spacing (by 0 < h - ho < 0.8 nm sufficient to accommodate 1 to 2 macromolecules) any larger expansion is forbidden by a rapid decrease of the total entropic change. Thus, in the absence of favourable energetic interactions any further polymer diffusion into the interlamellar galleries is not possible. As it will be evident from the experimental data presented in Section 4, many CPNC prepared by the melt or reactive intercalation method (e.g., in PS or PO matrix), show increased interlayer spacing, Δd001, by about 0.35 to 0.75 nm, which corresponds to single or two macromolecular layers inside the interlamellar gallery. On the basis of Vaia’s entropic calculation one may postulate that in these systems the energetic contribution to the thermodynamics is negligibly small. 265
Clay-Containing Polymeric Nanocomposites
Figure 61 Changes of entropy during diffusion of macromolecules into interlamellar galleries. Initially, the positive contribution from the intercalant compensates for the negative one from polymer. After expansion by ca. 0.8 nm, the entropic contribution becomes negative and, in the absence of negative enthalpic contribution, the diffusion is expected to stop. After [Vaia, 1995].
3.1.3.3 Interactions Vaia assumed that the energetic contributions originate in interactions between the three components of the system: the silicate, s, the tethered intercalant, a, and the polymer, p. Furthermore, the total change of the internal energy is taken as a sum of the energy change associated with each pairwise interaction, εjk, with the number of contacts expressed by the contact area. The pairwise interaction energy per contact was expressed in terms of the cohesive or interfacial energy per lattice site. Thus, the internal energy change of the system upon polymer intercalation is:
(
)
ΔE = Aspf ε sp + Aapf ε ap + Asaf − Asai ε sa A jkf and
A ijk is
where the total area of contact between components j-k for the final (f) and the initial (i) system, respectively. The dependence was simplified assuming that initially all the lattice sites within the interlayer gallery are occupied by tethered intercalant chain segments, thus Asp = Asaf − Asai
The next step in the derivation was a bit arbitrary, using van Opstal et al. [1991] expression for the internal energy change per interlayer volume that takes place during intercalation:
Δeν = ϕˆ 1ϕˆ 2 Δε =
266
(
2ϕˆ 1ϕˆ 2 ε sp, sa / ho + ε ap / r2
[
)
]
1 − ϕˆ 2 1 − (r1 / r2 ) − (r1 / ho )
(48)
Thermodynamics where Δε is the effective interaction energy per interlayer volume and ri is the radius of the interaction surface, related to the molar volume and segment length by the simple dependence: ri2 = ν i / πN A ai (where NA is Avogadro’s number). The change in interactions at the interlayer surface was written as εsp, sa = εsp – εsa. To encourage intercalation by diffusion of macromolecules the energetic contribution, Δev, thus the numerator in Equation 48, should be negative. Evidently, this can be expected when the interaction energies of polymer: εap < 0 and/or εsp < εsa – when polymer ‘likes’ to cohabit the gallery, interacting with intercalant chains and/or the clay surfaces. The total free energy change associated with polymer diffusion into interlamellar galleries and increased interlamellar spacing may be calculated from Equations 44-48. In Figure 62 the total free energy change during polymer diffusion into the gallery space is presented for εap ≅ 0 (van der Waals interactions between intercalant chains and macromolecules) and for four values of the interaction parameter, εsp, sa. Other parameters were taken from Table 40. As the author noted, the values for the interaction parameters, εij, are not readily available. Thus, he proposed that these might be replaced by interfacial energies between interacting species i-j, viz. εap ~ vap and εsp, sa ~ vsp – vsa. However, since the interfacial energies are also inaccessible, Vaia adopted the procedure proposed by van Oss [1994] of calculating the interfacial energy from the surface energies of the interacting species. The procedure considers that the interactions are composed of polar/associative ( ν ip ) and dispersive/dipolar ( ν id ) forces:
ν ij = ν ijp + ν ijd ν ijp = 2⎛ ν i+ − ν i− − ν −j ⎞ ⎝ ⎠ ν ijd = ⎛ ν id − ν dj ⎞ ⎝ ⎠
(49)
2
where ν i+ and ν i− are respectively, the electron acceptor and donor contributions. Luciani et al. [1996; 1997; 1998] showed that the interfacial properties are dominated by the polar and hydrogen bonding interactions, whereas the bulk properties are dominated by the dispersive forces. In consequence, another method for calculating the interfacial energy has been proposed. The method involves computation of the interfacial tension coefficient from the difference between the dipolar, polar and hydrogen bonding components of the two interacting species i-j, viz.: n 2 2 2 (50) ν ij = k1 (T )⎧⎨Θ δ i, d − δ j, d + δ i, p − δ j, p + δ i, h − δ j, h ⎫⎬ ⎩ ⎭ where k1(T) and n are equation parameters and Θ ≅ 0.4 is the relative measure of the importance of the dispersive contribution (its value for bulk interactions is 4). Since the solubility parameter can be calculated from the group contribution method [Hansen, 1967; 1994; 1995; 2000], Equation 50 is general.
(
) (
) (
)
3.1.3.4 Consequences of the Model Vaia’s model predicts that the only source of favourable entropic contribution toward intercalation originates in a greater number of possible configurations of the tethered intercalant molecules. This contribution reaches maximum when in the organoclay the intercalant has a highly ordered, pseudocrystalline structure 267
Clay-Containing Polymeric Nanocomposites
Figure 62 Changes of the free energy during intercalation/exfoliation by polymer for four indicated values of the interaction parameter between clay and polymer. With strong ‘specific’ energy of interaction the system is expected to exfoliate. For middle values it may intercalate, while in the absence of energetic contribution little change of Δd001 is expected. After [Vaia, 1995].
and it ‘dissolves’ into polymer diffusing into the interlamellar gallery. However, reduction of entropy by such a ‘melting’ process is relatively small. Thus, the theory predicts that the entropic effects are small and mainly detrimental to intercalation by diffusion of macromolecules; hence the only method for achieving intercalation is by ascertaining favourable energetic contributions. Adopting the derivation by van Opstal et al. [1991], Vaia simplified the energetic contribution given by Equation 48. The magnitude and sign of the effective interaction parameters is related to pairwise interactions between the three constituents, the accessibility of interaction sites, the interlayer packing density, and the size of the intercalant chain. In most organoclays the intercalants have apolar, paraffinic moieties, thus van der Waals dispersion forces dominate polymerintercalant interactions. In most cases, these interactions are characterised by a small positive value of εap. By contrast, the clay surface is polar, thus if the diffusing polymer has groups able to form polar or hydrogen bonds, the polymer-surface interactions may be more favourable to exfoliation, thus εsp,sa < 0. In the case of polar polymers, the favourable interactions may also be engendered by suitably modifying the intercalant molecules. However, such a modification must be carefully designed, as too strong interactions during an early stage of intercalation may form a tightly packed polymer layer around the clay particles, slowing down further polymer diffusion into the galleries. The enthalpic contributions in Equation 48, εsp,sa and εap, are scaled with ho and r2, respectively. Since, as discussed above, the former may be negative and the latter positive, contrary to the expectations, better chances for exfoliation may be found starting with organoclay having small (but adequate to polymer 268
Thermodynamics diffusion) value of the interlayer spacing, d001 ≅ h0 + 0.96. On the other hand, for reducing the small, parasitic contribution of εap the value of r2 should be maximised by selecting intercalant with large molar volume moieties. Thus, not only the magnitude of the energetic interactions, but the intercalant structure is important. When the interaction parameter, εsp, sa = εsp – εsa < 0 an intercalation, then at higher values, an exfoliation is possible. According to the adopted mechanism, it is not a binary interaction parameter, but the relative magnitude of two binary interaction parameters that counts. If exfoliation is to take place: εsp < εsa, i.e., clay must interact more strongly with the macromolecules than with the intercalant chains. Accordingly, the role of the latter seems to be to sufficiently open the interlamellar galleries facilitating the diffusion of macromolecules. Thus, again the thermodynamic argument favours organoclays with small, but sufficient interlayer spacing. Serious difficulties are expected, and indeed found, when trying to exfoliate organoclay with nonpolar olefinic macromolecules, e.g., PP. For these systems a compatibiliser strategy has been used. Thus, polar groups are introduced (e.g., by maleation) into macromolecules having the same molecular configuration as the nonpolar polymer. The key strategy is for these compatibilising macromolecules to engender strong polymer-clay interactions. The nonpolar part of these modified macromolecules must remain miscible with the nonpolar polymer, ascertaining a single-phase structure of the matrix. Since the magnitude of the Helmholtz free energy depends on the fine balance between small entropic and enthalpic contributions, these ‘compatibilising’ strategies may be suitable for the polymer-induced exfoliation. In summary, the mean-field statistical lattice model is a first approximation of the polymer melt intercalation. It is a simplified equilibrium thermodynamics model, which considers only diffusion of molten polymer into a single gallery formed by two organoclay platelets immersed in an infinite sea of polymer melt. Of the three components: clay (s), polymer (p) and intercalant (a), the binary s-s interactions have not been considered. Under normal circumstances these interactions are about 100 times stronger than those between organic segments, i.e., a-a or p-p. The s-s interactions are responsible for the difficulties in breaking the clay particles into short stacks, doublets and finally into individual exfoliated platelets. The model does not incorporate the ‘small system’ contributions introduced by Hill. One must also ponder whether for the nanoscale system the assumption of random mixing (e.g., absence of molecular adsorption on the surface of crystalline solid) is valid. Furthermore, the model assumes absence of external forces during melt exfoliation. In spite of the simplifications and omissions the model helps in understanding the mechanisms responsible for polymer intercalation. The insight gained leads to recommended procedures for enhanced intercalation hence improved CPNC performance. The recommendations are based on considerations of the entropic and enthalpic contributions to equilibrium thermodynamics. The free energy of the process is a fine balance of these two. The entropic penalty of macromolecular confinement within the interlamellar galleries may be, at least partially, compensated by the increased conformational freedom of the intercalant chain as the gallery expands. The model indicates that to be successful in forming exfoliated CPNC strong energetic interactions between the diffusing polymer and clay are required. Vaia is less enthusiastic about inducing interactions between intercalant and polymer, expecting hindered kinetics of polymer diffusion. 269
Clay-Containing Polymeric Nanocomposites 3.1.3.5 Model Prediction versus Static Intercalation Results Vaia [1995] carried out static melt intercalation of several organoclays. Thus, Li+-fluorohectorite (FH, CEC = 1.5 meq/g), Li+-saponite (S, CEC = 1.0 meq/g), and Na+-MMT (MMT, CEC = 0.8 meq/g) were intercalated with excess of dimethyl-dioctadecyl ammonium bromide (2M2ODA), trimethyl-octadecyl ammonium bromide (3MODA), or a primary alkylammonium chloride CnH2n+1NH3+Cl-, where n = 6, 9-16, and 18. To prepare the specimens, powders of dry organoclay (25 mg) and polymer (75 mg) were mixed by hand and compressed into a pellet, which was then statically annealed for up to 48 h under vacuum at T > Tg. XRD was employed to detect the equilibrium intercalation. PS (Mw = 30, 90 and 400 kg/mol), P4VP, poly-3-Br-styrene (PS3Br), and polyvinyl cyclohexane (PVCH) were used as the matrix polymers. Melt intercalation at 170 °C in PS30 of FH-ODA, FH-3MODA, MMT-2M2ODA, and S-2MODA resulted in intercalation with gallery height expanded from ca. 1.5 to 2.3 nm; hence Δd001 = 0.8 nm. It is noteworthy that FH-ODA having high CEC value was intercalated whereas its two analogues with lower charge density, MMT-ODA and S-ODA, were not. Replacing the primary octadecyl chain (in FH-ODA) by one in a quaternary onium (in FH-3MODA) resulted in a similar behaviour, but replacing it by two octadecyl chains in quaternary onium (in FH-2M2ODA) hindered diffusion of PS into galleries. Evidently, for melt intercalation the packing density within the galleries as well as the thermodynamic miscibility is important. Pellets of PS30/FHn were annealed at 120, 140 and 160 °C (FHn = FH intercalated with n–CH2 groups). No intercalation was observed for n ≤ 12 where ho ≅ 0.8 nm, whereas for n > 12 (ho ≥ 1.0 nm) the final gallery height was independent of T. However, at 180 °C intercalation takes place for all n ≥ 6 organoclays. A possible explanation for such behaviour is the formation of a pseudocrystalline alkyl phase near the clay surface, preferentially for shorter alkyls at T < 180 °C. The structure may be distorted by too long alkyls or by high temperature. Next, the influence of the matrix polymer was studied. Vaia observed that the MW of PS affected the FH-ODA intercalation kinetics, but not the final expansion of the gallery height. This could not be expected from the derived model, as the entropic contribution, ΔSvpolymer, contains the term dependent on the degree of polymerisation, m1. For PS30, PS90, and PS400 full intercalation at 160 °C was reached after ca. 6, 24 and 48 h, respectively. As shown in Figure 63, there is a strong correlation between the Mw and the time required to reach equilibrium intercalation, teq. Incorporation of MMT-2M2ODA into the styrene-derivative polymers resulted in a logical sequence of increasing gallery heights. Thus, Δh ≡ h – ho = 0, 0.82, 0.96 and 1.0 were obtained for PVCH, PS, PSBr and P4VP, respectively. Such ordering is to be expected when the critically important interactions between the intercalating polymer and the silicate substrate take place. The pendant ring polarisability increases in the order: PVCH > PS3Br, P4VP. The intercalation kinetics depend not only on the Mw, but also on the degree of interaction between the polymer and clay. At low clay loading, the large polarity of PS3Br and/or P4VP may lead to exfoliation. Thus, in accord with the thermodynamics calculations, formation of CPNC depends on the sign and magnitude of the interaction parameters (εjk) and on the initial interlayer structure of the organoclay. For the polymer to interact favourably 270
Thermodynamics
Figure 63 Empirical correlation between the PS matrix molecular weight and the time required for reaching equilibrium intercalation under static melt annealing at 160 °C. Data [Vaia and Giannelis, 1997b].
with the clay, εsp,sa < 0 and the polar contribution to the interfacial energy, ν ijp < 0; and ν ijp > ν ijd . According to Equation 48, the internal energy is zero when: ε sp, sa / ε ap = − h / r2 or when ε sp, sa / ε ap = ξc
(51)
where εjk are the interaction parameters (the subscripts indicate: the silicate, s, the tethered intercalant, a, and the polymer, p), h is the gallery height, r2 is radius of the interacting species 2, ξc is a characteristic parameter of the system. Substituting the corresponding surface energy contributions into this relation a ‘product map’ was constructed. The map (see Figure 64) defines the area where the favourable (negative) values of the energetic contribution are expected.
3.1.4 Computations of Polymeric Brushes Since the early 1990s the computer simulation of end-tethered molecules has been gaining attention. While in the context of this book, the end tethering to a plane is of central interest, the approach is more general – tethering to a point (star molecules), line (graft copolymers) and self-tethering (network formation) find applications in many fields [Grest and Murat, 1995]. The result of end tethering to a plane has often been labelled in the literature as a ‘polymeric brush’, ‘grafted layer’ or ‘hairy clay platelet (HCP)’. Two simulation methods have been used: Monte Carlo (MC) and molecular dynamics (MD). The MC method stochastically generates molecular configurations either on a lattice or off-lattice (in continuum). On the lattice the chain molecules are modelled using self-avoiding random walk (SAW). Adopting the non-reversal random walk reproduces the Rouse dynamics. However, the 271
Clay-Containing Polymeric Nanocomposites
Figure 64 A ‘product map’ showing that favourable energetic interactions (Δev < 0) are expected for polymers with dominating acid or base character. After [Vaia and Giannelis, 1997b], see text.
lattice (e.g., diamond or cubic) model limits the possible molecular configurations. It works well for low segment density, but it is difficult to use when the density is high, as for example in planes with end-tethered linear macromolecules. The simulations show that as the grafting density increases the macromolecules, to avoid overcrowding, stretch in the vertical or z-direction. In the absence of a solvent (‘dry brush’) the incompressibility of the polymer chain is responsible for the stretching. The off-lattice continuum models, MC or MD, perform better at high segment density. Here the bond angle and/or length are allowed to vary, i.e., in the ‘pearlnecklace’ model bond length is fixed while in the ‘bead-and-spring’ model it is not. MD solves equations of motions for each statistical segment, i:
(
)
m d 2 ri / dt 2 = −∇Ui − mΓ ( dri / dt ) + Wi (t )
(52)
where Ui is the total potential for segment i, Γ is the bead friction and Wi is a random force acting on each bead. The interactions are assumed to follow the Lennard-Jones potential within specified radial distances for the repulsion and attraction. The numerical method that is important in describing the configuration of polymeric brushes is based on the self-consistent mean-field approach.
3.1.5 Balazs Self-Consistent Field Approach The self-consistent field (SCF) lattice modelling of polymer adsorption on a solid surface is relatively recent [Fleer et al., 1993]. The 3D lattice facilitates counting of the possible number of conformations, for example by the step-weighed Markov process. This modern approach to thermodynamics heavily depends on 272
Thermodynamics computational techniques. The method is well described by Fleer et al., in the cited reference. The SCF approach, developed for describing polymers at interfaces, is particularly useful for modelling the thermodynamic behaviour of systems comprising nanoparticles. For example, Nowicki [2002] described the conformational properties of long polymer chains in the presence of nanoparticles. The author used a random self-avoiding walk (SAW) model on a 3D lattice. The computed properties included the conformational entropy, segment distributions in the not necessarily Gaussian coils, and dimensions of the coiled chain attached to a particle. Since 1997 Balazs and her colleagues have published several articles exploring the SCF model capabilities for the description of the equilibrium thermodynamic properties in CPNC. The authors modelled the systems by combining Markov chain statistics with a mean field approximation for free energy. The aim was not to obtain quantitative predictions, but rather to show how the system may be designed for the best performance. 3.1.5.1 Numerical Simulation To study the effects of interactions, the authors adopted a model system that resembles the one used by Vaia, i.e., a sandwich of two infinitely large clay platelets immersed in a ‘bath’ of molten polymer [Balazs et al., 1998]. For the computations a 3D lattice was divided into z = 1 to M layers. The lattice was assumed incompressible, with all sites occupied by statistical segments, without voids or small molecules. The intercalant chains were assumed to be linear alkyls, endtethered to the platelet surface. Initially they formed a melt pool inside the sandwich, i.e., between two confining walls, into which the polymer may diffuse under the influence of favourable thermodynamic interactions. The equations for such a model can be solved self-consistently. The self-consistent potential is a function of the polymer segment density distribution and the Huggins-Flory type binary interaction parameters between all components. Properties of the system are averaged over the x and y directions, but they change perpendicular to the clay platelet z-direction (hence ho ≤ z ≤ h). As in the earlier lattice models, here also the excess free energy is expressed by entropic and enthalpic terms, the first one computed from the configurational probability, Gi(z), the other expressing the energetic change caused by two interacting elements: F( z ) =
∑ φ (z) ln G (z) + (1 / 2)∑ χ ∫ η(z − z')φ (z)φ (z')dz' i
i
i
ij
i
j
(53)
ij
where φi is the polymer concentration, χij is the Huggins-Flory type binary interaction parameter between species ‘i’ and ‘j’ and η(z - z′) is the short-range interaction function. The connectivity of segments belonging to the same macromolecule was incorporated by means of Green’s function. Assuming the initial composition of the system, number of segments per species and the magnitude of the binary interaction parameters leads to determination of the self-consistent concentration profile as a function of the gallery height, z. The authors considered the binary interaction parameters between three components: clay platelets (s), polymer (p) and intercalant (a), identified as, χsa = χsp = 0; χap = χ (variable). As the polymer diffuses from the surrounding ‘bath’ into the interlamellar gallery the distance between the two platelets increases 273
Clay-Containing Polymeric Nanocomposites from the initial value ho to h. The free energy change during intercalation is computed as: ΔF = F(h) – F(ho). It is noteworthy that in these computations the interactions between the clay platelet and either intercalant or polymer were assumed neutral, non-perturbing. The only type of interaction of interest was the one between intercalant and polymer. This focused interest contrasts with the opinion expressed by Vaia. Balazs et al., admit that the intercalant-polymer interactions may unfavourably affect the kinetics of polymer diffusion into the galleries. However, there is nothing in the SCF methodology that would preclude systematic studies of the property changes when varying other interaction parameters. For the case of a non-intercalated clay sandwich immersed in a molten polymer (see Figure 65) computations of the free energy per unit area as a function of the interlamellar gallery height, ΔF/A = f(h), show that even for χsp = 0 the free energy is positive, and thus intercalation is unfavourable. The effect is entropic – the macromolecules in contact with the surface have reduced conformational probability. When to start with the clay is intercalated and the macromolecules are able to interact with the intercalant chains the enthalpic contribution may compensate for the confinement effects. The computations for χap = -0.01 to 0.02 show a systematic increase of the free energy, ΔF/A = f(h). For polymer with Np = 100 statistical segments, intercalant with Ni = 25 and the intercalant grafting density ρ = 0.04, the polymeric intercalation was predicted only for χap ≤ 0.005. However, when Np = 300 was assumed, no intercalation was predicted for χap = 0. In other words, the miscibility between intercalant and macromolecules is reduced with increased polymer molecular weight. To compensate for this effect one may want to increase the Ni value from 25 to 50 or 100 – indeed, the computations show that for χap = 0, this strategy leads to exfoliation, for χap = 0.01 it allows for the polymer to intercalate, but for χap = 0.02 it makes the matter worse. These observations based on the free energy changes are supported by the computed segmental density profiles φ = φ(z). While some results of the numerical calculations could be reasoned out from the thermodynamic principles of polymeric systems, the effects of changes on the clay platelets’ grafting density, ρ = 1/s (where s is the surface area per one intercalant molecule), would be difficult to predict. Note that this parameter is related to the intercalant concentration in the system, intercalant structure and to the CEC of the clay (e.g., for MMT with CEC = 1 the surface area per one ionic group is 1.244 nm2). The CEC varies (see Section 2.2) from 0.02 (kaolin) to 1 (MMT) to 2.5 (hectorite). SCF computations for ρ = 0.04 to 0.12 show that for the latter value ΔF is high, thus as the packing density within the gallery increases it becomes harder for the macromolecules to diffuse into it and mix with intercalant chains. In short, on the one hand it is hard for macromolecules to diffuse in between non-intercalated clay platelets, but on the other diffusion is difficult when the intercalation density is high, i.e., there is an optimum in grafting density for forming polymer/clay nanocomposites. According to these computations the optimum value of CEC (or ρ) depends on the interaction between the polymers and the intercalant, χap. In practice, the intercalant ions may be primary, secondary, tertiary or quaternary ammonium or phosphonium; they may have shorter or longer hydrocarbon chains, and (in the case of the most popular quaternary type) they may have one to four long chain groups. In the latter case, the results of SCF computations predict that the intercalant with a single long tail should be better than the one with two or 274
Thermodynamics
Figure 65 Free energy per unit area as a function of surface separation, h, for five different values of the polymer-intercalant interaction parameter, χap. Other parameters are: N = 100; Ni = 25 and χsa = χsp = 0. The grafting density ρ = 0.04 and 0.12 for Figure (a) and Figure (b), respectively. The cartoons (c), left and right show, respectively, the initial and final state, where the surfaces are separated by macromolecules [Balazs et al., 1998]. Reprinted with permission from [Balazs et al., 1998]. Copyright 1998 American Chemical Society.
275
Clay-Containing Polymeric Nanocomposites three. This may explain the behaviour observed for the melt exfoliation of PA-6 [Dennis et al., 2000; 2001]. The authors melt blended PA-6 with 5 wt% of either Cloisite® 15A or Cloisite® 30B. TSE compounding with the former organoclay resulted in intercalation, whereas that with the latter in exfoliation. Both are based on montmorillonite intercalated with quaternary ammonium ions: dimethyl dihydrogenated tallow quaternary ammonium chloride and methyl tallow bis-2-hydroxy ethyl quaternary ammonium chloride (MMT-2M2HTA and MMT-MT2EtOH, respectively). Thus a higher packing density is expected for Cloisite® 15A than for Cloisite® 30B. Evidently, the ability of macromolecules to diffuse into interlamellar galleries of pre-intercalated clay depends not only on the type of radicals (e.g., aliphatic versus aromatic) but also on the intercalant structure, viz. the number of long aliphatic chains attached to onium ion. Vaia [1995] observed that the kinetics of intercalation for FH clay intercalated with 2M2ODA is slower than that with either ODA or 3MODA. The SCF numerical computations were also carried out for CPNC systems with a compatibiliser. Two methods of compatibilisation were explored: (1) by addition of compatibiliser between the intercalant chains and the polymer, and (2) by addition of a compatibiliser, which is miscible with the polymer and able to bind directly to the clay surface. 1. To examine the efficiency of the first approach, the computations were carried out for the system with Na = 25, Np = 100, and χap = 0.01. Different amounts of a polymeric compatibiliser (subscript c; Nc = 100) were added. The interactions of compatibiliser with intercalant (subscript a) and polymer (subscript p) were assumed to be: χac = χpc < χap. The simulation showed, as expected, that addition of a compatibiliser improves the ability of polymer to intercalate. However, the process was found inefficient, requiring at least 10% of the compatibiliser that forms an interphase between the tethered intercalant layer and molten polymer. Evidently, the need for a large amount of compatibiliser would have a negative effect on CPNC cost and performance. Furthermore, enhanced interaction between the intercalant and polymer may hinder the intercalation/exfoliation process not only for kinetics reasons, but also owing to the loss of the conformational entropy. 2. To examine the efficiency of the second approach, computations were carried out for the system where bare clay platelets are dispersed in a mixture of a polymer with its homologue containing reactive group (called a ‘sticker’) at one chain end. Aside from the sticker, the functionalised chains were chemically identical to the matrix polymer. For the computations, the polymer and compatibiliser chain lengths were assumed to be: Np = 100 and Nf = 75, respectively. Strong interactions between the clay and the sticker group were assumed (χsf = -75), while the other interaction parameters were set equal to 0. Figure 66 indicates that this approach is highly promising. There is a substantial effect upon addition of 5% of the functionalised polymer, but little further gain upon addition of up to 70%. The dependence shows that the method leads to exfoliation, creating stable polymer/clay dispersions. Evidently, in practice one may start with pre-intercalated clay, but one having low grafting density, ρ. The key to the second approach is the presence of the ‘sticker’ groups at the chain ends able to strongly interact with the clay surface. In principle, such a functionalised compatibiliser could be either a homopolymer with a strong polar group or an AB
276
Thermodynamics
Figure 66 Free energy per surface area versus gallery height for clay-polymercompatibiliser (functionalised polymer) system. The latter polymer content (0 to 70 wt%) is indicated.
diblock copolymer with, e.g., a short hydrophilic A-block and B-block chemically identical to the matrix polymer. The large organophilic B-block would extend away from the surface, mix with the matrix and cause separation of the clay platelets. For the best performance the molecular weight of the B-block should be higher than the entanglement molecular weight of the polymer; Mn(B) > Me(B). Such a system would be sterically stabilised against re-aggregation. Other numerical calculations (SCF with Markov-chain statistics) focused on the effect of macromolecular architecture on the miscibility of the polymer-clay system [Singh and Balazs, 2000]. Again, two infinite, parallel, organoclay plates were immersed in a molten polymer. The value of the binary Huggins-Flory interaction parameters was assumed χij = 0, intercalant chain length, Na = 25, and its grafting density ρ = 0.04. The miscibility of the polymer/clay mixture was studied by increasing the number of branches for polymers of fixed molecular weight. The analysis showed that starting with linear macromolecules the free energy per unit area ΔF/A < 0, thus the system is miscible. As the number of branches increases the value of ΔF/A became more negative. The plot of ΔF/A versus h indicates that changes in the macromolecular architecture affect the CPNC morphology, viz. for a linear polymer there is a local minimum suggesting preference for the intercalated structures, while for a ten-armed star there is a preference for exfoliation. When the value of χap was assumed to be 0.01, ΔF/A shifted to higher values – only for the ten-armed star there was a local, negative minimum, indicating an intercalated structure. The enhanced miscibility between the organoclay and the polymers with higher number of branches is mainly due to the compactness of the macromolecules that can more easily interact with and interpenetrate the short, grafted layer.
277
Clay-Containing Polymeric Nanocomposites 3.1.5.2 Analytical Self-Consistent-Field Theory for Compatibilised Systems Since numerical analysis indicated that dispersion of clay in a mixture of functionalised polymer and its non-functionalised homologue shows great promise, the authors developed an analytical SCF theory [Balazs et al., 1998]. The adopted model is analogous to that evaluated by numerical means. Thus, the polymeric phase comprises a volume fraction, φ, of monodispersed, functionalised chains and (1 - φ) of polydispersed, non-functionalised chains. The functionalised and non-functionalised chains are chemically identical, differing by the presence of one terminal group in the functionalised chain. However, they may also vary by the number of segments, N, each of diameter a. The end-group is highly attracted to the clay surface, and reacting with it forms a HCP structure, effectively pushing the sheets apart. However, these functional groups do not interact between themselves or with the other components of the system. The bare clay platelets are modelled as planar surfaces, each having an area A. The degree to which the functionalised polymer binds to the surface is related to the gallery height, h. Within the gallery, the functionalised macromolecules (each of length N) are either attached to the surface (na) or free (nf). Similarly, the number of non-functionalised chains of length Pi is ni. Thus, the total number of segments between two clay platelets is given by: Ntotal = N na + n f + ni Pi = Ah / a3 (54)
(
) ∑
and the total free energy of the system is:
(
i
) ∑n μ
ΔF = ΔFbrush − na + n f μ −
i
i
(55) ΔFbrush = ΔFstretch + ΔFmix + ΔEa where μ is the chemical potential of the end-functionalised chain, μi is that of the non-functionalised, polydispersed polymer, ΔFstretch accounts for the stretching of the attached macromolecules, ΔFmix accounts for the free energy of mixing of functionalised and non-functionalised chains, and ΔEa is the energy of attaching the functional group to the surface. By substituting the appropriate terms the following expression for the total free energy of the system was derived: i
(ΔFtotal / RT )( N / Ntotal ) = h ⎡ ⎤ 2 2 + φ ⎢1 − exp − Nk 2 hmax Nk 2 h2 / 3 − hmax exp − Nk 2 x 2 dx / h ⎥ ⎢⎣ ⎥⎦ 0 h ⎡ ⎤ 2 2 2 2 + N (φi / Pi )⎢1 − exp − Pk h exp − Pk x dx / h⎥ i i max ⎢⎣ ⎥⎦ i 0
(
{
)
{
∑
}∫ {
}∫ {
}
(56)
}
where the maximum platelet separation before a homogeneous matrix polymer layer is formed is given by: hmax = ( 2a / π ) 2N (ε / kT + 1 + ln φ ) / 3
(57)
In these relations: k ≡ 3π /8a N , while ε = –ΔEa/na is the energy gain per one reactive group. Note that for h > hmax, a layer of bulk polymer appears between the outer edges of the two brushes. This however does not change the free energy of the 2
278
2
2
2
Thermodynamics system. Thus, at h = hmax, the gain in free energy due to intercalation/exfoliation by the polymeric matrix is at maximum. This limit depends only on the chain length of the functionalised polymer, the reaction energy and concentration – the latter in a logarithmic form, hence sensitive to small changes at low concentration, but insensitive at high. The adsorbed amount of functionalised polymer follows the dependence:
{
2 Θ (h) = h − φ exp − Nk 2 hmax
−N
∑ φ exp{ i
i
2 2 hmax − Pk i
h
}∫ exp{− Nk x }dx 2
2
0 h
}∫ exp{− Pk x }dx 2
(58)
2
i
0
Evidently, the amount of functionalised polymer that is bound to the surface depends on the gallery height, h, with the slope increasing with molecular weight of the functionalised (N) and non-functionalised (Np = Pi) chains. In Figure 67 the function Θ = Θ (h) is presented for various values of the length of non-functionalised chains, Np = 1 to 100. The length of the functionalised chains, Nf = 75 and its volume fraction φ = 0.05 were assumed constant. The interaction energy was set as ε = 12.5 kBT. Increasing Np reduces the miscibility of the functionalised with non-functionalised chains, what causes the functionalised chains to migrate to the strongly interacting clay surface. This in turn leads to increased adsorption of the functionalised chains on the clay surface. Up to ca. h = 20 the dependence Θ = Θ (h) is about linear. It is noteworthy that for Nf = 75 the dependence is common for Np = 50 and Np = 100, suggesting that as a rule of thumb Nf ≈ Np may offer the optimum condition for adsorption.
Figure 67 The amount of adsorbed functionalised polymer with Nf = 75 and concentration φ = 0.05 for four chain lengths of the matrix polymer, Np = 1, 10, 50 and 100. After [Balazs et al., 1998].
279
Clay-Containing Polymeric Nanocomposites It was found that the derived analytical expressions agreed well with the numerical predictions by SCF. For example, the dependencies shown in Figures 66 and 67 could be generated by either of these two models. Similarly, the numerical SCF data on the thickness of the unperturbed brush, hmax, also show satisfactory agreement with the analytical prediction. Recently, SCF was used to determine a range of independent variables that guarantee thermodynamic miscibility in a realistic model system, which comprises four components: solid clay platelets, low molecular weight intercalant, polymeric matrix, and an end-functionalised compatibiliser [Kim et al., 2004]. In the simulation, realistic values of the binary interaction parameters were used. The results showed that intercalation and exfoliation is expected within limited ranges of independent variables. Furthermore, it was found that the presence of bare clay surface (e.g., generated by thermal decomposition of intercalant) strongly hinders the clay dispersion. The 2D simulation successfully identified the most influential factors (e.g., compatibiliser type and concentration) and established their optimum ranges. 3.1.5.3 Phase Behaviour SCF methods have been used to compute the phase behaviour for CPNC model systems [Balazs et al., 1999]. The initial model considered a mixture of individual clay platelets dispersed in polymer melt. The platelets were rigid disks of diameter D and thickness L, while the polymer was made of flexible chains of length N. The volume fraction of the platelets and the polymer in the incompressible mixture was φd, and φp = 1 - φd, respectively. The interaction between the polymer statistical segment and a platelet site was expressed by the Huggins-Flory interaction parameter, χ. Two miscible systems were distinguished: isotropic with clay platelets randomly oriented in respect to each other, and a nematic with mutually aligned platelets. Depending on the CPNC applicability, e.g., for the mechanical or barrier properties, either one of these two structures may be preferred. The formation of the isotropic and nematic structures was predicted using a modified Onsager model for the nematic ordering of rigid rods. The following expression was derived for the free energy:
[
]
F = nd K + ln φd + σ − ln(1 − φd )(b / ν d )( ρ − 1) − χν d φd + n p ln φp
(59)
where K is a constant, np is the number of polymer molecules, nd is the number of platelets per unit volume, and vd = (πD2L)/4 is the volume of one disk. The parameters: σ and ρ describe the orientation of the disks with respect to, respectively, the nematic direction and each other. In the isotropic phase, σ = 0 and ρ = 1, while in the nematic phase:
[(
)
]
σ n = 2 ln −2ν d π −1 / 2 / b ln(1 − φd ) − 1 ρn ≡ ( 4 / π) sin γ = −2(ν d / b) / ln(1 − φd )
(60)
where γ is the angle between two disks. The analysis for disks with an aspect ratio p ≡ D/L = 30 predicts the existence of a phase diagram with three regions: immiscible, isotropic and nematic. The region of miscibility decreases with increasing N and χ. Furthermore; increasing p promotes nematic ordering and reduces the isotropic phase. 280
Thermodynamics This initial investigation was significantly expanded during the following years [Ginzburg and Balazs 1999; 2000; Ginzburg et al., 2000; 2001]. The new model predicts phase diagrams that include: isotropic, nematic, smectic, columnar, plastic (or house-of-cards), and crystalline structures. For these calculations, the authors combined the SCF model with the Somoza-Tarazona formalism of the density functional theory (DFT). The resulting free energy functional was minimised with respect to both the orientational and positional single-particle distribution function, potentially determining all phases and the coexistence regions. The free energy of a system was as a functional of a single-particle r r r written r r r distribution function, γ (r , n ) = ρ(r ) f (n ) , where r and n are the coordinate rand r the nematic director, respectively, while ρ(r ) is disk number density, and f (n ) is the Onsager orientational distribution function. The fluid was assumed incompressible; i.e., the sum of the volume fractions of polymer, φp, of intercalant, φi, and clay, φc, equals 1: thus φp + φi + φc = 1. The free energy consisted of three terms, Fid, Fster and Fint: 1. Fid – the free energy of an ‘ideal gas’ of polymer This term can be written as a sum of the translational and orientational contributions coming from the clay particles, and that originating from the polymer, Fid = Fc + Fp, respectively, the latter calculated from the Huggins-Flory theory: r r r r r r r r Fid / kT = ρ(r ) ln ρ (r )dr + ρ (r ) f (n ) ln 4πf (n ) dndr + (ν / Nν m )φp ln φp (61)
∫
(
∫
)
( )
r where N is the chain length of the polymer, ρ(r ) is the positional (or local) density of clay platelets, n is the total volume of the system, vm is the monomer volume and φp is the volume fraction of the polymer. For the intercalated organoclay platelets the effective thickness is: Leff = L + 2ρNi, and the effective volume of such a particle: Veff = (π/6)D2Leff. 2. Fster – the contribution due to the excluded-volume effects for clay platelets A semi-empirical steric interaction expression was used: r r r (62) Fster / kT = ρ(r ) Vexcl ( f ) / Vphe Ψhs Φ (r ) dr
∫
[
] (
)
where Ψhs is the Carnahan-Starling function that describes the excess freer energy density for hard spheres as a function of their packing fraction; Φ (r ) is the smoothed, local volume fraction of clay; Vexcl(f) is the average excluded volume per particle for a given orientational distribution; and Vphe is the excluded volume per particle for perfectly aligned ellipsoids. The sum of the free energy terms, Fid + Fster, describes an athermal dispersion of hard ellipsoids in a polymer matrix, capable of forming liquid crystalline (nematic) or crystalline phases. To generate smectic or columnar phases, strong anisotropic long-range interactions are required. These are provided by the Fint contribution. 3. Fint – the enthalpic interactions between clay platelets This term was derived from the pair correlation function for the particles, g(1,2); hence it is mostly determined by the excluded-volume effects: r r r r r r r r r r r r Fint / kT = (1 / 2) ρ (r1 ) f (n1 )ρ (r2 ) f (n2 ) × δ (1 − n1 n2 )g(1, 2)V (r1 − r2 )dr1 dr2 dn1 dn2 (63)
∫
281
Clay-Containing Polymeric Nanocomposites where for overlapping or not-overlapping particles the mean-field pair correlation function g(1,2) = 0, or g(1,2) = 1, respectively. The δ-function allows only parallel disk configurations. The potential function V(r1 – r2) is expressed as: 2 r (64) V (r ) = ( π / 4) D2 ⎡1 − (r⊥ / D) ⎤U ( z ) ⎣⎢ ⎦⎥ r where r = (x, y, z), and r⊥ = (x2 + y2)1/2. The interaction potential per unit area, U(z), has two components: U1(z) originating from electrostatic and van der Waals interactions between ‘bare’ clay particles, and U2(z) originating from the intercalant and polymer chain contributions within the interlamellar gallery. The attractive and short-ranged U1(z) term depends on the chemical structure of the clay, and a priori was written as:
[
{(
U1 ( z ) = Eo 1 − exp − z − Leff
)}]
(65)
where Eo is the interaction energy between clay platelets and Leff is an effective thickness of clay platelet. The second component, U2(z), was computed using the SCF method. The shape of this potential depends on N, Ni, ρ and the binary interaction parameters, χ. The equilibrium morphology was computed for the mixtures of a polymer (N = 300) with intercalated clay platelets, modelled as oblate disks having D = 30, L = 1, thus p = 30. Two values for the clay-clay interaction strength were used, E0 = 0 (no long-range attraction), and E0 = 0.1 kBT/a2 (strong attraction). All the binary interactions, except that between polymer and intercalant were assumed = 0; the latter χap = χ varied from –0.05 to +0.05. As far as the intercalation was concerned, the following systems were considered: (a) ρ = 0.2, Ni = 5, (b) ρ = 0.04, Ni = 25, (c) ρ = 0.02, Ni = 50, (d) ρ = 0.04, Ni = 50, and (e) ρ = 0.02, Ni = 100. Note that the total amount of intercalant in systems (a)-(e) was: θ = ρNi = 1, 1, 1, 2, and 2, respectively. The calculations of the phase diagram were carried out in two steps: 1. Using SCF the free energy profile U2(z) was computed for the five systems, varying the binary interaction parameter within the indicated range, first for E0 = 0, then for E0 = 0.1 kBT/a2. 2. The phase diagram was constructed as a map: χ versus φ (see Figure 68). The total volume fraction of the intercalated clay: φc ≤ 0.601. Schematic representation of the phases is shown in Figure 69. The phase diagrams for the considered systems are complex. For χ > 0, the systems are immiscible, showing coexistence between the polymer-rich isotropic phase and the clay-rich crystalline phase. For χ close to 0, the two-phase region becomes narrower and splits into isotropic-nematic and nematic-crystal phases. The triple point (I-N-Cr) and the width of the nematic phase depend on the system. The computed smectic phases were always metastable. For χ < 0, the isotropic-nematic and nematic-crystal coexistence regions become narrower and shift toward higher values of φ. When χ is strongly negative, the new ‘plastic solid’ (or house-ofcards) and columnar phases appear. Evidently, at low clay loadings, the strong repulsion between neighbouring disks forces them to adopt more energetically favourable ‘edge-to-face’ or ‘house-of-cards’ configurations. This leads to either gelation or crystallisation. At higher clay volume fractions, the steric excluded volume effects dominate the long range disk-disk repulsion, forcing the formation of columnar and crystal phases.
282
Thermodynamics
Figure 68 The map of phase behaviour for polymer-clay mixtures, obtained by varying the binary interaction parameter between polymer and intercalant, from χ = -0.05 to +0.05 and the inorganic clay content from φ = 0 to 0.2. The 5 systems are: (a) ρ = 0.2, Ni = 5, (b) ρ = 0.04, Ni = 25, (c) ρ = 0.02, Ni = 50, (d) ρ = 0.04, Ni = 50, and (e) ρ = 0.02, Ni = 100 (see text). The interaction potential between clay platelets, e0 = 0, was assumed. The phases are: I, isotropic; N, nematic; Cr, crystal; PS, plastic solid; Col, columnar. Points represent calculated coexistence densities; lines serve as a guide to the eye. Dashed lines represent approximate locations of phase transition boundaries (exact calculation was not possible) [Ginzburg et al., 2000]. Reprinted with permission from [Ginzburg et al., 2000]. Copyright 2000 American Chemical Society.
283
Clay-Containing Polymeric Nanocomposites
Figure 69 Schematics of possible phase structures for dispersed platelets in a polymer: (a) isotropic, (b) nematic, (c) smectic A, (d) columnar, (e) plastic solid
r
(house-of-cards), and (f) crystal. The nematic director n is aligned in the Z direction, while the platelets lie in the XY plane. Reprinted with permission from [Ginzburg et al., 2000]. Copyright 2000 American Chemical Society.
284
Thermodynamics The phase maps in Figure 68 show similarity for the same total intercalant loading, θ = 1 and θ = 2, viz. systems (a)-(c) and (d)-(e), respectively. However, for platelets intercalated with short densely grafted chains (system a), the immiscibility between the clay disks and polymer dominates the phase behaviour of the system. This observation confirms the previous SCF computations, which suggested that the present strategy of intercalation leads to an unfavourable thermodynamic environment. Instead of high saturation of clay surfaces with C12 to C18 alkyls an intercalation with fewer, strongly bound to the surface, long chain molecules should be carried out. Note that increasing Ni from 25 (system b) to 50 (system d) moves the isotropic-nematic-crystal triple point upward, extends the stability region of the nematic phase to higher χ values and narrows the two-phase isotropic-nematic region, indicating thermodynamically stable exfoliated composites. Comparing systems c and e leads to similar conclusions. One may object to the above computations on the ground that assumption of E0 = 0 for the clay particles is not realistic. It is known that the surface energy of a crystalline solid is high and that there is a strong van der Waals interaction between the clay platelets. Thus, it is interesting to see how the results are affected by the imposition of the strong clay-clay attraction, viz. E0 = 0.1 kBT/a2. For systems with θ = 1, the clay-clay attractions dominate the entropic contribution of the surfactant molecules. The phase diagrams show the immiscibility between polymer and clay within the full range of χ. For systems with θ = 2 (see Figure 70) the miscibility is significantly diminished in comparison to the corresponding E0 = 0 cases (compare with systems (d) and (e) in Figure 68). The isotropic-nematic-crystal triple point is shifted downward, the areas occupied by isotropic and nematic phases is reduced, and the nematic-crystal transition occurs at lower values of φ. 3.1.5.4 Contribution and Potential of the SCF Method The discussed SCF calculations start with a model of a two-platelet (intercalated or not) sandwich in a molten polymer and consider the influence of diverse parameters on the equilibrium thermodynamic properties, viz. miscibility and phase diagrams. The kinetic aspects of the polymer intercalation or exfoliation process are not being addressed. Furthermore, the effects introduced by energy input from external sources have not been considered. The analysis of kinetics is beyond the scope of the model, but the effects of the mechanical energy input (e.g., during compounding in an extruder) are not. Balazs and her colleagues applied the SCF model to study the influences of such parameters as the molecular weight of the polymeric chain, size and grafting density of the intercalant, magnitude of the binary interaction parameter for the polymer, intercalant and clay, influence of the clay-clay interaction, compatibilisation strategies using compatibiliser for the intercalant-polymer or for clay-polymer interfaces. One may argue whether the magnitude of the selected parameters has been optimal, but the numerical comparison of model predictions with experimental data has not been the goal. However, these are the most thorough fundamental studies of CPNC, offering guidance toward better methods of exfoliation in thermodynamically stable systems. One of the important findings of the SCF calculations demonstrated that increasing the length of the intercalant chain (beyond the typically used C12-C20) to oligomeric of polymeric values should enhance the thermodynamic stability of exfoliated systems. Another highly practical finding is that there is little to be 285
Clay-Containing Polymeric Nanocomposites
Figure 70 Phase diagrams of polymer-clay mixtures: (a) ρ = 0.04, Ni = 50, and (b)ρ = 0.02, Ni = 100. The interaction potential between clay platelets, e0 = 0.1, was assumed. The phase diagrams for the other three cases shown in Figure 68 indicate immiscibility. Reprinted with permission from [Ginzburg et al., 2000]. Copyright 2000 American Chemical Society.
gained from compatibilisation of the polymeric matrix with the intercalant chains – a large quantity of the compatibiliser would be required. By contrast, direct compatibilisation of the clay/polymer shows greater potential. For this purpose, the end-terminated macromolecular chains with single strongly bonding group (a ‘sticker’) are preferred. Their chain length should be smaller, but of the same order of magnitude as that of the matrix polymer. Evidently, for the practical reasons (kinetics) one would start with pre-intercalated clay, and use the functional compatibiliser with molecular weight just above the critical value for entanglement. More recently complex phase diagrams were computed. This was accomplished by combining the SCF-generated free energy profiles with the Somoza-Tarazona free energy functional. The computations demonstrated how variations of the principal parameters affect the formation of isotropic, nematic, smectic, columnar, crystal, and plastic solid (house-of-cards) phases. The calculated phase diagrams 286
Thermodynamics are in qualitative agreement with the predictions of the SCF model. As the length of the grafted chains and/or their density increases, the miscibility between the clay sheets and the polymer is improved, and the resulting mixture can exhibit exfoliated (isotropic or nematic) structures for a range of clay volume fractions. For short intercalant chains, the polymer is unable to penetrate the galleries and the system becomes immiscible. The computations demonstrated that the phase equilibrium of CPNC is sensitive to the specific features of clay-clay, clay-polymer, intercalant-polymer and clay-compatibiliser interactions. For quantitative prediction of the phase behaviour, one must correctly describe all these interactions, e.g., by using molecular simulation. There is a large difference between the idealised mathematical model of CPNC and the physical reality. For obvious reasons the ingredients used in the simulation are homogeneous, monodispersed in size and molecular properties. Furthermore, the simulation is based on equilibrium thermodynamics – the forces responsible for the evolution of idealised CPNC structures are exclusively thermodynamic and the final morphology is at thermodynamic equilibrium in the molten state. The reality is quite different as the ingredients are far from homogeneous, there are numerous process additives in each commercial resin, the clays are mineral products with a host of compositional and structural defaults, the system is mechanically dispersed which may either disturb the shape of clay platelets (attrition or bending) and/or induce platelet orientation, then usually cooled down to room temperature which induces vitrification or crystallisation that cause internal stresses, etc. These are some of the reasons that make direct comparison between the model and reality problematic. The model can and does serve as a guide – with future developments the guide will become progressively more realistic, able to account for the numerous contributions mentioned in the preceding sentences. However, considering the complexity of the CPNC systems, molecular modelling is the only method to examine the contributions of each element, the only way to understand the mechanisms involved, and the only logical approach to maximise nanocomposite performance.
3.1.6 Scaling Theory for Telechelic Polymer/Clay Systems Using numerical and analytical SCF calculations Zhulina et al. [1999] investigated the interactions in a system composed of polymer mixture with a sandwich of clay platelets dispersed in it. The mixture consisted of molten polymer with its end-functionalised, telechelic homologue, containing two reactive ‘stickers’, one at each end of the chain. These reactive groups were highly attracted to the platelets’ surfaces. The computed free energy profiles showed distinctive minima, even at low concentration of the end-functionalised chains. The studies showed that the telechelic chains could react with both clay platelets of the model sandwich, thus bridging the interlamellar gallery. The SCF computations showed that telechelic polymers may promote the formation of thermodynamically stable intercalated systems, but bridging the interlamellar gallery could prevent exfoliation. However, it was not clear whether the presence of telechelic chains necessarily prohibits exfoliation. To search for possible conditions under which thermodynamically stable exfoliated CPNC might be formed in the presence of telechelic polymers Kuznetsov and Balazs [2000a; 2000b] applied the scaling theory. The computations again started with a model of two clay platelets immersed in a polymer melt. Each platelet had the surface area A and it was pre-intercalated with short alkyl chains 287
Clay-Containing Polymeric Nanocomposites that reduced the clay-clay interactions to zero. The melt was a mixture of telechelic flexible N-polymers (their volume fraction = φ) and non-functionalised P-polymer (volume fraction = 1-φ). These polymers had P ≥ N >> 1 statistical segments, each of diameter a. The N-polymer had two ‘sticker’ terminal groups, one at each end of the chain. The groups could interact with the clay surface with the energy per sticker-clay contact of ε. Thus, the gallery between the platelets was filled by N- and P-polymers. A fraction of the N-chains was anchored to the platelets with the rest (as well as P-chains) being free (subscript f). The anchored chains could be attached to the surface by one (subscript t for ‘tails’) or two ends (subscripts l or b for ‘loops’ or ‘bridges’, respectively). The following values of the Huggins-Flory interaction parameters were assumed: χNN = χPP = 0, and χ = χNP. The free energy of the chains within the gallery was expressed as a sum of the individual contributions: F = Fads + Fint + Fcomp + Fel + Fent + Fdem
(66)
where Fads is the sticker-surface adsorption energy for N-chains, Fint is the interaction energy between the N and P chains, Fcomp is the compression energy for chains at relatively low interlamellar gallery height, h, and Fel is the elongation energy for loops, tails, and bridges, Fent is the entropic energy for mixing the different types of chain configurations within the brush layers, and Fdem is the demixing energy associated with extracting N and P chains from the surrounding melt and localising these chains in the confined layers. The free energy of the system, F, was written deriving the six components of Equation 66 and casting them in terms of the reduced gallery height: a = h/Na, αp= h/Pa, αtot = htot/Na and αl = hl /Na, where htot stands for the total height of the brush, htot ≤ h/2 (with h being the gallery height), and hl is the average distance that a loop molecule reaches from the clay surface. Its value depends on the energy per contact between the clay platelet and a functionalised chain end (ε/T), the volume fraction of functionalised chains in the bulk melt, N and P, the numbers of segments in each functionalised and non-functionalised chain, respectively, and the Huggins-Flory interaction energy parameter, χNP. Next, F was numerically differentiated to obtain the conditions for local minima in respect to individual volume fractions: φP, φl, φt, φb, and φf, calculated from the respective number of statistical segments:
φl, t, b, f = nl, t, b, f Na3 / Ah ; φ p = n p Pa3 / Ah ;
∑φ = 1 i
(67)
i
The free energy, F, was computed in its reduced form (in respect to temperature and surface area) as a function of the reduced gallery height:
(
)
Fa2 / AT = α Na3 / Ah ( F / T ) versus α ≡ h / Na
(68)
The initial computations were carried out for the dispersion of intercalated clay in a telechelic, bi-functionalised melt (φ = 1). The results are presented in Figure 71 as the reduced free energy and volume fractions of different conformations versus the reduced gallery height, α. Four regions are indicated: (I) The chains t, l, and b are highly compressed. The reduced free energy rapidly decreases with α to negative values (miscibility). (II) A shallow, local minimum of free energy, which originates in weak stretching of all the chains within the brushes. The number of bridges decreases and tends to zero on the border of this region. 288
Thermodynamics
Figure 71 The reduced free energy for different conformations versus the reduced gallery height, α. Four regions of behaviour are indicated. After [Kuznetsov and Balazs, 2000a]. See text.
(III) This is the ‘relaxation’ region, where the free chains appear. The free energy is further reduced with α. (IV)Separate brush layers are formed with unperturbed Gaussian free chains filling the space between them, thus exfoliation. The minimum in region (II) is thermodynamically metastable hence there is no optimal spacing (intercalation) other than exfoliation. This is surprising as it could be expected that the bridging chains give rise to a thermodynamic barrier to separating clay platelets beyond the point where the bridges are disrupted. Evidently, this energy cost is compensated by that gained by adsorption of free chains, which can penetrate the gap at larger α. The results imply that clay sheets can be exfoliated in a melt of telechelic chains. However, the metastable minimum in region (II) can lead to a kinetically trapped degree of dispersion. Additional computations show that the magnitude of the local minimum can be controlled by the sticker adsorption energy, ε/T, and the molecular weight of the N-chain. Evidently, increasing the former and decreasing the latter forces the reduced free energy deeper into the negative values, increasing the miscibility. However, for practical reasons this may not be the best solution. For sufficiently small values of the barrier height, an imposed stress or increased temperature can help overcoming the kinetic barrier and lead to exfoliated CPNC. In a system comprising intercalated clay dispersed in a mixture of functionalised (N) and non-functionalised (P) polymers the plots of the free energy per area (F/T)(a2/A) versus the reduced gallery height, α, are similar to that shown in Figure 71. However, at low volume fraction of N-chains, φ < 0.05, and high e/T-values, the minimum in region (II) is a global one, forcing the system into a 289
Clay-Containing Polymeric Nanocomposites thermodynamically stable intercalation. One solution is to use a higher concentration of the N-chains, another to increase interaction between N- and P-chains, i.e., to have χNP < 0. The phase diagram in Figure 72 summarises these findings. The data are plotted in terms of ε/T versus the volume fraction of polymer-N, φ, for the case of N = 1000 and P = 100. The curves (calculated by equating the free energies) between the intercalated, exfoliated and immiscible phases correspond to the first-order phase transitions. At low contact energies between the sticker group and clay surface, the system is immiscible, while at higher values of ε/T, it can either be intercalated or exfoliated (at high φ). In the intercalated state, loops, tails and bridges fill the gallery. In the exfoliated state loops and tails form the brushes. The transition from an intercalated to an exfoliated system is related to the increased number of tails, which in turn is facilitated by a higher value of φ and the formation of a free polymer layer between the brushes. The relative magnitude of the intercalatedto-exfoliated region very much depends on the N/P ratio. Figure 72 illustrates a perverted case with the functionalised chains being ten times longer than the matrix polymer. In practice, N ≤ P and under these conditions the intercalated area is quite small, relegated to ε/T < 5 and φ > 0.15 – the miscible clay/polymer systems are mainly exfoliated. To summarise, the scaling theory was shown to be appropriate for the analysis of free energy in three-component systems composed of pre-intercalated clay, bifunctionalised (telechelic) compatibiliser and its non-functionalised homologue within a wide range of parameters, viz. ε, N, P, φ and χ. The computations indicate that in the melt of only N-chains there are no thermodynamically stable intercalated states – the system is either immiscible or it exfoliates. It is easier to
Figure 72 Phase diagram for the polymer-clay mixture containing φ volume fraction of functionalised polymer (N = 1000) and non-functionalised polymer (P = 100). After [Kuznetsov and Balazs, 2000a]. See text.
290
Thermodynamics expand the gallery height in a mixture of telechelic and non-functionalised polymers. The scaling-theory shows a significant influence of the chain length ratio N/P on the location of the phase boundaries between the immiscible, intercalated and exfoliated states of the polymer-clay mixtures. Increasing N at fixed P with N ≤ P, causes the immiscible phase to transform into an intercalated or exfoliated one. At the same time, the intercalated-exfoliated transitional volume fractions weakly depend on N. When changing the value of P at constant value of N with N ≥ P, the intercalated phase expands at a cost of the exfoliated one. Theoretical analysis of the clay/polymer/functional polymer indicates that the most promising strategy for exfoliation is by using long-chain functionalised compatibiliser of smaller but comparable in magnitude molecular weight to the matrix polymer. Promising results were obtained for functionalised linear polymers having either one reactive chain-end, or two reactive chain-ends. Experimental confirmation of this conclusion can be found, for example, in the work by Hoffmann et al. [2000a]. Since multi-block copolymers form morphologies that may be controlled by the composition and molecular weight of the blocks, there is a concerted effort to explore the potential structure formation of nanocomposites with block copolymer as a matrix. Thus, for example, SCF was used to analyse the structure of nanocomposite thin films (of di-block copolymer with nanoparticles) confined between two walls [Lee et al., 2003]. In such restricted geometry, particles are driven to the wall, effectively modifying their chemical nature and the film structure. The changes are robust, relying on the entropic effects. They can be exploited to form nanoscale devices with particles assembled into nanowires. Another suggested application of the phenomenon is the use of such systems for coatings – the high concentration of solid particles near the surface makes such a coating more fracture and abrasion resistant.
3.1.7 Solid Surface Effects on Molecular Mobility The effects of a solid phase on organic chain mobility are important for understanding nanoparticle behaviour in a polymeric matrix. Crystalline solids show high surface energy and a tendency for aggregation. The key to the success of CPNC is the stable and uniform dispersion of exfoliated clay platelets or other nanoparticles (e.g., carbon nanotubes). 3.1.7.1 Surface Energy of Solids The surface energy depends on the nature and structure of materials. Its value is defined in terms of the thermodynamic equilibrium of the solid with its saturated vapour. Evidently, measurements of the surface energy of a liquid are significantly more straightforward than those of the solids. The surface energy, σs, is defined as:
σ s = U s − TS s = ν1 s +
∑μ n
s i i
(69)
i
where s is the surface area, the superscript ‘s’ indicates the surface quantities of total energy (U), entropy (S) and the number of moles (n) of species ‘i’ with chemical potential, μ i , adsorbed on the surface. Thus, the equilibrium thermodynamics predicts that the total surface energy may change due to temperature and composition. The variable, ν1, is the surface tension (subscript 1 indicates a single component in equilibrium with its vapours). In the absence of 291
Clay-Containing Polymeric Nanocomposites adsorbed foreign substance (ni = 0) the surface tension coefficient, ν1, is given by the ratio of surface energy to surface area; hence it may also be called the specific surface energy. The specific surface energy of solids, ν1, is difficult to measure as these are often subjected to 3D residual stresses, surface roughness all the way to nanoscale, readily adsorb environmental gas, vapour or liquid molecules and may be contaminated by compounds that are difficult to remove. The highest measured surface energy is for diamond: ν1(C-diamond) = 9820 mN/m. A high value was also determined for orthoclase, KAlSi3O8, viz. ν1 = 7770 mN/m [Brace and Walsh, 1962]. The specific surface energy of a metal in equilibrium with its own vapour is also high, viz. ν1(Cu) = 1,430, ν1(Au) = 1,510 mN/m, similarly for crystalline oxides, viz. ν1(CaO) = 1,310, ν1(MgO) = 1,090 mN/m. By contrast, amorphous bodies have significantly lower ν1 than crystalline. For example SiO 2 : ν1(amorphous) = 260, ν1(hydrated, amorphous) = 130, while ν1(quartz) = 1030 and ν1(hydrated quartz) = 422 mN/m [Condon and Odishaw, 1967; Stokes and Evans, 1997]. These values for crystalline solids should be compared with those of organic compounds. For example, the polymer surface tension coefficients at 20 °C range from 10 to 49 mN/m (for fluorinated acrylic to polyesters or polyamides, respectively) [Brandrup et al., 1999]. Similar values have been measured for organic liquids – the highest is ν1(glycerol) = 63 mN/m, with –OH groups on the surface. In short, the surface energy of a crystalline solid is about two orders of magnitude greater than that of organic liquids. Owing to morphological nanorugosity the surface of crystalline solids is usually highly complex. This is particularly true for mineral silicates where the surface may be defined only in statistical terms. The clays, due to variation of composition, show great variation of the surface heterogeneity [Papirer and Barard, 1998]. The latter also varies with the type of adsorbed compound. Evidently, adsorption affects the surface energy of crystalline solids. For example, the specific surface energy of freshly cleaved mica in vacuum is ν1 = 4,500 mN/m whereas that measured in air is 375 mN/m. One of the consequences of the surface energy contribution to the energetics of a body is the dependence of the transition temperature on the degree of dispersion. For example, the melting temperature of bulk PE is Tm ≅ 400 K, but PE-particles with diameter varying from 13 to 5 nm have Tm = 266 to 218 K. Another consequence of the high surface energy is the aggregation of solid powders. According to Johnson et al. [1971], the force between two interacting spheres (of radii R1 and R2) is given by:
ε = ν11πR1R2/(R1 + R2)
(70)
For R1 = R2 the relation simplifies to read ε = ν11πR/2. Equation 70 is general and independent of the elastic modulus. The attraction force is then proportional to the radius – the smaller the particle, the smaller is the force of individual contact. However, dividing the same mass of crystalline powder into the number of particles requires that NiRi3 = NjRj3, thus a change of the total aggregation energy, E = Nε, i.e.: Niεi/Njεj ≡ Ei/Ej = (Rj/Ri)2
(71)
For example, to disperse the same volume of CaCO3 particles of about 10 nm diameter would require 62,500-fold more energy than that needed for a standard filler powder 292
Thermodynamics with diameter of about 2.5 μm. Furthermore, since the dispersing force during compounding is given by the product of radius and stress, Rσij, to achieve the same efficiency the stress should be increased by a factor of 200. 3.1.7.2 Polymer Adsorption on Solid Particles As discussed by Fleer et al. [1993], any interface between two phases (solid, liquid or gas) induces changes to properties. For example, solid readily adsorbs macromolecules from a solution. The thickness of the adsorbed layer usually extends to the magnitude of the radius of gyration of the macromolecular coil, sθ2
1 /2
= 5 to 35 nm. The adsorption depends on the polymer (its chemistry,
conformation, molecular weight, polydispersity, etc.), on the solvent, temperature and the substrate surface energy. The model computations indicate that in the absence of direct chemical bonding, macromolecules interact with a solid surface by physical contact of several statistical segments labelled as ‘trains’ (see Figure 73). The trains create loops and tails. The relative proportion of these structures depends on the molecular weight, e.g., loops are absent for short chains, but dominant for long ones. Recently van der Gucht et al. [2002] proposed a lattice model for describing the behaviour of an ideal polymer solution at a surface. The ratio of the surface excess volume fraction, φex, of polymer molecules with N-statistical segments, to the bulk volume fraction of polymer, φb, was found to follow a simple dependence:
[ (
φ ex / φ b = A B + χ s − χ sc∞
)]
(72)
where A ≅ 5/6 and B ≅ 1/5 are constants, χs is the Silberberg adsorption parameter, and χ sc∞ is the critical adsorption energy for infinite chain lengths. With decreasing chain length the adsorption/depletion transition shifts to lower χs values. This effect is strongly enhanced if the end segments of the chain adsorb preferentially. Cosgrove et al. [1987; 1991] used neutron scattering to study polymer adsorption from dilute solutions on solids (mica or PS latex particles). The results presented in Figure 74 indicated that homopolymer concentration exponentially declines from the surface in the z-direction, while the copolymer concentration profile follows a parabolic dependence. The thickness of the adsorbed layer varied from one system to the next, but within a relatively narrow range of values. The root-mean-square (rms) thickness was slightly larger than the unperturbed radius of gyration, viz. 3.6 versus 4.8 nm for copolymer and 6.8 versus 9.0 nm for PS, respectively. After evaporation of solvent the thickness of the adsorbed layer was reduced by about 50% (viz. to 3.3 and 4.4 nm, respectively) indicating high
Figure 73 Schematic representation of adsorbed macromolecule. Loop, train and tail are shown.
293
Clay-Containing Polymeric Nanocomposites packing density in the original layer adsorbed from dilute solution. The rms thickness of the adsorbed layer is one measure of the phenomenon, another being the distance in the z-direction, zmax, over which the polymer concentration is measurably higher than the bulk concentration, φ > φb. For the systems studied by Cosgrove et al., zmax = 24 nm was reported. As shown in Figure 74, all other parameters being the same, the magnitude of zmax very much depends on the polymer molecular weight. Israelachvili et al. [1984] reported that when two saturated layers of PS on mica approach each other, starting at a characteristic distance, zc, they attract each other. The experiment was conducted in a cyclohexane solution of PS, near the Θ-condition using the surface force analyser (SFA; see Figure 75(a)) designed and originally built by Israelachvili in 1978. The distance zc = 60 to 120 nm depended on the PS molecular weight and the thermodynamic miscibility. The results were well reproduced in two different laboratories. 3.1.7.3 Nanoscale Rheology In the layer adjacent to the clay surface the adsorbed molecules have a tendency to be strongly adsorbed [Israelachvili, 1985; Horn and Israelachvili, 1998]. This solid-like behaviour of the first z = 2-9 nm thick surface layer is followed by another one, in which the viscosity progressively decreases in the z-direction, finally to reach the level of the bulk solution viscosity at z ≅ 120 nm. The force measurements for PDMS squeezed between two mica plates showed a solid-like behaviour at a distance of about 4 nm, and exponentially decreasing viscosity toward the bulk value at a distance of about 17 nm. For the measurements the Israelachvili’s surface force balance was modified to impose either a steady-state or dynamic shearing (see Figure 75).
Figure 74 Concentration profiles of adsorbed PEG on deuterated PS latex particles. Data [Cosgrove et al., 1987].
294
Thermodynamics
Figure 75(a) The surface force analyser (SFA) (a) Reprinted from Intermolecular and Surface Forces, J. Israelachvili. Copyright 1992, with permission from Elsevier.
Figure 75(b) The SFA was used to measure the viscosity of adsorbed PS-X from a toluene solution. The results indicate a solid-like behaviour for the first adsorbed layer: 2z < 220 ± 10 nm [Klein et al., 1993]. See text. (b) Reprinted with permission from [Klein et al., 1993]. Copyright 1993 American Chemical Society.
295
Clay-Containing Polymeric Nanocomposites Confirming the earlier observations, Klein et al. [1993] reported that nonfunctionalised anionic PS poorly adsorbs from dilute toluene solution onto a mica surface. The measurements gave the correct magnitude of the solution viscosity within a few nanometres from the surface. The situation was dramatically different when the PS chains were terminated with the zwitterionic group, -(CH3)2N+-(CH2)3SO3–, giving ammonium-terminated PS-X. The molecular weight of PS-X was Mw = 375 kg/mol, Mw/Mn = 1.03 and the radius of gyration in toluene, Rg = 57.5 nm. Figure 75(b) also shows variation of the ‘effective mobility’ parameter, G, in the z-direction, thus orthogonal to the mica surface. The ‘effective mobility’ parameter was defined as: G ≡ 6πR2 ω / K
( Ao / A)
2
− 1 = 2( z − z H ) / ηo
(73)
where R is the mean radius of curvature of the mica sheets, K is the spring constant, ω is the frequency, A and Ao are amplitudes of oscillation, z is the distance from the mica surface and zH is its value (known as the hydrodynamic thickness) at which the extrapolated value G = 0, and ηo is the zero-shear viscosity of the bulk liquid. With the low grafting density (mean spacing between attached chains of s = 14.5 ± 1.5 nm) the hydrodynamic thickness nearly doubles the radius of gyration. Furthermore, within the first layer from the mica surface (110 nm or so thick) the segmental mobility of PS macromolecules is practically zero, indicating a solid-like behaviour of the adsorbed chains. More recently, the surface force apparatus was redesigned and the nanorheological measurements were carried out on undiluted polybutadiene (PBD; Mw = 6.95 kg/mol; the entanglement molecular weight, Me = 1.85) [Luengo et al., 1997]. The initial measurements of the normal forces at very slow approach (2-3 min per point) indicated the presence of two regions: (1) a steeply repulsive force with an incompressible ‘hard wall’ at zhw ≅ 5 nm of solid-like PBD adsorbed on the mica surface, and a roughly exponential decay at larger distances. The exponentially decaying region had a decay length of about 3.5 nm, which is close to the unperturbed radius of gyration, Rg = 3 nm. These repulsive forces indicate strong binding of polymer to the surface, at least over the time scales of the force runs (1-2 h), and an immobilised layer of thickness of about 1.5 Rg per surface. The chain conformations in such confined surface layers may be in the glassy or rubbery state. Next, the dynamic oscillatory measurements were carried out at strains γ < 30%, frequency, ν (Hz) = 0.03 to 80 and with the separation distance, 2z = 10 to 250 nm. As the distance between two mica sheets decreased, three regions of the dynamic behaviour could be distinguished: (1) bulk, (2) intermediate, and (3) tribological. Figure 76 and Figure 77 present results of these measurements. (1) Bulk: For 2z > 200 nm the phase angle (between strain and stress signals) δ ≈ π/2, and the measured storage and loss shear moduli (G´ and G´´, respectively) followed the bulk behaviour. (2) Intermediate: At smaller z-gaps, δ < π/2 while the values of the shear moduli were higher, showing an ‘elastomeric-like’ plateau indicating a 3D structure. (3) Tribological: At the smallest gaps, 2z ≤ 12 nm, δ = 0, or a Hookean-type response of an elastic body was obtained. On the flow curve (log viscosity versus log deformation rate) the data followed a straight line with the slope = –1, i.e., η ∝ 1 / γ˙ ; or : G ′ ∝ ν 2 and G ′′ ∝ ν where ν is the radial frequency. 296
Thermodynamics
Figure 76 Effect of surface separation (distance D = 2z) on the storage (G´) and loss (G´´) moduli of PBD at T = 25 °C and frequency n = 13 Hz. Data [Luengo et al., 1997].
Figure 77 Effect of frequency on the shear moduli for PBD at T = 25 °C, D = 2z = 10, 30, 180 and 250 nm [Luengo et al., 1997] Copyright 1997 American Chemical Society.
297
Clay-Containing Polymeric Nanocomposites The authors also reported the presence of normal stresses during the steadystate shearing. Caused by them, initially the gap between mica sheets would increase toward an equilibrium separation, to return to the original separation in about one minute after stopping. The effect depended on the initial separation and the shear rate. Since a similar behaviour has been observed for ‘brush’ layers of zwitterion-terminated PS end-grafted to mica in toluene, the inference is that PBD coils are effectively bound to the surfaces not by ionic but van der Waals forces. It should be stressed that such behaviour is not limited to macromolecules. Low molecular weight organic solvents having either spherical or short-chain molecules (e.g., alkanes, octamethyl-cyclotetrasiloxane, oligomers, etc.) behaved similarly under confinement [Gee and Israelachvili, 1990; Granick, 1991; Hu and Granick, 1992]. For example, when dodecane was confined to a diminishing gap from ca. 5 to 2.8 nm it underwent an abrupt transition (similar to crystallisation) to a solid-like state. Virtually all confined liquids solidified on the mica surface. Since the process is rate-dependent (time scale ranges from minutes to hours), the thickness of the solidified layer varied from two molecular layers up. In some cases, with increasing confinement some liquids showed a smooth increase of viscosity by up to 7 orders of magnitude. For either of these two types of low molecular weight liquids the layer thickness within which the viscosity was above that in the bulk was quite large. 3.1.7.4 Molecular Modelling of Nanoconfined Molecules (Intercalation) The behaviour of molecules in the direct proximity of a smooth solid surface has been studied for a number of years. The research has been motivated not only by intellectual curiosity, but also by practical concerns in geology, biology, pharmacy, etc. Surface force measurements, surface force microscopy, surface tunnelling microscopy, X-ray or neutron scattering and other modern methods have replaced the classical measurements of the adsorption isotherms. However, the biggest progress in understanding the solid/liquid interaction has been gained from computer simulation techniques, using Monte Carlo (MC) or molecular dynamic (MD) procedures. This work is of particular interest to the CPNC technology as in these materials the specific surface area, A(MMT) ≅ 800 m2/g, has a strong multiplying effect, even at low clay loading. As it was discussed in the preceding parts, molecules (short chains as well as macromolecules) strongly adsorb on the crystalline surface either from solution or from melt. While the solid-like behaviour extends only to 2 to 9 nm, the reduced mobility may extend to over 100 nm. Considering that the thickness of a MMT platelet is 0.96 nm, even the minimum level of immobilisation has a large multiplying effect on the reinforcing solid content. Two excellent overviews on the molecular modelling of adsorption and ordering at solid interfaces are available [Hentschke, 1997; Binder et al., 2004]. The former author discusses both aspects of adsorption, chemosorption involving direct chemical bonding to substrate and physicsorption, involving repulsive, dispersive, multipole and induced interactions. Several examples of good agreement between the computational and experimental data are shown. By contrast the approach of Binder et al. is more pedagogical – a tutorial introduction to the techniques and the basic algorithms of MD. As an example, the authors computed self-diffusion constants, viscosity, and thermal conductivity of molten SiO2, as well as vitrification – a non-equilibrium behaviour. 298
Thermodynamics Computations using MC or MD methods, as well as various models have shown that macromolecules near solid surfaces are greatly affected – their molecular arrangements, conformations and dynamics are strongly perturbed with respect to the isotropic bulk [Brown et al., 2003]. Thus, the surface layers in contact with a solid are densely packed with chain segments. The ordered layers extend into the liquid phase for two to three times the transverse diameter of the macromolecular chain. The surface also distorts the Gaussian distribution of chain segments to a length of about the radius of gyration of the polymer. In consequence, the chains whose centre of mass is close to the surface are flattened with a substantial fraction of segments forming the first densely packed layers. Furthermore, the diffusivity is non-isotropic – hindered in the vertical z-direction and enhanced in the planar xy-direction. MC simulations are mostly carried out on a lattice. While the model greatly reduces the computing time, it is not totally realistic. The simulations have been limited to systems with polymeric chains represented as random, self-avoiding walks on simple lattices. The computations have been used for clarification of specific aspects, for indicating a solution to specific problems. However, the lattice models are not suitable for studies of packing and ordering. For this purpose out-of-lattice models are more realistic [Binder, 1995]. MD simulations were used to study the static and dynamic properties of 2:1 layered silicates (CEC = 0.8, 1.0 and 1.5 meq/g for two MMT clays and a hectorite, respectively) intercalated with alkyl-ammonium ion, having 6 to 18 -CH2- groups [Hackett et al., 1998]. The results indicated that within the gallery, the alkyl chains show strong layering tendencies. Shorter alkyl chains form stable ‘monolayers’ with d001 = 1.32 nm. The space between the clay surface oxygen atoms is about equal to the diameter of a single -CH2- group. As the alkyl chain length increases, a bilayer (d001 = 1.8 nm) and trilayer (d001 = 2.27 nm) disordered, intertwined structure was predicted. For most bilayer configurations, nearly half of the -CH2- groups are in the layer opposite the grafted ammonium group of that chain. In a trilayer configuration, -CH2- groups tend to jump to the middle layer, but few to the layer opposite to the grafted chain end. The trans-gauche conformer ratios as well as the trans-gauche transition rates were computed. The data were found to agree with FTIR observations by Vaia et al. [1994]. Thus, within the interlamellar galleries of alkyl ammonium intercalated clay the alkyls show dual structure – crystal-like for the surface-bonded trans-trans part and disordered, liquid-like away from the surface. According to a terse announcement, Amcol International developed a computer model that predicts the type of organoclay structure that will form with a given intercalant [Beall, 1999]. The model is reported to quantitatively predict the d001-spacings. It may also be used to compute the energetics of exfoliation by a polymer to form a CPNC with ionic or ion-dipole bonding. As an example, the predicted versus experimental d001 spacing of MMT was given. The values reproduced in Table 41 have a standard deviation of ± 0.06 nm. Molecular modelling has also been used to examine the validity of several theoretical assumptions, confronting them with directly performed numerical ‘experiments’. In 2000 Sumpter et al. [2000] used MD to simulate polymer nanoparticles comprising up to 300,000 atoms, with a variety of chain lengths. The results showed that the ratio of surface atoms to the total number of atoms for the nanoparticles is very large and the surface effects are responsible for the properties of CPNC being so different from those of the bulk polymer. 299
Clay-Containing Polymeric Nanocomposites
Table 41 Comparison between MD-computed and XRD-measured interlayer spacing, d001 ± 0.06 nm, for MMT-intercalant systems [Beall, 1999] Computed d001 (nm)
Measured d001 (nm)
Δd001 (nm)
(None)
0.91
0.96
0.06
Water
1.22
1.26
0.04
Caprolactam
1.61
1.68
0.07
Butyrolactam
1.82
1.90
0.08
2-Pyrrolidone
1.90
1.90
0.00
N-Ethyl hydroxy-pyrrolidone
2.00
2.00
0.00
Dodecyl pyrrolidone
4.50
4.40
-0.10
Monostearyl
1.73
1.79
0.06
Glycerol
3.84
3.94
0.10
Intercalant
The dynamics of the organic phase were studied using NMR surface-sensitive cross polarisation and bulk-sensitive spin-echo experiments. Synthetic, Fe3+-free fluorohectorite (FH) was intercalated with ODA, and then melt-annealed with PS under static conditions [Zax et al., 2000]. Within the gallery the polymer constituted 68 wt% (ODA the rest). The molecular dynamics simulation suggested the presence of a symmetrical five-layer structure of the organic phase. Thus, near the surface there was a preponderance of phenyl groups, next the backbone carbons intermixed with the alkyl groups. The highest probability of finding the intercalant carbon was within the central layer. The NMR results showed that the chain segments (both phenyl and -CH2- groups) that interacted with the surface were dynamically inhibited. By contrast, within the central layers the segments were more mobile than in the bulk at comparable temperature. The transition from the glassy to molten state took place over a wide temperature range – starting at lower T and ending higher than for the bulk PS. Furthermore, by contrast with the bulk behaviour, the PS inside the interlamellar gallery space did not show the customary, isotropic molten phase behaviour. Anastasiadis et al. [2000] used dielectric spectroscopy to study the segmental dynamics of 1.5 to 2.0 nm thick polymer film confined between clay platelets in a nanocomposite. For the study hectorite and MMT were intercalated with dimethyl-dioctadecyl-ammonium bromide (2M2ODA). Mixing the dry organosilicate with up to 30 wt% of poly(methyl-phenyl siloxane) (PMPS, Mw = 2.6 kg/mol, Mw/Mn = 1.2) resulted in formation of h = 1.5 to 2 nm thick polymeric films inbetween the organically modified silicates. The PMPS segmental motion is dielectrically active. Three processes were detected: (1) slow, resembling the one for bulk-PMPS; (2) intermediate, possibly related to orientational motions of the intercalant; and (3) fast due to the confined PMPS. 300
Thermodynamics Measuring the dielectric spectra of PMPS with 2M2ODA eliminated the possibility that the fast relaxation was due to the presence of intercalant. The other possible mechanisms include the restriction placed by the interlayer spacing on the cooperative volume. The effective confinement of the polymer chain was of the order of a few statistical segment lengths of PMPS. The effect of confinement leads to the local reorientation of macromolecules which in turn changes the local dynamics, evidenced by the presence of a new faster mode than that in the bulk polymer with weaker temperature dependence. Simulations showed that chains adopt a preferentially parallel configuration near a wall with oscillations in the monomer density profile. These lead to a dynamic anisotropy with enhanced parallel and reduced perpendicular monomeric mobility extending over distances. Under severe confinement this interphase is anticipated to extend over the whole film thus leading to fast relaxation. Interesting MD studies were carried out by Yu et al. [2000b]. For over 60 years methylene blue (MB) has been used to measure clay surface area and CEC. The authors computed adsorption of MB on a model beidellite, MMT and muscovite mica, in the presence or absence of water. A variety of configurations, including single and double layers, parallel or inclined to the basal surface were found in agreement with the XRD results. The MD computations indicated that this traditional method of clay characterisation has hidden dangers. As far as the CPNC technology is concerned these MD simulations demonstrated the possibilities of formation of diverse structures, dependent on the type of clay, intercalant loading, and possible contamination by other substances (e.g., moisture). Vacatello [2001a] carried out MC simulations for dense polymer melts with solid, spherical nanoparticles. The model incorporated off-lattice approximation and conformational distribution of the simulated chains, similar to that of real polymers. The results showed that at the interface the polymer segments are densely packed in the form of ordered shells around the nanoparticles, analogous to the layer formation near a planar solid surface. The thickness of the shells was approximately twice the transverse diameter of the polymer isodiametrical (to occupy a single lattice site) segments. The size of these (adopted after Flory et al. [1984]) was taken to be equal to about 3.5 -CH2- units hence having a diameter of ca. 0.45 nm. The nanoparticles behaved as highly functional physical crosslinks, reducing mobility of the polymer chains. Since the effect was related to specific surface area, it increased with 1/R2, where R is the radius of the particle. Furthermore, the conformational distribution of the polymer was perturbed by the presence of these solid nanoparticles. Thus, for example, the average dimension of chain segments was reduced, the polymer chains were either totally contained within the interface shell of a single particle or they formed bridges connecting different particles. Some macromolecules visited several nanoparticle shells, and each particle was in contact with many different polymer chains. These calculations simulate the structure and dynamics of a polymer/solid surface system well. The results, determined by topology and entropy, are applicable to diverse situations, including clay intercalation. The presence of preferential interactions between polymer and solid should not significantly change the computed structure, the order of the interface shells and the conformations of the polymer chains. Obviously, since so far computations have been carried out for binary clay/polymer systems, the presence of a third component will affect the structure. This would be expected for clays modified with quaternary onium 301
Clay-Containing Polymeric Nanocomposites ions having more than one type of substituent groups. However, the overall image that emerges from these models is expected to remain intact: reduced segmental mobility near the high-energy clay surface, increasing towards the characteristic bulk melt mobility at a distance of ca. 100 nm. Manias and Kuppa [2002] provided a short topical review, focusing on the simulation of structure and dynamics of PS macromolecules in slits. The MD results were compared to experimental results for CPNC containing onium-modified fluorohectorite dispersed in a PS matrix. Reasonable agreement was found.
3.1.8 Kinetics of Polymer Intercalation The term ‘kinetics of polymer intercalation’ has several meanings. To theoreticians considering a model sandwich in molten polymer ‘bath’ it describes how fast the two platelets will move apart under the influence of the matrix polymer diffusing into the interlamellar gallery. To an academic researcher investigating diverse aspects of intercalation/exfoliation (viz. effects of intercalant, type of clay, molecular parameters of the matrix polymer, processing conditions, etc.) under static conditions, it may mean the rate with which the position of the XRD peak is displaced. To most industrial researchers the term immediately conjures image of a compounding line – will the diffusion be fast enough to engender exfoliation during the available residence time? Evidently the three cases are related, but there is a level of increasing complications. The model sandwich is the simplest case. The static intercalation is complicated by the presence of stacks of organoclay platelets. The stacks have different number of platelets, up to 300 or so. While intercalation of the external layers may resemble that of the model sandwich, the internal layers will see a hindered process all the way to the centre. The simplest analogy would be a stack of cards subjected to simultaneous separation from the edges by balls having larger diameter than the thickness of the card. The process would lead to progressive bending of the cards from the centre toward the margin, requiring extra energy. Melt compounding in an extruder is further complicated by the presence of intricate stress fields that are expected to accelerate the static intercalation process, but at the same time may cause attrition and reaggregation under compressive stresses. 3.1.8.1 Macromolecular Diffusion As was mentioned in Section 3.1.3, diffusion of a polymer into the interlamellar gallery very much depends on the magnitude of the driving thermodynamic forces. Considering the results of the equilibrium thermodynamic computations one must realise that owing to the loss of entropy the diffusion must involve enthalpic interactions, hence it is not of the entropy-driven self-diffusion type. The required energetic interactions may take place between the polymer and either intercalant (χap), or clay (χsp) or both. For χap < 0 the polymer diffuses into the energetically favourable gallery and it interacts with the two intercalant layers, which could lead to a kinetically trapped intercalated state. If χap < 0 leads to the prediction of kinetically trapped intercalation, then a solution may be that reverse magnitude, χap ≥ 0, is needed. However, the SCF calculations indicated that when polymer and organoclay are immiscible the expansion of galleries by polymer diffusion would not take place. An elegant solution to the problem is to use a mixture of functionalised and non-functionalised polymers, where the end-functionalised 302
Thermodynamics macromolecules strongly interact with clay, i.e., χsp Me
(74)
The prediction of the Doi-Edwards theory was experimentally confirmed, e.g., by Klein et al. [1983] who showed that for linear PBD for logDs versus logMw, the slope: n = –1.95 ± 0.1. Similar values were reported for other polymers, viz. for PE and PS [Fleischer, 1987]. A wider range of the slope values, n = –1.66 to –2.25 was reported by von Meerwall [1983]. In the case of mutual diffusion of polymer M in a matrix of polymer P with energetic interaction between the two species, the chemical potential and the concentration gradient control the process. Thus, DM contains terms related to the binary interaction coefficient, χ12 ≤ 0, and the volume fraction of the components, φ [Brochard et al., 1983]:
(
)
DM = φ (1 − φ ) 2 χ12 Me / M DRouse
(75)
The dependence indicates that while DM is a function of other variables, the MW-dependence is stronger than in the self-diffusion case. The mutual diffusion is related to the molecular weight of the diffusing molecules, but not of the surrounding matrix, P. This conclusion originates from the assumption that the process is reptation-controlled and at large values of matrix molecular weights: P > 2Pe the reptation tube is independent of P. The relation is limited to miscible systems, χ12 ≤ 0, since when the interaction coefficient is positive the system phase separates and mutual diffusion does not take place. When the molecular weight P < Pe the tube undergoes continuous renewal – the surrounding molecules are releasing the confinement and rebuilding it. Within 303
Clay-Containing Polymeric Nanocomposites this region there is a strong dependence of DM on P-a, with a = 2.5 or 3.0, dependent on the assumption. At a still lower range of non-entangled matrix, P Tg = 100 °C, under vacuum. The kinetics of PS diffusion into the galleries was followed by XRD (in situ and ex situ). For FH-ODA with PS-30 the measurements showed progressive disappearance of the d001 = 2.13 nm peak and growth of another at d001 = 3.13 nm. The XRD peak intensity, Iij,(of the i-th peak and the j-th substance) is proportional to its weight fraction, wj: w j (t ) = Iij (t )ρj μ T* / Kij ⎫⎪ ⎬ ⇒ ζ(t ) ≡ w j (t ) / w j (∞) = Iij (t ) / Iij (∞) w j (∞) = Iij (∞)ρj μ T* / Kij ⎪⎭
(76)
where ρj, μ T* and Kij are, respectively, density, mass absorption coefficient and a constant. The time-dependent quantity ζ(t) is a fraction of the melt-intercalated organoclay. The kinetic curves, ζ(t) versus t, depend on composition (organoclay, polymer, its Mw), temperature (T), pressure (P), etc. The kinetic of intercalation curves were fitted to the relation derived by Breen et al., [1987] for vapour sorption of MMT:
ζ(t ) = 1 −
∞
∑ (4 / α ) exp{− Da m =1
2 m
−2
α m2 t
}
(77)
The latter authors modelled MMT as a cylinder of stacked circular disks, each of the radius α. The diffusion was assumed Fickian, with the diffusivity constant, D. Integration of the diffusion equation yields the above relation, where am is the m-th positive root of the zero order Bassel function. The predicted dependence is shown in Figure 78. Fitting the kinetic data to Equations 76 and 77 Vaia et al., were able to determine the effective diffusion rate parameter, D/a2. For the PS30/FH-ODA system the temperature dependence was of the Fulcher-type: log(D/a2) = ao – a1/(T – T∞), where a1 = 604 ± 40 K and T∞ = 322 K. For the series of narrow molecular weight distribution (MWD) PS the data followed Equation 74. In Figure 79 the self-diffusion data extracted from these measurements (FH-ODA with monodispersed PS resins) as well as for PS fractions [Fleischer, 1987; Antonietti 304
Thermodynamics
Figure 78 Kinetics of melt intercalation under static conditions displayed as a change of the fraction of melt-intercalated organoclay, ζ(t), with diffusion time. The Breen et al., function for D/a2 values from 0.002 to 0.03.
Figure 79 Molecular weight dependence of the effective diffusion coefficient, D/a2, for PS at 180 °C and for the self-diffusion coefficient of monodispersed PS resins with FH-ODA. After [Fleischer, 1987; Antonietti and Sillescu, 1985] – see text.
305
Clay-Containing Polymeric Nanocomposites and Sillescu, 1985] are plotted versus Mw. The solid lines are drawn with the theoretical slope of –2, predicted by the reptation theory. The coefficient governing the kinetic diffusion into galleries, Da-2, is ten decades larger than Ds for PS. Assuming that the diffusion into the galleries is the same as in the case of selfdiffusion, one obtains for the apparent radius of the FH platelets a = 214 nm. This seems to be the right order of magnitude. The reasonableness of this quantity implies that diffusion into a gallery with, h = 1.2 to 2.2 nm, is not particularly affected by the narrow slit geometry. This is striking, since the unperturbed radius of gyration of the PS samples used was rΘ2
1 /2
= 5 to 15 nm (for Mw = 30 to
300 kg/mol) hence significantly larger than h. Thus PS macromolecules do not diffuse as a Gaussian coil but as extended macromolecules with chain diameter controlled by the segmental motion. According to Flory, the average diameter of a paraffinic chain is equal to about 3.5 -CH2- units, thus having the diameter of ca. 0.45 nm, for PS this magnitude is about twice as large, ca. 0.90 to 0.95 nm. In another publication [Vaia et al., 1996] melt intercalation was carried out on synthetic FH pre-intercalated with either dodecyl (DDA) or octadecyl (ODA) ammonium ion. Two narrow molecular weight PS resins, with Mw = 30 and 400 kg/mol, and one poly-3-bromo-styrene (PS3Br) PS with Mw = 55 and Mw/Mn = 2 were used. XRD studies showed that in contrast with intercalated PS30/FH-DDA, the PS3Br/FH-DDA system was exfoliated. Clearly, the polar Br-group in the para position increased the polymer-silicate interactions to the extent that exfoliation was possible. The exfoliation was partially confirmed by TEM. Stacking of 5 to 20 layers with interlayer spacings of between 2 and 4 nm and the presence of the original crystallites or the primary particle were discernible. Individual silicate layers were observed near the edge whereas small coherent layer packets separated by polymer-filled gaps are prevalent toward the interior of the primary particle. The larger expansion of the gallery height and disordered structure were apparently responsible for the disappearance of the XRD peak. The heterogeneity of the CPNC structure, stacking in the primary particle implies that the process is more complex than simple sequential separation of individual layers starting from the surface of the primary particle. Long-range forces (e.g., stresses associated with gallery expansion and layer bending) will maintain stacking of an optimal size. Layer-by-layer delamination may take place near the edge of the primary particle but it is unlikely in the central part. The increased polarity of PS3Br (with respect to PS) will interact more with the polar FH surface. As a result, the frictional coefficient, associated with polymer diffusion into the gallery, should increase slowing the melt intercalation kinetics. Qualitatively, these effects were observed. Chen et al. [2003] followed the procedure used by Vaia et al. However, in this case poly(styrene-b-isoprene) block copolymers (PS-b-PI) diffused into interlamellar galleries of synthetic fluorohectorite, pre-intercalated with ODA. Intercalation at T = 100 to 170 °C was followed by XRD for up to 70 h. The copolymer diffused similarly on either side of the order/disorder transition temperature. The intercalation rate decreased as the PS block increased. 3.1.8.3 Simulation of Melt Intercalation Kinetics Melt intercalation can be visualised as a flow of molten polymer into narrow slits. As discussed before, the driving force for the process is not the configurational entropy, but enthalpic interactions with either low molecular weight intercalant, 306
Thermodynamics or (preferably) with clay. In other words, it is the gradient of chemical potential that drives the diffusion. In the presence of flow, there is also frictional resistance retarding the process. Thus, the flux of melt from the molten matrix reservoir into the gallery in the x-direction has diffusional and viscous components [Mason and Viehland, 1978]: −J =
c( x) Do ∂μ c( x) Bo ∂p + η ∂x k B T ∂x
(78)
where c(x) is molecular concentration in x-direction of flow, Do is the diffusion coefficient, μ is the chemical potential, Bo ≅ h2/12 is a geometric factor, η is viscosity and p is mean value of the driving pressure in the x-direction. From the Gibbs-Duhem equation and Fick’s law this dependence can be used to express the effective transport diffusion coefficient as [Nicholson et al., 1996]:
Dtrans =
c( x ) Bo ⎤ ⎛ ∂μ ⎞ ⎛ Do ⎞ ⎡ c( x )h2 k B T ⎤ ∂μ ⎡ Do + ⎢1 + ⎥ ⎟⎜ ⎢ ⎥ ⎜ ⎟ 12Doη ⎥⎦ η ⎦ ⎝ ∂ ln c( x ) ⎠ ⎝ k B T ⎠ ⎢⎣ ∂ ln c( x ) ⎣ k B T
(79)
For the non-entangled melts, the Stokes equation holds, hence Doη/kBT = 1/3πa (where a is segmental diameter). Thus, the diffusional component dominates for small slit height, h, and the viscous one when the gap is wide. The kinetics of melt intercalation by polymer diffusing into a clay slit under stationary (no mixing) conditions was modelled using MD [Lee et al., 1998a; 1999; 2000; Baljon, 1999]. The simulation model was a face centred cubic lattice oriented with planes parallel to the xy plane of the clay platelet. On each side of the lattice stack there was a reservoir of molten polymer, equilibrated at constant T and P. To start the process, two central lattice layers were removed to form a rectangular slit into which the polymer could diffuse. The size of the slit was fixed, non-expanding by the diffusion process. The system was strictly binary, without a low molecular weight intercalant. The polymer molecules were represented by anharmonic bead-spring chains. Polymer beads, separated by a distance r, interacted with the interaction energy ε according to the truncated, repulsive Lennard-Jones potential for r ≤ rc ≡ σ21/6. The finite extendable nonlinear elastic (FENE) model was used to express the bonding potential between nearest-neighbour beads along the chain, separated by distance r. The bead-lattice potential with the interaction energy εbl was given by truncated, attractive LennardJones potential at rc = 2.2σ. Thus, the bead-bead interactions were either zero or repulsive at short distance, while the bead lattice was attractive. The simulations were performed at constant T* =kBT/ε = 1, assuming that the ratio of the interaction parameters, ε/εbl = 1, 2 or 3 (in latter publications the ratio went up to 10), and the random force in y and z directions obeys the Gaussian statistics. An analytical model for diffusion of polymer into a slit gallery of length L in the x-direction was derived in the form of Laplace transforms: ζˆ (s) =
[
sinh Ψ
sΨ cosh Ψ + (din / Dout )1 / 2 sinh Ψ
]
(80)
Ψ ≡ sL / 4 Din 2
where Din and Dout are diffusion coefficients inside and outside the slit. Since the Laplace transform is a function of sL2 divided by s, the intercalation kinetics, 307
Clay-Containing Polymeric Nanocomposites
ζ(t), depend on tL-2. The superposability of the kinetic curves computed for different values of L would provide evidence that melt intercalation is a diffusioncontrolled process. The computations were performed for L = 43σ to 341σ. The results ζ(t) versus t/L2 indeed superposed onto a single curve. Assuming that Dout = ∞ the effective diffusion coefficient, Din = Deff was calculated from Equation 80 by fitting the dependence to the simulated kinetic curve. Excellent agreement between the simulated data and the analytical prediction with the fitted value of Deff was obtained. It is important to note that as the interaction ratio, ε/εbl, increases the computed kinetics slow down and the diffusion coefficient decreases. Since the diffusion driving force is proportional to this ratio, crowding at the entry to the slit is evidently slowing down the intercalation – the macromolecules want immediately to interact with the slit walls instead of diffusing further into the gallery in a more civil way. Crowding at the gap by macromolecules eager to interact with slit surfaces also slows the intercalation process. As a result, the polymer concentration decreases linearly in the x-direction and shows two local maxima in the z-direction. While the x-gradient is a measure of the non-equilibrium aspect of the process, the enhanced concentration of the interacting polymer segments near the silicate wall is not. Simulation was also carried out to determine the centre of mass diffusion of polymeric chain within the reservoir. The computations determined that the equilibrium, bulk self-diffusion coefficient, Ds = 0.006σ2/τ, but the effective diffusion coefficient governing the intercalation Deff = 3σ2/τ, hence 500 times larger. The authors argued that once the slit is open the process is no longer governed by the equilibrium properties. Computations of the centre of mass displacement of the 1st chain to enter the slit show that its diffusion coefficient is comparable to Deff, while the macromolecules that follow diffuse more slowly. Simulation of diffusion inside the gallery, in the y-direction gave very small values – about 1/10 of the equilibrium Ds. The amphiphilic intercalant may enhance the intercalation kinetics (relative to the case of homopolymer intercalants) and form novel structures. Thus, in the last publication of this series melt intercalation with symmetrical diblock copolymer was simulated [Lee et al., 2000]. The model slit surfaces were assumed grafted with low molecular weight intercalant. The system was maintained at constant pressure to permit the slit to open as polymer intercalates. The kinetics of intercalation were simulated for different values of surface-block-A and surfaceblock-B interaction parameters. It was concluded that suitable diblock copolymers might be used to intercalate clay. Starting with the Landau and Lifshitz one-dimensional equation of motion, Ginzburg et al. [2001] derived a simple relation that describes the dynamics of melt intercalation. The derivation is based on a model, which assumes the presence of ‘kinks’ – a sudden, local expansion of the interlayer spacing between two clay platelets. Accordingly, intercalation is a process of localised excitations that move the kink, opening the interlamellar space. The model stipulates that the kinks originate in the interplay between the double-well potential for the clay-clay long-range interaction, bending elasticity of the platelets, and sufficiently strong external shear force.
308
Thermodynamics
3.1.9 Pressure-Volume-Temperature Dependence for CPNC There are several reasons for the study of the pressure-volume-temperature (PVT) properties of a condensed system. These are pertinent for materials processing – the compressibility and the thermal expansion coefficients are required for the modelling of flow through virtually any processing machine. However, there is a stronger reason for these measurements – describing the experimental data by means of an adequate theory, provides an insight into intermolecular attractions and repulsions in a given system. Furthermore, a free volume parameter may be extracted from the PVT data, and then used to predict dynamic and kinetic properties, viz. flow, ageing, viscoelasticity. Finally, it is important to establish the effects of additives (gaseous, liquid or solid!) on all these properties. 3.1.9.1 Equations of State (eos) In his PhD thesis of 1873, van der Waals proposed the first equation of state (or eos). The relation is usually written in terms of reduced variables (indicated by tilde):
( P + a / V )(V − b) = T / R 2
(
)(
)
or P˜ + 3 / V˜ 2 3V˜ − 1 = 8T˜
V˜ = V / V*; T˜ = T / T*; P˜ = P / P * V* = 3b = Vc ; T* = 8a / 27 Rb = Tc ; P* = a / 27b2 = Pc
(81)
Pc Vc / RTc = 3 / 8 Of interest to van der Waals were low molecular weight liquids with a welldefined critical point – the coordinates of this point, Pc, Vc and Tc, were used as the reducing parameters. For T = 0 the hard-core volume, Vo = b. van der Waals considered that molecules move in ‘cells’ made by the surrounding molecules with uniform potential. The volume, within which the centre of a molecule can freely move, is what defines the free volume. The free volume fraction is usually defined as: f = Vf /V = V – Vo/V, where Vf = V – Vo. Detailed methods of computation of Vo from the chemical structure have been developed, viz. [Bondi, 1968; van Krevelen, 1993; Porter, 1995]. Several comprehensive reviews of the eos’s used for polymeric liquids have been published. For example, Rodgers [1993] collected PVT data for 56 polymers at P ≤ 200 MPa and melting T-range from 50 to 170 °C. The review presents fundamentals of the theories and it evaluates fit to experimental data. Five prominent eos were examined. These are listed in chronological order. 1. Flory-Orwoll-Vrij (FOV) [1964]: ˜ ˜ / T˜ = (1 − V˜ −1 / 3 ) −1 − 1 / VT ˜˜ PV ˜ ˜ ˜ V = V / V*; T = T / T*; P = P / P *
(82)
V* = ρ ; T* = s*η* /(c* k B ); P* = c* k B T * / V * 3 *
where: ρ*, s*, η*, c* are respectively: the ‘hard-sphere’ radius, number of contacts per segment, the segment-segment interaction energy, and the coordination number (kB is the Boltzmann constant). 2. Simha-Somcynsky [1969] (S-S) – will be discussed in the next section. 3. Sanchez-Lacombe [1978] (S-L):
309
Clay-Containing Polymeric Nanocomposites
[
]
˜ ˜ / T˜ = −V˜ ln(1 − ρ˜ ) + (1 − 1 / r )ρ˜ − 1 / TV ˜˜ PV
ρ˜ = 1 / V˜ = V * / V ; T˜ = T / T*; P˜ = P / P * where r = M w P * / RT * ρ*; T* = ε * / k B
(83)
where M w is the weight-average molecular weight, while ρ* is the characteristic density parameter. The parameter r represents the number of lattice sites occupied by the r-mer. Evidently, its presence in the eos negates the principles of corresponding states – it can be recovered only for r → ∞. 4. Hartmann and Haque [1985] (H-H):
˜ ˜ 5 = T˜ 3 / 2 − ln V˜ PV (84) V˜ = V / V0 ; T˜ = T / T0 ; P˜ = P / B0 The characteristic pressure reducing parameter, B0, has been identified as the isothermal bulk modulus extrapolated to T = 0 and P = 0. 5. Dee and Walsh [1988] (D-W):
(
˜ ˜ / T˜ = 1 − 0.8909qV˜ −1 / 3 PV
) − (2 / T˜ )(1.2045V˜ −1
−2
− 1.011V˜ −4
)
(85) q ≅ 1.07 ; V˜ = V / V*; T˜ = T / T*; P˜ = P / P * Within the low-pressure region Rodgers found that all these five eos are adequate – the worst performance was of S-L and the best of S-S theories. However, as the pressure increased, FOV and S-L dependencies started to perform poorly, while S-S and D-W continued to provide good description. Rodgers cited the following deviations: for FOV ΔV × 104 = ± 22 (7), for S-L ΔV × 104 = ± 33 (10), for H-H ΔV × 104 = ± 9 (6), for D-W ΔV × 104 = ± 6 (5), and for S-S ΔV × 104 = ± 7 (4). These differences were computed for the full pressure range; the values in parentheses are for the low pressure range, P ≤ 50 MPa. 3.1.9.2 Simha-Somcynsky (S-S) Equation of State The Simha and Somcynsky [1969] (S-S) theory is based on the lattice-hole model presented in Figure 80 for a binary mixture of two liquids with linear chain molecules (A and B) and with holes or vacancies (V). The model assumes that molecular segments occupy y-fraction of the lattice sites. The remaining part, h= 1 – y, is occupied by holes simulating the molecular disorder. The fraction h may be viewed as a particular measure of free volume content. The placement of the occupied and vacant sites is random. For a simple liquid (A = B) the configurational thermodynamic properties, such as the PVT (eos) relations, or the cohesive energy density, are characterised by three quantities: the maximum attraction energy, ε*, between a pair of chain segments, the corresponding segmental repulsion volume, v*, and the number 3c of volume dependent, external degrees of freedom. In terms of these quantities and the number of segments per chain, s, the characteristic pressure, temperature and volume parameters can be defined, viz.: P* = qzε*/(sv*); T* = qzε*/Rc; V* = v*/Ms hence: P*V*/RT* = c/sMs(86) with Ms the molar mass of the statistical segment, qz = s(z-2) + 2, the number of interchain contacts in a lattice of coordination number z = 12, and R the gas constant. The variables of state, P, T, V, are then scaled in terms of these. Since the theory is based on the corresponding-states principles, these reduced variables 310
Thermodynamics
Figure 80 Simha-Somcynsky lattice-hole model for two constituent liquid molecules (A and B) with vacancies (V). The energetic binary interaction parameters are also shown [Simha et al., 2001].
define a universal, reduced P˜ − T˜ − V˜ surface for all liquids. In other words, they generate master isotherms and isobars. Provided the theory is quantitatively successful, a superposition of experimental and theoretical lines will yield the scaling parameters and thus the material characteristic interaction parameters, ε* and v*. To derive the eos, S-S first calculated the partition function, Z, for all possible numbers of arrangements of occupied sites and empty holes in a lattice with z coordination number. The Helmholtz free energy is directly given as: F = -RT lnZ. In reduced variables the obtained free energy function F˜ has the form: F˜ = F˜ [V˜ , T˜ , h(V˜ , T˜ )] . Thus, in addition to the usual volume and temperature dependence, F˜ contains the hole fraction h, which in turn must vary as: h = h(V˜ , T˜ ) . This variation is obtained by minimising the free energy at a specified volume and temperature, i.e.: (∂F˜ / ∂V )V˜ , T˜ = 0
(87)
From the Helmholtz free energy the pressure is obtained as: P˜ = −(∂F˜ / ∂V˜ )T˜
(88)
From the above relations the S-S eos is derived in the form of coupled equations: ˜ ˜ / T˜ = (1 − η)−1 + 2yQ2 (1.011Q2 − 1.2045) / T˜ PV
(89)
3c[(η − 1 / 3) /(1 − η) − yQ (3.033Q − 2.409) / 6T˜ ] + (1 − s) − sln[(1 − y) / y] = 0 (90) 2
2
with Q = 1 /(yV˜ ) and η = 2-1/6yQ1/3. The two equations not only predict how the specific volume varies with pressure and temperature, but at the same time how the free volume parameter, h, changes with these. Of all the eos, only the one derived by S-S explicitly gives the hole fraction, h = 1 – y, which is directly related to the free volume fraction, f. Equations 89 and 90 provide a corresponding states description of the PVT behaviour of any liquid. Once the four characteristic 311
Clay-Containing Polymeric Nanocomposites parameters: P*, V*, T*, and 3c/s are known, the specific volume and all its derivatives are known in the full range of P and T. For linear molecules the external degrees of freedom are proportional to the number of segments: 3c = s +3. Thus, for linear polymers, where s >> 3, the external degree of freedom: 3c/s ≅ 1; hence for polymers, only three parameters: P*, V*, and T*, are required. The S-S theory also provides a simple expression for the reduced cohesive energy density (CED):
[
˜ = U˜ / V˜ = −( y / 2V˜ )( yV˜ )−2 2.409 − 1.001( yV˜ )−2 CED
]
(91)
where U is the internal energy of the system – one of the intensive thermodynamic quantities. From CED the solubility parameter can be calculated as ˜ × P * . Accordingly, δ is also an intensive function of δ = CED = CED independent variables: δ = δ(T, P), independent of other variables, viz. the type of solvent used to dissolve a given polymer. Recent analysis of δ for 38 different molten polymers shows that, in accord with van der Waals prediction, the product ˜ ˜ )1 / 2 is approximately constant, thus δ is proportional to polymer density, δV ∝ (UV ρ = 1/V, which in turn depends on T and P [Utracki and Simha, 2004; Utracki, 2004]. Comparison with the values listed in handbooks for δ indicate that these are significantly lower than computed from the polymer melt properties. The origin of this discrepancy has been traced to the free volume contribution – since the experimental values for δ are determined in solution, the macromolecular environment there (i.e., free volume) is equivalent to that found in molten polymer at high temperature, viz. T ≅ Tg + 300. Considering that different polymers have different compressibility and thermal expansion coefficients, the difference between the solubility parameters of two liquids (supposedly related to their miscibility) will change with T and P. Users beware! The full range of the reduced independent variables of polymer melts was calculated as [Utracki and Simha, 2001a]: 1.6 < 100 × T˜ < 7.1 and 0 < 100 × P˜ < 35 (92)
Owing to the overlapping properties of polymers, the range of variables in these inequalities is much wider than that experienced by a single resin. Within these limits the coupled eos in Equations 89 and 90 can be approximated by polynomials. The goodness of fit was judged by considering the values of the standard deviation (σ), the correlation coefficient squared (r2) and the coefficient of determination (Cd). The reduced specific volume follows the dependence:
[ (
) ]
ln V˜ = ao + a1 T˜ 3 / 2 + P˜ a2 + a3 + a4 P˜ + a5 P˜ 2 T˜ 2
(93)
The values of ai parameters and the statistics of fit are listed in Table 42. Considering the value of the standard deviation (σ = ± 0.18%) the fit is considered satisfactory. The following expression well approximates the hole fraction:
( )
h V˜ , T˜ = ao + a1 / V˜ + a2 T˜ 3 / 2 + a3 / V˜ 2 + a4 T˜ 3
(94)
Again, the goodness of the fit can be judged by the parameter values listed in Table 42. However, even with only three parameters (i.e., assuming that a3 = a4 = 0) the standard deviation of σ = 0.55% was computed.
312
Thermodynamics ˜ ˜ P) ˜ Table 42 Statistics of the polynomial data fit to S-S eos: V˜ = V(T, ˜ ˜; Equation 93 and 94, respectively. and h = h(V,T) Data [Utracki and Simha, 2001a]
Parameters
Values for log (V/V*)
Values for h
a0
-0.10346 ± 0.00034
1.203 ± 0.020
a1
23.854 ± 0.032
-1.929 ± 0.047
a2
-0. 1320 ± 0. 0012
10.039 ± 0.249
a3
-333.7 ± 2.5
0.729 ± 0.026
a4
1032.5 ± 23.6
-218.42 ± 10.29
a5
-1329.9 ± 52.8
—
_
0.00183
0.00253
r2
0.99973
0.99965
Cd
0.99967
0.99899
Comments
Most errors at
V˜ ≤ 0.93
˜ -values computed from the S-S eos were Furthermore, the scaled CED approximated by: ˜ = ao + a1 / V˜ 2 + a2 T˜ + a3 / V˜ + a4 VT ˜˜ CED
(95)
The reduced solubility parameter, δ˜ = δ˜ (T˜ , P˜ ) was well approximated by:
(
)
δ˜ = ao + a1 P˜ / T˜ + a2 + a3 P˜ + a4 P˜ 2 T˜
(96)
The numerical values of the fitting parameters for equations 95 and 96 are listed in Table 43. Evidently, the polynomial expressions well approximate the eos prediction, even with three parameters. There are several practical advantages of these polynomial relationships. In spite of their simple form they provide good approximation of thermodynamic properties in the molten state within the full range of P and T. Their derivatives, e.g., the thermal expansion coefficient and the compressibility factor:
α ≡ (∂ ln V / ∂T ) P−const and β ≡ (∂ ln V / ∂P)T =const respectively, can easily be calculated for any temperature and pressure. Furthermore, fitting Equation 93 to experimental data results in rapid determination of the reducing parameters, P*, V* and T*. When highly accurate values of these parameters is required (viz. for determining the binary interaction parameters, ε* and v*), the latter values may be used as the initial estimates for the iterative fitting of the data to the coupled Equations 89 and 90. This approach greatly shortens the iteration process and provides highly reliable solutions.
313
Clay-Containing Polymeric Nanocomposites
Table 43 Algebraic representation of the reduced cohesive energy ˜ , and reduced solubility parameter, δ˜ . density, CED Data [Utracki and Simha, 2001a] ˜ Parameter Reduced CED Reduced δ˜ Version #1
Version #2
Version #1
Version #2
a0
0.1323 ± 0.0044 –1.053 ± 0.061
0.9505 ± 0.0013
0.95788 ± 0.00088
a1
0.5659 ± 0.0033 –0.592 ± 0.056
az
–0.00085 ± 0.00005
a2
–0.8892 ± 0.0340
–4.97 ± 0.26
–4.782 ± 0.031 –5.108 ± 0.020
a3
az
2.36 ± 0.12
6.091 ± 0.090
10.73 ± 0.16
a4
az
3.74 ± 0.24
az
–13.20 ± 0.49
σ
0.00392
0.00189
0.00673
0.00289
r2
0.99996
0.999991
0.99993
0.99999
Cd
0.99876
0.999714
0.99153
0.99844
Comments
Excellent fit Most errors at with 3 constants V˜ ≤ 0.95
Good fit with 3 Most errors at constants P≅0
az = assumed zero
It is evident that the reducing parameters of the S-S eos, P*, V* and T* (or the interaction parameters incorporated there) must depend on the polymer molecular characteristics (its conformation, configuration, molecular weight, etc.) as well as on the additives, present in all commercial polymers [Utracki, 2002b]. As an example Table 44 lists the values of the reducing parameters for several PS resins. The molecular weight of resins marked Z (for Zoller) is Mw = 110 to 0.9 kg/mol. Nova PS1301 (Mw = 270 kg/mol) is a ‘crystal’ (transparent) PS for extrusion, thus probably relatively free of additives. Dow 667 is an injection moulding grade containing 2.5 wt% mineral oil. The other commercial resins, BASF 1424 and Monsanto HH105, have unknown additives. The indicated authors computed the last four sets of parameters from the same set of experimental data (measured by Quach and Simha in 1971), but using different procedures. Evidently, the characteristic parameters depend not only on polymer MW and diverse additives, but also on the routine used to compute them, viz. sequential or simultaneous, the criteria used for minimisation of deviations, etc. In many cases, only a relative change of polymer specific volume is of interest and the use of listed in the literature values of P*, V* and T* is acceptable. In this case it is possible to compute the PVT surface and obtain all the derivative properties (viz. thermal expansion coefficient, compressibility, etc.) using tabulated parameters. When more fundamental information is required, repetitive tests of a given resin must be conducted with precision ΔV < 0.2%. 314
Thermodynamics
Table 44 Interaction parameters computed for different PS resins PS Resin
P* (MPa)
T* (K)
V* (ml/g)
Mo (g/mol)
v* (ml/mol)
ε* (kJ/mol)
Z-110
738.7
11920
0.96310
46.49
44.78
33.07
Z-34
786.1
11836
0.96530
43.42
41.92
32.94
Z-9
794.5
11670
0.96920
42.74
41.42
32.87
Z-09
779.1
10504
0.98750
44.41
43.85
33.75
Nova PS1301
743.5
11723
0.95259
45.87
43.69
32.49
BASF 1424
813.4
11379
0.94770
40.98
38.83
31.53
HH105
811.1
11265
0.93390
41.21
38.49
31.22
Dow-667
774.0
11728
0.95021
44.19
41.99
32.50
Quach and Simha [1971]
745.3
12680
0.95980
49.12
47.15
35.14
Utracki and Simha [2001a]
699.5
12684
0.96230
52.22
50.25
35.15
Hartmann et al., [1989]
715.7
12791
0.96260
51.45
49.53
35.45
Rodgers [1993]
715.9
12840
0.96340
51.59
49.70
35.58
Note: The first 8 sets of parameters in this table were computed by the author (LAU)
3.1.9.3 Extension of S-S eos to Binary Miscible Systems Extension of S-S theory to CPNC systems can be carried out in three logical steps: (1) Extension of the original theory for one-component liquids to two-component miscible systems, (2) Extension of the latter theory to suspensions of solid particles in polymeric liquid and (3) Development of a specific model for CPNC. During the last 30-odd years S-S eos has been used to describe a variety of equilibrium and non-equilibrium thermodynamic problems in single and multicomponent systems, viz. solutions, blends, foams and composites. In this section the use of S-S eos for bicomponent, miscible systems will be outlined. A major assumption of these derivations is the mean-field approach – the miscible components (e.g., polymer segments or solvent molecules) are randomly placed on the lattice. 315
Clay-Containing Polymeric Nanocomposites Jain and Simha [1980; 1984] demonstrated that for such binary mixtures the S-S theory is directly applicable provided that the reducing and interaction parameters are treated as averages. Thus, fitting Equations 89 and 90 to experimental data yields the composition-dependent averages indicated by the angular brackets, viz. X . Extension of the S-S theory has to account for the presence of component-1 (mole fraction x1) and component-2 (x2 = 1 – x1):
(
)
P * = qz ε * / s ν * ; T * = qz ε * / R c ; V * = ν * / Mo
(
)(
P * V * / R T * = c / s 1 / Mo
)
(97)
c = c1 x1 + c2 x2 ; s = s1 x1 + s2 x2
Mo = ( Mo1 s1 x1 + M02 s2 x2 ) / ( s1 x1 + s2 x2 )
Similarly, the interaction parameters ε* and v* are becoming compositional averages over 11, 22 and 12 interactions. The interacting site fractions, X1 and X2 = 1 – X1, are determined by the lattice interchain contact, zqi, and the mole fractions xi, thus:
Xi = qi zxi /
∑ q zx i
(98)
i
where: zqi = si(z – 2) + 2 with z = 12. Then the averages and are related to the individual interactions as follows:
ε* ν*
p
* *p * *p *p = X12 ε11 ν11 + 2 X1 X2 ε12 ν12 + X22 ε 22 ; p = 2, 4
(99)
The two values of p reflect the assumed Lennard-Jones 6-12 pair potential. The two relations in Equation 99 have six parameters, four of which are accessible from the PVT measurements of the neat components (i.e., polymer-1 and polymer-2), * * and ν12 can be determined. Their thus the two cross-interaction parameters, ε12 value may indicate attractive or repulsive cross-interactions, but as experience with liquid mixtures showed, the strong attractive energetic interactions usually lead to a smaller value of the volumetric interaction parameter. To facilitate computations, Equation 99 can be transformed into:
ν*
2
=
Ξ2 ≡
X12
Ξ4 ≡
X12
Ξ4 Ξ2 *2 * × ν11 ; and ε * = 2 × ε11 Ξ2 Ξ4 +
2 2 X1 X2 e12 ν12
+
4 2 X1 X2 e12 ν12
+
X22 e22 ν222
2 ⎛ ν * ⎞ ⎛ ε* =⎜ * ⎟ ⎜ * ⎝ ν11 ⎠ ⎜⎝ ε11
⎞ ⎟ ⎟ ⎠
+
X22 e22 ν422
4 ⎛ ν * ⎞ ⎛ ε* =⎜ * ⎟ ⎜ * ⎝ ν11 ⎠ ⎜⎝ ε11
⎞ ⎟ ⎟ ⎠
(100)
with the following definitions: * * * * * * * * e12 ≡ ε12 / ε11 ; e22 ≡ ε 22 / ε11 ; ν12 ≡ ν12 / ν11 ; ν 22 ≡ ν 22 / ν11 ;. The outlined extension of the S-S theory to binary systems has been successful in describing thermodynamics, e.g., phase equilibria, and solubility of gases [Xie and Simha, 1997; Simha and Moulinié, 2000]. As shown in Figure 81 the
316
Thermodynamics assumption of random mixing is also valid for miscible polymer blends of polyphenylene ether (PPE) with PS [Utracki and Simha, 2001a]. Furthermore, the hole fraction, h, computed from PVT data was found to correlate with the dynamic properties, e.g., flow of low molecular weight liquids and their solutions, that of polymers, their blends and foamable compositions [Utracki, 1986; Utracki and Simha, 2001b]. Thus, the random mixing assumption was proved adequate for describing the compositional variation of PVT surfaces in a wide range of miscible, binary * * , were extracted liquids. Several sets of binary interaction parameters, ε12 , ν12 from the experimental data. Analysis indicates that they may be well approximated by simple geometric (Berthelot’s) and algebraic (in radius) averages, respectively:
(
* * * ε12 = δ e ε11 × ε 22
)
1 /2
(101) 3 1 /3 ⎤ ⎡ * 1 /2 * * ν12 = δ ν ⎢ ν11 + ν 22 8 ⎥⎦ ⎣ In these dependencies the two adjustable parameters are close to unity, δe, δv ~ 1 (maximum departure from unity was reported by Simha and Moulinié [2000] to be ca. 13%). Note that the Berthelot’s geometric mean rule with δe = 1 has been shown to be valid in many applications. The next step is extension of this binary treatment to suspensions.
( )
( )
/
3.1.9.4 Extension of S-S eos to Suspensions The key in extending the S-S multicomponent theory to suspensions is the assumption that the solid particles can be treated as giant, rigid molecules, differing
Figure 81 Scaling volume, V*(ml/g) T*(k), and temperature parameters as functions of composition (mol fraction) in PPE/PS blends: the correlation coefficient, r, is also given [Utracki and Simha, 2001a]. Reproduced with permission, Copyright 2001 Wiley-VCH-Macromol.
317
Clay-Containing Polymeric Nanocomposites from the matrix macromolecules by size and the interaction parameters. Both components, the matrix and the solid particles, are placed on a lattice, assuming * * ~ ν 22 . that hard-core specific volumes of the components are about equal: ν11 For the flexible polymeric chain the external degree of freedom, 3c, is assumed to be proportional to the number of statistical segments, s, and for macromolecules 3c1/s1 → 1. By contrast, for solid particles this parameter tends to zero, 3c2/s2 → 0. Furthermore, since now the PVT behaviour of component 2 (solid particles!) is not accessible, additional assumptions are needed to solve Equation 99. The validity of these assumptions was demonstrated by Simha et al. [1984; 1986] and by Papazoglou et al. [1989]. The aim of these studies was to examine the suitability of the S-S eos theory for describing the observed changes in the system modulus, thermal expansion coefficient and the compressibility engendered by the addition of solid particles. The authors computed these functions for specific ratios of the interaction parameters, viz.:
(
⎧⎡ * * * * * * ν 22 / ν11 = 1 : 1; ν12 / ν11 = ⎨⎢1 + ν 22 / ν11 ⎩⎣
ε
* 22
/ε
* 11
= 5; ε
* 12
/ε
* 11
)
3
⎫ ⎥⎦ / 2⎬ = 1.0492 ⎭
1 /3 ⎤
(102)
= 1.2 − 10
It has been shown that the extended S-S theory was able to account for different experimental sets of data and demonstrated that for an optimised ‘adhesion ratio’, ∗ ∗ ε12 /ε11 , good agreement between the S-S eos and several continuum-based theories of the system modulus was achieved. It is noteworthy that the relationships in the first line of Equation 102 are consistent with the fundamental assumption of the lattice theories that the cell size for any component in the system should not differ by more than 10% from the average. The Berthelot’s geometric mean rule was also assumed valid, thus: * * * * ε12 / ε11 = ε 22 / ε11 ≅ 2.24 .
3.1.9.5 Extension of S-S eos to Nanocomposites The effect of nanoparticles on CPNC behaviour seems to be disproportionate to the clay content. For example, addition of 0.64 vol% of MMT increased the HDT of a PA-6 matrix by 70 °C, tensile modulus by 70%, and flexural modulus by 130%, but reduced oxygen permeability to 50%, etc. While XRD, TEM and SEM provide an assessment of the inorganic phase dispersion in a polymeric matrix they do not provide information on the interactions involved. Rheology (discussed in the following part of this book) is a sensitive tool for observation of the interaction and structural effects on the behaviour of multiphase systems, but owing to its sensitivity to structure and orientation it cannot be used for the direct determination of interactions [Utracki and Kamal, 2002a]. In this part the utility of the PVT measurements for this purpose will be examined. In the simplest case, dilute CPNC may be visualised as a three-component mixture: polymer, clay particles (stacks of platelets or aggregates), and exfoliated clay platelets with attached polymer chains. However, in many systems, low molecular weight intercalants as well as standard additives (stabilisers, lubricants, etc.) are also present.
318
Thermodynamics The PVT behaviour represents a volume-average response hence the computed parameters, P * , V * , T * , ε * and ν * , are volume averages. Furthermore, the data fitting to eos makes it possible to compute the hole fraction within the full range of independent variables. This function is also an average response of the matrix and dispersed phases. Up to this point the procedure is straightforward, identical to that applied for a single-component system. Thus, for each composition one obtains predictive behaviour for V = V(P, T), h = h(P, T) as well as the average values for the reducing and interaction parameters. A plot of these parameters versus composition offers an opportunity to predict the PVT behaviour of intermediate compositions as well. However, to calculate the magnitude of the individual binary interaction parameters, ε ij* , ν ij* , i , j = 1, 2, the system structure must be modelled. The model is an idealisation of the perceived CPNC structure, formulated assuming a binary system. The assumptions introduced ought to be simple but realistic. There is a multitude of CPNC types (intercalated and exfoliated; end tethered and with an intermediate layer of low molecular weight intercalant molecules forming a cloud around clay platelets; systems with compatibiliser or without; etc.). Furthermore, the structure may vary with concentration and process variables, viz. exfoliated in diluted systems under large strains, forming short stacks at higher concentration and/or high stress and temperature. For these reasons, different models may be required to interpret different CPNC behaviour. The simplest would be the one for randomly dispersed clay platelets in a molten matrix – expected to be valid for diluted, exfoliated nanocomposites. 3.1.9.5.1 Diluted, Exfoliated CPNC – Simplified Approach The PVT behaviour of PA-6 and a CPNC containing this matrix polymer may serve as an illustration of eos applicability to CPNC. Two commercial resins from Ube Ind., PA-1015B and PA-1015C2 were used. According to the manufacturer’s information, the former is a neat PA-6 and the latter is a nanocomposite based on PA-6 containing 2 wt% organoclay, i.e., 1.6 wt% (or 0.64 vol%) of clay. In the following text these resins will be labelled PA and PANC, respectively. Their PVT behaviour was measured within the ranges T = 300-580 K and P = 0.1 to 190 MPa. Prior to testing the material was dried for 48 h at 80 °C. Since the instrument used measures only the incremental changes of the specific volume as a function of P and T, first an absolute value of the specific volume at ambient condition was measured in a glove box with an accuracy of Δ ≤ ± 0.001 ml/g [Simha et al., 2001]. For each resin at least six runs were carried out, two isobaric the others isothermal. Furthermore, some runs were repeated up to three times. Owing to thermal decomposition at higher temperatures, the reproducibility between these two types of measurements was poor, worse for PANC than for PA. For this reason only single sweep, isothermal data were considered reliable. The specific volumes of molten PA and PANC at ambient pressure and T = 240 °C were determined as: V240 = 1.0182 ± 0.0154 and 1.0009 ± 0.0150, respectively. The observed reduction of the specific volume, V, (by 1.7%) is greater than that expected on the basis of additivity (1.2%). Examples of the experimental data and the ability of S-S eos to describe the behaviour are displayed in Figure 82. Evidently, eos describes the observed behaviour for molten PA and PANC well.
319
Clay-Containing Polymeric Nanocomposites
Figure 82 Comparison of experimental (solid circles) and computed data from Simha-Somcynsky eos (open circles) isotherms in the molten state for two Ube resins; PA-6 and PANC within the range of pressure from ambient to 200 MPa (every 10 MPa). Reprinted with permission from [Utracki et al., 2003]. Copyright 2003 American Chemical Society.
320
Thermodynamics The deviations from the theoretical dependence at ambient P and the highest T are caused by thermal decomposition and degassing. The reducing parameters (P*, V*, T*) were obtained by the simultaneous least squares fit of all data points to the S-S eos (Equations 89 and 90). Their values, the statistics of fit, and interaction parameters computed from Equation 86 are listed in Table 45. To directly compare the specific volume variations, V = V(P, T), for PA and PANC the predicted dependencies were computed by substituting the reducing parameters from Table 45 into Equations 89 and 90. To avoid crowding the figure, only five pressures are shown. Computations simultaneously provided h = h(P, T). The results are presented in Figure 83. In Figure 84 the ratios V(PA)/V(PANC) and h(PA)/h(PANC) at the same T and P are shown. While the addition of organoclay to PA reduced the specific volume by about 1% (insensitive to P and T) the effect on the hole fraction is more pronounced (by ca. 15%) and more dependent on the independent variables. Up to this point the treatment of the PVT data for PA and PANC were identical – for the former it resulted in the material-characteristic reducing and interaction parameters, whereas for the latter only average values were obtained. To calculate the individual binary interaction parameters of Equation 99, the site fractions, Xi, must be calculated from a CPNC model. The CPNC from Ube is known to be exfoliated, free of interfering low molecular weight intercalant and the clay concentration is low. Thus, the first proposed model is that of randomly dispersed clay platelets in a molten PA matrix. Accordingly, the clay platelets may be approximated by disks of diameter d = 200 nm and thickness t = 1 nm, density, ρ = 2.5 g/ml, with ‘molecular’ mass, M = NAρπd2h/4 = 47,296 kg/mol and ‘molecular’ volume Vp = M/ρ = 18.92 × 106 ml/mol. This volume is assumed to occupy s2 lattice sites. In the lattice model of a mixture, the hard-core volumes of the constituents should not differ too much. For the assumed 6-12 potential the factor 21/6 relates the positions of potential
Table 45 Eos fit to PVT data for PA-6 and PANC from Ube. Data [Simah et al., 2001] Parameter
PA-6
PANC
Correlation coefficient squared, r2
0.999999
0.999998
Standard deviation of data, σ
0.00112
0.001271
Coefficient of determination, CD
0.998602
0.998550
P* (MPa)
1257 ± 8
116 4 ± 13
11134 ± 32
1130 7 ± 54
8919.3 ± 9.6
8888.6 ± 16.2
27.512
30.288
T* (K) 4
10 V*(ml/g) Ms (g/mol)
ε11 (kJ/mol)
ε11* = 31.23 ± 0.09
= 31.71 ± 0.15
v11 (ml/mol)
v11* = 24.54 ± 0.03
= 26.92 ± 0.05
321
Clay-Containing Polymeric Nanocomposites
Figure 83 Specific volume (a) and hole fraction (b) for PA-6 and PANC versus temperature at five pressures: ambient, 50, 110, 150 and 190 MPa. Note that a small difference in density translates to a substantial loss of the free volume parameter, h. Reprinted with permission from [Utracki et al., 2003]. Copyright 2003 American Chemical Society.
* (PA) = v*/21/6 = 21.86 ml/mol. The minimum and onset of repulsion. For PA ν hard * = 7.8677 ×105. chain length of PA-6, s1 = 801, and that of clay, s2 = Vp / ν hard With these numbers and the inorganic clay content of 1.6 wt% and the mole fractions, xi, then the site fractions X1 = 0.99263, X2 = 1 – X1 are calculated.
322
Thermodynamics
Figure 84 The V and h data from the preceding figure are plotted as ratios at the same T and P (for each point). Note the reverse P-dependence – while the ratios of h increase with P, those of V decrease. Reprinted with permission from [Utracki et al., 2003]. Copyright 2003 American Chemical Society.
As stated before, Equation 99 contains six interaction parameters, viz. two * * and ν11 and four unknown: experimentally determinable quantities, ε11 * * * * that must be determined. However, the lattice model ε12 , ν12 , ε 22 and ν 22 * * * * ~ ν 22 = 1.11ν/ 211 requires that ν11 , thus as before ν 22 may be assumed. The use of * * * Berthelot’s geometric mean rule, ε12 = ε11ε 22 , reduces the problem to two * * and ε 22 unknowns ν12 . The adopted model leads to the following results (‘11’ represents polymerpolymer, ‘22’ represents clay-clay and ‘12’ represents polymer-clay interactions) [Simah et al., 2001]:
(
)
* * * ε11 = 32.09 ; ε12 = 313.54 and ε 22 = 3, 063 (kJ / mol) * * * ν11 = 24.89 ; ν12 = 33.53 and ν 22 = 24.89 (ml / mol)
In summary, the model of bare clay platelets randomly dispersed in the PA-6 * * , i.e., that ε 22 , is about two orders matrix resulted in determining that ε ∗22 ≅ 95ε11 ∗ . This large of magnitude larger than the liquid-liquid interaction parameter, ε11 difference is consistent with the values of the specific surface energy discussed in Section 3.1.7.1. However, the model resulted in inconsistency as far as the repulsion volume parameter is concerned. Even assuming near identity of the * * = ν 22 , the computed value for the 12 interaction, interacting volumes, ν11 * * * ν12 ≅ 1.35 × ν11 , is much larger than the theoretical requirement of ν ij* ≤ 1.1 × ν11 . To examine the validity of the additive model an intermediate composition was prepared by mixing equal portions of PANC and PA. The PVT behaviour was determined and the values of the average parameters, ε * and ν * , were extracted. The use of the simple model failed to yield consistent values for the * * > 1.1 × ν11 binary interaction parameters – again ν12 was obtained. 323
Clay-Containing Polymeric Nanocomposites * , Besides the unexplained large value of the volume repulsion parameter, ν12 the disproportionately large reduction of the matrix free volume by addition of 0.64 vol% of clay is inconsistent with the model – the model simply does not offer any mechanism that could justify these results. The assumption of naked clay platelets randomly floating in molten PA-6 contradicts the high energy of crystalline solids discussed in Section 3.1.7.1, the results computed from MD adsorption of organic molecules on the clay surface, and the direct measurements of solidification of macromolecules (by means of surface force apparatus and neutron scattering) on the high-energy solid surface (see Section 3.1.7.2). Thus, the simple, additive model had to be modified.
3.1.9.5.2 Dilute, Exfoliated CPNC – Gradient Mobility Approach Again, two commercial resins from Ube, PA and PANC, were used [Utracki et al. 2003]. An intermediate composition of PA:PANC = 1:1 was prepared. Since the diluted composition was extruder-blended in a TSE, the original resins were also re-extruded under the same processing conditions. As shown in Table 46 the PVT dependence of the three compositions was well described by S-S eos. The new scrupulous measurements confirmed the previously reported large reduction of the free volume parameter, h = h(P, T). The new CPNC model used for the interpretation of the experimental values of the average binary interaction parameters: ε * and ν * , is based on three sets of information:
Table 46 Statistics of fitting S-S eos to the experimental data of PA, PANC and their 1:1 mixture [Utracki et al., 2003] Parameter
PA
PA/PANC = 1:1
PNC
Correlation coefficient, r2
0.999999
0.999998
0.999998
Standard deviation 0.00112 of data, σ
0.00145
0.00127
Coefficient of 0.998602 determination, CD
0.998501
0.998550
P* (MPa)
1257 ± 8
1186 ± 11
1164 ± 13
T* (K)
11134 ± 32
11364 ± 49
11307 ± 54
10 V* (ml/g)
8919.3 ± 9.6
8943.1 ± 14.3
8888.6 ± 16.2
Ms (g/mol)
27.512
29.700
30.288
ε11 (kJ/mol)
ε11* = 31.23 ± 0.09
= 31.49 ± 0.29
= 31.71 ± 0.15
v11 (ml/mol)
v11* = 24.54 ± 0.03
= 26.56 ± 0.36
= 26.92 ± 0.05
4
324
Thermodynamics (1) The reduction of molecular mobility near the crystalline surface computed from MD and measured experimentally. This has been discussed in Section 3.1.7. Several researchers reported an increase of viscosity as the distance from the clay platelet decreased toward the last 4 to 6 nm – at that stage the viscosity reached an immeasurably high value, taken as an indication of solidification. For example, the dynamic oscillatory measurements identified three regions: bulk behaviour for z > 100 nm, intermediate, and solid-like for z > 6 nm where Hookean-type stress-strain behaviour was observed [Luengo et al., 1997]. (2) The molecular structure of Ube PANC. According to an early patent, the preparation of exfoliated PNC involved three steps [Deguchi et al., 1992]: (a) intercalation of MMT with ω-dodecanoic acid ammonium chloride (ADA), (b) mixing the intercalated clay with ε-caprolactam and water, and (c) compounding the adduct in a TSE with ca. 70% of PA-6. Chain-ends titration indicated that about 1/3 of the macromolecules were end-tethered to the clay surface. (3) The observed PVT behaviour of PA, PANC and their 1:1 mixture. The PVT data were collected with painstaking care, investigating all possible influences, sources of error and ad nauseam verifying the reproducibility. The new sets of data for PA and PANC confirmed the accuracy of the preliminary findings, and lead to similar values of the interaction parameters. However, calculation of the binary interaction parameters using different pairs of the three polymeric compositions (PA, PNC and their 1:1 mixture) resulted in different values of the binary interaction parameters. The assumption of random mixing is probably not correct for the CPNC system studied, and a different model must be proposed. The new model considers that in these diluted and exfoliated PNC: 1. Clay platelets are exfoliated and at the concentration of 0.64 vol% (hence φ = 0.0064) they are randomly distributed in the PA-6 matrix. 2. 1/3 of clay cations are used for end-tethering the PA-6 macromolecules. 3. Contrary to the previous assumptions, the clay platelets are not bare, but rather they are enrobed in two PA-6 layers of different structure and properties. 4. Directly on a clay platelet there is the first or inner layer of solidified PA-6, zsw ca.6 nm thick. As the distance from the clay platelet surface exceeds this limit, an intermediate layer starts. This layer, where PA-6 chain mobility progressively increases, extends from ca. 6 to ca. 100 nm from the platelet surface, thus 6 ≤ zin (nm) ≤ 100. At still larger distance, z > 100 nm, the segmental mobility of this exterior layer is the same as that in the bulk polymeric matrix. This model of ‘hairy clay platelets’ (HCP) [Utracki and Lyngaae-Jørgensen, 2002] implies that only at φ < 0.005, where the clay platelets are more than 200 nm apart (the assumed diameter of the model disk is 200 nm), are the values of the interaction parameters expected to be constant. At higher concentrations, φ > 0.005 or w > 1.2 wt%, the HCPs overlap, PA-6 macromolecules with bulk properties are absent, and the interactions depend on composition. Thus, the adopted model considers individual clay platelets to be covered by solidified polymer and dispersed in a matrix composed of PA macromolecules of the intermediate and exterior layers. It 325
Clay-Containing Polymeric Nanocomposites is important to note that in this new model the random placement of binary component has been assured by virtue of redefining the dispersed particles as HCP. In the calculations it was assumed that the ‘12’ parameters characterising the hetero-contacts are approximated by averages – the repulsive volume by the algebraic average of the radii, the attractive interactions by Berthelot’s geometric mean rule, i.e., that in Equation 101: δe, δv = 1. Applying the model to Equation 99 resulted in the sets of binary interaction parameters listed in Table 47. The model considers the clay platelets to be covered by two layers of PA-6 with the segmental mobility increasing with the distance z from the platelet surface. In consequence it is expected that the interaction parameters will be similar * * ε11 ≈ ε12 ≈ ε *22 . This is indeed shown by data listed in Table 47. Thus, for the simplest diluted and exfoliated CPNC systems, the PVT measurements provide well-defined functions of V = V(P, T) and h = h(P, T). Furthermore, from the reducing parameters the average values of the energetic and volumetric parameters can be unambiguously calculated. However, the calculations of the individual interaction parameters ‘11’, ‘12’, and ‘22’ very much depend on the adopted model – on the definition of what physically are the interacting species ‘1’ and ‘2’. Thus, when ‘1’ is taken as an average statistical segment of the PA macromolecule and ‘2’ an inorganic ‘bare’ clay platelet the * * ≅ 95ε11 are found. On the other hand, when the model energetic interactions: ε 22 specifies that species ‘1’ are statistical segments of PA in the melt and ‘2’ are inorganic clay platelets embedded in layers of PA with reduced chain mobility, * * ≅ ε11 . the energetic interactions change dramatically to ε 22 On face value both approaches are correct and the relative magnitudes of the energetic interactions have physical sense. However, while the former model was unable to explain the significant reduction of free volume, the latter model provides a direct and unequivocal response: it is precisely the strong interaction between the clay surface and PA-6 segments that causes the reduction of segmental mobility and the reduction of free volume. The latter model also agrees with the welldocumented adsorption of organic molecules of inorganic, high surface energy substances. It is consistent with our understanding of this type of CPNC.
Table 47 Interaction parameters in PA, PANC and their 1:1 mixture. Data [Utracki et al., 2003] System Contacts
PA-6
PA/PANC = 1:1
PANC
ε*ij
(kJ/mol)
v (ml/mol)
ε*ij
(kJ/mol)
v ij (ml/mol)
ε*ij
(kJ/mol)
v*ij (ml/mol)
11
31.23
24.54
31.21
24.54
(26.69)
24.54
12
—
—
31.53
25.75
(29.16)
25.75
22
—
—
31.86
26.99
31.86
26.99
* ij
*
Notes: In the PA/PANC mixture and in PANC the variations of v*ij are as postulated in the text. Values of ε*ij in parentheses were computed for matrices of different mobility hence they are fitting averages. The experimental error for ε*ij and v*ij is ± 1.5 and ± 1.0%, respectively.
326
Thermodynamics 3.1.9.5.3 Intercalated CPNC – Concentration Gradient Polystyrene-based CPNC serve as an example of the third type of PVT behaviour [Tanoue et al., 2003a]. Commercial polystyrene (PS1301; Mw = 270 kg/mol, melt flow rate MFR = 3.5 g/10 min) was compounded with organoclay (Cloisite® 10A = MMT intercalated with 1.25 meq/g of 2MBHTA d001 = 1.92 nm). The compounding was carried out in a Leistritz co-rotating TSE-34 (L/D = 40) at uniform barrel temperature of 200 °C, screw speed of 200 rpm, and feed rate of 5 kg/h (the average residence time t = 240 s). The organoclay (according to subsequent TGA: 0, 1.4, 2.8, 5.7, 10.6 and 17.1 wt%) was added to molten polymer using a side-feeder. XRD analysis showed that the melt compounding resulted in a partial intercalation (interlayer spacing increased to d001 = 4.2 to 4.8 nm) accompanied by a simultaneous collapse of the interlamellar gallery spacing (d001 decreased from 1.92 to ≤ 1.7 nm) caused by the thermo-oxidative decomposition of the organoclay. TEM showed the presence of short stacks with few exfoliated platelets. Enhancement of the mechanical properties by addition of the organoclay was poor. The PVT data were collected within the range of temperatures (T = 300-520 K) and pressures (P = 0.1 to 190 MPa). For each composition two to five isothermal runs were carried out. A high degree of reproducibility was obtained. As before, the data were fitted to S-S eos, Equations 89 and 90. The non-linear least square fit yielded the reducing parameters, P*, V* and T*, as well as the free volume function, h. The parameters and the statistics of the fitting procedure are listed in Table 48. Using Equation 86 the average interaction parameters, ε * and ν * , were calculated from the computed values of P*, V*, and T*. As an example, the h = h(P, T) dependence for neat PS 1301 is shown in Figure 85 – the ones for other compositions were similar. As expected, h increases with T and decreases with P.
Figure 85 Hole fraction as a function of temperature and pressure for neat PS 1301.
327
328 749 ± 5 11938 ± 46
743 ± 3
11723 ± 22
9525.9 ± 4.7
P* (MPa)
T* (K)
104 × V* (ml/g) 46.079 33.08 ± 0.14 44.16 ± 0.48
32.49 ± 0.06
43.69 ± 0.23
(kJ/mol)
(ml/mol)
9583.7 ± 9.6
45.869
Mo (g/mol)
0.9990610
0.9994420
Correlation
0.9999980
0.9999987
r2
0.0014205
1.4
HS16A
0.0011119
0
PS1301
σ
C10A (wt%):
Parameter
44.53 ± 0.30
33.28 ± 0.08
46.537
9567.8 ± 5.9
12010 ± 29
747 ± 3
0.9995868
0.9999991
0.0009326
2.8
HS17
43.92 ± 0.50
33.19 ± 0.13
46.433
9458.8 ± 9.8
11975 ± 48
756 ± 6
0.9989405
0.9999977
0.0014970
5.7
HS18
43.90 ± 0.77
32.80 ± 0.20
47.407
9259.6 ± 15.1
11834 ± 74
747 ± 8
0.9971886
0.9999939
0.0023897
10.6
HS19
43.05 ± 0.78
32.45 ± 0.20
47.770
9011.2 ± 14.6
1171 1 ± 73
75 1 ± 9
0.9967893
0.9999924
0.0025876
17. 1
HS20
Table 48 Statistics of fitting the Simha-Somcynsky eos to experimental data and the computed parameters for PS, and PNC containing 1.4 to 17.1 wt% Cloisite‚ 10A [Tanoue et al., 2003a; 2004]
Clay-Containing Polymeric Nanocomposites
Thermodynamics The most interesting question was how the addition of different amounts of organoclay affected the free volume fraction, i.e., the h-function. A plot of the relative hole fraction, h(PNC)/h(PS) at constant pressure P = 1 MPa and three temperatures, T = 360, 460 and 560 K, is shown in Figure 86. A similar plot of the relative specific volume, V(PNC)/V(PS), versus w linearly decreased in the full range of composition, consistently with the constant ratio of densities in the molten state: ρ(organoclay)/ρ(PS) = 1.49. The free volume in PANC with 2 wt% organoclay dispersed in PA-6 at T = 300-590 K and P = 0.1 to 200 MPa was lower by 12 to 17% than that of the matrix (see Figure 84). Since during the reactive exfoliation the intercalant (ω-amino dodecyl acid (ADA)) became the first mer that end-tethered PA-6 macromolecules to the clay surface, there was nothing in these nanocomposites to prevent adsorption of PA-chains on the high surface energy clay platelets. The adsorption and resulting reduction of chain mobility caused such a large loss of free volume. The dependencies displayed in Figure 86 are most interesting since in these PS-based CPNC the degree of intercalation/exfoliation is incomparably lower than that in PA-6/MMT. Furthermore, here the 2MBHTA intercalant is in stoichiometric excess (over MMT CEC), thus it forms an intermediate layer between the inorganic, high-energy clay surface and PS matrix. As a consequence, one would not expect PS macromolecules to be adsorbed and loose mobility. Apparently, in spite of these obstacles, the clay is able to reduce PS segmental mobility. Furthermore, the data show that initially the relative loss of free volume is proportional to the clay content – more clay, less free volume. Figure 86 indicates that h(PNC) < h(PS) for organoclay content from 0 to ca. 16 wt%. However, the dependence reaches a local minimum at ca. 4 wt% organoclay (4 to 6% loss of the free volume) then starts to increase with the organoclay loading. The latter effect is related to progressively poorer dispersion of clay platelets.
Figure 86 The relative hole fraction of PS-based CPNC versus organoclay concentration at P = 1 MPa and T = 360, 460 and 560 K.
329
Clay-Containing Polymeric Nanocomposites The dependencies in Figure 86 seem to suggest that should the concentration increase still further the hole fraction of CPNC would exceed the value of the PS matrix. Olson et al. [1997] used positron annihilation lifetime spectroscopy (PALS) for CPNC containing 75 wt% organoclay (FH-ODA) within the range of temperature from –60 to 140 °C. The CPNC was prepared by dry blending the component powders, compressing and annealing it for 48 h at 170 °C. Annealing increased the organoclay interlayer spacing from d001 = 2.253 to 3.031 nm. From the PALS data the free volume fraction, f, can be calculated as proportional to the product of the positron lifetime, τ3, and intensity, I3. Accordingly, the ratio: f(CPNC)/f(PS) was calculated from the cited data for T= –60 and 140 °C – the ratio was found equal to 0.535 and 0.544, respectively. Thus, in these non-degraded, highly loaded CPNC the free volume fraction is about 46% smaller than that in PS. Even correcting for the reduction of the organic phase content (in these CPNCs there is about 40 vol% clay) these measurements still indicate a reduction of free volume by about 10%, both in the glassy as well as in the molten state of PS. Using Equation 86 the average interaction parameters, ε * and ν * , were calculated (see the last two rows of Table 48). Their parallel variation with composition is noteworthy (they are linked through the Lennard-Johnson potential). In Figure 87 the relative free volume at three temperatures (the same data as shown in Figure 86) is plotted versus ε * . The dependence is linear with a high value of the correlation coefficient r2 ≥ 0.99992. In the lower part of the figure the organoclay loading is indicated – evidently the reduction of h is related to the average interaction parameters, thus indirectly to the clay loading.
Figure 87 Relative hole fraction of PS1301 with 0 to 17.1 wt% of Cloisite® 10A at T = 360, 460 and 560 K as a function of the average energetic interaction parameter, ε * . In the upper part the linear regression results are shown; in the lower part the organoclay concentration for each data point is given. Data [Tanoue et al., 2003a; 2004].
330
Thermodynamics 3.1.9.5.4 PVT – Concluding Notes PVT measurements offer unique insight into the structure and interactions of CPNC. The experimental data can be well described in terms of the SimhaSomcynsky equation of state (S-S eos). The relation provides two sets of valuable information; the functional dependence of the free volume parameter, h, and the average interaction parameters, ε * and ν * . From the admittedly limited information on the PVT behaviour of CPNC it seems that h provides a relatively simple and accurate measure of the average degree of exfoliation – as the degree of exfoliation increases the free volume content in CPNC decreases. Since the mechanism is based on the degree of macromolecular adsorption on the clay surface, the decrease must be related to the magnitude of the clay-polymer interactions and composition. The latter variable has two influences as initially its increase provides a larger surface for the polymer to adsorb onto, and at higher concentration it reduces the interlayer spacing, thus reducing the total surface available for adsorption. Recently, from analysis of the PVT and PALS data for PP the authors determined that at ambient pressure, within the range of T = 300 to 400 K the hole volume increases linearly from vf = 0.12 to 0.22 nm3, while the number of holes per gram remains constant Nh ≅ 0.38 (±0.02) × 1021, or ca. 0.34 holes per nm3 [Kilburn et al., 2003]. Analysis of the concentration dependence of the average interaction parameters, ε * and ν * offers invaluable insight into the interactions inside CPNC. The calculated values of the individual binary interaction parameters very much depend on the adopted model. So far the adopted models have been simple. The insight gained suggests that for the interpretation of CPNC data the present version of the S-S theory (derived for binary, randomly mixed systems) should be extended to non-random mixtures. Xie et al. [1992] have made preliminary efforts in this direction.
331
Clay-Containing Polymeric Nanocomposites
332
Thermal Stability
3.2
Thermal Stability
3.2.1 Thermal Stability During Processing The stability of the clay-intercalant complex is essential for the melt exfoliation process to succeed. Experience shows that the lack of thermal stability complicates the production and forming processes of CPNC not only for engineering and speciality resin (which require high processing temperature), but also for CPNC based on PO or PS. Owing to the lack of polar groups these polymers need wellintercalated clay and a compatibiliser (a functionalised polymer or copolymer). The stability aspect is also important for the characterisation of CPNC, viz. PVT or rheological measurements where CPNC may be exposed to high temperature for long periods. Several researchers reported that MMT intercalated with a long chain aliphatic quaternary ammonium cation is unstable at temperatures above 180 or 200 °C. A brief summary of the work on the thermal stability of organoclays is provided in Table 49 and Table 50. Early studies on the thermal decomposition of n-alkyl amines were published by Ogawa et al. [1992] and by Menéndez et al. [1993]. The former team intercalated MMT with quaternary ammonium chloride, (CH3)3CnH2n+1N+Cl(where n = 8, 12, 14, 16, and 18), while the latter used primary amine, CnH2n+1NH2 (n = 1 to 6) to intercalate γ-titanium phosphate in either vapour or aqueous phase: γ-Ti(H2PO4)(PO4)×2H2O, (TiP) to obtain TiP with 1.3 mol of alkyl amine. In both cases the interlayer spacing increased linearly with n: d001 = ao + a1n (nm)
(103)
For the MMT with quaternary ammonium ions the parameter values were: a0 = 0.7311 and a1 = 0.0851 with the correlation coefficient squared r2 = 0.9947, while for the TiP systems a0 = 1.12 and a1 = 0.230. The n-alkyl amines formed bimolecular layers. To follow the thermal degradation processes, the intercalated TiP was heated up to 600 °C. Two weight loss processes were identified: (1) at T = 80-110 °C loss of crystallisation water; and (2) a complicated thermal decomposition occurring in discreet steps, evidenced as peaks in DSC thermograms. The first step at T = 130 to 195 °C was caused by a loss of ca. 30% of n-alkyl amine and a reduction of d001; now a0 = 1.12 and a1 = 0.189. The second step at T = 190-270 °C involved loss of another 30% of n-alkyl amine, reduction of d001 to a0 = 1.12 and a1 = 0.157, but the bimolecular structure survived up to this point. The third step at T = 335-375 °C was caused by a loss of another 40% and a reduction of d001 that involved both parameters of Equation 74; now a0 = 0.89 and a1 = 0.044. Other peaks followed. Maxfield et al. [1995; 1996] reported that incorporation of a traditional ‘sizing agent’, e.g., silane or titanate, enhances the thermal stability. More recent work by Yu et al. [2003] places some doubts on the generality of such a statement. 333
Clay-Containing Polymeric Nanocomposites Table 49 Reports on the thermal stability of organoclays Polymer
Intercalant
Comments +
Reference
None
MMT(CH3)3CnH2n+1N ; DTA: exothermic peak n = 8, 12, 14, 16, & 18 started at ca. 240 °C
None
γ-titanium phosphate intercalated by n-alkyl amines; n = 1-6.
Thermal decomposition Menéndez et in several steps at 130 al., 1993 to 500 °C
None
3M-alkyl amine added (CnH2n-1); n = 6-12
d001 linearly increased with n; degradation starts at ca. 200 °C.
PE
MMT intercalated with Compounding 5 min at Heinemann et di-methyl stearyl benzyl 190 °C reduced d001 al., 1999 ammonium ion from 1.94 to 1.41 nm
None
MMT with [Ni{di(2aminoethyl)amine}2 or [Ni(2,2´:6´,2´´terpyridine)2]
Thermal stability of Ni- Bora et al., amines was 50 and 150 2000 °C higher than that of metal-free amines
PS
2M2ODA
Degradation during extrusion at 185 °C
Gilman et al., 2000
PP/PP-MA
Cloisite® 20A or 30B
Loss of about 10 wt% of the organic constituent
Lee et al., 2000a
None
MMT with tri-methyl dodecyl or octadecyl, 3MDDA or 3MODA
TGA: T = 200 to 300 °C the decomposition of C12 to C17 unsaturated hydrocarbons
Xie et al., 2000
PA-6
MMT with quaternary alkylammonium
TGA: lower stability than neat PA-6; 5 wt% loss at 271 °C.
Cho and Paul, 2001
PS
Ammonium or phosphonium cation
See Table 50
Zhu et al., 2001a
PS
Ammonium or phosphonium cation
T > 200 °C to d001 reduced to 1.58 nm
Zhu et al., 2001b
Styrenic copolymers
2MDBHTA quaternary Heating at 210 °C Yoon et al., ammonium ion reduced d001 from 3.5 to 2001 1.9 nm
PS, PS-co-MA, PE-g-MA
2M2HTA quaternary ammonium ion
Polyimide
Quaternary ammonium See text with tallow radical
Delozier et al., 2002
PS
Cloisite® 10A (2MDBHTA)
Tanoue et al., 2003a; 2003b
334
10 min at 210 °C reduced d001 from 3.18 to 2.94 nm
Oxidative decomposition of organoclay at 200 °C
Ogawa et al., 1992
Ogawa, 1994
Lim and Park, 2001
Thermal Stability The authors did not observe any significant changes to the thermal decomposition of organoclay treated with t-butyl dimethyl chlorosilane. In air, the decomposition started with a small TGA peak at about 150 °C and a major one centred at 250 °C. Under nitrogen the decomposition started at ca. 180 °C and proceeded in three steps with peaks at 254, 320 and 438 °C. In their patent of 1998, Christiani and Maxfield noted that a MMT-ammonium complex with secondary amine was more thermally stable (onset of decomposition at T = 275 °C) than that with either tertiary or quaternary ones. Thus, for melt exfoliation in a PA matrix, they preferred MMT pre-intercalated with dipentyl ammonium chloride. Gilman et al. [2000a; 2000b] reported that during extrusion compounding of PS with MMT-2M2ODA (dimethyl dioctadecyl ammonium chloride) at 185 °C the intercalant was thermally degraded. The authors pointed out two factors causing this unwanted effect: increased temperatures and the presence of air. The oxidative degradation also affected the matrix PS, broadening its polydispersity, Mw/Mn from 1.8 to 2.40. The degradation was found detrimental to the flame retardancy of CPNC, as the re-aggregated clay offered less protection to the polymer. Since extrusion of PS with inorganic NaMMT, under the same processing conditions, did not affect the PS, it was concluded that the free radicals formed during the intercalant degradation were to blame. The authors proposed the mechanism, where the byproducts of the Hofmann elimination mechanism [Hofmann, 1851] of quaternary ammonium reacted with oxygen to give peroxy radicals, which in turn attacked the PS macromolecules. Tanoue et al. [2004a; 2004b] reported similar observations for CPNC prepared by melt compounding PS with MMT-2MBHTA (Cloisite® 10A; C10A) at 200 °C. FTIR transmission spectra were measured on a Thermo Nicolet spectrometer at a resolution of 4 cm-1 and with an accumulation of 128 scans. To enhance detectability of the chemical changes, difference spectra were calculated for the extruded samples by subtracting out the spectrum of neat PS and comparing the results with the spectrum of C10A. The analysis identified characteristic peaks of the dimethyl benzyl amine (2MBA) - a byproduct of the Hofmann decomposition [Aldrich Co., 1997; Bellamy, 1975]. In addition to these peaks, there are strong carbonyl peaks at 1705 and 1736 cm-1 in most of the extruded samples containing organoclay, which corresponds to =C=O groups formed by oxidation of the double bond in the long-chain olefin - another byproduct of the Hofmann decomposition. Finally, there is evidence of a ketone carbonyl group present in PS. The rheological measurements of the re-extruded CPNC showed that the zero-shear viscosity, ηo, was reduced by about 30% each time. This is particularly noteworthy since increasing the residence time in a TSE by a factor of up to 10 did not further reduce the CPNC matrix viscosity. Similarly, extrusion of neat PS had only a small (ca. 2%) effect on ηo. Thus, the combination of oxygen, temperature and organoclay was responsible for the degradation of PS. CPNC of PP (Mw = 278, Mn = 72 kg/mol), PP-MA (Mn = 23 kg/mol; 2 wt% MAH) and an organoclay (Cloisite® 20A or 30B; C20A = MMT-2M2HTA or C30B = MMT-MT2EtOH) were prepared by melt blending to have about 5 wt% of clay [Bellamy, 1975]. These dry blends were compounded at 210 °C for 10 min in an internal mixer at 50 rpm. During the process both organoclays lost a proportional amount (to the initial) of the constituent elements (e.g., lost of about 10% of C), which indicates that thermal decomposition took place even under these relatively mild conditions. This confirmed the suspicions that thermal 335
Clay-Containing Polymeric Nanocomposites desorption of intercalant takes place during thermal treatment, leading to a decrease of the interlayer spacing, viz. for C20A from d001 = 2.49 to 2.40 nm and for C30B from d001 = 1.84 to 1.50 nm, respectively. Bora et al. [2000] studied the thermal decomposition of MMT intercalated with either [Ni{di(2-aminoethyl)amine}2 (Ni1) or [Ni(2,2':6',2'-terpyridine)2] (Ni2). The thermal stability of these Ni-containing amines was 50 and 150 °C higher compared to their metal-free compounds. On heating, the interlayer spacing of the first compound decreased from d001 = 1.45 (at 50 °C) to 1.35 (at 250 °C) while the second decreased from d001 = 1.94 (at 50 °C) to 1.90 (at 350 °C). DTA of Ni1 had 5 peaks indicating loss of water at 140 °C, and then a stepwise exothermic (oxidation) decomposition of the amine at 273 °C (5.5% mass loss), at 354 °C, 456 °C and 557 °C. Ni2 also showed five DTA peaks, first at about 100 °C (loss of water), and then a series of exothermic (oxidation) peaks at ca. 340 °C, 400 °C (3.5% loss), 505 °C (8.5%), and 645 °C. XRD showed a constant position for the interlayer spacing up to about 350 °C, starting to decrease at ca. 400 °C. Zhu et al. [2001a] described the properties of PS-based CPNC prepared by bulk polymerisation in the presence of onium salts: N,N-dimethyl-n-hexadecyl(4-vinyl benzyl) ammonium-Cl (VB-16), N,N-dimethyl-n-hexadecyl-(4-hydroxy methyl benzyl) ammonium-Cl (OH-16) and n-hexadecyl triphenyl phosphoniumCl (P-16). XRD showed that the interlayer spacing of Na-MMT (d001 = 0.96 nm) increased after intercalation by VB-16, OH-16 and P-16 to, respectively, 2.87, 1.96 and 3.72 nm. After polymerisation the first system was exfoliated whereas the latter two intercalated with d001 = 3.53 and 4.06 nm, respectively. Accordingly, TEM showed discrete clay layers for CPNC with VB-16, only intercalated short stacks for CPNC containing OH-16, and a mixture of both intercalated and exfoliated structures for CPNC with P-16. TGA/FTIR studies indicated that both ammonium and phosphonium compounds degrade by a Hoffmann elimination mechanism, schematically shown in Figure 88. As shown in Table 50, the MMT-P-16 system was the most thermally stable - the onset of degradation took place at a temperature ca. 30 to 50 °C higher than that of the MMT-ammonium. This may be useful when CPNC must be processed at temperatures up to ca. 240 °C. TGA also showed that CPNCs start degrading at higher temperatures than the matrix polymer. Cone calorimetry data revealed a significantly reduced rate of heat release for the nanocomposites. X-ray photoelectron spectroscopy (XPS) was used at T ≤ 500 °C to examine these PS-based CPNCs [Wang, et al., 2002a]. Degradation of the organic components was detected at T = 100 to 200 °C, while at T = 200 to 250 °C MMT started to decompose into SiO2 and Al2O3. Similarly, Du et al. [2002] used XPS to investigate the thermal degradation of PMMA nanocomposites containing
Figure 88 Hofmann elimination mechanism for the thermal decomposition of an ammonium cation. After [Saunders, 1965].
336
Thermal Stability
Table 50 Stability of MMT intercalated with ammonium or phosphonium salts – TGA/FTIR. Data [Zhu et al., 2001a] MMT-VB-16
MMT-OH-16
T (°C)
wt% loss
T (°C)
wt% loss
T (°C)
wt% loss
185-235
1
210-250
2
240-290
2
hexadecene
235-340
10
250-340
6
290-340
9
hexadecene, hexadecanal, or hexadecanone
340-400
7
340-370
8
340-490
19
hexadecene, (PPh3)
400-600
8
370-560
16
490-600
5
Me2NCH2-C6H4-CH= CH2 or PPh3
Total
26
32
MMT-P-16
Decomposition products
35
Notes on the intercalating salts: VB-16 is N,N-dimethyl-n-hexadecyl-(4-vinyl benzyl) ammonium chloride; OH-16 is N,N-dimethyl-n-hexadecyl-(4-hydroxy methyl benzyl) ammonium chloride; and P-16 is n-hexadecyl triphenyl phosphonium chloride.
MMT pre-intercalated with VB16, hexadecyl-allyl dimethyl ammonium chloride (Allyl16), and hexadecyl vinyl-benzyl dimethyl ammonium chloride (Bz16). The three CPNCs were prepared by bulk polymerisation. For both series of CPNC with PS or PMMA as the matrix, the authors demonstrated that during the thermal degradation, the clay accumulates on the surface. Its composition changed at T = 200 to 300 °C. The change (presented as changes of the Si/Al ratio) was found to depend on the type of intercalant, viz. 2MBHd, 2M-allyl-Hd, or 2M-styryl-Hd (Hd = hexadecyl). Zhu et al. [2002] also studied the thermal degradation and flame retardancy (by TGA and Cone Calorimetry, respectively) of the latter systems (PMMA/MMT-VB16, -Allyl16 or -Bz16). Using XRD and TEM the authors found that the nanocomposites obtained by bulk polymerisation of MMA in the presence of organoclays having reactive groups (C=C double bonds) showed better dispersion of platelets than those obtained with MMT-Bz16, but all these systems were only intercalated. The thermal stability (by TGA) was found to be improved by incorporation of 3 wt% of clay; the temperature for 10 wt% loss for neat PMMA, and nanocomposites with MMT-Bz16, MMT-VB16 and MMT-Allyl16 were recorded as: 246, 235, 271, and 295 °C, respectively. Recent publications from the same laboratory describe the thermal stability and flame retardancy (by TGA and Cone Calorimetry, respectively) of nanocomposites in a PE [Zhang and Wilkie, 2003] and PP [Wang and Wilkie, 2003] matrix. These intercalated nanocomposites have been prepared by melt blending in an internal mixer. The PO-organoclay systems showed mixed immiscible-intercalated structures that improved when 337
Clay-Containing Polymeric Nanocomposites PO-MA graft copolymer was incorporated. The results from cone calorimetry suggest that nanocomposite formation has occurred, since there is a significant reduction (by 30 to 40%) in the peak heat release rate (for more information on flame retardancy see Section 5.2). However, these data do not differentiate between the stability of the clay-intercalant complex and that of the matrix. Delozier et al. [2002] performed the TGA analysis of their organoclay (MMT with a quaternary ammonium containing tallow group) in air and nitrogen. The organoclay started to degrade in air at 150 °C, lost approximately 1% weight at 200 °C, 5.5% at 250 °C, and 12% at 300 °C. Under nitrogen the organoclay was more thermally stable, viz. at 250 °C the weight loss was 4% and at 300 °C it was 9%. The effects were specifically blamed on the degradation of the long hydrocarbon tails. The process is known to begin near 200 °C and it accelerates as the temperature increase [Xie, et al., 2000]. Degradability of CPNC, comprising PA-6 and 5 wt% of MMT pre-intercalated with ω-amino dodecyl acid (ADA) was studied [Davis et al., 2003]. Compounded and dried specimens of neat PA-6 and CPNC were injection moulded at 300 °C. Under identical conditions, CPNC degraded (Mn decreased by ca. 40%), while PA-6 did not. ε-Caprolactam and terminal vinyl groups were detected in the product. The Hofmann elimination mechanism was used to explain the discoloration, molecular weight decrease, and reduction of mechanical performance of CPNC with PA-6 as the matrix [Fornes et al., 2003; 2004]. The authors compounded the CPNC in a TSE at 240 °C and then injection moulded at 260 °C. The relative (to extruded PA-6) colour index decreased (i.e., colour intensity increased) with organoclay loading. However, the value for CPNC (proportional to that of neat organoclay) decreased with the reduction of matrix molecular weight. Intercalants with more unsaturation, -C=C-, and/or hydroxyl-ethyl groups, -C2H4-OH, developed stronger discoloration. The authors concluded that more thermally stable intercalants are needed. Similar observations were reported for CPNC with PC as the matrix [Yoon et al.; 2003]. Evidently, thermal degradation of the organic cation poses a serious problem when preparing and processing CPNC. The experience shows that when blanketing by nitrogen is not used, the degradation mainly depends on the number of reprocessing steps and to a minor extent on the total residence time at higher temperature. The organoclays containing quaternary ammonium intercalants are thermally unstable at temperatures above 150 to 165 °C. However, the onset of degradation can be pushed up to above 200 °C by ascertaining the absence of O2 and controlling the composition, the method of treatment (e.g., extent of shearing) and time of heating. Furthermore, the decomposition of intercalant in case of highly polar macromolecules that may directly interact with clay surface, e.g., PA, is not as critical as that in the case of hydrophobic polymers with low polarity, such as PO or styrenics. The use by Inoue et al. [2002] of melamine as an intercalant is an important development. This the first time a compound known as a flame retardant has been used as intercalant, replacing the notoriously unstable quaternary ammonium. This promising route is very much worth pursuing. From the chemical point of view, the least stable are the quaternary ammonium ions with a structure: Ri4N+, where Ri is a straight paraffinic chain. Reduction of the degree of substitution is expected to improve the thermal stability, i.e., RiH3N+ > Ri2H2N+ > Ri3HN+ > Ri4N+ (but at the same time the ionic strength 338
Thermal Stability may decrease). Another method for improving the thermal stability involves double alkylation of the β-carbon(s), e.g., replacing straight paraffinic chain, viz. RCH2CH2-N+ by RC(CH3)2CH2-N+ [Starnes, 2002]. The enhancement of MMTammonium stability by the use of ammonium-Ni complexes has been reported; the decomposition temperature was 50 to 150 °C higher than that without the Ni-complex. Another method of enhancing thermal stability is by the use of the ‘sizing agents’, e.g., silanes or titanates [Maxfield, 1996]. Much less information is available on the stability of phosphonium systems. Stability up to 370 °C was claimed by Ellsworth [1999], but more recent work by Zhu et al. [2001a] indicates a more modest effect, viz. stability of MMT-phosphonium higher by ca. 30 to 50 °C than that of MMT-ammonium. One possible source for such discrepancy is the method used for defining the onset of decomposition. When the degradation products have low vapour pressure and high solubility in the degrading specimen the TGA analysis may be misleading - degradation may have taken place, but the byproducts did not evaporate. For this reason, XRD/FTIR followed by evaluation of the key properties after heating is essential. During the last 10 years or so, the search for more thermally stable organoclays has intensified (see Sections 2.3.2 to 2.3.7). The research on inorganic intercalants was a significant step in this direction, but owing to the complexity and associated costs so far this approach has not gained industrial approval. Observations that clay readily interact with complex, aromatic salts resulted in repeated attempts to incorporate these, more thermally stable molecules as intercalants. Thus, replacing ammonium salt by that of a cyclic tertiary amidine leads to greater thermal stability and more efficient intercalation of clay [Zilg et al., 1999]. Since complex, aromatic molecules are often either functional or coloured (conjugated double bonds), Eastman and TNO patented the use of dyes as intercalants [Barbee et al., 2000c; Fischer et al., 2001]. For similar purposes, Imai et al. [2002] developed intercalant (dimethyl isophthalate substituted with a triphenylphosphonium group) able to react with PET, bind to clay and be stable up to at least 275 °C. Evidently, other stable organic compounds may be used for this purpose limitations are only imposed by availability and costs.
3.2.2 Flame Retardancy and High Temperature Stability The advantageous properties of CPNC include increased thermal stability and reduced flammability [Gilman et al., 1998; 2001]. In 1998 NIST organised an industrial consortium to study the flammability of polymers and their CPNCs. The results of the first year of studies were summarised as follows. (1) In CPNC with 2% to 8% MMT, flammability was reduced by 50 to 80%, without an increase of smoke. (2) The reduced flammability was achieved along with improved physical properties. (3) Incorporation of MMT enhanced char formation, which acts as an insulator and a mass transport barrier. (4) Imidazolium treated MMT showed greater thermal stability and formed finely dispersed nanocomposites with PS and PA-6. The flame retardancy is often accompanied by increased high temperature stability as measured by thermogravimetric analysis (TGA). One of the reasons for such a correlation is the amount of volatiles - the higher their volume, the greater is the weight loss and the greater the flame intensity. However, in the case of CPNC, this relationship very much depends on the relative flammability, the amount of the matrix polymer and the organoclay intercalant [Zhu et al., 2001a, b]. 339
Clay-Containing Polymeric Nanocomposites The TGA results are usually presented in the form of three parameters: T10 (°C), T50 (°C), and char content (wt%); the first two indicating the temperature at which 10 or 50% weight loss was recorded, and the last represents the fraction of material which remains at 600 °C. It is evident that even the initial 10% weight loss takes place at much higher temperatures than the thermal degradation of organoclay (see Section 3.2.1). Recently, MMT was intercalated with oligomeric molecules, bringing the organic content in the organoclay to 65-73 wt% [Su et al., 2004]. In this case, both flammability, and T10 decreased with organoclay loading.
3.2.3 Photo-Oxidative Stability To study the effect of organoclay on the photo-oxidative degradation of PP, four film specimens were extruded, and then irradiated in the presence of oxygen at λ > 300 nm, 60 °C for up to 120 h [Morlat et al., 2004]. Besides neat PP, the specimens comprised organoclay (MMT with dimethyl ditallow ammonium), or maleated-PP, and both of these additives. The authors reported that while the mechanism of PP photo-oxidation did not change, the induction period and the rate of oxidation were modified, resulting in a decrease in material stability, but the same degradation products were formed, in the same relative concentrations. While the reason for the reduced photo-oxidative stability was not identified, several contributing mechanisms have been postulated. (1) Hofmann elimination process during extrusion that engenders unsaturated end groups, which could combine with oxygen during irradiation. (2) MMT may catalyse polymer degradation, especially in the presence of Fe2+ and Fe3+ ions, detected in the organoclay. (3) Maleated-PP readily photo-oxidises, forming volatile byproducts. The catalytic and sometimes inhibiting (catalyst poisoning) activities of clays have been known for a long time, e.g., as suspected agent of life on Earth [Ferris, 2002], in the petroleum industry [Hettinger, 1991], and in general chemical reactions [Newman, 1987]. The catalytic activities of MMT and its derivatives have been reported frequently [see for example: Okada et al., 1988; Usuki et al., 1989; Lan and Pinnavaia, 1994; Biswas and Ray, 1998; Heinemann et al., 1999; Agag and Takeichi, 2000; Pullukat and Shinomoto, 2001; Lü et al., 2001; Wypart et al., 2002; Sun et al., 2002]. Stackhouse et al. [2001] investigated self-catalysed in situ intercalative polymerisation within Na-MMT. The authors have shown that catalysis occurs at the clay lattice-edge where -OH groups and exposed Al+3 ions act as strong Brønsted and Lewis acid sites, respectively.
340
Rheology
3.3
Rheology
Modern rheology is expanding into new domains of research, which include parallel observations of flow behaviour along with other, flow-imposed physical properties, viz. orientation, electrical conductivity, magnetic properties, birefringence, etc. For example, small-angle neutron scattering (SANS) has been used to characterise clay dispersion in non-aqueous media [Ho et al., 2001]. More recently the focus has been on rheology performed on a progressively smaller scale [Mukhopadhyay and Granick, 2001]. The use of surface force apparatus makes it possible to study the steady-state and dynamic shear behaviour of liquid films 0.3 nm to 1 μm thick. However, the credit for the ultimate reduction of scale goes to Sakai et al. [2002]. The authors carried out steady-state and dynamic rheological measurements on a single telechelic, HS-terminated PS macromolecule (Mw = 39, Mn = 22 kg/mol). The force-extension data followed the worm-like chain model, giving the ultimate extension length, lmax = 20 to 60 nm, which agrees with the theoretical value of the macromolecular contour length. The dynamic measurements performed at frequency ν = 0.3 to 30 Hz and different extensions of the macromolecule offered a unique method for determining the inter- and intra-molecular interactions and their effects on segmental mobility.
3.3.1 Introduction Before discussing the flow behaviour of CPNC it is worth summarising information on the flow of clay suspensions. •
•
The early application of clay suspensions was for increasing the viscosity of aqueous or organic liquids, to ‘give body’ and facilitate application, e.g., in paints or greases. This has been efficiently accomplished by the ability of clay to form face-edge interactions that leads to ‘house-of-cards’ 3D (gelled) structures. The suspensions showed strong thixotropic effects, rapidly increasing with concentration up to full gelation of the system at about 5 vol%. The computational and experimental data show that molecules form layered structures on the face of crystalline solids. The first 2-3 layers show solidlike behaviour and the solid surface effect stretches a long distance in the z-direction. The z-distance where the viscosity is equivalent to that of neat liquid depends on the molecular mobility; on average it stretches to 1012 layers, but in the case of macromolecules the distance may be as large as 100 to 120 nm.
341
Clay-Containing Polymeric Nanocomposites •
The crystalline solid-induced low mobility is reflected in the reduced free volume (in comparison to the neat matrix polymer at the same T and P conditions) that leads to increased viscosity, Tg, stiffness, etc. • In CPNC there are at least three material components: polymeric matrix, clay and intercalant. The degree of dispersion and the type of structure very much depend on the interactions between the components (equilibrium thermodynamics), the intercalation kinetics, and imposed stresses/orientation (non-equilibrium effects). • Owing to anisometry of the clay platelets six types of morphology have been predicted: isotropic, nematic, smectic, columnar, house-of-cards, and crystalline. The phase diagram at thermodynamic equilibrium depends on the concentration and relative strength of interactions between the three components. Evidently, the external stress field (imposed by flow, sonification, pressure cycling, etc.) affects these structures. Once disturbed, the system takes a long time (from minutes to hours) to return to equilibrium. • Depending on the method of preparation, the CPNC is either end-tethered or non-tethered. The former resemble highly branched, ‘hairy clay platelet’ structures with ca. a thousand macromolecules attached to the clay platelet through the initial intercalant. The latter systems resemble a composite: polymer reinforced with plate-like solids whose inorganic-phase dimensions are enlarged by the intercalation and adsorption of organic molecules. • Evidently, the degree of dispersion (intercalated versus exfoliated) and clay content affects CPNC performance. • Owing to the thermo-oxidative degradability of the ever-popular quaternary ammonium intercalants at T > 200 °C, the time the material spends at these temperatures may seriously affect its composition and structure, and hence the flow behaviour. It is known that the flow of multiphase polymeric systems is sensitive to the dispersed phase size, shape, and surface characteristics as well as to the transient geometric structures and interactions between the components, i.e., the matrix polymer, clay and intercalant [Utracki, 1989; 1995; Utracki and Kamal 2002b]. Thus, rheology not only complements the traditional methods of CPNC characterisation (such as XRD and TEM), PVT measurements, permeability and mechanical testing, but in addition it provides information on the dynamic behaviour of these complex materials.
3.3.2 Multi-Phase Flow Behaviour – An Overview Rheology is a part of continuum mechanics hence the principles of continuity, homogeneity and isotropy are normally assumed to hold. The continuity principle requires that there is no discontinuity of material properties from one mathematical point to another; homogeneity demands that there is no concentration gradient, and isotropy implies that the flow does not impose orientation on the flow elements. In molten multiphase systems (i.e., in polymer alloys, blends, composites, foams and nanocomposites) these three principles are rarely valid. Thus, the rheology of the multiphase systems follows its own principles, extending the use of the general rheological dependencies. Obviously, the basic definitions of rheological functions, e.g., viscosity, η, dynamic shear moduli, G´ and G´´, dynamic shear compliance, J´ and J´´, etc., are identical. However, owing to the numerous 342
Rheology influences, viz., concentration, morphology, flow geometry, timescale, type of flow field, thermodynamic interactions between the phases, and many others, it is difficult to relate the measured rheological functions to the intrinsic physical properties of the CPNC. The major distinction between single-phase and multiphase rheometry is the effect of the flow field on the rheological response. Depending on the type and intensity (strain) of the flow field the morphology of the tested fluid may be modified. Since the structure modification is related to strain, it is to be expected that responses at high and low strains will differ. For this reason, the selected test method should reflect the final use of the data. Because of the sensitivity of morphology to the test conditions, there is a serious disagreement between predictions of the continuum-based theories and experiments, summarised in Table 51. Since morphology is the characteristic property of a given material (e.g., it affects the performance), testing the same multiphase system under different flow conditions is equivalent to testing different morphologies, thus different materials. When simulation of flow through a die is attempted, large strain capillary flow is useful, but if the material characterisation is important, low strain dynamic testing should be used [Utracki, 1988]. Flow fields affect the morphology of individual dispersed particles (orientation, dispersion, coagulation, etc.) as well as the structure of the whole deformed body (e.g., skin-core effect, weld-lines, flow encapsulation). The most efficient orientation fields are extensional. Using convergent and divergent flow one may control orientation of anisometric particles, e.g., in fibre-filled materials, in semicrystalline polymer melts, in liquid crystal polymers or in CPNC. There is less information on the flow-induced orientation of clay platelets, since due to their size these particles are less susceptible to orientation than fibres. Furthermore, their orientation seems to depend on whether they are end-tethered or not. In a strong extensional flow field the former may orient perpendicularly to the stretching direction (entangled matrix stretching), whereas the latter will normally orient parallel to the stresses. A two-stage orientation mechanism was observed in converging flow, but studies of the effect of flow on the orientation of nanofiller have just begun.
Table 51 Comparison of continuum-based predictions for simple fluid with experimental observations for multiphase fluids. Data [Utracki, 1995]. Rheological function Viscosity at vanishing deformation rate
Simple fluid
Multiphase fluid
lim η(γ˙ ) = lim η ′(ω ) = lim η E (ε˙ ) / 3
γ˙ →0
ω →0
ε˙ →0
lim η(γ˙ ) ≠ lim η ′(ω ) ≠ lim η E (ε˙ ) / 3
γ˙ →0
ω →0
ε˙ →0
Extensional viscosity
ηE = ηE (entrance pressure) [Cogswell, 1972]
ηE ≠ ηE (entrance pressure)
First normal stress difference
N1 = N1 (extrudate swell) [Tanner, 1970]
N1 ≠ N1 (extrudate swell)
343
Clay-Containing Polymeric Nanocomposites The time-temperature superposition principle, t-T, has been a cornerstone of viscoelastometry. It has been used to extend the customary 2-4 decades of frequency measurements to the whole range of variability of at least 15 decades. The t-T principle breaks down in any mixture with more than one type of relaxation time distribution, viz. in polymer blends (miscible or not), in LCP over the range of transition temperatures, or in polymer-filler system with temperature-activated association modes. In CPNC the t-T principle is expected to hold for the end-tethered systems over a limited (by the transition temperature) range of conditions. Evidently, any change of miscibility between matrix polymer, intercalant and/or compatibiliser will affect the t-T behaviour.
3.3.3 Rheology and Microrheology of Disc Suspensions The simplest rheological dependence for Newtonian suspensions is that of relative viscosity ηr, versus volume fraction of the suspended particles, φ:
( )
ηr = 1 + [η]φ + k1 [η]φ
2
([ ] )
+ K + kn−1 η φ
n
(104)
where the intrinsic viscosity, [η], depends on the rigidity and shape of the suspending particles. The first of the equation parameters, k1 = kH is known as the Huggins constant, indicating the interaction between the dispersed phase and the matrix. In shear fields the particles rotate with the period dependent on the rate of shear, γ˙ , and the aspect ratio, p: t = ( 2π / γ˙ )( p + 1 / p)
(105)
There are two conventions for defining the aspect ratio of ellipsoids. According to one school, there is a continuity of transition from one type of ellipsoid of rotation to the other and the aspect ratio is defined as a ratio of the major to minor axes; p´ = a1/a2. Thus, for prolate ellipsoids p´ > 1 and for oblate p´ < 1 with spheres being the intermediate step with p´ = 1. In this convention t-dependence predicted by Equation 105 is perfectly symmetrical, giving the same period of rotation for the rodlike ellipsoid with p´ = pr and for the disk-like ellipsoid with p´ = pd = 1/pr [Goldsmith and Mason, 1967]. However, this symmetry is not observed for the intrinsic viscosity ([η] is a measure of hydrodynamic volume). According to the second convention p is defined as the ratio of the largest to the smallest dimension, thus for rods: p = length/diameter > 1, whereas for discs, p = diameter/thickness > 1. The latter system is used in this book, thus p = p´ for fibres and spheres, whereas p = 1/p´ for discs. Owing to the rotation, discs generate the maximum resistance to flow in one position (perpendicular to the flow direction) and the minimum in another. For this reason [η] is a periodic function of two angles of orientation in flow [Goldsmith and Mason, 1967], from which one can compute the upper and lower bound as well as the time-averaged value. In their excellent chapter on microrheology the authors wrote the time-averaged intrinsic viscosity as:
[η] = ( p / 0.74)[3 ln(2 p) − 5.4]
(106)
The experimental data of [η] versus p for anisometric particles gave empirical relationships for rods, discs or hard spheres [Utracki, 1989]. For discs with p ≤ 300 the following relation was found:
[η] = 2.5 + a( p
344
b
)
−1
(107)
Rheology where a = 0.025 ± 0.004 and b = 1.47 ± 0.03 with the correlation coefficient squared, r2 = 0.9998 and standard deviation, σ = 0.622. Brodnyan [1959] adopted Mooney’s relation between relative viscosity and volume fraction of hard spheres and proposed an experimentally modified formula for prolate ellipsoids. At infinite dilution this formula has the same form as the relation between [η] and p given in Equation 107 for discs, but with a = 0.399 and b = 1.48. For freely rotating prolate ellipsoids (rods) with p > 20 the following dependence was derived [Simha, 1940; Simha, 1952; Frisch and Simha, 1956]: 2
p + [η] = 14 15 5
⎡ ⎤ 1 1 + ⎢ ⎥ ⎢⎣ 3 ln( 2 p) − 4.5 ln( 2 p) − 0.5 ⎥⎦
(108)
while for the oblate ellipsoids (discs) with p >> 1 Simha [1940] derived:
[η] = (16 / 15)( p / arctan p)
(109)
Figure 89 shows that [η] very much depends on the shape and the aspect ratio of anisometric particles. As is to be expected, the strongest enhancement of [η] with p is observed for rods (prolate ellipsoids or fibres). Oblate ellipsoids (disks) show intermediate behaviour between that of prolate ellipsoids and hard spheres. For disks the prediction by Simha virtually superimposes on that by Kuhn and Kuhn [1945] – they both form the upper bound for [η]. Goldsmith and Mason [1967] proposed a relation that gives a lower bound, while the experimental values are located in-between. The [η] is sensitive to polydispersity as well as to the regularity of shape.
Figure 89 Intrinsic viscosity for rods and discs versus aspect ratio. In the latter case the theoretical, (time-averaged) and the experimental dependencies are shown. The Einstein value for monodispersed hard spheres, 2.5 for p = 1 is illustrated by the horizontal line.
345
Clay-Containing Polymeric Nanocomposites The plotted dependencies are averages – when initially the particles are all aligned (e.g., in an electric field) and upon imposition of flow they start to rotate, the rheological response (e.g., shear viscosity or [η]) will vary with the period given by Equation 105. Ivanov et al. [1982] reported on the periodic oscillatory nature of suspension viscosity in shear flow. The above relations are valid within the high dilution region where the free rotation of anisometric particles is feasible. Detailed analysis of flow within this region is available [Goldsmith and Mason, 1967; van de Ven, 1989]. As the concentration increases the particle-particle interaction becomes stronger and the higher terms in Equation 104 have to be included. The relation breaks down once the concentration of particles exceeds the limit of possible free rotation. This limit, φ > φmax, can be calculated assuming that the discs are circular and the volume they require for free rotation is that of the encompassed sphere. For the monodispersed hard spheres and oblate ellipsoids the (experimental) maximum packing volume fraction is φm ≅ 0.62 and φmax = 0.62/p. Empirically, for p < 50 a weaker dependence was found:
φmax = (1.55 + 0.0598 p)
−1
(110)
As shown in Figure 90, the empirical dependence for φmax predicts higher values. The most probable reason for the disparity is the tendency of discs to align parallel to each other, even at low concentration (the same reason why, in suspensions, the spheres readily make doublets), thus prematurely forming the structures expected within the φ > φmax region, where the discs must adopt locally parallel orientation with the spacing dependent on concentration, d001 = a0 + a1/φ (see Figure 44 in Section 2.4). These simple relations are valid for suspensions without strong interactions. Dispersion of clays in aqueous media is complicated by the presence of electrostatic
Figure 90 Maximum packing volume fraction for freely rotating discs versus aspect ratio. Lower solid line – theoretical, assuming monodispersed particles, upper broken line – empirical, Equation 1.
346
Rheology forces, which results in the formation of 3D structures (e.g., house-of-cards) or aggregation [Jogun and Zukoski, 1996].
3.3.4 Similarity Between CPNC and Liquid Crystal Flow As discussed in Section 3.1.5.3, the equilibrium thermodynamics predicts that depending on the interactions and concentration in CPNC six phase structures are expected: isotropic, nematic, smectic-A, columnar, house-of-cards, and crystal. A similar list of phase structures is known for liquid crystal polymers (LCP): isotropic, nematic, cholesteric, smectic-A and smectic-C. However, while in CPNC platelet orientation creates these diverse structures, in LCP the focus is on the rod-like molecules or side groups with the transition between the phases dependent on temperature. Liquid crystals are defined as systems ‘that are molecularly ordered yet possess mechanical properties resembling those of fluids’ [Porter and Johnson, 1967]. Considering that it is more difficult to build disc-shaped molecules than rigidrod type, the focus of LCP has been on the latter systems. Furthermore, there are significant difficulties in developing theoretical treatment for LCP with disc moieties [Ciferri, 1991]. To simplify interpretation of the flow behaviour, rheological tests are usually carried out either within a specific phase or at the phase boundary. There are numerous publications describing behaviour of nematic LCP, modelled as a system of rods, which during the shear flow may undergo periodic rotation. Rheology of the smectic (layered grease structure) and cholesteric phases is not as well documented. However, it seems that there are similarities between the flow behaviour of nematic LCP and CPNC, especially these containing end-tethered macromolecules.
Figure 91 Three regions of flow of nematic LCP. The viscosity versus shear rate dependence as generalised by Onogi and Asada [1980].
347
Clay-Containing Polymeric Nanocomposites In nematic LCP, the morphology is characterised by local orientation within each domain, evident in rheo-optical studies. As shown in Figure 91, there are three regions of flow for nematic LCP: (I) low deformation rate shear-thinning region; (II) the (Newtonian?) plateau region; and (III) the power law shear thinning region [Onogi and Asada, 1980]. Within region I, the polydomain structure dominates the flow. The authors remarked that here the rheological response is highly variable, depending on the history of the sample, method of preparation, wall temperature, etc. Depending on these, the size of the domains and the extent of the interdomain interaction varies. Rheo-optics indicates that as the rate of shear increases the intensity of transmitted polarised light also increases. However, only within region II does the polarised light show systematic oscillations of intensity, indicating rotation of the nematic domains. Thus, while in region I there is strong interaction between nematic domains that rheologically mimic the yield stress, within region II the domains are dispersed in a monodomain continuous matrix, which also dominates region III. In this region, initially domain flow proceeds by the tumbling motion, but as the deformation rate increases the tumbling is replaced by flow alignment. One of the most intriguing characteristics of nematic LCP is the behaviour of the first normal stress difference, N1. Experiment and theory indicate that for the flow of nematic LCP there is a region of the deformation rates where N1 < 0 [Kiss and Porter, 1980; Marrucci, 1991]. The change from the tumbling to flow-aligned stationary monodomain flow takes place at shear rates in the middle of the negative N1 range. Theory also predicted negative values of N1 for the transitory response after start-up at low shear rates. Owing to the diversity of CPNC structures one must be careful in comparing their flow to that of nematic LCP. In LCP the structure is defined by the molecular constitution and temperature, while in CPNC it depends on composition, degree of tethering and exfoliation. Thus, similarity of flow behaviour is expected for the nematic LCP and diluted end-tethered CPNC, but not for the CPNC where platelets enrobed in intercalant are dispersed in a polymeric matrix. The three regions of flow observed in LCP have also been observed in end-tethered PA-6/MMT systems, but the negative values of N1 were not found [Utracki and LyngaaeJørgensen, 2002]. The reason may be greater difficulty in tumbling flow for the hairy clay platelets, HCP, the heterogeneity of sizes and structures responsible for a broad transition range, inherently weaker effects of the aspect ratio for platelets than that for rods, unsuitable timescale of the experiment, experimental difficulties, etc. Another similarity in the rheological response between CPNC and LCP is the transient behaviour at start-up [Metzner and Prilutski, 1986]. During the steadystate shearing of hydroxypropyl cellulose solution in a cone-and-plate geometry the shear stress, σ12, overshoot reached a value of 20 Pa then decreased to 12.5 and after 250 s of shearing it increased to a steady-state value of 22 Pa. Even stronger and slower responses were reported for N1. Over the years many studies of stress overshoot have been reported. It was noted that the magnitude depends on the shear history – the longer the specimen is undisturbed the larger is the stress overshoot (up to a limit). Two mechanisms could account for this: (1) orientation of anisometric particles and (2) interaction between the domains. However, the consensus seemed to emerge that the latter process, a disruption of the polydomain structure during flow, is the correct one [Viola and Baird, 1986]. The latter authors carried out in 348
Rheology parallel rheological and WAXS studies of stress growth, interrupted stress growth, stress growth after flow reversal and stress relaxation. In spite of large rheological effects, WAXS provided strong evidence that orientation effects were absent during the shear flow. The extensional flow produced orientation, but its changes occurred on a different timescale to those in the stress field. Thus, the disruption of the interactive domain structure was proposed as the stress overshoot mechanism. Pitches are notoriously complex comprising at least 461 identified chemical components, with molecular weights ranging up to 2 kg/mol [McNeil, 1983]. Owing to unsaturation, some of the molecules have a rigid rod structure, while some others (formed by fusing 4 to 10 aromatic rings) are discotic, hence their structure may be considered intermediate between the rod-like LCP and CPNC. The presence of these anisometric molecules is responsible for the nematic pitch behaviour. The anisometric compound content may be adjusted by extraction or by addition of synthetic compounds. Fleurot and Edie [1998] studied the rheology of three mesophase pitches used for production of carbon fibres. The steady shear data placed them within regions I and II of the Onogi and Asada classification. The transient shear response was determined by pre-shearing the samples at γ˙ = 0.5 s-1 for 2 min then letting the sample rest for 8 to 4000 s, and then restarting the flow. The results are shown in Figure 92. For all three samples the stress growth depended on composition. The normalised peak height increased with the rest time after pre-shearing, but for different pitches the form of the dependence was different and a master curve could not be generated. The determined domain size was large, decreasing with shear rate ( γ˙ = 0.5 to 10 s-1) from a = 16 to 6 μm. The authors applied the domain theory of LCP with a qualitative success.
Figure 92 Normalised stress growth peak for three mesophase pitches versus rest time before re-shearing at the rate of shear of 0.5 s-1. After [Fleurot and Edie, 1998].
349
Clay-Containing Polymeric Nanocomposites Marrucci [1984] formulated the domain flow theory assuming a balance between the alignment tendency of the velocity field and the elastic resistance to deformation of the director field. The Eriksen distortion stress, σE, was taken as proportional to the elastic constant, K, and inversely proportional to the domain size. Noting the domain size within the polydomain region I as ao and during the flow as a, the stress associated with the recoverable (elastic) energy was calculated as:
(
σ E ~ K a −2 − ao−2
)
(111)
While the average stress is given by this dependence, the flow depends on the local orientation – high velocity is expected in the region where the orientation director and velocity vector are parallel to each other, and low velocity for the opposite case. As a result, the rheological relation between the shear stress and the deformation rate was scaled down to read:
σ = ηequilγ˙ao / ( ao − a)
(112)
Equating these two equations gives the relation between the reduced viscosity, η/ηequil, and the reduced deformation rate, γ˙ redu = γ˙ ao2 ηequil / K (see Figure 93a). Figure 93b shows the complementary dependence of the relative domain size reduction versus the relative deformation rate. The derived relation has two asymptotic limits. At low deformation rates the initial slope is predicted to be –1/2 (in contrast to the value of –1 predicted for the yield stress). This slower decrease of the reduced viscosity with shear rate is in good agreement with many observations of LCP flow. The second asymptote is obtained at higher rates, where the plateau value of region II is recovered. Thus, as simple as the derivation is, it provides sound explanations for the shape of the flow curves of LCPs. Since the basic assumption was the presence of large domains with aligned particles, this treatment may be useful for interpreting the flow behaviour of CPNC systems.
(
)
3.3.5 End-Tethered versus Non-Tethered CPNC The rheological studies on CPNC are quite recent. A relatively early publication reported on the steady-shear flow of MMT-2M2T dispersed in either diphenyl dimethyl siloxane or in polydimethylsiloxane (PDMS) [Krishnamoorti et al., 1996]. In the former system clay was intercalated, whereas in the latter it was exfoliated. Characteristically, the flow curves for both systems showed a large Newtonian plateau. As expected, the zero-shear viscosity, ηo, increased with organoclay loading, but at high deformation rates matrix viscosity was recovered. The plot of ηo versus concentration is presented in Figure 94 and that of relative viscosity ηr versus φ in Figure 95. The presence of a Newtonian plateau indicates that there is no 3D structure in these systems; hence the interactions between organoclay particles are weak. However, the ηr versus φ dependence in the latter Figure shows that while for the exfoliated system (up to 13 wt% clay) the dependence is typical for diluted suspensions with high value of the intrinsic viscosity, [η] = 9.3, the intercalated system gives evidence of interactions at a loading of ca. 4 wt%. Furthermore, its value of [η] is lower than that calculated for the exfoliated system, indicating lower apparent aspect ratio, hence stack formation. Such behaviour is typical for suspensions. The nanoparticles, exfoliated or not, behave as a filler – no surprise here. Krishnamoorti et al. also carried out preliminary studies on the flow behaviour of CPNC with PA-6 as matrix. More information on the topic was given in the 350
Rheology
Figure 93 Predicted by the Marrucci’s domain flow theory (a) apparent reduced viscosity and (b) relative domain size versus shear rate. See text.
following publication [Krishnamoorti and Giannelis 1997]. Flow of three resins from Ube was studied. These contained 0, 2 and 5 wt% of organoclay. The authors also measured the dynamic flow behaviour for CPNC of poly-ε-caprolactone (PCL) with up to 10 wt% organoclay. For the latter system validity of the time-temperature superposition was reported. The Arrhenius plot of the frequency shift factor, aT versus 1/T did not depend on the clay loading. Thus, excepting the apparent yield behaviour at low frequencies, the matrix polymer dominated the flow. Both these systems with PA-6 or PCL as a matrix were prepared by polymerisation in the presence of intercalated clay, which engendered direct bonding between the MMT surface and the macromolecules – a hairy clay platelet (HCP) structure. The authors labelled these systems as ‘end-tethered polymer layered silicate nanocomposites’. The G´ and G´´ moduli were found to increase with clay loading. Their power law dependence in the terminal zone was different from that observed for homopolymers. At low frequencies (ω < 10 rad/s) the 351
Clay-Containing Polymeric Nanocomposites
Figure 94 Zero-shear viscosity of CPNC of dimethyl ditallow ammonium MMT dispersed in diphenyl dimethyl siloxane (intercalated) and in PDMS (exfoliated). Data [Krishnamoorti et al. 1996].
Figure 95 The data from Figure 94 re-plotted as ηr versus φ. The exfoliated clay has higher aspect ratio hence higher [η] and less particle-particle interaction than the intercalated one. Data [Krishnamoorti et al., 1996].
352
Rheology initial slope of G´ and G´´ was found to decrease with increasing concentration of organoclay. Thus for 0, 2 and 5 wt% loading: g´´ ≡ (d log G´´/d log ω)ω < 10 = 0.93, 0.80 and 0.70, respectively. These values should be compared to the expectations: for neat polymer g´ = 2 and g′′ = 1, for the LCP domain flow g´´ = 1/2, while for systems with yield stress g´´ = 0. Thus, these CPNC behaved like a solution of LCP in a molten polymer, and not like filled melts with yield stress. At high strains the platelets may be aligned, which would greatly reduce the rheological response. In short, these end-tethered systems behaved differently than could be expected from diluted suspensions (see also: Section 3.2.7). A slightly different approach was used by Wagener and Reisinger [2003]. Instead of using the initial slope of G´ and G´´ the authors calculated the ‘shear thinning exponent’ from the initial slope of the complex viscosity versus frequency dependence: n ≡ d logη*/d log ω. Evidently, the parameter is a reflection of the 3D structure. There are many mechanisms responsible for such effects (e.g., compatibilisation of immiscible mixture of polymers [Utracki, 2002a]), but in CPNC, when other factors are constant, the exponent n may reflect the degree of platelet dispersion. Indeed, for PBT with 4 wt% organoclay, empirically n correlated with Young’s modulus: Y(GPa) ≅ 2.5 – 1.2n. It is to be expected that the rheological behaviour of the end-tethered systems will differ from that where such direct bonding is missing [Hoffmann et al., 2000a]. The authors prepared two types of CPNC based on PS. In the first, the clay was intercalated with phenyl-ethyl amine, and then dispersed in PS, while in the second the clay was exfoliated with amine-terminated PS. A plot of the storage shear modulus, G´ versus reduced frequency ωaT for the neat PS and PS with intercalated clay nearly superimposed one on top of the other – addition of organoclay did not change the basic flow properties of the PS matrix. By contrast, the exfoliated, end-tethered CPNC (with amine-terminated PS) showed large increases of G´ at low ω, indicating a network formation. In the next publication, the melt flow behaviour of two exfoliated CPNCs was studied. Two types of clay: synthetic fluoromica, FM, and mineral MMT, were used [Hoffmann et al., 2000b]. First, the clays were intercalated using either protonated ω-amino dodecanoic acid (ADA) or water, then dispersed in ADA, which in turn was polymerised into PA-12. The use of ADA-intercalated clay resulted in the formation of end-tethered CPNC with platelets chemically bonded to the matrix, but with surprisingly small interlayer spacing, d001 ≤ 2 nm. The use of water as a swelling agent resulted in exfoliated nanocomposites, but without the chemical bonds between clay and the polymer chains. A stress-controlled rheometer was used in the parallel plate geometry at ω = 1 to 25 rad/s assuring linear viscoelastic behaviour. Comparison of the neat PA-12 matrix flow with those for the two CPNC showed significant differences. The presence of a superstructure was deduced from the low frequency behaviour. The slopes of G´´ in the terminal region were: g´´ = 0.4 to 0.6 for the end-tethered and 0.9 for the not-tethered CPNC, thus lower than those of PA-12 with similar molecular characteristics. Furthermore, flow of CPNC with exfoliated but nontethered clay platelets, was dominated (at least up to 4 wt% organoclay loading) by the matrix behaviour, with only a minor contribution from clay. Tethering dramatically enhanced the storage and loss shear moduli, by one and by one-half decade, respectively. This observation is particularly important since XRD showed 353
Clay-Containing Polymeric Nanocomposites that the end-tethered CPNC reached only a moderate level of intercalation. Evidently, the end-tethering has stronger influence on flow than exfoliation. During injection moulding of these CPNC the clay platelets became oriented in the injection direction [Kim et al., 2001b]. There is also evidence that significant modification of the flow behaviour (in comparison to classical filled systems) is possible without end-tethering of the matrix macromolecules. One of the examples comes from Ren et al. [2000]. The authors dispersed up to 9.5 wt% MMT (intercalated with 2M2ODA) in a matrix of polystyrene-b-polyisoprene (PSIR, Mw = 17.7 kg/mol). Linear viscoelastic flow was studied at T = 80-105 °C. The t-T superposition required simultaneous horizontal and vertical shifting of the shear moduli: bTG´ and bTG´´ versus aTω. While aT for all compositions followed the same WLF dependence, bT did not – there was a dichotomy between the matrix copolymer and CPNC sets of data, indicating different macromolecular structure engendered by the organoclay. XRD indicated intercalation with d001 expanded from 2.25 nm (in organoclay) to 3.45 nm. Thus, these nanocomposites are neither end-tethered nor exfoliated. The stress relaxation data in the terminal zone (see Figure 96) showed a solidlike behaviour of these CPNC samples. The effect was particularly pronounced at ≥ 6.7 wt% organoclay, resembling that observed for the exfoliated end-tethered nanocomposites. The solid-like behaviour was evident plotting G´ versus ω – at low frequencies the storage modulus was nearly constant with g´ ≅ 0. The authors attributed this behaviour to the presence of stacks of intercalated clay platelets, each stack randomly oriented, but forming a 3D network. A large-amplitude oscillatory shear was able to orient these structures and reduce the solid-like behaviour. However, that explanation may be partial. Dispersion of the same organoclay (MMT-2M2ODA) in polyisoprene (IR) did not change d001; hence there was no diffusion of IR macromolecules into the interlamellar galleries. Evidently, during the dispersion of MMT-2M2ODA in PSIR it was the PS block that diffused into the interlamellar galleries, expanding them by Δd001 = 1.2 nm. This intercalation may suggest that there is a certain degree of interaction between
Figure 96 Stress relaxation, G(t), for PSIR and its CPNC. The solid lines were computed from dynamic moduli, G´ and G´´. Reprinted with permission from [Ren et al., 2000]. Copyright 2000 American Chemical Society.
354
Rheology MMT and the aromatic rings of PS. As discussed above, the non-tethered clay platelets in the PS matrix showed limited influence on the CPNC flow. The dramatic modification of the rheological behaviour shown in Figure 96 suggests that in the copolymer clay is preferentially dispersed in PS-block domains, phase separated from those of IR-blocks. The data in Figure 96 also illustrate that the simple relation developed by Ferry [1980] for a single component melt is also valid in these rheologically complex systems: G(t ) t =1 / ω = G ′(ω ) − 0.4G ′′(0.4ω ) + 0.014G ′′(10ω )
(113)
Sometimes, the end-tethering may be replaced by strong interactions between polar groups and the clay surface. Schmidt et al. [2000, 2002a] measured birefringence and small angle neutron scattering (SANS) during Couette shear flow of clay suspensions in aqueous polymer solution. Thus, 3 wt% of synthetic hectorite (FH; platelets with diameter d = 30 nm and thickness h ≅ 1 nm) was dispersed in an aqueous solution of 2 wt% polyethylene glycol (PEG, Mw = 103 kg/mol) at pH = 10 and a NaCl concentration of 1 mmol/L. The birefringence indicated mechanical coupling between clay platelets and PEG. At low rates of shear, γ˙ < γ˙ critical ~ 30 s −1 , birefringence was dominated by the clay platelets, but at high by the polymer chains stretched in the flow direction. SANS data indicated that at γ˙ > γ˙ critical the flow was strong enough to induce chain orientation. The clay platelets (within aggregates with diameter d = 32-233 nm) were oriented in the flow direction with the surface normal in the neutral (not radial) direction. In a later publication the authors used SANS to study the effects of shear on clay platelet orientation [Schmidt et al., 2002b]. As the rate of shear increased the clay platelets progressively oriented. At the highest flow rates the macromolecules became stretched in the flow direction. Lim and Park [2000; 2001] prepared CPNC by dispersing up to 10 wt% of MMT (intercalated with 2M2HT; Cloisite® 6A) in either PS or in PS-co-MA (containing 7 wt% MAH). Compounding in an internal mixer increased d001 from 2.94 nm (Cloisite) to 3.45 and 3.37 nm for PS and PS-co-MA, respectively. In spite of the small difference of the interlayer spacing the low frequency dynamic moduli showed large differences, viz. G´(10% clay, ω = 0.02) ≅ 0.1 and 6 kPa, with the corresponding differences in the initial slope of g´ = 0.59 and 0.24, respectively. Shearing these highly loaded samples at γ = 120%, ω = 1 rad/s for 30 min reduced the effects of association, but did not eliminate them entirely. The cited examples demonstrate that instead of a sharp distinction between the classical, filler-like influence of organoclays and the effects assigned to the end-tethered CPNCs, there is a continuous spectrum of rheological behaviour. The rheological response is associated with the ability of dispersed particles or aggregates to interact with each other. In the presence of electrostatic interactions in aqueous media the platelets readily form edge-face ‘house-of-cards’ structures, even at low loading of ca. 5 wt%. However, in organic media there is a double layer of semi-solid organic molecules that shields the silicate surfaces. Here, the formation of 3D structures is possible by two mechanisms – crowding, as in classical composites, or by entanglements. In the latter case the strongest effects are to be expected when the terminal groups directly bond macromolecules to the clay surface. However, entanglements may also be promoted by the use of block copolymers or by forming associations between the clay surface and polar groups of the polymer chain. Further examples of the effects of tethering will be found in the following parts. 355
Clay-Containing Polymeric Nanocomposites
3.3.6 Fourier-Transform Rheology of CPNC Application of the Fourier transform methods in rheology is at least 30 years old [Wapner, 1971; Wapner and Forsman, 1971]. While the authors were interested in extracting the true linear viscoelastic response from vibrating reed experiments, their approach was general, incorporating all possible modes. During the last few years Fourier-transform rheology (FTR) has gained more prominence mainly as a tool for the analysis of non-linear viscoelastic response in polymeric systems subjected to large amplitude oscillatory shear (LAOS). FTR is capable of detecting and measuring the higher harmonics that in the past were only qualitatively characterised by, e.g., the Lissajou stress-strain loops [Wilhelm, 2002]. In view of the complex rheological behaviour of CPNC it seems that application of this method may provide a suitable tool for the characterisation of the non-linear viscoelastic response in CPNC as well as for measurements of the kinetics of orientation effects. FTR analysis starts with writing the absolute magnitude of the shear rate in the dynamic flow field, with frequency ω and strain amplitude Ao, as:
γ˙ (t ) = ωAo cos ωt
(114)
The Fourier transform of this dependence yields an expression with only even harmonics:
γ˙ (t ) ∝ ao + a1 cos 2ωt + a2 cos 4ωt + K
(115)
Thus, the stress being a product of rate of shear and shear-dependent viscosity is expressed as a sum of odd harmonics:
σ ∝ A1 cos ωt + A3 cos 3ωt + A5 cos 5ωt + K
(116)
The simplest method for the analysis of the FTR signal is to plot the relative magnitude of the odd harmonic peaks divided by the first: Rn(ω) = I(nω)/I1(ω), with n = 3, 5, 7, … For the linear viscoelastic fluids Rn(ω) = 0, whereas for nonlinear materials Ri(ω) = 1/n, thus the strongest third harmonic, R3(ω), contains all the important information. Wilhelm demonstrated that the strain dependence of R3(ω) follows two simple relations:
[
{
}]
R3 (ω ) = R 1 − exp −( γ − γ L ) / k ; γ > γ L
(117)
⎡ ⎤ 1 ⎥; γ > γ L R3 (ω ) = R ⎢1 − (118) ⎢ 1 + ( Bγ )C ⎥ ⎣ ⎦ In these relations R is a measure of the maximum intensity, γ is the applied strain, γL is the maximum strain for the linear viscoelastic response, while k, B and C are parameters characterising the relative intensity change of the 3rd harmonic peak. The former relation, Equation 117, has been used to determine the limit of the viscoelastic linearity, whereas the latter provides a better fit to data within the full range of strains.
3.3.7 Rheology of CPNC with PA Matrix Krishnamoorti et al. [1996] studied the flow of two end-tethered CPNCs prepared by in situ polymerisation (in the presence of pre-intercalated MMT) of 356
Rheology
ε-caprolactone (PCL) and ε-caprolactam. In these systems the polymer chains were end-tethered to the silicate surface via cationic intercalants. The linear viscoelastic properties, the large amplitude oscillations and orientation were examined. Unfortunately, the molecular weight varied with silicate loadings which made evaluation of the data a bit difficult. The t-T superposition principle was found to be valid and the master curves could be prepared in the whole range of concentration (up to 10 wt% clay). The authors reported that the nanocomposites were readily aligned at large amplitude oscillatory shear. The alignment resulted in a change of slope within the terminal zone as well as in reduction of the dynamic moduli by about one order of magnitude. Furthermore, a steep increase of the complex viscosity, η*, was reported during strain sweeps at low frequencies, ω = 1 or 3 rad/s. Since the tanδ = G´´/G´ was reduced, the effect was related to increases of G´, caused by enhanced interactions between the flow domains. The industrial production of CPNC started with commercialisation of PA-6 nanocomposites by Ube (on a license from Toyota). In spite of thirty-odd years of technology developments there have been relatively few publications dedicated to melt flow studies of these systems. By contrast, several processes for the manufacture of PA-6 nanocomposites have been published [Okada et al., 1988; Deguchi et al., 1992; Okada and Usuki, 1995; Ube Ind., 2000]. The Toyota/Ube process involves high temperature ring-opening polymerisation of ε-caprolactam in the presence of MMT pre-intercalated with ω-amino dodecanoic (lauric) acid (ADA). The molecular weight of the PA-6 matrix is about Mn = 22 kg/mol and the organoclay content 2 wt%. The composition and properties of natural MMT may vary with geographical location and local strata. However, an idealised Na-MMT unit cell can be written as: Triple layer sandwich of two silica tetrahedron sheets and a central octahedral sheet with 0.67 negative charge per unit cell
[ Al
Aqueous interlamellar layer containing 0.67 Na+ cations per unit cell
(n × H2 O) Na0+.67
3.33 Mg0.67
](
− 0.67 )
Si8 O20 (OH )4
c
Thus, the molecular weight of a MMT unit cell is Mu = 734 + water. The cation exchange capacity of the idealised MMT is CEC = 0.915 meq/g, with Na+ spaced ca. 1.2 nm apart. Assuming that every clay plate and every cation on its surface is available for the exchange reaction of Na+ for ADA and that each lauric acid group starts the polycondensation, leads to the conclusion that near the clay platelets the endtethered PA-6 chains are densely packed. Since the molecular weight of PA-6 is Mn = 22 kg/mol, the resulting clay content of the fully grafted hairy clay platelets (HCP) should be 4.74 wt%. As the clay platelets are 0.96 nm thick and their aspect ratio is p = 287 (see below), each platelet would have about 140,000 endtethered macromolecules, i.e., Mn ≅ 3×106 kg/mol of a single HCP. Thus, polycondensation that involves every Na+ on the Na-MMT surface results in formation of an entity that resembles a star-branched or dendritic macromolecule 357
Clay-Containing Polymeric Nanocomposites of a very high molecular weight. Near the clay surface the macromolecules are crowded and immobilised. Polycondensation results in CPNC containing exfoliated clay particles with PA-6 macromolecules tethered to their surface. For the reason of dyeability these CPNC need to be diluted with about twice the amount of neat PA-6. Alternatively, the grafting efficiency of the clay platelets may be reduced and HCP with Mn ≅ 1 to 3×106 kg/mol are produced. To form these materials, standard processing methods, e.g., extrusion or injection moulding, may be used. The improvement of the rigidity, tensile and flexural strength, heat distortion temperature and gas barrier properties over neat PA-6 have been rationalised on the basis of the surface effects, change in crystallinity as well as by the orientation effects imposed by the processing flow field. Kojima et al. [1994] studied the orientation and crystallinity of extruded PA-6 CPNC calendered films containing 0.18, 0.46 and 0.74 vol% of clay. XRD, TEM and DSC measurements were carried out. It was found that clay platelets as well as the γ-crystals of PA-6 had planar orientation. The orientation direction was independent of the clay content, but the degree of orientation increased with it. It seems that during the flow through a T-die followed by subsequent stretching between chilled rolls the clay platelets became in-plane oriented, then the matrix polymer crystallised with the chain axes parallel to the clay surface. The γ-form of PA-6 has four monomeric units in the monoclinic unit cell with hydrogen bonding between parallel chains. This form is often associated with the formation of extended chain crystals during processing involving elongational flow. In the following paper the authors reported on the orientation engendered by injection moulding of CPNC into an end-gated, 3 mm thick mould cavity [Kojima et al., 1995]. Here the MMT content was 2.2 vol%. XRD and TEM measurements were carried out. In Figure 97 the 002-reflection peak intensity of γ-PA-6 versus injected bar thickness, z, is shown. Three regions of orientations are evident: skin with in-plane orientation for the PA-6 macromolecules as well as for the clay platelets, intermediate with chains perpendicularly oriented to the clay platelets and the flow direction, and the centre layer where the macromolecular chain orientation is perpendicular to the clay surface, and clay platelets are primarily oriented perpendicular to the flow direction. It is interesting that during the flow through rolls that engenders film stretching or during fountain flow in the mould cavity the macromolecular chains of PA-6 get oriented in-plane and crystallise with the chain axis parallel to the clay surface. However, at lower stresses the lamellae are oriented along the clay surface with chains oriented perpendicular to it. Kojima et al. speculated that this is the natural behaviour of PA-6 chains, highly crowded near the clay surface caused by the high grafting density. More recently, Medellin-Rodriguez et al. [2001] studied SAXS/WAXS orientation of the same Ube CPNC during steady-state shearing between parallel plates at γ˙ = 60 s-1, and T = 240 °C for up to 20 min. Compression moulded film specimens, 0.25 mm thick and containing 0, 2, and 5 wt% of MMT were used. At this relatively low shear rate and at a temperature near the melting point there was a gradual change of clay platelet alignment (see Figure 98). Owing to the vorticity component in the shear stress matrix, the end-tethered clay platelets are expected to tumble with the period given by Equation 105. The increasing scattering intensity (in the direction perpendicular to the shear field) indicates
358
Rheology
Figure 97 XRD peak intensity of γ-PA-6 versus injected bar thickness, z. Three regions of lamellar orientation are shown: skin with in-plane orientation, intermediate with perpendicular orientation and central (see text). Data [Kojima et al., 1995].
Figure 98 MMT platelet orientation in PA-6/MMT-ADA nanocomposites with 0, 2 and 5 wt% of MMT-ADA as a function of the steady-state shearing time at γ˙ = 60 s -1, and T = 240 °C. Data [Medellin-Rodriguez et al., 2001].
359
Clay-Containing Polymeric Nanocomposites progressive orientation in the shear vector. Evidently, at φ < φmax the tumbling motion of the clay platelets continues, but since the motion is periodic with long residence time in the preferred direction the overall platelet orientation is in the flow direction (e.g., see Goldsmith and Mason [1967]). The authors also reported that randomisation of the clay orientation after cessation of shear is slow, substantially slower than the relaxation of polymer chains. Thus, at T = 240 °C it takes at least 12 min to randomise the platelets, whereas the PA-6 relaxation time is about 0.4 s. In consequence, crystallisation during normal processing conditions is bound to preserve the flow-induced clay orientation in the product. 3.3.7.1 Effects of Moisture The Ube PA-6 and PANC containing 0 and 2 wt% organoclay, respectively, were thoroughly studied in dynamic and steady-state shear flow [Utracki and LyngaaeJørgensen, 2002]. Prior to compounding or testing the material was dried for 48 h at 80 °C under vacuum. As shown in Figure 99, these conditions were sufficient to achieve about 97% of the equilibrium complex modulus, G*. The two commercial resins were blended in proportions of: 0, 25, 50, 75 and 100 wt% PANC. The studied samples (two ‘as received’ and five extruded) are listed in Table 52. The rheological properties of polyamides are known to depend on the measurement time, e.g., see [Khanna et al. 1996] and references cited therein. This reversible change is related to the variation of the moisture content and changes of molecular weight associated with it, i.e., to reversibility of the polycondensation-hydrolysis reaction. For these reasons, first the time sweeps were measured under a blanket of N2 at T = 240 °C, frequency ω = 6.28 red/s and strains γ = 10 and 40% for 1 h. After testing, the specimens did not show signs of oxidative reactions – they remained off-white, indicating that the increase of the
Figure 99 Complex shear modulus as a measure of the PA-6 drying time under vacuum at 80 °C. Data [Utracki and Lyngaae-Jørgensen, 2002]. The line follows the exponential dependence: log G* = ao – a1 exp{-a2/t} with ao = 2.87 ± 0.01; a1 = 0.43 ± 0.01; a2 = 1.4 ± 0.1 and the standard deviation σ = 0.02; r2 = 0.99995.
360
39.39
82.29
152.9
242.6
321.3
35.84
559.7
25
50
75
100 (PANC)
0 (PA-6)
100 (PANC)
G′′0
0 (PA-6)
PANC wt%
0.1774
0.04744
0.09600
0.09321
0.07995
0.06113
0.04082
G′′1
-1.0141e-05
0.0000
-4.2305e-06
-3.8080e-06
-2.6699e-06
0.0000
0.0000
G′′2
0.99962
0.99995
0.99971
0.99978
0.99987
0.99993
0.99996
rG′′
3773.2
1918.2
3133.3
2898.0
2560.4
2194.0
1926.9
G′′′′0
0.54600
0.65015
0.37305
0.44210
0.50421
0.48794
0.55258
G′′′′1
-2.8488e-05
-4.9298e-05
-2.2062e-05
-2.7627e-05
-3.4721e-05
-3.1387e-05
-4.5154e-05
G′′′′2
0.99888
0.99958
0.99859
0.99939
0.99970
0.99977
0.99983
rG′′
Table 52 Polynomial fit parameters for the time sweeps. The first five samples were re-extruded or compounded, the last two 'as received'. Data [Utracki and Lyngaae-Jørgensen, 2002]
Rheology
361
Clay-Containing Polymeric Nanocomposites shear moduli was caused by polycondensation. The response was found to be independent of frequency (ω = 0.1 to 100 rad/s) and strain (γ = 10 or 40%). The data were fitted to a second order polynomial, with t being the sweep time (in sec):
G ′ = Go′ + G1′t + G2′ t 2 ; R ′ ≡ dG ′ / dt = G1′ + 2G2′ t G ′′ = Go′′ + G1′′t + G2′′t 2 ; R ′′ ≡ dG ′′ / dt = G1′′+ 2G2′′t
(119)
In spite of the fact that before loading into a rheometer the specimens were well-dried, during the 1 h in the rheometer the shear moduli, G´ and G´´, increased by a factor of 2 to 5 and 1.4 to 1.8, respectively. From these changes the rates of G´ and G´´ increase, R´ and R´´, were calculated (see Equation 119, Table 52 and Figure 100). As shown in Figure 100, the rates R´ and R´´ are both positive hence the two moduli increase with time, but increasing the organoclay content caused an increase of R´ and a decrease of R´´. In other words, addition of organoclay accelerates the time-induced increase of G´, but it slows down that of G´´. This dual effect is indicative of two parallel mechanisms: (1) polycondensation and (2) interaction between the hairy clay platelets (HCP). Owing to the presence of the residual low molecular weight amines (introduced during the intercalation and/or polymerisation steps) an addition of organoclay slows down polycondensation. As the concentration of clay in CPNC increases, the interactions between the HCPs increase as well. Since these time effects were independent of the test conditions, it was concluded that the explored variables had negligible effects on orientation – the effects were chemical and thermodynamic in nature. As a consequence, the parameters listed in Table 52 were used to recalculate all the subsequent rheological data to the initial time. In consequence, the strain sweep, frequency sweep and other data reported below are freed of these time effects.
Figure 100 Concentration dependence of the initial rate (data extrapolated to t = 0) of G´ and G´´ changes. Lines represent the least-squares fit. Data [Utracki and Lyngaae-Jørgensen, 2002].
362
Rheology 3.3.7.2 Strain Effects The strain sweeps were conducted in the range γ = 0 to 100% at ω = 6.28 rad/s for about t ≤ 10 min. At this relatively high frequency the compositions containing 100, 75 and 50% of CPNC show viscoelastic non-linearity at strains of γ ≥ 12 to 20%. The strain effects on G´ and G´´ were well approximated by the KBKZ-type non-linearity expression:
[
]
G ′′(γ ) = Go′′ / 1 + G1′′γ 2f − G2′′γ 3f ; strain fraction : γ f + γ / 100
(120)
Knowing the parameters of this relation permits calculation of the strain dependence of G´ and G″ for any composition (at ω = 6.28 rad/s). Evidently, the largest strain effects are expected for the sample with the highest concentration of clay, i.e., for PANC. However, even here at γ = 50% the reduction of G´ and G″ is relatively small: 15 and 7.5%, respectively. 3.3.7.3 Dynamic Flow Curves To determine whether CPNC obeys the t-T superposition principle, the samples were first pre-sheared then scanned from ω = 100 to 0.1 rad/s at T = 230 to 260 °C. As shown in Figure 101, good superposition was achieved. Neglecting the lowest frequency data for G´ (at the limit of the equipment sensitivity) the slopes for G´ and G´´ are 1.18 and 0.90, respectively, hence already within the power law region. According to Ferry [1980] the frequency shift factor, aT, depends on the free volume fraction, f:
⎡1 1 ⎤ log aT = B⎢ − ⎥ ; f ≈ h fo ⎦ ⎣f
(121)
Figure 101 Time-temperature superposition for Ube PANC at the four indicated temperatures
363
Clay-Containing Polymeric Nanocomposites where B is the equation constant, fo is the free volume fraction at a reference temperature, To. It has been shown that f is well approximated by the SimhaSomcynsky hole fraction, h. Under ambient pressure, the S-S coupled equations yield the following dependence [Utracki and Simha, 2001a]: h = −0.0921 + 4.89T˜ + 12.56T˜ 2 ; T˜ ≡ T / T*; r 2 = 0.99999
(122)
The value of the reducing parameter for CPNC in Table 46 is: = 11307 ± 54. Substituting Equation 122 into Equation 121 gives: ⎡ ⎤ 1 1 log aT = B⎢ − 2 2 ⎥ ˜ ˜ ˜ ˜ −0.0921 + 4.89To + 12.56To ⎥⎦ (123) ⎢⎣ −0.0921 + 4.89T + 12.56T
(
or log aT = ao + a1 / 1 + a2 T + a3 T 2
)
It is noteworthy that for a2T >> 1 (i.e., well above the glass transition temperature), the dependence leads to the Arrhenius equation. Substituting numerical values into the Equation 123 (B = 0.3817 was used) gives the prediction shown as a dotted curve in Figure 102. Both dependencies (the experimental and calculated) seem to follow the Arrhenius dependence with the activation energy of flow:
ΔHη = R[d log aT/d (1/T)] = 18 kJ/mol.
Figure 102 Temperature dependence of the frequency shift factor for PANC (see Figure 101). Comparison between the experimental and computed (from Equation 123) dependencies.
364
Rheology The highest value of log aT corresponds to T = 230 °C, i.e., about 10 °C above Tm(PA-6) = 220 °C, thus some pre-crystallisation and/or stress-induced structural change is to be expected. The frequency dependence of the dynamic viscosity (η´ = G´´/ω) is shown in Figure 103. The frequency sweeps were conducted at T = 240 °C, strains γ = 40%, from ω = 0.1 to 100 or from 100 to 0.1 rad/s. For PA and the mixture containing 25 wt% PANC, the shear moduli did not depend on the scan directions. Thus, scanning by either increasing or decreasing frequency, or scanning in the same direction but starting from different frequency, produced different rheological responses and the observed differences increased with clay loading. These shear history effects were well reproduced within ± 2% by two operators who used two rheometers and a variety of specimens. A Newtonian plateau was observed only for PA and the mixture containing 25% PANC – the higher is the clay concentration the higher is the negative slope at ω < ωc. However, even for neat PANC the slope is small, ca. –0.1, smaller than what could be expected for the domain flow. Evidently there are interactions between HCP in this region, but at a clay loading of 0.64 vol% their strength and/or intensity are relatively low. To analyse the frequency dependence of G´´ the data at γ = 10% scanned in both directions and these at 40% scanned from 100 to 0.1 rad/s were fitted to the Krieger-Daugherty type dependence, rewritten for dynamic flow [Utracki, 1988]:
[
]
(
)
− m2
2 ; ψ ≡ Gcorr (124) ′ / ω 2 = ψ o ⎡⎢1 + G ′′ / Gψ ⎤⎥ ⎦ ⎣ The dependence was derived for linear viscoelastic, pseudoplastic systems; hence it is unable to describe the yield stress. The least-squares fit of Equation 124 to data is shown in Figure 103. Excepting PA all compositions showed some ‘solidlike’ behaviour at ω < ωc ≅ 1.4 ± 0.2 rad/s. Thus, only data above ωc could be used to determine the parameters of Equation 124 (see Table 53).
η ′ ≡ Gcorr ′′ / ω = ηo 1 + G ′′ / Gη
− m1
Figure 103 Frequency scans for extruded PANC/PA mixtures at 240 °C at γ = 40% from 100 to 0.1 and from 10 to 0.1 rad/s (different specimens); points – experimental, lines – Equation 124 [Utracki and Lyngaae-Jørgensen, 2002]. Reproduced with permission, copyright Springer 2002.
365
366 0.1344±0.0567 1.3947±0.0198
0.121±0.251 1.8404±0.0019
0.9999998
0.993585
r2
CD
0.99816
0.996539
r2
CD 0.9928
0.99423
0.0209
0.6876±0.0532
1.2471±0.0298
log Gψ
0.00759
0.2577±0.0103
0.2277±0.0096
m2
σ
0.3893±0.0122
0.996715
0.9999997
0.00027±0.00323
log (ψ0)
0.00137
σ 0.00161
334±3
302±1
η0
m1 log Gη
25% PANC
PA-6
Parameter
0.99511
0.997888
0.025106
0.421±0.085
0.702±0.030
0.912±0.024
2. ψ = G′′/ω 2
0.999231
0.9999998
0.00159
0.124±0.033 0.8321±0.0136
434±4
1. η′ η′ = G′ ′ /ω ω
50% PANC
0.996069
0.9990709
0.0282760
-0.143±0.118
0.752±0.025
1.494±0.053
0.999078
0.999995
0.00230
0.136±0.046 0.5411±0.0560
495±10
75% PANC
Table 53 Frequency effects for PA-6/PANC mixtures at T = 240 °C and γ = 40%. Data [Utracki and Lyngaae-Jørgensen, 2002]
0.995322
0.998994
0.036663
-0.3892±0.134
0.816±0.026
1.864±0.069
0.999441
0.9999996
0.00203
0.150±0.056 0.4672±0.0416
543±18
100% PANC
Clay-Containing Polymeric Nanocomposites
Rheology The presence of the critical frequency, ωc, at which the HCP associations vanish, is worth commenting. The transition is independent of the clay content. Thus, it is related to the relaxation time of the aggregates, or by analogy to the LCP-type flow, to the domain size. The effect is associated with the formation of a structure that, at the selected level of T and γ, breaks at ω = ωc. This corresponds to the relaxation time of τaggr = 4.5 s, i.e., significantly longer than that for PA-6: τPA = 0.2 s. Identifying ωc with the transition from Region I to Region II of the LCP-type behaviour and using the Marrucci [1984] relation:
γ˙ ≅ K / ao2 ηequil
(125)
with the elastic constant K ≅ 10 (N), leads to the estimated value of the domain size for the three compositions (PANC, 75 and 50%) of ao = 310 nm. This may be a coincidental agreement, but as will be shown below the aspect ratio, p = 296, was calculated from the barrier properties and suspension viscosity – the value gives the bare platelet diameter d = 275 ± 9 nm, quite close to the domain size in CPNC flow. It can be shown that for incompressible, linear viscoelastic liquids there is an interrelation between G´ and G´´ through the relaxation spectrum, H(λ): -11
∞
G′ / ω 2 =
λH (λ )dλ
∫ 1 + (ωλ )
2
∞
; G ′′ / ω =
0
H (λ )dλ
∫ 1 + (ωλ )
2
(126)
0
Since these functions are valid in the whole range of relaxation times, they are also valid within narrow ranges inside this interval, say from λ = t to (t + Δ), thus:
() (G ′ / ω ) = Δ 1 + ωξ ξH ξ
2
( )
2
= ξ (G ′′ / ω ); t ≤ ξ ≤ t + Δ
(127)
The mean value theorem only requires that the integrals be continuous and the interval Δ small, hence the proportionality between the two moduli should be valid in the full range of the relaxation time and frequency. For Maxwell fluids this means that:
(G ′ / ω ) = λ(G ′′ / ω ) 2
or ψ (ω ) = λη ′(ω )
(128)
where λ is the main relaxation time. Thus, the simple Maxwell model represents the flow behaviour for fluids in which the ratio of parameters in Equation 124: m2/m1 = 2. A similar prediction can be obtained using other fluid models proposed by, e.g., Oldroyd, Spriggs, Bird-Carreau, Bogue, Meister, and others. However, as the data in Table 53 show, this condition is approximately observed only for PA-6 and the 25% PANC mixture – for the other systems m2/m1 ≅ 5.5, hence these systems are rheologically complex. The deviation from linear viscoelastic behaviour at ω < ωc ≅ 1.4 ± 0.2 rad/s is dramatically evident in Figure 104, where the ratio: G´/G´´ is plotted as function of G´´. The plot is suggested by the Doi and Edwards [1978] theory predicting that the log G´/G´´ versus log G´´ dependence for ‘well behaving’ liquids should follow a straight line with the slope of 1: ln(G ′ / G ′′) = ln G ′′ + ln(6Me / 5 ρRT )
(129)
367
Clay-Containing Polymeric Nanocomposites
Figure 104 Inverse loss tangent versus G´´ for PA/PANC mixtures at T = 240 °C and ω = 0.1 to 100 rad/s. Solid symbols for γ = 10%, open symbols for g = 40%. The predicted slope = 1 is also shown. [Utracki and Lyngaae-Jørgensen, 2002]. Reproduced with permission, copyright Springer 2002.
Figure 104 displays the log G´/G´´ versus log G´´ relations for all PA/PANC compositions, frequencies and strains. The predicted dependence is close to the observed only for neat PA-6. For the clay-containing samples the deviation from the expected linearity increases with the nanoparticle content, hence it most likely originates from the interparticle interactions that more strongly affect the storage than the loss modulus. In simple words, addition of tethered clay particles to a PA-6 matrix results in higher values for the stored energy than expected from a second order fluid. 3.3.7.4 Apparent Yield Stress The increasing value of G´´ as the frequency decreases below ωc = 1.4 ± 0.2 rad/s is related to the formation of a 3D structure, that may be treated as an apparent yield stress. One can extract the yield function: Y ≡ η’exp/η’lin = G´´exp/G´´lin, from which the apparent yield stress can be calculated as: σy(ω) = (Y - 1)G´´lin. Some years ago, for compatibilised polymer blends, a theory for the dynamic yield stress was proposed [Utracki 1989]. The conceptual model assumed formation of dynamic aggregates of the dispersed drops. The strength of the drop-to-drop interaction was σ yo , the relaxation time of the dynamic aggregate was τy, and the exponent u accounted for the aggregate polydispersity:
[
{
σ y (ω ) = σ yo 1 − exp −τ yω
}]
u
(130)
This model may be also applied to CPNC. For end-tethered CPNC the interacting entity is the HCP. Here the polydispersity of size is related to polydispersity of the aspect ratio, which for CPNC is narrower than that found in polymer blends hence u ≅ 1 may be postulated. 368
Rheology The yield stress function at ω < ωc is presented in Figure 105. The data are well represented by Equation 130 with σ yo = σy(φ), τy = 0.59 ± 0.03 and u ≅ 1. The solid-like structure formation starts at about 20 wt% PANC reaching a maximum value of σ yo = 23 Pa for neat PANC. It is noteworthy that the onset of the yield stress takes place at a MMT concentration of about 2.5 times lower than that calculated for the platelet maximum packing fraction. Thus, the 3D structure formation originates from the presence of the end-tethered macromolecules – it is of the chain entanglement type. 3.3.7.5 Zero-Shear Viscosity and the Clay Aspect Ratio The zero-shear viscosities (ηo) in Table 53 are a function of the (computed from PVT data) hole fraction, h. The dependence (see Figure 106) follows the general relation [Utracki, 1983; Utracki and Simha, 2001b; Utracki, 2002a]: ln ησ = ao + a1Ys ; YS ≡ 1 / ( a2 + h)
(131)
where: ησ indicates constant stress viscosity, ai are parameters and h is the hole fraction in the Simha-Somcynsky eos. For n-paraffins a1 = 0.79 ± 0.01 and a2 = 0.07 was found. For CPNC analysis the parameter a2 (which is only needed to linearise the dependence) was assumed to be zero. The data in Figure 106 show about the same rate of viscosity increase with decrease of h for CPNC as that found for n-paraffins. The values of ηo have been also used to determine the intrinsic viscosity from Equation 104. Its value, [η] = 105.5 ± 22.5 (k1 = kH = 0.52 ± 0.058, and the measures of fit: σ = 0.0674 and r2 = 0.9983) were then used to calculate the aspect ratio, p for the MMT from Equation 107, p = 287 ± 9. The value is in good agreement with that calculated from the relative permeability for oxygen, p = 286 for the PANC resin.
Figure 105 Frequency dependence of the yield stress function Y ≡ η’exp/η’lin = G´´exp/G´´lin for PA/PANC mixtures at T = 240 °C. Lines computed from Equation 130. [Utracki and Lyngaae-Jørgensen, 2002], copyright Springer 2002.
369
Clay-Containing Polymeric Nanocomposites
Figure 106 Constant stress dynamic viscosity of PANC mixtures with PA as functions of the (computed from PVT) hole fraction – see text. The zero-shear viscosity as well as dynamic viscosity at constant stress (G´´ = 50 MPa) show similar behaviour.
3.3.7.6 Flow-Induced Orientation At frequency ω > ωc and strain γ = 10% the rheological signals were the same for scans up or down the frequency. This however was not the case for γ = 40% – at these higher strains the orientation effects became important. The highest degree of orientation was expected at the highest frequency where all the flow curves for CPNC collapsed into a single dependence. To verify these expectations TEM was used on two PANC specimens, one dynamically sheared at ω = 100 rad/s and γ = 40% for 15 min, and another also inserted into a rheometer, but not sheared (see Figure 107). To determine the clay orientation, the specimens were microtomed close to the disc border (maximum shear strain) in the planar and circumferential directions of the moulded disc [Perrin, 2002]. In the first specimen the clay platelets (uniformly dispersed, about 1 nm thick) were found oriented parallel to the disc thickness (the lowest resistance to the dynamic shearing). In addition to single platelets, small stacks of 2 to 3 platelets were also observed – in the micrographs these resembled tree branches with a single Y- or a double ψ-branching, which may suggest a crystallographic fault of the mineral clay. The slices microtomed parallel to the surface showed mainly cross-cuts of the MMT platelet and few in-plane platelets (average size: 470 x 220 nm) suggesting that platelets were not perfectly aligned. The TEM of a not-sheared specimen showed a random orientation. Characteristically, in this compression moulded specimen most of the MMT were bent – more so than those in the sheared specimen. The orientation plays a major role at the start-up of rheological tests, e.g., in a steady-state or dynamic shearing. At low deformation rates (see Figure 108 (a)), e.g., γ˙ = 0.003 or 0.01 initially there is a rapid increase of the shear stress follow by a moderate increase caused by polycondensation. At higher deformation 370
Rheology
Figure 107 TEM of PANC from Ube shows ‘in plane’ orientation of MMT platelets. Orientation in not sheared specimens was random, with many bent clay platelets [Perrin, 2002].
rates (Figure 108 (b)) the signal goes through a local minimum followed by the polycondensation effect. As was the case for LCP, interrupted stress growth studies have also been carried out on these CPNC systems. The work is usually performed in three stages: (1) pre-shearing; (2) allowing the system to rest for a specific time; (3) shearing either in the same or the opposite direction. The most interesting results are obtained within the Onogi-Asada plateau Region II. Figure 109 shows the stress growth functions for two PANC specimens. Both were identically pre-sheared for 300 s at γ˙ = 0.1 s-1. Next, the shearing stopped for either 600 s (specimen (1)) or 1200 s (specimen (2)), and then they were sheared for 300 s, at γ˙ = 0.1 s-1, but in the opposite direction. The data were fitted to a distribution curve:
η = ηequil + ao t b ( a1 )
t
(132)
The parameter b provides a measure of the breath of distribution: tw/tn = 1 + 1/b or tz/tn = 1 + 2/b, etc. The parameter ao is a measure of the orientation effect, while a1 is a measure of the rate with which the overshoot dissipates, i.e., the platelets re-align with the field. The dependence very well described the observed curves – it is difficult to distinguish the experimental points from the computed lines. The maximum values from the stress overshoot experiments are plotted versus the rest time in Figure 110. The quality of fit as well as the fitting parameters are listed in Table 54. Differentiating Equation 132 provides the conditions for the local extremum of the function at tmax, from which a simple relation for the incremental increase of shear viscosity is obtained:
(
)
ln Δη ≡ ln η − ηequil = ln ao + b(ln tmax − 1); tmax = − b / ln a1
(133)
Thus, for rest times of trest = 0, 600 and 1200 s, values of Δη = 0, 124 and 171 were obtained. As shown in Figure 110, the data follow a single exponential dependence. As the peak height increases it becomes narrower, thus the value of b increases and the dispersion parameter, tw/tn decreases. 371
Clay-Containing Polymeric Nanocomposites
Figure 108 Stress growth function of PANC at T = 240 °C and at γ˙ = 0.01 (a) or 0.1 (b). In the first case shearing in the cone-and-plate tooling was for 10 s only, in the second for t = 1000 s.
3.3.7.7 Steady-State Flow Curves – Shear History Effects During the dynamic flow studies it was noted that the rheological responses varied not only with strain and frequency, but also with the frequency scanning direction. Since the effects of polycondensation were extracted prior to data plotting, these additional variations reflected changes in orientation and domain size. 372
Rheology
Figure 109 Two PANC specimens pre-sheared at γ˙ = 0.1 s-1 for 300 s: (1) relaxed for 600 s then sheared for 300 s, at the same rate, but in the opposite direction; (2) had the rest time twice as long. The curves computed from Equation 132 are virtually indistinguishable from the experimental dependence.
Figure 110 The incremental increase of shear viscosity for PANC versus the rest time after pre-shearing. As shown, the increase follows a single exponential.
The steady-state shearing at γ˙ ≤ 10 s-1 was carried out in a cone-and-plate geometry, increasing the shear rate from the initial value with ‘wait time’ between the consecutive data acquisition of Δt = 360 s. Starting at γ˙ = 0.001 s-1, initially the viscosity increased, then it decreased to the plateau value before reaching the power law dependence. When the sweep started at γ˙ = 0.01 s-1 the viscosity values 373
Clay-Containing Polymeric Nanocomposites
Table 54 Interrupted stress growth functions with flow reversal for PANC at T = 240 °C. Data [Utracki and Lyngaae-Jørgensen, 2002] Parameter
Sample (1); rest 600 s
Sample (2); rest 1200 s
Pre-shearing
Recovery
Pre-shearing
Recovery
Std. Dev., σ
0.673
3.78
1.423
1.670
Corr.Coeff. Sq., r2
0.999999
0.999960
0.999994
0.999992
Coeff. Determin.
0.99959
0.99041
0.99878
0.99881
ηequil
544.1±0.1
557.7±0.3
538.3±0.1
557.0±0.1
ao
51.29±0.30
36.82±1.25
67.91±0.66
20.35±0.40
a1
0.9663±0.0001
0.9641±0.0005 0.9708±0.0002 0.9594±0.0002
b
0.4689±0.0027
0.6467±0.0147 0.4085±0.0043 0.9459±0.0078
tw/tn=1+1/b
3.133
2.546
3.448
2.057
immediately started to decrease toward the plateau then power law regions. However, it is important to note that in these two cases the plateau levels were different by about 10%. Apparently, the initial shearing of the specimen in the first experiment for 1975 s more than in the second one was responsible for the development of larger domains yielding a higher plateau value. This effect was additional to that of the time-dependent polycondensation. Next, the flow curves were determined starting at the same γ˙ = 0.01 s-1, but changing the wait time between data points, Δt from 0 to 540 s (see Figure 111). The flow behaviour systematically varied with Δt. The slower was the sweep time, the better defined and the higher was the Region II plateau. As the figure illustrates, similarly well-defined plateaux were found by plotting N1/ γ˙ versus γ˙ for these five rate sweeps. It is noteworthy that proportionality between N1 and γ˙ was observed for LCP at low deformation rates for a concentrated LCP solution in cresol [Kiss and Porter, 1980; Moldenaers and Mewis, 1992]. Proportionality between N1 and γ˙ was also observed for colloidal suspensions, block copolymers (especially those with higher molecular weight blocks) and multi-branched star polymers [Kotaka and Watanabe, 1987; Masuda et al., 1987]. The Larson and Doi [1991] theory for polydomain flows predicts that:
(
)
N1 / σ 12 = 2 λ2 − 1
−1 / 2
(
)(
)
; λ = pd2 + 1 / pd2 − 1
(134)
where λ is a characteristic ‘tumbling’ parameter given by the domain aspect ratio, pd. For thermotropic LCP the value of this parameter is λ = 1.01 to 1.05 indicating that N1 > σ12 [Ugaz et al., 2001]. This is not the case for CPNC in Region II where the ratio N1/σ12 < 1 and consequently λ = 4.4, hence the equivalent ellipsoid aspect ration pd ≅ 1.6. At the higher shear rates, in the beginning of the power 374
Rheology
b
Figure 111 (a) Shear viscosity and (b) a ratio of the first normal stress difference and the deformation rate, N1/ γ˙ , for PANC at 240 °C. A series of the indicated wait time between the data points, Δt = 0 to 540, was used. See text.
law Region III, for LCP negative first normal stress was observed; this behaviour was not found in CPNC. It has also been reported that in solutions of hydrophobically modified alkalisoluble emulsions G´ is nearly a linear function of ω, and in concentrated suspensions of non-colloidal spheres N1 is proportional to γ˙ [Brady and Bossis, 1985; English et al., 1997]. Proportionality of N1 to σ12 (with the proportionality 375
Clay-Containing Polymeric Nanocomposites factor being characteristic of the system) was also reported for the latter system [Zarraga et al., 2000]. 3.3.7.8 Fourier Transform Analysis of CPNC The FTR method was used to analyse the rheological signals from an ARES rheometer for the two resins from Ube, PA-6 and PANC, at 240 °C. The following independent variables were used: strain, γ = 20-70%; frequency, ν = 0.1, 1.0 and 10 Hz; and the shearing time, t ≤ 1100 s. For the analysis the computer program developed by Manfred Wilhelm was used. Within this range of variables PA-6 was found to follow linear viscoelastic principles, thus the tests focused on the PANC resin. Figure 112 shows the raw data for PANC as a function of frequency. The imposed test frequency was ν = 10 Hz. As evident from the figure, the intensity of the first peak, I1, is about 100 times stronger than that of the third harmonic, I3. From such plots the relative intensity, R3(ω) = I(3ω)/I1(ω) was calculated. The resulting map is shown in Figure 113. The dependence shown in Figure 113 will not be a surprise to rheologists – it has been known that non-linearity increases with strain and frequency (note that for CPNC at 0.1 Hz the non-linearity was so small that I3 was difficult to measure). However, the value of FTR is in quantification of these influences, as well as in showing the influence of the shearing time on the evolution of 3D structure. In nanocomposites as in LCP, the term ‘structure’ encompasses two types of contribution: orientation and association. The time dependence of R3 shown in Figure 113, most likely originates from the latter effects.
3.3.8 Rheology of CPNC with PO Matrix Lim and Park [2001] studied the dynamic flow behaviour of exfoliated (according to XRD) CPNC with PE as the matrix. The system was prepared by mixing in an internal mixer poly(ethylene-g-maleic anhydride) with up to 10 wt% of Cloisite® 6A (MMT-2M2HTA) for 10 min at 210 °C. The PE-MA resin had MW = 212 kg/mol and contained 0.8 wt% MA. The dynamic flow data showed systematic increases of the dynamic moduli as well as of their slopes within the terminal region. As shown in Figure 114, the ratio G´(CPNC)/G´(matrix) strongly increases with clay loading and decreases with test frequency. The authors also oriented the CPNC (with 10 wt% organoclay), shearing it at γ = 120% (LAOS). The tests were conducted until a plateau was achieved. The results for the PE-MA system were compared with those obtained for PS and PS-MA with the same amount of Cloisite® 6A. The principal difference between these three CPNC systems was the degree of exfoliation – at 5 wt% organoclay loading PE-MA was exfoliated whereas the other systems were intercalated with d001 = 3.45 and 3.37 nm, respectively. The time to reach the plateau in the LAOS experiments with PE-MA was tplat = 7200 s, whereas for PS and PS-MA system it was four times shorter. Furthermore, the frequency scan after alignment showed that G´ for PS-system approached the matrix dependence, whereas for PS-MA the G´ versus ω dependence was intermediate between that for non-aligned CPNC and that of the matrix. The authors also reported that G´ in aligned or non-aligned CPNC with PE-MA as the matrix showed a similar dependence, especially at low frequency. As the frequency increased, the dependence slowly drifted toward that for the matrix. Judging by the relatively fast alignment of the two styrenic CPNCs, it seems that 376
Rheology
Figure 112 Fourier transform rheology, FTR, data for PANC at 240 °C, after 600 s of shearing at 70% strain with imposed frequency of 10 Hz. The harmonic peak (I3) at 30 Hz is evident.
Figure 113 Relative intensity of 3rd harmonic peak, I3/I1, for PANC at 240 °C. The plot illustrates how the non-linearity varies with strain, frequency and shearing time.
377
Clay-Containing Polymeric Nanocomposites
Figure 114 Relative storage moduli of PE-MA with Cloisite® 6A at T = 210 °C. Curves for frequency ω = 0.12 and 119 rad/s are shown. Data [Lim and Park, 2001].
in spite of high clay content the dispersed assemblies were relatively large and had low aspect ratio. The high value of tplat agrees with the image of delaminated platelets hindered in their rotation by crowding of the encompassed volumes. Furthermore the observed drift of G´ versus ω dependence after alignment toward that of the matrix indicates that alignment is easier at higher frequencies than at ω = 1 rad/s, which was used in LAOS. Considering the long period of platelet rotation (see Equation 105) the frequency most likely affected the extent of interaction between platelets without forcing them to rotate. Similar CPNCs, containing PE-MA and silicates of different aspect ratios, were studied by Wang et al. [2002]. The PE-MA was LLDPE grafted with 0.85 wt% MAH. The silicates were two organoclays (Cloisite® 20A, C20A and Laponite® SCPX2231, SCPX) and a synthetic SiO2. The organoclays were both intercalated with 2M2HTA and had the nominal aspect ratio, p = 100 to 200 and 20 to 30, respectively. The silica had spherical particles with diameter of ca. 1.8 μm. XRD did not show diffraction peaks down to 2θ = 2°; hence d001 > 4.4 nm. As it is to be expected, the initial slope, g´ = d lnG´/d lnω, a measure of deviation from linear viscoelasticity, decreases with nanofiller content and aspect ratio. Note that the matrix alone showed a lower value than that expected for a linear viscoelastic liquid (g´ = 2), indicating that the system is phase segregated, with MA domains acting as local crosslinks. The data in Figure 115 indicate that g´ decreases with clay aspect ratio and concentration. Other factors, such as the degree of exfoliation and potential association of the intercalated clay platelets have also been identified [Lepoittevin et al., 2002a,b]. Association is suspected in the system containing MMT intercalated with MT2EtOH (Cloisite® 30B). 378
Rheology
Figure 115 The initial slope of the storage modulus versus frequency dependence, g′≡ (∂lnG′/∂lnω)T for LLDPE-MA with two organoclays C20A and SCPX (with aspect ratio: p ≅ 100-200, and 20-30, respectively) and with silica (p = 1). Data [Wang et al., 2002].
While the end-tethered CPNC with PA-6 as a matrix showed rheological behaviour similar to LCP, other CPNC may show behaviour resembling that of filled polymers with yield stress. The review by Giannelis et al. [1999] provided several examples of such behaviour. For example, the steady-state shearing of a poly(dimethyl0.95 diphenyl0.05 siloxane) containing up to 35 wt% of MMT intercalated with 2M2HTA (Cloisite® 6A) showed a steep increase of low-shear viscosity with clay loading [Krishnamoorti et al., 1996]. In this case exfoliation (suppressed at high loadings) seems only to intensify the known filler effects. However, the authors also studied end-tethered PCL and PA-6 nanocomposites with exfoliated MMT. Judging by the reported log G´ versus log ω plots, at a MMT loading of 1 to 3 wt% the behaviour resembled that described in the preceding parts for PA-6 based CPNC (LCP-type), but at a clay loading of 5 and 10 wt% the CPNC behaved as a filled-system. Note that at these high loadings the platelets are crowded, unable to tumble along the flow lines. Galgali et al. [2001] prepared CPNC by melt blending PP (82-97 wt%; MFI = 3 and Mw = 300 kg/mol) with maleic anhydride grafted PP (PP-MA, Polybond 3200) as compatibiliser and Cloisite® 6A (MMT-2M2HTA). The ratio of PP-MA to organoclay was either 0:1 or 1:1. The mixtures were characterised by TEM and XRD at T = 200 °C. Dynamic flow measurements were performed at T = 200 °C and ω = 0.06 to 100 rad/s on samples annealed in a controlled stress rheometer for t = 0 to 3 h. The ‘terminal’ slope of log G´ versus log ω at the lowest frequencies, ω = 0.06 to 1 rad/s, depended on the compatibiliser and clay content as well as on annealing time (see Figure 116). The data were analysed using the linear function: go′ ≡ lim ( d log G ′ / d log ω ) = ao + a1 w PP− MA + a2 wclay + a3 t ω →0
(135) 379
Clay-Containing Polymeric Nanocomposites
Figure 116 The initial slope of the storage modulus versus frequency dependence, g′≡ (∂lnG′/∂lnω)T for PP/PP-MA/ Cloisite® 6A. Upper line for CPNC without compatibiliser (PP-MA), lower for PP-MA content the same as that of Cloisite® 6A. Data from [Galgali et al., 2001].
The data fit generated the following set of parameters: ao = 1.48 ± 0.09; a1 = -0.085 ± 0.013; a2 = -0.043 ± 0.015; and a3 = -0.028 ± 0.027 with the standard deviation σ = 0.15 and the correlation coefficient squared, r2 = 0.986. Evidently, the dependence is dominated by PP-MA then by the organoclay content, but (within the statistical error) not by the annealing time. This strong variation of the terminal slope is noteworthy considering that, according to XRD, the system was only intercalated with d001 = 3.3 nm. On the other hand, TEM showed the presence of large clay aggregates, probably along with a few exfoliated platelets. Since addition of PP-MA did not change the interlayer spacing, the compatibiliser most likely coated the exterior of the intercalated stacks of MMT platelets. Thus, the reduction of the terminal G´ slope is most likely related to the formation of interactive 3D structures, but not to LCP-like behaviour. Next, creep experiments were carried out at T = 200 °C. For CPNC with 9 wt% MMT, changing the shear stress from σ12 = 10 to 2000 Pa reduced the viscosity by nearly four decades – in the vicinity of σ12 = 1 kPa. The latter value was identified as an apparent yield stress – apparent, since even at the lowest imposed stress the viscosity was measurable (η ≅ 107 Pas). This behaviour is consistent with a 3D structure formed with interacting domains, forming aggregates with their own relaxation times (see Equation 130). Creep measurements were also carried out at σ12 = 10 Pa (for 3% and 6% clay) and 50 Pa (for 9% clay). The data were collected every 15 min for a period of 3 h during which the sample was annealed inside the rheometer. The creep compliance was significantly lower for CPNC specimens containing MA-PP. For the specimens containing 6 and 9 wt% of PP-MA and organoclay the effect increased with the annealing time. Evidently some structure build up was taking place under the low stress creep. It is noteworthy that dynamic yield stress has 380
Rheology been observed for compatibilised blends that, prior to compatibilisation, showed a regular pseudoplastic behaviour with an upper Newtonian plateau. The mechanism responsible for the CPNC dynamic yield stress may be similar to that for blends – enhanced interactions between the dispersed domains that formed a dynamic, percolating 3D network. Extracted from the creep data the zero shear viscosity, ηo, increased with the PP-MA and clay content by two to three orders of magnitude over that of the matrix resin, but the activation energy of flow, Eη = 33.4 kJ/mol, was found to be unchanged. The compatibilised CPNC also showed a solid-like response that apparently originated from the interactions between compatibilised clay aggregates. The presence of MA-PP increased the probability of 3D structure formation (the solid-like rheological response) similar to that observed for the end-tethered chains. Solomon et al. [2001] reported on the linear and non-linear rheology of CPNC with PP as the matrix. The samples were prepared by melt mixing of organoclay, PP (Mw = 246 kg/mol and Mw/Mn = 6.1), and compatibiliser (PP-MA, Mw = 92 k g/mol, Mw/Mn = 2.6 MA = 0.43 wt%). The weight ratio of organoclay to compatibiliser was 1:3. The organoclay was prepared by cation exchange of Na-MMT (d001 = 1.1 nm) with stearyl amine (C-18), tri-decyl amine (C-13), or di-tri-decyl amine (2C-13). The inorganic content in these CPNCs was varied from 1.30 to 6.17 wt%. Melt mixing was carried out in an internal mixer under N2 at T = 175 °C for 40 min. XRD measurements indicated that intercalation about doubled the original interlayer spacing in Na-MMT (from d001 = 1.1 to about 2.3 nm), but compounding with PP and PP-MA induced only small changes. As shown in Table 55, at a loading of 4.8 wt% clay, depending on the intercalant the interlayer spacing changed by Δd001 = 0 to 0.8 nm. Addition of clay significantly increased G´ and G´´, but the time-temperature superposition principle was found to be obeyed with the same horizontal shift factors (aT) as those for PP. Evidently, increasing the amount of organoclay increased the magnitude of the dynamic moduli. Furthermore, as one might have expected, the rheological response depended on the type of intercalating onium salt, but it is surprising that (see Figure 117) the increase correlates so well with the interlayer spacing. It is difficult to comprehend how such a small change of the interlayer spacing (Δd001 ≤ 0.9 nm!) could be responsible for the 40-fold increase of G´, especially since these systems were only ‘mildly’ intercalated. It may be that at the high loading (4.8 wt% inorganic clay) the interactions between clay and
Table 55 Interlayer spacing of neat organoclays and in mixtures with PP/PP-MA. Data [Solomon et al., 2001] Intercalant ammonium salt of:
Interlayer spacing, d001 (nm) organoclay
CPNC
C-13
1.9
2.2
2C-13
2.5
2.5
50/50 C-18/2C-13
2.6
2.8
C-18 (ODA)
2.1
2.9
381
Clay-Containing Polymeric Nanocomposites
Figure 117 Storage shear modulus at constant frequency (G´ at T = 180 °C and ω = 10 rad/s) versus interlayer spacing, d001, for PP with 4.8 wt% MMT intercalated with (from the bottom left) tri-decyl amine (C13), di-tri-decyl amine (2C13), a mixture of these (C13+ 2C13), and stearyl amine (C18). Data [Solomon et al., 2001].
intercalant parallel those between intercalated stacks in the PP matrix – the latter are responsible for the rheological response. To study the viscoelastic non-linearity Solomon et al. used shear flow reversals. First, a CPNC specimen was sheared for 300 s at γ˙ = 0.005 to 1.0 s-1 recording the value of the shear stress (σ12), then the flow was stopped for a time, trest, and the specimen was re-sheared at the same rate of shear, but in the opposite direction. As has been observed for LCP and CPNC with PA-6 as the matrix, the magnitude of the stress overshoot (σmax) increases with trest. The authors constructed master curves by plotting: (σmax/σ∞ - 1)/c versus trest (where σ∞ is the value of σ12 at long time and c is MMT loading). The linear scaling with c indicates that during nonlinear deformation the stress response originates in the individual clay domains, thus the network is destroyed by deformation. The stress overshoot is related to the distribution of particle orientations in CPNC and the stress-induced destruction of the network. Furthermore, the linear scale suggests that the domain size is independent of loading. It is interesting to note that, not unexpectedly, the magnitude of the stress overshoot depended on the rate of shearing (see Figure 118). The anisometric structure of the CPNC was considered responsible for the non-linear flow behaviour. CPNC of PP with organoclay and PP-MA was prepared by melt compounding [Li et al., 2003]. The ratio of compatibiliser (containing 0.31 wt% of MAH) to organoclay was 3. Strong non-linear viscoelastic behaviour was observed at organoclay loadings exceeding 2 wt%, i.e., 8 wt% of organoclay + PP-MA. The interlayer spacing only slightly expanded from that of neat organoclay, viz. from d001 = 3.5 to 4.2 nm. The strong non-linearity in the dynamic response indicated the formation of a 3D structure readily altered by pre-shearing. Similarly, as for PA-based CPNC, (see Figure 109 and Figure 110) stress overshoot was observed. 382
Rheology
Figure 118 The ratio of the maximum stress overshoot to its plateau value versus the rate of shear, γ˙ . Data [Solomon et al., 2001].
Considering the low degree of clay dispersion, it is unlikely that the origin of the observed non-linearity is the platelet/platelet interaction. Similarly, since the degree of PP-MA maleation was low, it is doubtful that it phase-separated from the PP matrix. Thus, the most probable structure responsible for the non-linear viscoelastic behaviour is the interaction between domains of organoclay tactoids embedded in a cloud of the compatibiliser, the latter bonded to the clay platelets. Okamoto et al. [2001a,d] studied the rheology of CPNC with PP-MA as the matrix. The system was prepared by melt compounding at T = 200 °C maleatedPP (PP-MA, 0.2 wt% MAH) with 0, 2, 4 and 7.5 wt% of MMT intercalated with stearyl ammonium ion (ODA). TEM showed fine dispersion of the silicate stacks ca. 193 to 127 nm long and about 5 to 10.2 nm thick, respectively. According to XRD the interlayer spacing of the organoclay increased from only d001 ≅ 2.31 to 3.24, 3.03 and 2.89 nm, respectively. The PP crystalline lamellae thickness and spherulite diameter did not depend on clay content. For the rheological study the rotating clamps, Meissner-type elongational RME rheometer was used at T = 150 °C and Hencky strain rates ε˙ = 0.001 to 1.0 s-1. The stress growth function in elongational flow, log η E+ versus log ε˙ for the three CPNC compositions showed strain hardening (SH). The latter function is defined as a logarithm of a ratio of the stress growth function in elongation to three times that for the linear viscoelastic response in shear, with both values taken at the same deformation time, t, viz.:
(
SH ≡ log η E+ / 3ηS+
)
t
(136)
SH has been observed for entangled polymers (e.g., LDPE), for highly polydispersed resins (e.g., some LLDPE where Mw/Mn = 10 to 50) as well as for the new bimodal metallocene PO. Partial crosslinking as well as dissolution of ultrahigh molecular weight polymer into its standard resin are also known to induce SH. The phenomenon is essential for several processing operations, viz. 383
Clay-Containing Polymeric Nanocomposites foaming, film blowing, blow moulding, wire coating, etc. As exemplified in Figure 119 by data for the branched polycarbonate of bisphenol-A, PK, for a single-phase polymer, plot of SH versus Hencky strain, ε = ε˙t , does not depend on the strain rate and the value of its slope is characteristic for the material. There is a striking difference in the SH behaviour for a single-phase polymer and CPNC. In the latter case (see Figure 120) only the high strain rate results follow the customary straight-line dependence. However, even in this range there is no superposition of data taken at different Hencky strain rates, ε˙ . For lower rates of straining the SH is significantly higher – it almost seems that for each strain rate there is a specific polymeric system with its own structure and rheological response. A cross-plot of SH at constant Hencky strain versus strain rate gives a simple dependence: SHε=const = ao + a1log ε˙ . One may speculate that the flow disrupts a network of interacting domains, and then it orients them. A similar enhancement of SH was reported by Kotsilkova [2002] for CPNC of PMMA with 10 or 15 wt% of smectite pre-intercalated with methyl diethyl propylene glycol ammonium ions. The nanocomposites were prepared by radical polymerisation of MMA in the presence of organoclay. The elongational viscosity and birefringence were measured at 180 °C, using a rotating clamp, optorheometer designed by Kotaka at strain rates, ε˙ = 0.01 to 1.0 s-1 at Hencky strains ε = 0.8 to 3. The author noted a correlation between SH and birefringence – as the former increased so did the latter. Owing to disturbance of the stress distribution pattern by suspended solid particles, the extensional flow of classical composites shows not SH, but its opposite, strain softening [Takahashi, 1996]. The SH reported by Okamoto et al. for CPNC signalled the basic difference in rheological behaviour between macro- and nanocomposites. Evidently, the nanosize of clay platelets may be one of the elements that differentiate these systems, but the end-tethering (engendered by interactions between clay and maleic anhydride moieties) might also play a role. Furthermore, as the authors observed, there are significant structural changes during flow. At low extensional or shear flow rates, e.g., ε˙ = γ˙ = 0.001 s-1, a ‘house-of-cards’ structure was observed under TEM. At high extensional flow rate, ε˙ = 1.0 s-1, the platelets were found oriented perpendicularly to the stretch direction. Formation of either structure requires energy input; hence increased viscosity is to be expected. Another important observation of Okamoto et al. is that the linear viscoelastic envelope of the stress growth function in shear was about one decade lower than that measured in elongation. This type of behaviour is expected from multiphase systems with yield stress [Utracki, 1995]. Furthermore, unlike single-phase polymer melts, the low deformation rate ( γ˙ = ε˙ = 0.001 s-1) stress growth functions of CPNC (in shear or elongation) increase with test time, t ≤ 300 s, not showing a tendency to reach a steady-state. The authors concluded that flow-induced internal structure is different in shear than in elongation.
3.3.9 Foaming of CPNC It has been recognised that SH stabilises flow during the processing steps that involve elongational flows, particularly during extrusion foaming. Owing to the poor SH of classical isotactic PP, the resin was difficult to foam. The CPNCs of PP-MA (with 0, 2, 4 and 7.5 wt% organoclay) discussed above were autoclavefoamed with supercritical CO2 at P = 10 MPa and T = 130.6 °C to 143.4 °C, which was well below the melting temperature of CPNC and PP-MA [Okamoto 384
Rheology
Figure 119 Typical strain hardening plot for a single-phase polymer, viz. for branched PC at 270 °C [Utracki and Sammut, 2000; unpublished].
Figure 120 Strain hardening parameter, SH, versus Hencky strain at indicated rates of extension, ε˙ , for CPNC of PP-MA with 4 wt% of C18-MMT (d001 = 3.03 nm). Data [Okamoto et al., 2001]. The absence of superposition of data for constant values of ε˙ indicates the presence of a 3D network of interacting organoclay particles.
385
Clay-Containing Polymeric Nanocomposites et al., 2001a, d; 2002; Nam et al., 2002]. The foams of PP-MA and CPNC with 2 wt% organoclay had closed cell structures with pentagonal and/or hexagonal faces. For the higher clay concentrations, the cells had closed spherical structure. The foamed CPNCs had a high cell density of 107-108 cells/ml, the homogeneity of cell size was in the range of 20-120 μm, the cell wall thickness was 5-15 μm, and the low mass density was 0.05-0.3 g/ml. Within the cell walls clay particles were biaxially aligned along the cell boundary. TEM showed that MMT platelets were oriented parallel to the wall surface. In the junction of three adjacent cells the platelet orientation near the walls was parallel to the proximate one with a random orientation in the centre of the three-cells junction. However, it seems that SH is not the only mechanism responsible for enhanced foamability of nanocomposites – it has been shown that CO2 foaming of either PMMA-based CPNC or neat PMMA geometrically constrained 75-100 μm thick films follows a similar mechanism and leads to a decrease of cell size and increase of cell density [Siripurapu et al., 2002]. The batch foaming of PP-based CPNC (clay content 2, 4, and 7.5%) was carried out in a high-pressure cell equipped with a microscope and high-speed digital video recording camera [Taki et al., 2003]. Significant reduction of the bubble size (from 155 to 34 μm) and increase of cell density (from 2.5 to 220 cells/nL) has been observed. Furthermore, the foam compression modulus increased by a factor of 6.These observations were confirmed in the sequel publications. Thus, preparation of micro- and nanocellular foams in PLA-based CPNC has been accomplished [Fujimoto et al., 2003; Ray and Okamoto, 2003]. Foaming of PLA/organoclay nanocomposites has been conducted using supercritical CO2. Compared to neat PLA foam, the CPNC foam showed smaller cell size and higher cell density, confirming that the dispersed silicate particles acted as nucleating sites for cell formation and the increase in SH stabilised the bubble growth. The biaxial flow-induced alignment of the platelets along the cell boundary also had a positive effect on the strength of the bubble wall, hindering coalescence. PCbased CPNC were also successfully foamed [Mitsunaga et al., 2003]. The relatively easy foamability of CPNC with supercritical CO2 to give materials with small cell diameter (nanofoams with 200 nm diameter bubbles have been produced) may be in part due to the high solubility of CO2 in most polymers. Even in the case of insolubility (e.g., PEG swells but does not dissolve in CO2) its presence increases the free volume content, thus macromolecular mobility, which on the one hand leads to the expansion of interlayer spacing [Zhao and Samulski, 2003] and on the other to higher foaming rates. As recent events demonstrate, CPNC foaming is close to commercialisation. For example, PU-based CPNC foams with submicron size cells combining exceptional mechanical, thermal, and barrier properties have been identified as prime candidates for high efficiency insulation in refrigerators [Domszy, 2004]. The patenting activities also reflect the high expectations from this technology. For example, a general patent for foaming CPNC was applied for [Lee et al., 2003]. Its claims are quite broad, specifying a diversity of polymers (PS, PMMA, PP, PA, PU, elastomers, and their blends), of organoclays (of MMT, HT, FH, saponite, laponite, and beidellite), incorporated in amounts ranging from ≤ 0.5 to ≤ 20 wt%, and using from ≤ 1 to ≤ 7 wt% of a blowing agent (supercritical CO2). The resulting closed or open cell foams should have cells with diameter ranging from 15 to 20 μm, and cell density from 106 to 109 cells/ml. Extrusion or batch foaming quality depends on the CO2 content, melt temperature, and pressure drop rate. 386
Rheology
3.3.10 Rheology of CPNC with PS and Styrenics Matrix CPNC with PS or a styrene copolymer as a base have been prepared by polymerisation (in solution, suspension or bulk) or by melt processing methods. Because of the amorphous matrix these CPNCs are advantageous as models for studies of, e.g., mechanical or barrier properties, without the complications due to crystallinity. However, as will be evident from the few cases discussed below, exfoliation of clay in a PS matrix is difficult and it requires careful consideration of chemistry. Furthermore, as discussed in Section 3.2, the work with PS nanocomposites is seriously complicated by the thermal decomposition of the quaternary intercalant, which in the presence of oxygen leads to formation of peroxy radicals that in turn cause degradation of the PS matrix. Thus, in spite of extensive academic studies these systems are not commercialised. Hoffmann et al. [2000a] prepared two types of CPNC by intercalating synthetic fluoromica (FM; Somasif ME-100; CEC = 0.7-0.8 meq/g) with either amine-terminated PS (ATPS; Mn = 5.8 kg/mol) or 2-phenyl-ethyl amine (PEA; Mn = 121 g/mol). The intercalation was carried out in a THF/H2O solution at 40 °C. The dried organoclays were compounded with the PS at 200 °C in a microcompounder for 5 min. According to XRD, the interlayer spacing of FM, d001 = 0.95 nm expanded after intercalation with PEA to 1.4 nm and with ATPS to > 4 nm. TEM of the CPNC showed the presence of large clay aggregates in CPNC with PEA (CPEA), and full exfoliation in CPNC with ATPS (C-PS). In the latter system it was estimated that the clay platelets of ca. 1 nm thick were about 600 nm long and 100 nm wide. The dynamic rheological measurements showed a sharp difference in behaviour between these two systems. The presence of 5 wt% clay in C-PEA caused G´ = G´(ω) to parallel the dependence of the matrix, with only a slight increase caused by the presence of organoclay acting as filler. In the exfoliated CPS system the low frequency slope in the terminal zone was about 0.5 (instead of 2, as in the former case). The authors concluded that in the presence of ATPS the clay platelets formed a network. The observed large difference in behaviour was ascribed to the length of the intercalating compound. Accordingly, one may control the degree of intercalation/exfoliation by varying the chain length of the intercalant chain. A more recent publication [Meincke et al., 2003] extended this work to a series of compositions containing 0 to 10 wt% of FM pre-intercalated with either PEA or ATPS. For either system good time-temperature superposition was obtained, with virtually identical WLF c1 and c2 constants as those for the PS matrix. Analysis of the dynamic data showed a dramatic change in the van Gurp plot (plot of arctan G´´/G´ versus G*) – the dependencies for PS and for PS with FM-PEA were virtually identical, quite different from that of FM-ATPS. Furthermore, it was observed that the plateau modulus follows the dependence: n GN0 ∝ 1 / φ PS , where φPS is the PS matrix volume fraction, and the exponent n = 0 when FM-PEA was used, and n = 2 when FM-ATPS was incorporated. The hairy platelet model (HCP) was postulated. This experimental finding confirms the theoretical conclusions (see Section 3.1.5) by Balazs and her colleagues [1999; 2000]. Their theoretical analysis of clay dispersed in a mixture of polymer with its functionalised homologue showed that the most promising strategy for developing an exfoliated system is by using long-chain end-functionalised compatibiliser. Its chain length should be smaller but comparable to that of the matrix polymer. 387
Clay-Containing Polymeric Nanocomposites The rheology of CPNC with a di-block copolymer of PS and IR (PS-IR; Mw = 17.7 kg/mol; 44 wt% PS) was of interest to Ren et al. [2000]. The authors dispersed organoclay (MMT with CEC = 0.90 meq/g, intercalated with a dimethyl dioctadecyl ammonium ion, 2M2ODA) in a toluene solution of PS. The five CPNCs had clay content ranging from 0 to 9.5 wt%. According to XRD, intercalation increased the interlayer spacing of MMT from d001 = 0.95 to 1.3 nm, whereas solution blending of PS or PS-IR increased it further to d001 = 2.1 to 2.5 nm, independently of the clay content. Thus, these CPNCs were only intercalated. Dynamic shear tests indicated that to achieve time-temperature superposition both horizontal (aT) and vertical (bT) shifting was necessary. While a plot of aT versus T was common for all compositions, a similar plot for bT degenerated into two dependencies: for PS-IR and for CPNCs. The authors also reported significant difference in the rheological responses for specimens oriented or not. Considering the small degree of interlayer expansion, at first it is surprising that the dynamic data show concentration dependence of the G´ and G´´ slopes in the terminal zone. However, the reported observations indicate that only the PS part of the copolymer can be inserted into the interlamellar galleries, with the PI block being outside. Thus, the structure of these CPNCs is of stacks of MMT platelets intercalated with 2M2ODA and PS-blocks surrounded by IR-blocks. Formation of the 3D structures responsible for the rheological behaviour in the terminal zone is due to the interaction between IR clouds. This situation very much resembles that of compatibilised immiscible polymer blends. Finally, it is worth recalling that, as shown in Figure 96, the stress relaxation data computed from the dynamic moduli span five decades and still follow the simple Ferry’s relation. The work was extended to systems containing up to 5 wt% of an organoclay, either a fluorohectorite (FH) modified with 3MODA, MMT intercalated with 2M2ODA, or laponite intercalated with 2M2ODA. According to XRD the degree of dispersion increased in the order of decreasing aspect ratio – from FH to MMT to laponite. Similarly, as for PA-6 systems [Utracki and Lyngaae-Jørgensen, 2002], here also the critical frequency, ωc, was observed – at ω < ωc viscoelastic non-linearity was observed. Fu and Qutubuddin [2000, 2001] started their work by preparing a polymerisable intercalant, vinyl-benzyl-dimethyl dodecyl ammonium (2MVBDDA), which was ion-exchanged with either Na-MMT or Ca-MMT (d001 = 4.62 and 4.0 nm, respectively). The purified and dried organoclay was dispersed in styrene and copolymerised. XRD and TEM of the samples indicated full exfoliation. However, only limited data on the rheological and mechanical behaviour were reported. The samples had a viscous gel structure with yield stress and shear thinning behaviour. Okamoto et al. [2000] used Na-MMT (CEC = 0.866 meq/g) intercalated either with oligo(oxy-propylene) diethyl methyl-ammonium chloride, [(C 2 H 5 ) 2 (CH 3 )N + (O ¯ iPr) 25 ]Cl - , or methyl- trioctil-ammonium chloride, [CH3(C8H17)3N+]Cl-. The intercalated clays (SPN and STN, respectively) were dispersed in MMA or styrene (St) via ultrasonication at 25 °C for 7 h then the monomer was polymerised. The organoclay content was 10 wt%. The mean interlayer spacing of the neat organoclays was: d001 = 4.20 and 1.81 nm for SPN and STN, respectively. The spacing expanded in monomer suspension. However, polymerisation in a MMT matrix resulted in PMMA/STN nanocomposite with 388
Rheology d001 = 2.66 nm, smaller by about 0.3 nm from the value in monomer suspension. SPN suspended in MMT was fully exfoliated, while PMMA/SPN nanocomposite showed a small shoulder at d001 ≅ 4.55 nm. For the St/SPN suspension, a small shoulder was found at d001 ≅ 3.85 nm, i.e., reduction of the interlayer spacing of neat SPN. The PS/SPN nanocomposites showed strong diffraction peaks, demonstrating that in this case polymerisation leads to ordered intercalated nanocomposites. The frequency sweeps for the suspensions showed two types of rheological behaviour. The suspensions of intercalated MMA/STN and St/SPN systems indicated solid 3D, gel-like behaviour with frequency-independent dynamic moduli. By contrast, the exfoliated MMA/SPN suspension showed strong frequency dependence. The data indicate that at 10 wt% loading in the monomer the expanded, intercalated clay stacks strongly interact with each other. However, surprisingly, the exfoliation into individual platelets seems to eliminate the interactions. For the polymerised CPNC systems only a viscoelastic temperature sweep was carried out. Kim et al. [2002] emulsion polymerised PS in the presence of Na-MMT highly swollen in water. CPNC containing 0, 2, 5 and 10 wt% of clay were prepared. Polymerisation increased d001 from 1.195 nm determined for dry Na-MMT to 1.511 nm measured for the three CPNCs. Thus, in these nanocomposites the interlamellar galleries of MMT were expanded to h = 0.55 nm hence slightly larger than estimated from the Flory diameter of the paraffinic chain (0.45 nm), but significantly smaller than that of PS (0.83 nm). As a result, the clay was poorly dispersed in the PS matrix. The flow curves, log η versus log γ˙ , showed a simple pseudoplastic behaviour, following the dependence: 1− n η = ηo ⎡1 + (λγ˙ ) ⎤ ⎣⎢ ⎦⎥
(137)
Here, ηo is the zero shear viscosity, λ is the longest relaxation time, and n is a power law exponent. From the values of ηo the intrinsic viscosity listed by the authors, [η] = 112 and the aspect ratio p = 270 was calculated. Thus, the assemblies of intercalated MMT platelets are highly anisometric. In consequence, scans conducted by increasing and decreasing the rate of shear resulted in a hysteresis loop, with the former scan data forming the upper and the latter scan data forming the lower part of the loop. The size of the loop increased with clay concentration. Three PS grades from Nova Chem with Mw = 310, 270, and 230 kg/mol were melt compounded in a TSE (T = 200 °C, screw speed of 200 rpm, feed rate of 5 kg/h) with 1 to 20 wt% of Cloisite® 10A (MMT-2MBHTA) [Tanoue, et al., 2003b]. It was found that each re-extrusion of CPNC reduced the zero-shear viscosity, ηo, by about 30%. This is particularly interesting since increasing the residence time in a TSE by a factor of up to 10 did not reduce further the CPNC matrix viscosity. Similarly, extrusion of neat PS had only a small (ca. 2%) effect on ηo. Thus, the combination of oxygen, temperature and organoclay was responsible for the degradation of PS. The rheological properties of these CPNCs were measured in the steady shear and dynamic mode at 160, 200 and 240 °C. The time-temperature (t-T) superposition was found valid, with the horizontal and vertical shift factors being nearly independent of the organoclay content and PS grade. The extrapolated zero-shear viscosity and zero-shear storage modulus slightly increase with organoclay content. According to the results of strain sweeps (γ = 0~100%, 389
Clay-Containing Polymeric Nanocomposites T = 200 °C, ω = 6.28 rad/s) for γ > 40%, the storage and loss moduli of all specimens (including PS) decreased with strain. The frequency sweeps showed that the storage and loss modulus increase with organoclay content. At low frequency (ω < 0.1rad/s), the initial slopes of the log G´ or log G´´ versus log ω (g´ or g´´, respectively) decreased with organoclay content, e.g., for 0 to 10 wt% organoclay g´ decreased from 1.9 to 1.6, and g´´ from 0.99 to 0.95. In steadystate shear flow tests in a capillary rheometer, the power law index of the shear viscosity decreased with increasing organoclay content. The extensional flow behaviour of the PS matrix and the CPNC were also studied. It was found that incorporation of the degraded organoclay resulted not in strain hardening, but in a small strain softening effect, similar to that observed by Takahashi [1996] for polymeric composites with solid particles. Sohn et al. [2003], studied PS with Cloisite® 25A (MMT-2MHTL8). A solvent (chloroform) casting method was used to prepare CPNC with 0, 2 and 10 wt% of organoclay. Upon mixing with PS the organoclay interlayer spacing increased from d001 = 1.94 to 3.27 nm. The dynamic flow measurements were carried out at 200 °C. In spite of the achieved expansion of the interlamellar galleries, the rheological data show small increases of both, G´ and G´´ moduli – a behaviour typical to composites, but not nanocomposites. Thus, similarly as in the work by Ren et al. [2000], here also only a monolayer of PS could be inserted into the organoclay gallery space. As these studies on PS-based CPNC indicate, preparation of CPNC with PS as the matrix has been largely unsuccessful. The exceptions are the systems where amine-terminated PS was used as intercalant (as in publications from Friedrich’s or from Qutubuddin’s laboratories), to directly bond to the clay surface and to form a miscible blend with the matrix. Evidently, PS is immiscible with alkylintercalants. The macromolecular diffusion into interlamellar galleries most likely originates in the tendency of aromatic molecules (i.e., benzene rings) to complex with surface cations. However, once the first macromolecule is inserted, on the one hand the attractive sides are shielded and on the other the interaction between benzene side groups and clay expose the paraffinic chain. As a consequence, the intercalated stacks of organoclay particles phase-separate from the matrix. The situation is worse when during melt compounding the intercalant decomposes and bare clay platelets re-aggregate into stacks with mineral clay spacing, e.g., d001 ≈ 1.4 to 1.7 nm.
3.3.11 Rheology of CPNC with Other Polymer Matrix Types Owing to solubility in water and polarity of statistical segments, polyethylene glycol (PEG), poly-ε-caprolactone, polyacrylics, polyvinyl esters and their hydrolysed versions, e.g., poly(ethylene-co-vinyl alcohol), have been frequently used for the preparation of CPNC in aqueous media. Recently, interest in these systems was sparked by their potential application as solid-state electrolytes. Hyun et al. [2001] studied the flow behaviour of PEG/MMT. The authors prepared these systems by solvent casting. Three organoclays: Cloisite® 15A (C15A; d001 = 2.96 nm), Cloisite® 20A (C20A; d001 = 2.47 nm), and Cloisite® 25A (C25A; d001 = 2.02 nm) comprised MMT (CEC = 0.95 meq/g; d001 = 1.23 nm) modified with either 2M2HTA (32% excess in 15A and no excess in 20A) or 2MHTL8 (25A – stoichiometric). XRD showed that dispersion of these organoclays in PEG resulted in expansion of the interlayer spacing by Δd001 = 0.52, 1.2 and 1.03 nm, respectively. Thus the resulting CPNCs were intercalated, but not 390
Rheology exfoliated – concentration had only a small effect on d001. TEM showed highly anisometric aggregates, ca. 200 nm thick and several micrometres long. Rheological properties of these systems were measured in the steady-state and dynamic shearing modes in parallel-plate geometry at 120 °C. The steadystate data were fitted to the generalised Carreau-Yasuda equation: 1− n ) / 2 2 ( (138) η = ηo ⎡1 + (λγ˙ ) ⎤ ⎣⎢ ⎦⎥ During steady-state shearing the systems showed a pseudoplastic behaviour. Addition of organoclay increased the shear viscosity – the strongest effects were observed for C15A, then C20A and the weakest for C25A. Evidently, increasing the concentration of organoclay also increased the strength of the rheological signal. From the cited values of ηo for the CPNC with C25A, the intrinsic viscosity [η] = 44 and the aspect ratio, p = 155 were calculated, confirming the TEM observations of anisometric aggregates. Shearing while increasing and then decreasing the deformation rate produced hysteresis loops, with the maximum loop height observed for the highest hs material near the rate of shear: γ˙ ~ 1/λ. The concentration dependence of the relative shear viscosity was fitted to the Krieger-Daugherty [1959] equation:
ηr ≡ η / ηφ →0
σ 12 = const
[
= 1 − φ / φmax
][]
− η φmax
(139)
For the multiphase systems, this relation must be taken at constant stress, σ12 = const. For monodispersed particles, the maximum packing volume fraction, φmax, has a unique, theoretically predicted value. For polydispersed suspensions, its value depends on the polydispersity of size, shape and orientation of the dispersed particles, thus it has to be treated as an adjustable parameter. Dynamic measurements were carried out to analyse the CPNC structure within the linear viscoelastic region at strain γ = 0.03. G´ and G´´ were measured for nanocomposites containing C25A at clay content 0 to 17 wt%. The moduli showed a monotonic increase at all frequencies. The initial slope of the storage modulus, g´ ≡ d log G´/d log ω < 2 was observed for all concentrations (including neat PEG). A solid-like behaviour was particularly pronounced at w ≥ 9 wt% of organoclay. This may indicate that within this region, the clay aggregates are unable to rotate hence they are prevented from relaxing. It was reported that organoclay enhanced thermal stability of the nanocomposites. Gelfer et al. [2002] prepared CPNC by melt blending commercial organoclays with either poly(ethylene-vinylacetate) (EVAc; 3.1 and 8.15 mol% VAc) or neutralised poly(ethylene-methacrylic acid) (PEMA; 3.1 mol% methacrylic acid). The reason for using these copolymers originated from the observation that organoclay is easier to disperse in highly polar polymers. Thus, EVAc and PEMA were used as models for studying the structure, property and processing relationships in CPNC. The organoclays were: Cloisite® 6A (C6A; d001 = 3.59 nm), Cloisite® 20A (C20A; d001 = 2.47 nm), and Nanomer I30 E (I30). They all contain MMT intercalated with (1) 2M2HTA (6A and 20A) or (2) N-tallow alkyl trimethylene diamine chloride (?) (I30 usually comprises MMT-ODA). The intercalant content was 45 wt% in C6A, and 30 wt% in C20A and in I30. XRD of the CPNC indicated intercalation with d001 = 2.1 to 4.1 nm. In all systems, Tm and crystallinity were not significantly affected by the presence of organoclays, 391
Clay-Containing Polymeric Nanocomposites suggesting that clay particles were shielded from the matrix by the intercalant molecules and predominantly confined to the amorphous phase. Small strain oscillatory experiments were carried out at constant strain amplitude (γ = 0.06), frequency 0.1 < ω < 100 rad/s and T = 120-200 °C (above Tm). Under these conditions the specimens remained stable for t < 30 min. The CPNC with EVAc as a matrix showed solid-like behaviour at small-strain oscillatory shear, but it was able to yield and flow under a steady shear – the characteristic performance of physically crosslinked systems. In contrast, the CPNC with PEMA as matrix exhibited a melt-like rheological behaviour, with a minimal contribution by the organoclays. The controlled stress rheometer was used to determine the yield stress. The authors speculated that the carbonyl groups of VAc in EVAc interact with the clay surface, resulting in physically crosslinked structures. By contrast, the interactions between PEMA and the clay were considered weak (due to repulsion between carboxyl anions and the negatively charged clay surface) preventing the formation of structures in these systems. The extrapolated ηo versus T showed an upswing at T > 200 °C, indicating physical crosslinking. The time-temperature superposition has not been discussed, but judging by the reported rheological data, its applicability to the studied systems is dubious.
3.3.12 Rheology of CPNC – A Summary In summary, the rheological studies of CPNC in shear and elongation demonstrate that even at low clay loading the flow is frequently complex. One source of complication that seldom is even considered by rheologists are the chemical changes within the system, caused by the decomposition of intercalant, which in turn affects the interlayer spacing and thermomechanical degradation of the matrix polymer. As discussed on the preceding pages, CPNCs show a range of performance that starts with the traditional behaviour of filled systems and ends with endtethered nanocomposites showing quite distinct flow characteristics. For the endtethered CPNC, at low or moderate concentration of clay platelets, the shear flow may be interpreted using the LCP theories. Following Onogi and Asada classification, three regions of flow can be clearly identified: (1) At low deformation rates – a solid-like yield stress behaviour, caused by a 3D structure. (2) At middle strain rates assemblies of clay platelets undergo either a tumbling (in shear) or stretching (in elongation) motion. (3) At high rates of deformation, the platelets become oriented in the shear direction, which causes the shear viscosity to decrease nearly to the level of the matrix (the effect is particularly evident in the steady-state flow). The end-tethered systems show the formation of 3D structures at a concentration of about 0.5 vol% clay. These structures are responsible for the non-linear viscoelastic flow behaviour, characterised by classic rheological tests or Fourier transform rheology [Wilhelm, 2002; Debbaut and Burhin, 2002]. This method is particularly well suited for quantification of the non-linear effects as a function of composition, strain rate, strain, temperature, etc. The unique character of CPNC is evident in the extensional flows. The studies on these flows lead to the conclusion that the presence of exfoliated clay platelets 392
Rheology able to interact with the matrix (e.g., end-tethered systems) results in significant enhancement of strain hardening. This effect agrees very well with the ‘hairy clay platelet’ (HCP) model of CPNC. Thus, in analogy to improved processability of some resins by blending them with branched homologues (e.g., common industrial blends of LLDPE with LDPE) one may use CPNC technology to improve film blowing, blow moulding or foaming (and microfoaming) of difficult to process resins. At high extensional flow rates, the platelets may be oriented perpendicular to the stretch direction, which causes the transient viscosity to move into the strain hardening region. Both effects are stronger for the end-tethered than the free platelets systems, especially at higher clay loading.
393
Clay-Containing Polymeric Nanocomposites
394
Nucleation and Crystallisation
3.4
Nucleation and Crystallisation
3.4.1 Introduction Polymers may crystallise when: (1) their molecules have sufficiently regular structure and mobility, (2) the temperature is: Tg < T < Tm, (3) there are nuclei present and (4) the rate of crystallisation is sufficiently high. Nucleation is the initial stage of the phase separation during which a new phase is formed on a minute amount of substance that acts as a nucleus for subsequent phase growth. In Nature, fog, rain or snow are formed through this process. Nucleation of crystals requires seed crystals, viz. self-formed nuclei, dust particles or nucleating agents. Thus, crystallisation may take place via a homogeneous crystallisation mechanism, where the molecules self-assemble into ordered entities having critical size for the crystal growth, or via a heterogeneous crystallisation mechanism, where the molecules assemble on the surface of a foreign body. In polymer technology, especially for injection moulding, the desired high crystallisation rates often require the addition of nucleating agents. A nucleating agent is a substance that forms nuclei for the growth of crystals in a supercooled polymer melt. Virtually any solid body with a high energy surface may act as a nucleating agent. However, for efficient nucleation it is necessary that the crystalline structure of the nucleating agent closely matches the crystalline structure of the polymer [Vesely, 1996]. Several specific crystalline or crystallisable substances have been developed, viz. 4-biphenyl carboxylic acid, aluminium salts (benzoate, phenyl-acetate, tert-butyl-benzoate), antimony compounds (trioxide, phosphates), sodium salts (4-methyl-valerate, benzoate, β-naphthoate, caproate, cinnamate, succinate), pigments, etc. These substances generate from 2 to 20 million nuclei per 1 mm3. The nucleating agents may preferentially induce a specific crystallographic form of the polymer. For example, addition to isotactic-PP of either 1,2,3,4-bis(3,4-dimethyl-benzylidene sorbitol) or N,N´-dicyclohexyl-2,6-naphthalate dicarboxamide, preferentially generate α-iPP or β-iPP, respectively. The nucleating efficiency depends on several independent variables, such as temperature, pressure, stress, part thickness as well as the presence of other processing additives. For this reason, during the last few years combinatorial methods have been used to study polymer nucleation and crystallisation. As stated above, an efficient nucleating agent must have a high-energy surface – the larger the specific surface, the more efficient it is expected to be. Nucleation involves initial adsorption of macromolecules on the surface. The process is particularly efficient if the foreign body is able to provide an energetic matrix for the formation of thermodynamically favourable crystalline forms. Alternatively, the crystalline cell type and size of the nucleating agent may induce a transitory crystalline polymer form that upon annealing transforms into a stable form of 395
Clay-Containing Polymeric Nanocomposites higher packing density. Several researchers have reported this behaviour for CPNC with PA-6 as the matrix. The nucleating efficiency, ϕ, may be expressed in terms of the energy ratio required to generate a nucleus in a heterogeneous nucleation over that in a homogeneous one [Dobreva and Gutzow, 1993]. The authors assumed that nucleation is the rate-determining step, while the extent of crystallinity is constant. Thus, the nucleation rate, r, depends on the degree of supercooling, ΔTp = Tm - Tc:
{
}
r = A exp − En / ΔTp2 ; En = ωσ 3 Vm2 / nk B Tm ΔSm2
ϕ ≡ En (hetero) / En (homo) T ≈Tm
(140)
where En is the energy required to form a nucleus, ω is a geometrical factor, σ is the specific surface energy, Vm is the molar volume of the crystallising substance, ΔSm is the melting entropy, n is the Avrami exponent and kB is the Boltzmann constant. According to Equation 140, the nucleating activity factor (ϕ) is given by the ratio of the slopes of ln r versus 1/ΔTp2. Note that 0 ≤ ϕ ≤ 1, with ϕ = 0 indicates the highest activity of the nucleating agent and ϕ = 1 indicates total lack of it. Nanofillers such as clays may have strong nucleating effects. The wide variety of intercalants, intercalating methods and compatibilisers may form a barrier between the high-energy clay surface and the semicrystalline polymer matrix. In this section the recorded effects of nanoclay on the crystallinity of CPNC will be summarised. Since CPNC with PA and PP matrices are of great academic and industrial interest the focus will be on these two classes of nanocomposites.
3.4.2 Fundamentals of Crystallisation The growth rate of lamellar crystal is controlled by the degree of undercooling, ΔT = Tm - Tg and the macromolecular diffusion rate towards the crystal growth front. A maximum rate of crystal growth takes place near Tmax ≈ (Tg + Tm)/2. In other words, the rate of crystal growth, G, is governed by the activation energy required to transport crystalline molecules across the solid-liquid interface (ΔE) and the work necessary to form a critical nucleus (ΔF*). The crystallisation growth rate is usually expressed as [Turnbull and Fisher, 1949]:
{
}
G = Go exp − ΔE / R(Tc − To ) exp{− ΔG * / k B T }
(141)
where Go is a constant, To is the temperature at which motions necessary for the transport of molecules through the liquid-solid boundary cease, and Tc is the temperature of crystallisation. At low supercooling the growth rate is nucleation controlled, while at high supercooling it is diffusion controlled. Inherent to this model is an assumption that ΔG* depends on the size and shape of the homogeneously formed nucleus. The crystallisation may take place if the nucleus reaches a critical size [Hoffman et al., 1976]:
[
]
ΔG* = 32σ e σTm (Tm + T ) / 2Δh f (Tm − T )
2
(142)
where σ and σe are the side and end free energies of the crystal, Δhf is the free enthalpy of fusion and Tm is the equilibrium melting temperature. The critical size for homogeneous nuclei is about 100 nm3, comparable to the macromolecular chain size. Binsbergen [1973] considered the second type of nucleation, the heterogeneous one. The author assumed that a foreign substance, e.g., a dust particle or a nucleating agent, facilitates formation of a polymeric nucleus, similar to that formed in
396
Nucleation and Crystallisation homogeneous nucleation excepting the presence of a high energy solid surface that lowers the magnitude of ΔG*. As a result the critical size of the nucleus is reduced, which in turn leads to crystallisation at lower undercooling, thus:
[
]
ΔG* = 16σ ( Δσ )σ e Tm (Tm + T ) / 2Δh f (Tm − T )
2
(143)
where Δσ is the specific interfacial free energy difference between the nucleus and the nucleating agent. The overall crystallisation kinetics of blends is often described by the Avrami equation [Avrami, 1939]:
{ }
α (t ) = 1 − exp − kt n
(144)
α(t) is the weight fraction of a crystalline part at time t, whereas n and k are equation parameters. The Avrami index (n) depends on the type and geometry of nucleation and the crystal growth, thus it may be written as n = nnucleation + ngrowth. The Avrami rate parameter k, is often expressed in terms of the crystallisation half-time, t1/2: k = (ln 2) / t1n/ 2
(145)
Derivation of the Avrami equation is based on several assumptions, such as the shape constancy of the growing crystal, the constant rate of radial growth, lack of induction time, the uniqueness of the nucleation mode, complete crystallinity of the sample, random distribution of nuclei, constant radial density, primary nucleation process (no secondary nucleation) and absence of an overlap between the growing crystallisation fronts. Furthermore, the often-cited form of the dependence is simplified – as Ga ęski [1995] pointed out, the relation is only valid for sporadic or instantaneous nucleation. The derived expressions are different for crystallisation in films and in the bulk. Pérez-Cardenas et al. [1991] developed a modified Avrami expression, separating the primary and the secondary (subscripts ‘p’ and ‘s’, respectively) crystallisation effects. Thus, the crystalline weight fraction, α, was written as:
α = αp + αs
(146)
Accordingly, the crystallisation proceeds in three steps: (I) the initial primary crystallisation, (II) mixed primary and secondary crystallisation and (III) pure secondary crystallisation. The authors expressed the weight fraction of the polymer crystallised by primary and secondary crystallisation as ζ. In consequence, the crystallisation may be described by two equations: t ⎡ ⎤ 1 − α = exp − kt n − k ′t n′ ⎢kn(1 − ς ) xp kt n + k ′t n′ τ n−1 dτ + 1⎥ ; α ≤ ς (147) ⎢⎣ ⎥⎦ 0 1 − α = (1 − ς ) exp k ′t * n′ exp k ′t n′ ; α > ς (148)
(
)
∫ (
{
)
} { }
Now, six parameters are required to describe the process: k and n (the primary crystallisation parameters) depend on crystallisation temperature, the nature of primary nucleation and the fast growth; the secondary crystallisation parameters, k´ and n´ depend on the conditions under which the slow crystallisation of the remaining amorphous regions takes place, ζ, indicates the weight fraction of material crystallised up to the moment the primary crystallisation ends and t* indicates the start of the pure secondary crystallisation. Furthermore, Avrami theory is limited to isothermal processes. Since polymer processing is mostly non-isothermal, the theory has been extended by Ozawa
397
Clay-Containing Polymeric Nanocomposites [1971] who considered the Avrami constant k to depend on temperature, k = k(T), and the crystallisation time t as dependent on the cooling rate: t = Φ–m (m is the Ozawa exponent). A more general approach to non-isothermal crystallisation was developed by Kamal and Chu [1983]: ⎧ t ⎫ α (t ) = 1 − exp ⎨− k (T )nt n−1 dt ⎬ ⎩ 0 ⎭ n −1 ⎧⎪ T ⎛T −T⎞ dT ⎫⎪ α (T ) = 1 − exp ⎨− k (T )n⎜ o ⎬ ⎟ T R⎪ ⎝ R ⎠ ⎭ ⎩⎪ o
∫
(149)
∫
where k(T) and n are Avrami’s isothermal parameters. For DSC scans the following expression has been successful:
[α ′ (T p
p
)] [ (
− Tons / β 1 − a p
)] = (n − 1) − [ E(T
p
)] [
− Tons / RTp2
where : Tp = Tons + βt and α ′p ≡ dα / dt = (1 − α )nkt (
]
n −1 )
(150)
where β is the heating rate, Tons is the onset temperature, Tp is the peak temperature reached after time t, and E is the activation energy of the crystallisation process. In a semicrystalline homopolymer, the change in free energy of melting per mole of monomer is given by:
ΔGu(T) = ΔHu - TΔSu
(151)
where ΔHu and ΔSu are the enthalpy and the entropy changes on melting, respectively. For an infinitely thick crystal at equilibrium melting temperature Tm, ΔGu(Tm) = 0 and: T m = Δ Hu / ΔS u
(152)
The value of Tm can experimentally be determined from Hoffman-Weeks plots, where the experimental melting point is plotted as a function of the crystallisation temperature, Tc. Extrapolation of experimental data to the Tm = Tc line gives the equilibrium value of Tm. Xu et al. [2001] studied the nonisothermal crystallisation kinetics of CPNC with POM or with PP [2002] as the matrix. Two CPNC with POM were prepared in a roller mill at T = 175 to 180 °C; the first comprised 5 wt% of Na+-MMT and the second MMT-3MHDA. The interlayer spacing in these systems was d001 = 1.92 and 3.52 nm, respectively. The nonisothermal crystallisation kinetics were investigated by DSC at 5 to 40 K/min cooling rate. The difference in the values of the Avrami exponent n between POM and the nanocomposites suggests a tri-dimensional growth with heterogeneous nucleation. At a given cooling rate the crystallisation of neat POM was slightly slower than that of either nanocomposite. The activation energies, 387, 330, and 329 kJ/mol were determined for the nonisothermal crystallisation of POM, POM/Na+-MMT nanocomposite, and POM/MMT-3MHD, respectively. A similar process was used to prepare nanocomposites with PP with 3 wt% MMT-3MHDA. A 3 mm thick sheet was compression moulded at 180 °C. The interlayer spacing was d001 = 3.88 nm. The crystallisation rate increased with increasing cooling rates for both PP and its nanocomposite. The crystallisation rate of CPNC was higher by 40% at a cooling rate of 5 K/min, but virtually 398
Nucleation and Crystallisation identical at the fastest cooling rate of 40 K/min. The activation energies were estimated as 189 and 156 kJ/mol for PP and nanocomposite, respectively. Thus, incorporation of 3 wt% of organoclay into PP affected the matrix crystallisation more than 5 wt% in POM. General procedures for evaluating crystallinity in polymeric systems by XRD and DSC have been developed, e.g., by Murthy and Minor [1990] and by Khanna and Kuhn [1997], respectively. Additional information on the topic can be found in [Mathot, 1994; Chan et al., 1995; Hammami et al., 1995], etc.
3.4.3 Effects of Clay on Crystallisation of PA-6 Matrix Nucleating PA by means of standard nucleating agents (e.g., silica, talc, Na-phenyl phosphate, etc.) leads to higher crystallinity that translates into higher modulus, hardness, yield strength, HDT, improved abrasion and water absorbance, but reduced elongation at break and impact strength. The characteristic properties of PA-6 and PA-66 are listed in Table 56. Both these polymers are polymorphs, the former crystallising mainly in α-form, γ-form and a metastable β-form. Historically, PA-6 is more important for CPNC technology. There is a large body of information in the patent and open literature for these systems. As far as the effects of clay addition on crystallisation of PA-6 are concerned, the focus has been on nucleation, crystalline structure and total crystallinity. In PA-6 the stable monoclinic
Table 56 Characteristic parameters of polyamides Property
PA-6
PA-66
Condition
Value
Condition
Value
Dry 50% RH
320-330 276-293
Dry 50% RH
351 308
Melting point (eq.), Tm (K)
Dry
493
Dry
542.2
Density, ρ (kg/m3)
Dry
1130
Dry
1220-1250
Typical crystallinity, α (%)
Dry
50
Dry
40-60
Unit cell dimensions (nm)
α-form γ-form
0.956×1.724×0.801 0.933×1.688×0.478
αI-form β-form
1.57×1.05×1.73 0.49×0.8×1.72
Unit cell angles (# mers)
α-form γ-form
67.5 (4) 121 (4)
αI-form β-form
73 (9) 77 (2)
all forms
Monoclinic
αI-form
Monoclinic and triclinic Triclinic
Glass transition temp., Tg (K)
Lattice
β-form
399
Clay-Containing Polymeric Nanocomposites
α-form has a planar zigzag chain conformation, the metastable (pseudo-) hexagonal γ-form has a twisted chain, while the β-form is less well identified and often considered to be an intermediate stage between the two others. The α-form is most common, readily obtained during melt processing. A nearly pure α- and γ-form may be obtained by, respectively, treating the specimen in superheated water vapour at 150 °C, and by reversibly complexing it with iodine from KI aqueous solution [Penel-Pierron et al., 2001a,b]. Uniaxial drawing increases the α-form content and reduces that of the γ-form. The tensile strength is higher for the α-form, while the γ-form shows better toughness. The solid/solid transition of α-crystal into γ-crystal, observed during heating or cooling, is known as the Brill transition. At the crystallisation temperature, Tc > 450 K PA-6 crystallises in a mixture of α- and γ-forms, whereas above this limit essentially in nearly a pure α-form [Privalko et al., 1979]. Thus, the α-form is more stable at higher temperatures. The authors reported that addition of glass beads or aerosil to PA-6 did not change the molecular mechanism of crystallisation. PA-66 crystallises in at least six forms, of which the αI- and β-form dominate. Incorporation of organoclay into PA-6 increases the nucleation density, the proportion of the γ-form in the crystal as well as modifying the total crystallinity content. While the former two effects are generally accepted the latter is more controversial – some authors report enhancement, some others reduction while newer publications indicate that both may be present within specific ranges of conditions, e.g., using a different cooling rate or annealing method. Kojima et al. [1994] used XRD and DSC to study orientation and crystallinity of the exfoliated PA-6/MMT system. The platelets of MMT and the γ-form of PA-6 crystallites were reported parallel to each other. The α-form was not detected. Incorporation of 0.2 to 0.8 vol% MMT increased the degree of crystallinity of PA-6 (to 31%) by 5 to 7%, but the effect did not depend on composition. In the following publication it was reported that in injection moulded tensile bars the orientation depended on the depth, but the γ-form of PA-6 still dominated the crystalline phase [Kojima et al., 1995]. In the surface layer of the moulding, MMT platelets and chain axes of crystalline PA-6 were parallel to the surface. In the intermediate layer the MMT platelets remained parallel to the surface of tensile bar, but the chain orientation rotated 90o (hence orienting perpendicularly to both, bar surface and MMT). In the central layer the platelets were randomly oriented along the flow axis with chain axes perpendicular to them. Usuki et al. [1995] studied CPNCs of PA-6 containing ca. 5 wt% of MMT, synthetic mica, saponite or hectorite (CEC = 1.2, 1.0, 1.0 and 0.5 meq/g, respectively). Prior to reactive exfoliation in ε-caprolactam, the clays were pre-intercalated with ω-amino lauric acid (ADA). Owing to end-tethering, the concentration of –COOH macromolecular chain ends was about twice as large as that of –NH2. The authors established correlations between the performance (e.g., tensile strength, modulus, HDT), the strength of the clay-polymer interaction (determined by the 15N-NMR chemical shift, CP) and the degree of crystallinity, X (wt%). The latter was calculated from the DSC-measured heat of fusion, taking the literature value for the heat of fusion of PA-6 crystal as 188 J/g. Surprisingly, it was found that X decreased upon incorporation of organoclay. At the same time, the mechanical performance (e.g., tensile modulus) increased with CP from 1.11 for PA-6 to 2.02 GPa for CPNC with mica. These observations are summarised in Figure 121 – note that excepting the degree of crystallinity, X (wt%), the other performance characteristics for CPNC diverge from those of PA. 400
Nucleation and Crystallisation Liu et al. [1999] prepared CPNC by melt blending PA-6 in a TSE with 1 to 18 wt% of MMT (CEC = 1.0 meq/g) intercalated with ODA. After compounding at T = 180 to 220 °C, the CPNC with more than 10 wt% of organoclay showed intercalated structures with d001 increased from 1.55 to 3.68 nm. However, at low MMT content the samples were exfoliated. In accord with the Usuki et al. [1995] results, the authors observed a difference in crystallisability between PA-6 and its CPNC – addition of any amount of clay reduced the supercooling effect to a common level, indicating enhanced nucleation. However, by contrast with the Usuki et al. [1995] results, here all specimens (with 0 to 6.7 wt% MMT) showed about the same crystallinity. The difference most likely originates in the different internal structure of these CPNC – in the former work the reactive exfoliation resulted in end-tethering, whereas compounding PA-6 with pre-intercalated MMT generated exfoliated CPNC (at lower clay content), but without endtethering. XRD of PA-6 showed a virtually pure α-crystallographic form, whereas the polymer in CPNC crystallised mainly in the γ-form. Wu and Liao [2000] prepared CPNCs of exfoliated synthetic saponite in a PA-6 matrix using the reactive route with a unique pre-intercalation step; synthetic clay was dispersed in water/ε-caprolactam/phosphoric acid solution, then the mixture was polymerised at T = 80 to 270 °C under pressure. The authors do not provide information about the platelet dispersion. XRD and DSC were used to study the effect of clay on crystallinity, which in turn was related to performance. The specimens containing 0, 2.5 and 5 wt% clay were pressed into sheets at 240 °C, then either slowly cooled or quenched to room temperature. Slow cooling
Figure 121 Crystallinity, relative tensile strength, TS/TS(PA), and relative HDT (expressed in Kelvin), HDT/HDT(PA) as functions of the 15N-NMR chemical shift, CP. Systems: PA-6 with 5 wt% of exfoliated clay – see text. Data [Usuki et al., 1995].
401
Clay-Containing Polymeric Nanocomposites resulted in nearly pure γ-form crystals in the sample containing 2.5% clay, nearly pure α-form in the sample containing 5 wt% clay and polymorphic crystals in PA-6. By contrast, quenching yielded a nearly pure γ-form for the sample with 5 wt% clay, predominantly α-form for PA-6 and intermediate polymorphous composition for the sample containing 2.5 wt% clay. To transform γ- into α-form required 2 h heating at 210 °C. Addition of synthetic saponite increased the crystallisation rate, but the crystallinity was not determined. The tensile modulus and HDT showed significant improvements. Akkapeddi [2000] used a melt compounding method for the manufacture of PA-6 based CPNC. Thus, organoclay of MMT or hectorite was dispersed in PA-6 at a level of 2 to 5 wt% of clay. The author reported that clay platelets had strong surface nucleation effects that promoted faster crystallisation and a higher level of crystallinity in relation to that of neat PA-6 (particularly at the surface and in thin-wall injection mouldings). For example, in 1 mm thick PA-6 injection moulded bars the crystallinity was 16 to 18%, while in CPNC (4 wt% clay) the crystallinity was 50% throughout the specimen. It is noteworthy that whilst in PA-6 specimens the skin had lower crystallinity than the core, in CPNC the situation was reversed. Giza et al. [2000] reported enhanced crystallinity in spun fibres of PA-6/MMT (from Ube). While the crystallinity content in drawn fibres of PA-6 was about 51%, in those containing 2 and 5 wt% clay it was 59 and 63%, respectively. Part of the reason for the difference may be the higher drawing temperature for CPNC (170 °C) compared with PA-6 (150 °C). Furthermore, the high-speed drawing that produced the highest crystallinity was also instrumental in transforming the γ-form into predominantly α-form. Increased crystallinity translated into higher modulus, viz. the tensile modulus for CPNC with 0, 2, and 5 wt% clay was about 8, 9 and 10 GPa, respectively. Intercalated or exfoliated CPNC were prepared by extrusion compounding of PA-6 with ca. 3 wt% of pre-intercalated MMT, Cloisite® 30B (C30B); d001 = 1.8 nm or Cloisite® 6A (C6A); d001 = 3.6 nm [Varlot et al., 2001]. The compounds were injection moulded into 4 mm thick bars, which subsequently were characterised by XRD. In accord with Kojima et al. [1995], the authors reported significant variation of crystalline morphology with depth. Thus in the three specimens: PA-6, PA-6 + C30B, and PA-6 + C6A, the surface layer was rich in γ-form. The central, bulk layer of the injection moulded bar of PA-6 showed preponderance of α-form, whereas γ-form dominated the two other specimens. The CPNC showed MMT platelets aligned with the flow direction. The average distance between MMT platelets in a PA-6A sample was determined from the SAXS spectra to be ca. 35 nm. The same value was calculated from the composition, assuming perfect exfoliation and uniform distribution of the platelets. VanderHart et al. [2001] used solid-state proton and 13C NMR to study 17 CPNC compositions based on PA-6, provided by Ube Ind. and SCP. Except three samples that contained laponite clay, the CPNC comprised natural, pre-intercalated MMT dispersed either by in situ polymerisation or by melt compounding. The clays were found to promote growth of the γ-form of PA-6 in all CPNC samples, while the α-form characterised the neat PA-6. The total crystallinity in neat PA-6 was 40% – all in α-form. In CPNC the crystallinity was ca. 33 to 45% with the αform ranging from 2 to 75% (after injection moulding and annealing). However, most specimens annealed at Ta = 214 °C showed only a partial conversion of γ-form 402
Nucleation and Crystallisation to α-form (e.g., from 4 to 19%). The CPNC showed higher Tc and Tm than that of PA-6. Since the main objective was to study paramagnetic effects there is little more information on crystallinity. To prepare CPNC with PA-6 as a matrix, Liu and Wu [2002a,b] used a three-step method. Thus, Na-MMT (CEC = 0.8 meq/g) was intercalated with trimethyl hexadecyl ammonium bromide (3MHDA). The precipitate of the reaction in H2O was dried under vacuum then compounded in an internal mixer with the diglycidyl ether of bisphenol-A (MW = 360 g/mol). In the last step the doubly intercalated MMT was dispersed in molten PA-6 by means of compounding in a TSE at ca. 230 °C. The product containing 5 wt% clay was used in crystallographic studies published in several articles, summarised below. More recently, the work was extended to a PA-66/MMT system [Liu and Wu, 2002a; 2003]. A higher crystallisation temperature for CPNC than that for neat PA-66 (Tc = 237 instead of 226 °C), reduction of spherulite size, the presence of a γ-phase instead of an α-phase were reported. Similar behaviour was observed for an exfoliated PA-66/MMT system [Zhang et al., 2003]. The DSC and XRD studies on the influence of the cooling rate on the crystallinity in PA-6 and CPNC demonstrated that the two materials behave differently [Wu et al., 2001a]. Thus, as the cooling rate increases the rate of crystallisation of PA-6 decreases while that of CPNC increases. In Figure 122 two separate sets of experiments are presented. The DSC data represent experiments with well-defined cooling rates, whereas those measured by XRD are plotted versus estimated ones. The dependencies are similar. The XRD data show even stronger divergence of behaviour between PA and CPNC than DSC. Furthermore, WAXS data indicated that PA-6 was highly polymorphous – only at the slowest cooling rate was nearly pure α-form obtained. By contrast, the
Figure 122 Crystallinity versus cooling rate for PA-6 and PA-6 containing 3 wt% of doubly pre-intercalated MMT. Two methods were used: DSC with well defined cooling rates and XRD with roughly estimated cooling rates. Data [Wu et al., 2001]. See text.
403
Clay-Containing Polymeric Nanocomposites CPNC diffraction showed nearly pure γ-form – only at the slowest cooling rate did the α2-peak-emerge. A partial explanation for the diverse behaviour of PA-6 as a neat resin and as a nanocomposite matrix can be gleaned considering the results of an annealing experiment, displayed in Figure 123 [Liu and Wu, 2002a]. Granules of PA or CPNC were placed between two glass slides, heated in an oil-bath up to 250 °C for 10 min, and then quenched in liquid N2. The films were annealed in an oilbath at different Ta for 1 h, re-quenched and their crystallinity was determined by XRD. In the Figure the α- and γ-form crystalline content is plotted versus Ta – the total crystallinity is a sum of these two forms. During the compounding and pelletising some crystallinity in PA-6 and CPNC certainly developed. However, this crystallinity was to be eliminated by keeping the specimens for 10 min at 250 °C. Thus, the subsequent quenching generated 0 and 48% crystallinity in PA and CPNC, respectively. The 1 h annealing of PA-6 at T ≥ 100 °C characteristically formed g-crystals that subsequently converted to α-form. The total crystallinity reached a maximum value of 36% at 180 °C. By contrast, CPNC maintained ca. 49% of crystallinity in the γ-form up to ca. 130 °C. Above this temperature γ- progressively converted to α-form – the latter reached its maximum content of ca. 19% at 180 °C. Thus, the presence of organoclay increased the total crystallinity of the PA-6 matrix well above the level observed for the neat resin, and it had stabilised the γ-form to higher temperatures before the onset of the solid-solid phase Brill transition from γ- to α-form. It is noteworthy that the DSC-determined Tm of PA-6 and its CPNC was 223 and 219 °C, respectively, in good agreement with the known melting points for the α- and γcrystals. FTIR studies of PA-6 and CPNC led to the conclusion that the presence of organoclay weakens the hydrogen bonding in PA-6. The γ-form was found
Figure 123 Determined by XRD crystallinity of PA and CPNC (containing 5 wt% of doubly pre-intercalated MMT) as a function of the annealing temperature. The α- and γ-form crystalline content is shown – see text. Data [Liu and Wu, 2002].
404
Nucleation and Crystallisation preferentially located near the clay platelets, whereas the α-one preferentially in the bulk [Wu et al., 2002a]. Murase et al. [2002] used FTIR, XRD and DSC to study the structure of PA-6 with 2 wt% of either synthetic mica or natural MMT. The PA-6/mica was prepared by melt polymerisation of ε-caprolactam while the PA-6/MMT was industrial resin from Ube Ind. The films were prepared by pressing at 250 °C for 2 min followed by quenching into an ice-water bath and then annealing at 110 °C for 6 h. In PA-6 only the α-form was found, in the Ube nanocomposite both α- and γ-crystalline forms, and in the mica-containing CPNC only γ-form crystals were detected. The latter composition also showed the highest degree of crystallinity of about 58% followed by that of PA/MMT and PA-6 (37%). Judging by the reported results, it seems that mica in the studied CPNC was only intercalated, whereas MMT in the Ube nanocomposite was exfoliated. Thus, it is difficult to speculate whether the observed difference in the composition of the crystalline phase is due to the nature of these two clays or to the degree of exfoliation. Kamal et al. [2002] studied the crystallisation kinetics of PA-6 and its CPNC (Ube PA-1015B and –1015C2, respectively) under pressure P = 50 to 200 MPa. Isobaric volume changes associated with the crystallisation process were determined at constant T and P, from which the crystallinity was calculated as: X(%) = 100(V0 – Vt)/(V0 – V∞)
(153)
where V0 is the initial (melt) volume, Vt is the volume at time t, and V∞ is the volume at the end of crystallisation. Isobaric heating or cooling was used to determine the melting or crystallisation temperatures of PA-6 in its neat state and in CPNC. The data were fitted to a linear dependence: Tm,c(°C) = ao + a1P(MPa)
(154)
The numerical values of the parameters ai, along with the correlation coefficient squares, r2, are listed in Table 57. It is noteworthy that the Tm for both resins follows the same dependence, whereas Tc of the CPNC is lower by about 6 °C than that of neat PA-6. It is also interesting that within the experimental error of these measurements, the pressure gradient of Tm and Tc, dTm,c/dP = a1, is about the same for PA-6 and CPNC. The kinetics of crystallisation were studied at constant T and P. For both resins, the data plotted according to the linearised Avrami equation showed two intersecting linear dependencies. Apparently, the crystallisation started with the γ-form, and then proceeded to the α-form. Thus, within the first zone the γ-form
Table 57 Pressure dependence of the melting point, Tm °C, and the crystallisation temperature, Tc °C, for PA-6 and its nanocomposite. Data [Kamal et al., 2002] Parameter
PA-6
CPNC
Tm
Tc
Tm
Tc
ao
214
184
214
178
a1
0.21
0.23
0.21
0.22
r2
0.986
0.997
0.989
0.996
405
Clay-Containing Polymeric Nanocomposites was dominant while within the second zone the α-form dominated. The Avrami exponent n for the γ-form was between 1.0 and 3.2 in PA-6 and between 0.9 and 2.6 in CPNC. Its value for the α-form was between 1.0 and 2.1 in PA-6 and between 1.2 and 2.6 in PNC. In Figure 124 and Figure 125 the crystallisation rate of, respectively, PA-6 and PANC is expressed as the crystallisation half-time, t1/2, versus the free volume parameter, h = h P˜ , T˜ , where P˜ , T˜ are reduced pressure and temperature, computed for each resin from the Simha-Somcynsky eos. Evidently, the kinetics of crystallisation not only depend on h, but on P as well. To more clearly see the effects of the independent variables, the experimental data were fitted to a linear dependence, t1/2 = t1/2(h, P). The correlation coefficient squared, r2 = 0.74 to 0.95 was obtained. At the same value of h and P, the rate of crystallisation of the α-crystals is systematically slower (t1/2 larger) than that of the γ-crystals. PA-6 in CPNC crystallises faster than in its neat form. Surprisingly, in these isothermal and isobaric experiments the pressure seems to play a secondary role to h as far as the crystallisation kinetics is concerned. This was not the case for the isobaric scans (at constant scanning rate) where high pressure slowed down either melting or crystallisation in PA and CPNC. Bureau et al. [2002] reported on the crystallinity and mechanical properties of compression moulded PA-6 and its CPNC (Ube Ind., PA-1015B and 1015C2). Specimens were moulded at 250 °C for 90 sec at P = 0.7 MPa, then either cooled to room temperature under pressure at a rate of ca. 30 °C/min, or quenched in ice water. Some quenched specimens were subsequently annealed at 80 °C for 24 h under vacuum. It was expected that these procedures would produce specimens containing, respectively: the α-form (cooled), amorphous PA-6 (quenched), and the γ-form (annealed). The DSC scans of PA-6 specimens showed a crystallisation peak followed by an α-form melting peak at Tm = 221 °C. The crystalline content was estimated as: 31% (cooled), 7% (quenched) and 21% (annealed). DSC scans of cooled or quenched CPNC showed a double melting point, Tm = 213 and 221 °C, indicating a polymorph with either γ- (in the cooled sample) or α-form (in the quenched sample) predominating. The crystalline content in both these two samples was ca. 25%, thus lower for the cooled specimen (than that in neat PA-6) and significantly higher in the quenched specimen, indicating rapid nucleation by MMT. Identification of the a- or g-form was confirmed by XRD and FTIR. The tensile test results are summarised in Table 58. There is strong correlation between the mechanical behaviour and morphology. All PA-6 specimens showed necking with strain at break varying between 120% and 670%, depending on the moulding condition. For the cooled CPNC specimens, higher Young´s modulus and yield stress (with brittle fracture without necking) was observed. The tensile strength and the Young´s modulus of the CPNC were ca. 15% higher than those of PA-6. 25% polymorphous crystallinity in CPNC yielded a higher tensile modulus than PA-6 with a 31% α-form crystalline phase. The data indicate that improvements of rigidity and strength caused by addition of MMT are related to the reinforcing effects of the nanofiller and not to the increased γ-crystals content. The effect of matrix molecular weight on the kinetics of isothermal and nonisothermal crystallisation in a PA-6/MMT system was studied by Fornes and Paul [2003]. Incorporation of ca. 0.5 to 1 wt% clay resulted in the highest PA-6 crystallisation rate. While for neat PA-6 the crystallisation rate decreases with molecular weight, in the case of CPNC the maximum crystallisation rate was
( )
406
Nucleation and Crystallisation
Figure 124 Half-time for crystallisation of the γ- and α-forms in PA-6 versus the free volume fraction parameter, h. Solid points are experimental, open are computed from the linear fit to Equation 94. Data [Kamal et al., 2002]. See text.
Figure 125 Same dependence as in Figure 124 but for a CPNC sample of 2 wt% organoclay in a PA-6 matrix (Ube PA-1015 C2). Data [Kamal et al., 2002]. See text.
407
Clay-Containing Polymeric Nanocomposites hardly affected by it. Furthermore, the largest enhancement of the crystallisation kinetics was reported for systems with the highest molecular weight (the authors postulate that this was related to the highest degree of exfoliation). However, the degree of crystallinity decreased with the matrix molecular weight.
3.4.4 Clay Effect on Crystallisation of Other Polyamides Kuchta et al. [2000] studied crystallisation of PA-11 in its CPNCs. The authors prepared several compositions by in situ polycondensation or melt compounding. To start with, MMT was pre-intercalated with ω-amino-undecanoic acid. Polycondensation yielded CPNC with 2.4 to 19.7 wt% MMT, while compounding of PA-11 with organoclay (10 min in a recirculating mini-TSE) yielded CPNC with 1.7 to 31 wt% clay. TEM and synchrotron radiation indicated uniform dispersion of MMT in the PA-11 matrix. By contrast with PA-6 CPNC, where the γ-form has been observed already at low clay loadings, in the PA-11 nanocomposites the α-form remained stable up to 20 wt% MMT. In the sample containing 5.6 wt% MMT the α- to γ-form Brill transition was observed at T ≅ 100 °C, the same as in neat PA-11. Thus, the influence of MMT on the solid-solid transition in PA-11 is less significant than that in PA-6. Heating PA-11 and its CPNC resulted in different extents of lamellar thickening. While in neat PA-11 the lamella thickness at 180 °C is 72% higher than that at room temperature, in CPNC the constraint by the MMT layers reduced the increase to 17%. Reactively prepared CPNC showed a reduction of crystallinity at MMT content above 20 wt%. The CPNC had enhanced thermal stability, tensile modulus and increased elastic behaviour over a broader temperature range than the neat resin. Zhang et al. [2004] used the same method for the preparation of neat PA-11 and its CPNC containing 5 wt% MMT. However, the γ-form of PA-11 was induced and stabilised by MMT. Furthermore, the hydrogen bonding in neat resin and its CPNC was quite different. The crystallisation behaviour of PA-1010 and its CPNC (containing up to 10 wt% MMT) were studied using DSC and the results were interpreted with Avrami’s equation [Zhang and Yan, 2003]. Addition of 1 wt% MMT increased
Table 58 Summary of test results for PA-6 and CPNC. Data [Bureau et al., 2002] Material
PA-6
CPNC
408
Molding conditions (structure)
Crystallinity (%)
Young's modulus, E (GPa)
Yield stress,
σY
(MPa)
Strain at break, εb (%)
Cooled (α-form)
31
4.0±0.3
93±1
140±40
Quenched (amorphous)
7
2.4±0.2
69±2
670±60
Annealed (γ-form)
21
2.8±0.1
91±2
120±30
Cooled (γ- and α-form)
25
4.7±0.2
106±3
3.8±0.3
Nucleation and Crystallisation the crystallisation rate, but at higher loading (i.e., 6 and 10 wt% MMT) the rate decreased. The thermograms for all specimens showed multiple melting peaks. However, such crystallisation parameters as the Avrami’s exponent, n, the crystallisation temperature, and the heat of crystallisation were virtually independent of the MMT content. In the following publication [Liu et al., 2004], CPNC of PA-1010 was prepared by melt compounding with up to 10 wt% of MMT pre-intercalated with either methyl tallow dihydroxyethyl ammonium (MT2EtOH) or trimethyl hexadecyl ammonium (3MHDA) salts. XRD showed significantly better clay dispersion for the MT2EtOH series (than 3MHDA), which in turn resulted in superior mechanical performance. At 7 wt% MMT loading the glass-transition temperature increased by 8.5 and 1.0 °C for the specimens with MT2EtOH and 3MHDA, respectively. Addition of the former organoclay accelerated the crystallisation rate of the matrix and changed its crystallisation behaviour. Wu et al. [2002b] studied the nucleating effect of MMT on the crystallisation of PA-1212. The exfoliated CPNC was prepared by melt compounding of preintercalated MMT with PA-1212. The film (ca. 0.5 mm thick) was moulded at 10.0 MPa and 200 °C for a few minutes, then quenched. Non-isothermal crystallisation and melting tests were conducted in a DSC under N2. The sample was heated to 220 °C and kept at this temperature for 10 min, then cooled at a rate of 5 to 40 °C/min to 50 °C, and then re-heated to 200 °C at a rate of 10 °C/min. For PA-1212 and its CPNC, an increased cooling rate shifted the Tc to lower temperatures, as well as reducing and broadening the Tc peak. Compared to PA-1212 the CPNC had narrower exothermic peaks at higher temperatures. In Figure 126 the peak crystallisation temperature of PA-1212 and its CPNC, Tc, is plotted versus the cooling rate, r. The overall time of crystallisation, tc ≡ (Ti-Te)/r versus r is also presented (the subscripts i and e indicate the initial and final temperature). Thus, addition of organo-MMT shifts the Tc to a higher temperature and shortens the crystallisation time, tc. Both these functions indicate enhanced nucleation. The authors also calculated the time for the initiation of crystallisation (ti) as well as the nucleating activity coefficient, ϕ = 0.71. As expected, these parameters also indicate enhanced nucleation by MMT. Optical microscopy under polarised light, POM, confirmed this conclusion – the spherulites in neat resin were about 15¯25 times larger than those in its nanocomposite.
3.4.5 Crystallisation of PO Matrix PP has a complicated crystalline microstructure, which depends on the mechanism and the rate of crystallisation. The four principal crystalline structures of PP are: monoclinic (α), hexagonal (β), triclinic (γ), and smectic or quenched polymorphic. The monoclinic α-form is usually obtained under typical industrial and laboratory processing conditions. The β-form is thermodynamically less stable, and the shear stress or specific nucleating agents enhance its formation. The other forms are quite rare. The characteristic parameters of PP and HDPE are listed in Table 59. When injection moulding thin-walled articles of PP, it is customary to add a nucleating agent. The addition reduces the moulding cycle and improves optical as well as mechanical properties. Inorganic compounds such as talc, silicates, clays, and carbon black have poor nucleating ability. The preferred nucleating agents are organic salts, viz. Na-succinate, Na-glutarate, Na-caproate, Na-(4-methyl) valerate, Na-benzoate, Na- β-naphthoate, Al-benzoate, Al-tert-butylbenzoate, bisbenzylidene sorbitol, etc. Pigments are also known to nucleate PP. For example, 409
Clay-Containing Polymeric Nanocomposites
Figure 126 Non-isothermal crystallisation of PA-1212 and its CPNC as a function of the cooling rate from 220 °C. Data [Wu et al., 2002].
red pigment (Quinacridone) is an efficient a-crystal nucleating agent, while white pigment (White PE MB) was found to nucleate the β-form. By contrast with PP, PE crystallises rapidly, thus additives are rarely used to nucleate it. However, K-stearate has been used to reduce spherulite size. Nanocomposites with PE as matrix are still rare. Recently CPNCs, containing PE-MA and silicates of different aspect ratios, were studied by Wang et al. [2002]. The PE-MA was LLDPE grafted with 0.85 wt% MAH. The –MA groups were randomly placed along the LLDPE macromolecules. The silicates were two organoclays (Cloisite® 20A, C20A and Laponite®SCPX2231, SCPX) and synthetic SiO2. The organoclays were both intercalated with 2M2HTA and had the nominal aspect ratio, p = 100 to 200 and 20 to 30, respectively. The silica had spherical particles with diameter of ca. 1.8 μm. XRD did not show diffraction peaks down to 2θ = 2°; hence d001 > 4.4 nm. Surprisingly, addition of any of these three silicates reduced the total crystallinity, X, of PE-MA by an amount proportional to the silicate volume fraction, φ: X(%) ≅ 41.5 - 120φ. Probably the reaction of the randomly grafted –MA groups on the silicate surface disrupted formation of the ordered macromolecular crystals. Young´s modulus and yield stress increased with increasing crystallinity and increasing α-form content, while elongation at break increased with increasing β-crystal content [Mubarak et al., 2000]. Addition to PP of 0.05 wt% 1,2,3,4bis-(3,4-dimethyl-benzylidene sorbitol) or N,N´-di-cyclohexyl-2,6-naphthalate di-carboxamide, generated the α- or β-form, respectively [Ellis et al., 2001]. Infrared microscopy has been used to study the crystalline morphology of these two crystalline forms. CPNC with PP as a matrix are more recent than these of PA [Kurokawa et al., 1996; 1997]. The authors used an elaborate process involving pre-intercalation of smectite clay with ammonium ion, polymerisation of di-acetone acryl amide on the 410
Nucleation and Crystallisation
Table 59 Characteristic parameters of polyolefins Property
HDPE
PP
Condition
Value
Condition
Value
Glass transition temp., Tg (K)
Dry
140-150
Dry
280
Melting point (equilibr.), Tm (K)
Dry
419
Dry
459
Density, ρ (kg/m3)
Dry
920-990
Dry
900-910
Typical crystallinity, α (%)
Dry
35-90
Dry
50-70
Unit cell dimensions (nm)
α-form β-form
0.742×0.495×0.255 0.809×0.479×0.253
Unit cell angles (# mers)
α-form β-form
- (4) 107.9 (4)
α-forms
98.67 (3)
Lattice
α-form β-form
Orthorhombic Monoclinic
α-forms β-form γ-form
Monoclinic Hexagonal Orthorhombic
αI-form 0.667×2.08×0.65 αII-form 0.665×2.07×0.65 β-form 1.103×1.103×0.649 γ-form 0.854×0.993×4.241
organoclay, and then mixing the doubly intercalated clay with maleated-PP and finally with PP. Exfoliated and well-dispersed clay platelets were observed under the TEM. Kato et al. [1997] simplified this procedure by melt-mixing PP with oligopropylene modified by MAH (PP-MA) and MMT intercalated with stearyl-ammonium salt (ODA). CPNC with a high degree of exfoliation was obtained when PP-MA was able to (1) bond to the clay and (2) be miscible with the PP matrix. There are several variants of these procedures, with different amounts of the PP-MA compatibiliser, of different structure, molecular weight and MAH-content. Since these variables control the mutual miscibility of PP with PP-MA, this type of CPNC shows a wide range of properties. For example, pre-intercalation and compatibilisation may results in a system where the clay particles are shielded from the matrix, preventing direct PP-clay contact, hence the matrix crystalline structure entirely depends on the behaviour of the PP/PP-MA blend. In such a CPNC the effect of clay on matrix crystallisation may be negligible. There are two major differences between CPNC with PA and that with PO matrix. (1) ω-Amino acid may be used to intercalate clay, thus after polymerisation the clay is in full contact with the matrix. By contrast, owing to the strong hydrophobicity of PO-macromolecules such a structure could not form in the latter systems. (2) The crystallinity of PA rarely exceeds 50%, thus is significantly 411
Clay-Containing Polymeric Nanocomposites lower than that in isotactic PP where X = 60 to 70% is often observed. Since the intercalated clay platelets cannot enter the crystalline domains, they must be concentrated in small pockets of the amorphous or meso-crystalline phase. This means that the local concentration is about three times higher than the average, thus (excepting low clay loading of ca. w ≤ 0.5 wt%) the platelets are unable to freely rotate and form parallel stacks with local ordering – full exfoliation is difficult to achieve. The National Chemical Laboratories (NCL) in Pune conducted systematic studies of clay effects on crystallisation of PP and the resulting performance of the CPNC. Hambir et al. [2001] prepared PP-based CPNC with MMT (CEC = 1.35 meq/g) intercalated with octadecyl amine (ODA) and compatibilised with maleated PP (MA-PP). XRD gave the interlayer spacing of d001 ≅ 2.5 nm. The isothermal crystallisation of the PP and PP/MMT-ODA systems was carried out by DSC. The thermograms indicated that the clay presence narrows the crystallisation peak and reduces the crystallisation time hence it accelerates PP crystallisation. As shown in Figure 127, crystallisation in neat PP and in its CPNC may have different activation energy, but this conclusion hinges on the validity of two data points at low temperatures (T = 100 and 136 °C) where reduced chain mobility may already start slowing down the crystallisation rate. Optical microscopy under polarised light showed a dramatic change of morphology – large, well-defined spherulites in specimens crystallised at 122 °C PP, but crystallites in the form of a fibrous mat in those crystallised at 142 °C CPNC. In the following publication from the same laboratory [Kodgire et al., 2001] PP was compounded in a single-screw extruder (SSE) with 12 wt% MA-PP and 4 wt% of commercial organoclay, either Cloisite® 6A (C6A; MMT, CEC = 1.4 meq/g intercalated with 2M2HTA), or Nanomer I.30 (MMT, CEC = 1.35 meq/g intercalated with ODA). The XRD-determined interlayer spacing in Na-MMT was d001 = 1.2 nm, in C6A (two peaks) d001 = 1.2 and 1.8 nm and in I.30 d001 = 2.45 nm.
Figure 127 Crystallisation time as a function of crystallisation temperature in nonisothermal DSC scans for PP and PP/MMT-ODA. Data [Hambir et al., 2001].
412
Nucleation and Crystallisation Compounding with PP increased d001-values, but the system containing C6A was exfoliated only after compounding with MA-PP, while that with I.30 intercalated (d001 = 2.75 nm). However, the enhancement of mechanical properties was similar for either CPNC (hence independent of the degree of clay dispersion) – ca. 35% increase in the tensile modulus and about a 10% increase in the tensile strength. In accord with the results reported in the preceding publication, the Tc of PP increased after addition of clay from 122 °C (for neat PP) to 142 °C. Furthermore, in CPNC the crystallites formed a fibrous mat that coarsened with time, but did not transform into spherulites. Thus, clay nucleated a very specific PP morphology. Another publication from this laboratory provides additional information [Hambir et al., 2002]. This time the ODA intercalated clay was prepared in house and used along with C6A. Furthermore, three PP resins (Mw = 164, 241 and 362 kg/mol) and two MA-PP resins (Mw = 188 kg/mol, MAH = 1%; and 151 kg/mol, MAH = 0.64%) were used. The isothermal crystallisation of PP and PP/clay was studied by DSC at Tc = 115 to 131 °C. The PP/clay systems showed a sharper crystallisation peak than that of PP. Thus, crystallisation of PP was accelerated by organoclays. Furthermore, the crystallisation half-time depended on the organoclay content, Mw of PP as well as on the presence of MA-PP. However, the change of Mw from 241 to 362 had a more pronounced effect than addition of up to 4 wt% of organoclay. The crystallisation activation energy seems to be more influenced by the Mw of PP than the other variables. Thus, incorporation of organoclay resulted in enhanced nucleation and faster crystallisation of the PP matrix. The clay also changed the macromorphology of the crystalline phase – in CPNC instead of spherulites fibrillar structures have been observed. However, since the authors did not calculate the unit cell, it is unknown whether the addition of organoclay modified the crystal cell structure, e.g., from the common α-monoclinic into the more desirable β-hexagonal one. Saujanya and Radhakrishnan [2001] studied the crystallisation of PP in the presence of calcium phosphate nanoparticles (CaP). These were prepared from CaCl2 and Na3PO4 in a MeOH solution of PEG. Varying the concentration ratio of PEG to CaCl2 from 0 to 32, resulted in reduction of β-ortho CaP (monoclinic) particle diameter from d = 82 to 7 nm. It is noteworthy that in the presence of CaP PP crystallises in spherulitic form with a monoclinic α-crystalline cell structure. At 2 wt% CaP the t1/2 and the ultimate spherulite size decreased with the particle size from 45 to 20 μm. The strong nucleation efficiency of CaP originates in high surface energy and large surface area, viz.: ln(G/Go)∝1/d
(155)
where G is the crystallisation rate in the presence of CaP and Go without it. The optical transparency of the PP/CaP-nanoparticles system was higher than that containing conventional CaP at the same concentration. The DSC data showed that incorporation of CaP increased the crystallisation temperature of PP from 110 to 114 °C and reduced the width of the crystalline peak as compared to PP. The heat of fusion of neat PP was determined as 77.9 J/g whereas that of the nanofilled PP as 132.9 J/g, indicating that in the presence of 2 wt% CaP the crystallinity of PP nearly doubled. Such a large effect has not been observed for compatibilised PP/organoclay systems. The most likely reason is the presence of an organic phase around the clay particles. Since the intercalant molecules are immiscible with PP, they form a barrier, separating the crystallisable matrix from the high-energy clay surface. 413
Clay-Containing Polymeric Nanocomposites
3.4.6 Crystallisation of PEST Matrix Several patents on the preparation of PET-nanocomposites claim that addition of clay increases the crystallisation rate. When organoclay is used, the crystallisation temperature and the melting point increase. Again, since there are different methods for the preparation of these materials as well as different intercalants, it is difficult to predict the extent of these effects. Tsai [2000] grouped the processing methods described in the patent literature into three categories: melt compounding with organoclay, polymerisation in the presence of organoclay and polymerisation in the presence of synthetic, not pre-intercalated clay (e.g., fluoromica dispersed in ethylene glycol prior to polycondensation with p-terephthalic acid). However, in these publications crystallinity has been of little concern. The crystallisation process and crystal morphology in PET/organoclay CPNC was studied by Ke et al. [1999]. The nanocomposites were prepared by in situ polymerisation of PET in the presence of up to 5 wt% organoclay (MMT with CEC = 0.7 to 1.1 meq/g intercalated by a ‘proprietary’ method). At low magnification SEM and TEM showed a large number of solid particles. The interlayer distance, d001 = 1.4 to 3.5 nm, was determined by TEM and XRD, hence only intercalation was achieved. In consequence, while the modulus increased with clay loading the stress at break decreased. The authors used the Avrami equation for the analysis of non-isothermal crystallisation. The pertinent information is summarised in Table 60. As observed for other CPNC systems, also here addition of organoclay increased the crystallisation temperature and the crystallisation rate. However, in the studied system, the melting temperature and the heat of melting decreased with organoclay content. The most likely explanation is an excess of the non-identified, intercalating agent. CPNC with 5% of clay had a broad distribution of particle sizes from ca. 10 to 1,000 nm, with a maximum at about 80 nm. It was difficult to detect spherulites in these PET/clay systems. The macromolecular chains diffused into the interlamellar galleries. A more recent publication focuses on poly(ethylene terephthalate-co-ethylene naphthalate), PETN [Chang and Park, 2001a,b]. First, Na-MMT (CEC = 1.19 meq/g) was intercalated with hexadecyl ammonium salt (HDA). PETN (92 mol% ethylene terephthalate) was solution-blended with the organoclay then cast into film of ca. 10 to 15 μm thick. The interlayer spacing, as measured by XRD, increased
Table 60 Crystallisation temperature, Tc, melting point, Tm, crystallisation half-time, t1/2, and heat of fusion, ΔHm, for PET/MMT systems. Data [Ke et al., 1999] Tc (°C)
Tm (°C)
t1/2 (s)
ΔHm (J/g)
0
174
259
108
50.0
0.5
201
258
–
–
1.5
206
257
43
43.5
2.5
209
254
48
44.0
5.0
208
252
36
45.2
MMT (wt%)
414
Nucleation and Crystallisation from d001 = 1.199 (for Na-MMT) to 2.596 nm (for MMT-HDA). The PETN containing 2 to 4 wt% MMT-HDA was only intercalated (d001 = 3.104 nm). Furthermore, as the concentration of the organoclay increased to 6 wt% the interlayer spacing decreased to d001 = 2.662 nm, indicating that at higher concentration the organoclay aggregates (also observed in SEM). DSC measurements of these CPNCs have shown that the glass-transition temperature (Tg), melting temperature (Tm), and heat of fusion (ΔHm) are nearly independent of the organoclay loading (see Table 61). Thus, the total crystallinity of PETN does not seem to be affected by MMT-HDA. The structure of these CPNCs is a phase-separated matrix with intercalated, internally ordered organoclay domains dispersed in it. Nevertheless, incorporation of the organoclay did enhance the thermal stabilities and mechanical properties (maximum in the tensile strength and modulus was obtained for 4 wt% loading). Imai and his colleagues from AIST prepared PET-based CPNC by melt, and then solid-state polymerisation in the presence of ca. 3.7 ± 0.2 wt% of clay [Imai et al., 2003; Saujanya et al., 2003]. Fluoromica (FM), FM pre-intercalated with triphenyl dodecyl phosphonium bromide (3PPC12), or FM pre-intercalated with triphenyl di(methoxycarbonyl)phenoxy decyl phosphonium bromide (3PPPC10) were used. The interlayer spacings were: d001 = 1.33, 1.82, and 1.86 nm, respectively. For the crystallisation studies neat PET (PET-0), CPNC with FM (PET-1) and CPNC with FM-3PPPC10 (PET-2) were used. The isothermal crystallisation tests showed that the crystallisation rate, and crystallinity increase in the order from PET-0 to PET-2; the opposite tendency was observed for the activation energy. The results seem to indicate that, at least for the crystallisation of PET, the degree of dispersion is more important than the surface energy of the dispersed clay. This observation may explain the small effect of organoclay on PET crystallisation, reported by Wang et al. [2003b; 2004], and by Ou et al. [2003]. Recently nanocomposites with poly(trimethylene terephthalate) (PTT) have been prepared using the solution intercalation or melt compounding method, and their crystallisation behaviour was studied [Ou, 2003; Liu et al., 2003]. The first author used MMT pre-intercalated with cetyl-pyridinium chloride (CPC), whereas the others used MMT-MT2EtOH. For both systems XRD indicated
Table 61 Glass transition temperature, Tg, melting point, Tm, and heat of fusion, ΔHm, for PETN/C16-MMT nanocomposites. Data [Chang and Park, 2001] Tg (°C)
Tm (°C)
ΔHm (J/g)
0
76
236
32
1
74
237
31
2
73
238
31
3
75
238
30
4
75
238
32
6
74
236
30
MMT-HDA (wt%)
415
Clay-Containing Polymeric Nanocomposites intercalation with short stacks dispersed in the matrix. From the DSC thermograms the authors concluded that organoclay behaves as a nucleating agent, enhancing the crystallisation rate of PTT. As in other systems, here also a maximum crystallisation rate was observed at relatively low organoclay loadings.
3.4.7 Crystallisation of Syndiotactic PS Matrix Wu et al. [2001] analysed the effects of MMT on the chain conformation and crystallisation of syndiotactic polystyrene (sPS) using FTIR, XRD, and TEM. The CPNC was prepared by solution dispersing of 5 wt% of MMT or organoclay (MMT pre-intercalated with cetyl-pyridinium chloride, CPC) in a dichlorobenzene solution of sPS at 140 °C for 24 h. For the first system TEM showed large aggregates (dimensions from a few tenths to 100 nm), but a high degree of intercalation (size 1-2 nm) in the second system: sPS/MMT-CPC. sPS has four crystalline forms: α-, β-, γ-, and δ-, which can be divided into two groups: (i) γ- and δ- forms possessing a helical trans-trans and gauche-gauche (TTGG) chain conformation and (ii) α- and β- forms possessing a planar all-trans (TTTT) ‘zigzag’ conformation. Addition of clay changes the sPS chain conformation from TTGG to TTTT. This phenomenon leads to a change in the mechanism of molecular packing during melt-crystallisation. The clay plays a vital role in facilitating the formation of the thermodynamically favoured all-trans β-crystal, particularly in thin films. Crystallising the sPS/MMT system at a cooling rate of 1 °C/min from 320 °C resulted in the highest absolute crystallinity of β-form of 56%. This is to be compared with 49% obtained for sPS/organoclay and 42% for neat resin. Thus, clay may significantly affect the chain conformation and crystallisation of sPS, but the effect very much depends on clay surface energy. It is noteworthy that the effect of poorly dispersed clay with significantly reduced surface contact area affected sPS more than the finely dispersed cetyl-pyridinium covered clay platelets. Another team of researchers [Wu et al., 2002a] confirmed these observations. The nanocomposites were prepared by dispersing 0.5, 1 and 5 wt% organoclay in sPS/xylene solution. The organoclay was MMT pre-intercalated with 3MHDA and then treated with a mixture of monomers (styrene and methacrylate in the ratio 4:1), which were then polymerised into copolymer having short PS blocks on the outside. Both XRD and TEM showed exfoliation and random dispersion of clay platelets in the matrix. XRD also indicated polymorphism of sPS, which strongly depended on thermal history and clay content. Isothermal crystallisation at higher Tc favoured formation of β-crystals. Thus, during the process the α-crystals could be transformed into β-form. During crystallisation, the sPS chains achieved better alignment at higher T c . In the presence of MMT or at lower T c conformational defects may preclude perfect chain alignment. At the present state of knowledge, it is impossible to make generally valid statements about the influence of organoclay on the polymeric matrix crystallinity. There are many factors that influence the outcome. The presence of a high-energy clay surface most often leads to an enhanced crystallisation rate, which in turn results in a different, less thermally stable crystalline form (viz. γ-form in PA-6, β-form in PVDF [Priya and Jog, 2003], etc.) crystallising at higher Tc and melting at higher Tm. Notwithstanding the higher crystallisation rates, the total crystallinity content may increase or decrease with organoclay content. These effects are controlled by the equilibrium thermodynamics (type and intensity of interaction) as well as dynamics (orientation and annealing).
416
Mechanical Behaviour
3.5
Mechanical Behaviour
One of the reasons for the interest in CPNC is enhancement of the mechanical performance of the matrix polymer at low clay loading. For example, addition of 2 wt% of organoclay to PA-6 increased the flexural modulus by 26%, and tensile strength by 14% [Ube Ind., 2000]. Similarly, dispersion of 2 and 5 wt% of vermiculite in PP increased the tensile strength by 18 and 30%, the tensile modulus by 20 and 54%, and the storage modulus by 204 and 324%, respectively [Tjong et al., 2002]. A thorough review of the theories of mechanical behaviour in a multiphase polymeric system can be found in the PhD thesis of D. Colombini [1999].
3.5.1 Micromechanics of CPNC The assumed model of a nanocomposite is highly simplified, containing either exfoliated clay platelets or intercalated short stacks, aligned with the stress tensor. The particles are assumed to perfectly adhere to the matrix. The simplest model of perfectly aligned large reinforcing particles leads to the volumetric rule of mixtures for the tensile modulus in the stress and transverse direction: in stress : Ec = Emφ m + f L E f φ f transverse : 1 / Ect = φ m / Em + φ f / E f
(156)
where E is tensile modulus, φ is the volume fraction, φm = 1 - φf ; subscripts c, m and f stand for composite, matrix and filler, respectively; and fL is the filler particle length correcting factor. The dependence predicted by Equation 156 is presented in Figure 128. Note that the predictions are supposed to be valid for infinitely large filler particles with perfect adherence to the matrix. For the computations the moduli ratio was assumed to be Ef /Em = 100 or 1000. Several modifications of Equation 156 have been published:
(
Nicolais & Narkis : Ec = Em 1 − αφ bf
( ) ) / (1 − Bφ );
Faber & Farris : Ec = Em 1 − φ f
(
Kerner : Ec = Em 1 + ABφ f
)
b
f
Gc / Gm = Ec / Em
A = (7 − 5ν m ) / (8 − 10ν m )
(
)(
(157)
) ) / (1 − ΨBφ ); ) / φ ]φ
B = E f / Em − 1 / E f / Em + A
( Ψ = 1 + [(1 − φ
Neilson : Ec = Em 1 + ABφ f f ,max
f
2 f ,max
Gc / Gm = Ec / Em
f
where νm is the Poisson’s ratio. 417
Clay-Containing Polymeric Nanocomposites Halpin-Tsai (H-S) rigorously derived the tensile modulus in the stress direction, Ec:
(
)(
Ec / Em = 1 + 2 pκφ f / 1 − κφ f
)
κ = ( Er − 1) / ( Er + 2 p); Er ≡ E f / Em
(158)
Here p is the aspect ratio defined in this book as p = (platelet diameter)/(platelet thickness). For the modulus in transverse direction p = 1 may be assumed. For p → ∞ Equation 158 converts into Equation 156 with fL → 1. The predicted behaviour is illustrated in Figure 129. According to Brune and Bicerano [2002] the above relationships may be used only for predicting the modulus of fully exfoliated CPNC with oriented, clay platelets perfectly adhering to the matrix. When the clay platelets are only intercalated then the variables: p, Ef, and φf obtained from fitting the equations to data must be consider empirical. On the other hand, the authors derived the following relationships between these variables for exfoliated and intercalated (symbols with prime) systems:
[
]
p ′ = ( p / N ) / 1 + (1 − 1 / N )( s / t )
[
]
φ ′ = φ 1 + (1 − 1 / N )( s / t ) Er′ =
(159)
Er + (1 − 1 / N )( s / t ) 1 + (1 − 1 / N )( s / t )
In this relation the aspect ratio, p´, the volume fraction, φ´, and the tensile modulus, E´, refer to the platelet stack composed of N-clay layers, each with thickness t and the interlamellar gallery spacing, s ≅ d001 – 0.96 nm. However, since the
Figure 128 Relative tensile modulus for Ef /Em = 100 or 1000 in the stress and transverse to it directions. See Equation 156.
418
Mechanical Behaviour
Figure 129 Predictions of the Halpin-Tsai Equation 158 for relative modulus in the stress direction for Ef /Em = 100 or 1000 and p = 10 or 1000.
dependence does not have the correct upper limit for the s/t ratio, the parameter N in the dependence should be replaced by:
φ (160) 1−φ Evidently, Equations 157 to 159 are valid for idealised, well-oriented CPNCs where the platelets are perfectly bonded to the matrix. The important teaching of this theory is the strong influence of the degree of dispersion on the relative modulus, Ec/Em = E´/Em. In Figure 130 variation of the relative modulus for intercalated CPNC is plotted as a function of the number of platelets in the stack, N, and the size of the interlamellar gallery expressed as: s/t = (d001/t) – 1 ≅ d001 – 1. The role of this parameter is to express the modification of the macromolecular behaviour within the galleries. It is noteworthy that the steepest decline of performance is predicted for stacks containing 2 to 3 platelets, i.e., for these most often observed in CPNC. In other words, the modulus of a truly exfoliated system is at least twice as high as that obtained for the most common ‘exfoliated’ ones. Brune and Bicerano [2002] proposed a further refinement to the above treatment for disk-shaped filler particles. The aim was to correctly predict the effects of the aspect ratio. The derived equation is in a scaled form: N ′ = N + (1 − N )( s / t )
Es ≡
Ec ( p) − Ec ( p = 1)
Ec ( p → ∞) − Ec ( p = 1)
=
2( p − 1)
( Er − 1)(1 − φ f ) + 1 + 2 p
; Er ≡
Ef Em
(161)
The relation virtually traces the prediction of the Halpin-Tsai relation (see Figure 131). Within the customary range of the clay content, φf < 5 vol%, it is insensitive to composition, but it strongly depends on Er and p. It is important to note that when a high Er value is required, platelets with larger aspect ratio ought to be used. Considering that E(clay) ≅ 170 GPa and the modulus of unoriented, unfilled engineering plastic ranges from 2 to 4 GPa the clay platelets 419
Clay-Containing Polymeric Nanocomposites
Figure 130 Prediction of the Brune-Bicerano theory (see Equation 159) of the relative modulus of intercalated CPNC versus the average number of platelets in the stack for the interlayer spacing, d001 ≅ (s/t) + 1. For the most common value of d001 ≅ 3 nm stacks of ca. three platelets show only 40% of the modulus expected for exfoliated CPNC.
Figure 131 Scaled tensile modulus versus aspect ratio according to Halpin-Tsai (HT) and Brune-Bicerano (B-B) equations. Moduli ratio: Er = Ef /Em = 100 or 1000, filler volume fraction, φf = 0.01 or 0.05. Both relations provide equivalent dependence, insensitive to φf but sensitive to Er.
420
Mechanical Behaviour with p ≅ 200 would reach about 90% of the theoretically possible modulus. For CPNC with more ductile polymeric matrices clay with still higher p should be used. Hui and Shia [1998] derived a simplified H-S theory for the tensile modulus of oriented fibre or flake reinforced composites. The simplification involved the assumption that the Poisson’s ratio, νm = 0.5, is the same for both components and that there is perfect adhesion between filler and matrix. For flakes oriented in the stress direction the modulus is given by: ⎧⎪ 3 ⎤ ⎫⎪ φ ⎡1 Ec = Em ⎨1 − ⎢ + ⎥⎬ 4 ⎣ ξ ξ + Λ ⎦ ⎪⎭ ⎪⎩
ξ ≡φ+
−1
(
)
(1 − g) − gp2 / 2 1 + 3(1 − φ ) Er − 1 1 − p2
Λ ≡ (1 − φ )
(
)
3 1 + 0.25 p2 g − 2 1− p
2
(162)
; g ≅ π / 2p
Predictions of Equation 162 are higher and are more sensitive to the aspect ratio than those from Halpin-Tsai (H-T) Equation 158. The Hui-Shia (H-S) theory was evaluated using the tensile modulus data for CPNC with PDMS as the matrix [Shia et al., 1998]. The experimental results did not follow the theoretical predictions of H-T and H-S theories. To explain the discrepancy the authors postulated that the difference originates in imperfect bonding between the matrix and clay. This introduced two ‘effective’ quantities into the H-S model: the aspect ratio and clay volume fraction. Both were assumed to depend on the interfacial shear stress, σi, determined by fitting the data to the equation – the calculated σI-value was in the range of kPa. The interfacial shear stress was further decomposed into an intrinsic and a frictional shear stress, the former decreasing with the increasing volume fraction of the inclusion and the latter increasing linearly with strain. For the CPNC studied, the effective friction coefficient was calculated as 0.0932. However, as shown in Figure 132, in the whole range of concentration the data are well described by Brune-Bicerano Equations 159-161, with Ef = 170 GPa and Em = 1 MPa. The numerical values of the remaining variables were taken from Shia et al. [1998]. The least squares fit gave a reasonable value of the aspect ratio, p = 239. The number of platelets in the stack increased from the initial value of N = 2 to 4 at the highest concentration. The standard deviation and the correlation coefficient squared were: σ = 0.18 and r2 = 0.995, respectively. Ji et al. [2002] expanded Takayanagi’s two-phase model into three-phases: the matrix (m), interphase (i), and filler (f), randomly distributed in the matrix. The filler particles may be spherical, cylindrical or lamellar. The incorporation of an interphase is particularly important for the nanoparticles. According to the model, the three phases are connected to each other in series and in parallel. The realistic representation of a filled system is then reduced to three idealised regions, A, B, and C connected in series. The general form of the derived relation is:
1 1− β β −ϕ ϕ + = + Ec Em (1 − α )Em + αEBi (1 − α )Em + (α − λ )ECi + λE f
(163)
421
Clay-Containing Polymeric Nanocomposites
Figure 132 Relative modulus versus clay content for CPNC with PDMS as the matrix. Experimental data [Shia et al., 1998] and theoretical prediction from Brune-Bicerano (B-B) theory. See text.
where α, β, ϕ and λ are idealised dimensions of the interphase and filler regions determining the volume fraction of each of the three components, viz. Vf = λϕ; Vi = αβ – λϕ; and Vm = 1 – Vf - Vi. The moduli of the composite, matrix and filler are given as: Ec, Em, and Ef. The interphase modulus is defined for the model regions B (in series) and C (in parallel) as EBi and ECi , respectively. When particles are relatively large the interphase contribution is negligible, Vi → 0 and Takayanagi’s equation is recovered. The principal advantage of this theory is incorporation of the interphase contribution. As stated before, the authors consider two types of interphases connected to other phases either in parallel or in series. In the case of CPNCs these may be treated as the interphase inside the stack of clay platelets, and that between the stacks and the matrix. The two types depend on the distance from the filler surface respectively, linearly and exponentially: (164) ln( E0 i / Em ) 1 = ; ECi = E0 i + Em / 2 with : E0 i / Em ≡ k E Bi E0 i − Em
[
]
where E0i is the interphase modulus on the filler surface at τ = 0 and k is one of the model parameters. The platelet and the interphase shapes were assumed to be square, thus:
α=β=
[2(τ / t ) + 1]V
f
; ϕ = λ = Vf
Substituting these relations into Equation 164 yields:
422
(165)
Mechanical Behaviour
1 1− = Ec +
[2(τ / t ) + 1]V
f
Em
+
[2(τ / t) + 1]V − V [2(τ / t) + 1]V ⎫⎬⎭E + [2(τ / t ) + 1]V (k − 1)E f
⎧1 − ⎨ ⎩
f
m
f
f
m
/ ln k
Vf ⎧1 − ⎨ ⎩
[2(τ / t) + 1]V ⎫⎬⎭E f
m
[
]
+ ⎧⎨ 2(τ / t ) + 1 V f − V f ⎫⎬( k + 1) Em / 2 + V f E f ⎭ ⎩
(166) For CPNC the controlling parameters are: the dispersed particle size: t 1 would be expected. The experimental value of k = E0i/Em = 12 is high (especially since Ef/Em = 40 was assumed), but it may be reasonable (the ratio of the crystalline to the amorphous modulus for the same polymer ranges from 2 to 6 [Ferry, 1980]). In this book there are several tables listing the tensile modulus versus clay loading (inorganic part). These values were used to generate Figure 133. The aim was not to prove or disprove any of the listed theories, but to show how the best CPNC performs in an amorphous polymer matrix (PS), in a semicrystalline endtethered exfoliated matrix (PA-6), and in a semicrystalline not tethered matrix with a compatibiliser (PP). For these systems MMT was used with an average aspect ratio p = 200 to 300. The results indicate that independently of the matrix the CPNC modulus doubles at clay content of about 5 wt%. CPNC comprising fluoromica (FM) instead of montmorillonite (MMT) has lower modulus, most likely caused by lower aspect ratio, p. Lower modulus is also observed for systems with poorly dispersed clay platelets. In dilute systems the tensile modulus linearly increases with clay loading following the relation:
[]
E R ≡ E / Em = 1 + η φ = 1 + aw ; w ≤ 8wt%
(167)
423
Clay-Containing Polymeric Nanocomposites
Figure 133 Relative tensile modulus versus clay content for CPNC of PA-6, PP, PLA and PS with montmorillonite (MMT), as well as for PP with fluoromica (FM). See text.
Here the reinforcing factor [η] is the hydrodynamic volume, for monodispersed hard spheres (p = 1) it is equal to the Einstein value of 2.5. The parameter a = η / 100 ρ f / ρ m ~ η / 314 . The lowest predicted value for hard spheres is a = 0.008. For poorly exfoliated PP, PA-6 and polylactic acid (PLA) the values of ER increase with the slope a = 0.07. Substituting the equivalent value of [η] into Equation 107, gives the aspect ratio: p = 93. On the other hand, fully exfoliated, best performing systems follow the dependence with a = 0.20 equivalent to p = 200. The latter value is close to that expected for the commercial organoclays with MMT. It is noteworthy that the values for PP reinforced with FM can also be approximated by Equation 167 with a = 0.13 equivalent to p = 150. This value is close to p = 132 calculated from data by Zilg et al. [1999a]. It has been universally accepted that the values measured for CPNC following well-established procedures for polymeric specimens represent the inherent material behaviour. However, the same PS/organoclay or PP/organoclay/compatibiliser system tested in two laboratories resulted in not only different values, but also different tendencies for increasing clay content. Part of the answer can be found in the recent work by Uribe-Arocha et al. [2003]. The authors prepared CPNC of PA-6 (with 0 and 5 wt% clay) by injection moulding the specimens with thickness of 0.5, 0.75, 1.0 and 2.0 mm. These were subsequently tested by DMA and in a tensile tester. Both methods showed only a small effect of thickness on the performance of PA-6. However, the effects for CPNC with 5 wt% clay were severe. As an example, Figure 134 displays the dependence of the tensile modulus (E) on specimen thickness (t). The data indicate that increasing t from 0.5 to 2 mm does not affect the tensile modulus of neat PA-6, while it reduces E(CPNC) by ca. 25%. Similarly, for PA the stress at 2% strain ranged from ca. 46 to52 MPa, whereas for CPNC specimens with thickness: t = 0.5, 0.75, 1.0 and 2.0 mm the stress at 2% strain was: 84, 84, 72 and 61 MPa, respectively. Other measured parameters showed similar dependencies.
[](
424
) []
Mechanical Behaviour
Figure 134 Tensile modulus versus specimen thickness for PA-6 and its CPNC with 5 wt% clay. Data [Uribe-Arocha, 2003].
Explanation for this behaviour rests in the skin-core structure of injection moulded specimens. The data show only limited effect of skin on PA-6 behaviour – the molecular alignment near the mould surface was not too severe. In the case of CPNC, the orientation of clay platelets near the wall causes significant improvement of modulus. It is noteworthy that theoretically the ratio: E(perfect orientation)/E(random orientation) = 1/2 thus the observed decrease by 50% is reasonable. Specimens with thickness t ≤ 0.75 mm had similar stress-strain dependence, whereas there is significant reduction of yield stress, σy, for thicker specimens. SEM of these specimens tested under uniaxial loading showed multiple voiding in the core. After this initial stage (evidenced by necking) the skin with aligned clay platelets maintained the load – the specimen failed when the skin broke. During the last few years a new method of computation of mechanical properties of nanocomposites has been developed. This multiscale approach is most advanced for systems containing carbon nanotubes [Odegard et al., 2002]. The method consists of three-steps: (1) atomic/molecular dynamics (MD) modelling of the nanoparticle and its interactions; (2) construction of a representative volume element (RVE); and (3) computation of macroscopic behaviour of the nanocomposite from RVE, using either classical continuum expressions or a finite element method (FEM). The latter method was used by Sheng et al. [2004]. For CPNC the authors applied a multiscale modelling strategy assuming that: (1) at a length scale of millimeters, the structure is one of high aspect ratio particles within a matrix; (2) at micron-scale, the clay particles are either exfoliated or in the form of intercalated stacks; (3) at a nanoscale the interactions between the matrix and nanoparticles must be computed. In addition, the matrix may be amorphous or semicrystalline. In the latter case the effects of 425
Clay-Containing Polymeric Nanocomposites polymer lamellae orientation and transcrystallisation must be accounted for. Starting with the structural parameters (extracted from XRD and TEM) and platelet modulus obtained from MD, a layer of matrix surrounding the nanoparticle was modelled. Next, numerical or analytical models based on the ‘effective clay particle’ were used to calculate the CPNC elastic modulus. The new model correctly predicted the elastic moduli for CPNC with MXD6 or PA-6 matrix. The model predicted only a moderate increase in the overall modulus for exfoliated clay as compared to that comprising stacks of intercalated clay. Sheng et al. even speculated that experimentally (due to curving of clay platelets) full exfoliation might not result in the highest modulus.
3.5.2 Prediction of Tensile Strength Prediction of the tensile strength, σ, for composites or nanocomposites is more difficult than that for modulus. Strength involves transmission of stresses through the tested specimen. Thus, strength must depend on the degree of exfoliation, the type of bonding between clay platelets and the matrix, as well as on the compatibiliser. The simplest case is for continuous particles that perfectly adhere to the matrix. The tensile strength in the stress direction is expected to follow the volumetric rule of mixtures:
σ c = σ mφ m + σ f φ f
σ R ≡ σ c / σ m = 1 + φ f (σ r − 1)
(168)
σr ≡ σ f / σm where, as before, the subscripts c, m and f stand for composite, matrix and filler, respectively. An example of experimental findings is shown in Figure 135. In the figure, the relative tensile strength versus volume fraction of organoclay is shown.
Figure 135 Relative tensile strength for a biodegradable polyester of glycol + butanediol + succinic + adipic acids (Mw = 60 kg/mol) with Cloisite® 30B (circles) and Cloisite® 10A (squares). Data Lee et al. [2002b].
426
Mechanical Behaviour The CPNC contained a biodegradable polyester of glycol + butanediol + succinic + adipic acids (Mw = 60 kg/mol) as a matrix in which either Cloisite® -30B or -10A (C30B or C10A, respectively) were dispersed [Lee et al., 2002]. XRD data indicated that the former system was well dispersed, especially at high clay concentration. By contrast, the system with C10A was intercalated with d001 = 3.25 nm. Evidently, the hydroxyl group in C30B provided superior interaction between the intercalated organoclay and the matrix, while the benzyl of C10A failed. As a result the tensile strength of the C30B system followed the volumetric rule of mixtures up to at least 10 vol%, while the C10A system followed it only up to ca. 4 vol%. The latter type of behaviour with a local maximum is most common. The secret of success is to find at which composition the maximum performance is reached and optimise the process to obtain reproducible performance. According to Equation 168, the slope of the straight line in Figure 135 provides a low value for the relative strength: σr = σf/σm = 6.39, giving σf = 6.39 × σm = 83 MPa. By analogy to well described micas, the flexural strength of clay platelets should also depend on the aspect ratio, p. The tensile strength of mica with p = 100 is about σf = 240 MPa [Milewski and Katz, 1987], i.e., about three times higher than the value calculated for the clay platelets. The reduced tensile strength (as well as moduli) of the reinforcements may originate from: stress concentration at the edges, imperfect alignment with the stress direction, imperfect spacing in the matrix, imperfect interaction between platelets and the matrix, interaction between the flakes, etc. Furthermore, the failure can take place by diverse mechanisms, e.g., by fracturing the flake or by pulling it out. The full reinforcing effect can be reached only if the width of the reinforcing platelet, W, is [Clegg and Collyer, 1986]: W>Wc = tσf /σfm
(169)
where t is flake thickness and σfm is the interfacial shear strength between clay platelet and matrix. These considerations lead to revision of Equation 168 into:
σ c = σ mφ m + FRσ f φ f ; σ R = 1 + φ f ( FRσ r − 1) ⎡ tanh(u) ⎤ FR = ⎢1 − ⎥ 1 − sec h(u) u ⎥⎦ ⎢⎣
[ (
u = p Gmφ f / E f 1 − φ f
(170)
)]
where Gm is the matrix shear modulus, Ef is the flake tensile modulus and FR is the filler strength reducing factor. For the aliphatic polyester (APES)/clay system [Lee et al., 2002b] the calculated value of FR = 0.667 leads to a value for the tensile strength of clay platelets σf = 125 MPa, approaching the expected strength, but still smaller by a factor of about 1.8. It should be noted that the definition of FR in Equation 170 predicts that: 0.666 ≤ FR ≤ 0.900. Kerner’s composites theory provides the following relation for the tensile strength:
(
)(
)
σ c = σ m 1 + ABφ f / 1 − Bφ f ; where : A = (7 – 5ν m ) / (8 − 10ν m )
(
)(
B = σ f /σm −1 / σ f /σm + A
(171)
) 427
Clay-Containing Polymeric Nanocomposites The dependence is identical to that shown in Equation 157 with strength (σ) replacing the modulus (E). Finally, an interesting relation was proposed by Turcsáyi et al. [1988]. The authors corrected for the decreasing load bearing cross-section of a specimen and for the effects of the interfacial interactions:
( )( B = (1 + A ρ l ) ln(σ / σ )
) { }
σ R ≡ σ c / σ m = 1 − φ f / 1 + 2.5φ f exp Bφ f ; where : f
f
i
(172)
m
In Equation 172 Af and ρf are the specific surface area and density of the reinforcing particles, while l and σi are the thickness and strength of the interphase. It is worth noting that B < 0 for σi < σm – the other variables only scale the effect. The dependence predicted by Equation 172 is illustrated in Figure 136. For the yield stress of composites to be about equal to that of neat polymer the value of the B-parameter should be about 3. Substituting the largest possible values (Af = 750 m2/g; ρf = 5000 kg/m3 and l = 100 nm) indicates that this condition would be met if σi > 1.01σm. Thus, for nanocomposites able to adsorb and solidify the macromolecules one may expect that the tensile strength will increase with loading of clay platelets.
3.5.3 Fatigue Resistance of CPNC Properties such as modulus and strength provide information on the material behaviour at relatively low deformation under steady-state loading. In particular, the modulus (measured as the initial slope of the stress-strain curve) reflects the initial structure of the material. The other type of mechanical testing addresses the ultimate properties – how the material breaks. The three principal test methods
Figure 136 Relative tensile strength of CPNC as a function of the clay platelet volume fraction and the interaction parameter B from Equation 172 [Turcsáyi et al., 1988].
428
Mechanical Behaviour used by fracture mechanics are impact tests, steady-state deformation (at stresses exceeding the strength of the material) and fatigue [van der Giessen and Needleman, 2002]. In the latter case the fatigue crack grows when the stress intensity (cycling between σmin and σmax) is much smaller than that needed for crack growth under steady-state loading. The cycling causes failure at stresses well below the tensile strength level. Analysis of this behaviour is done by determining the stress versus number of cycles to failure (S-N diagram) curves, usually followed by studies of the mechanism of crack propagation and determination of the fracture energy. Fatigue crack growth will take place if and only if the following three conditions are met: (1) the stress amplitude, σa = σmax – σmin exceeds a critical value; (2) the maximum stress intensity σmax exceeds its critical value; and (3) there is an energy dissipation process during the cycle (fatigue fracture cannot occur in a fully elastic system without memory). Above the threshold stress value, the crack growth rate, ∂l/∂N, is a function of the stress amplitude, σa. To generate an S-N diagram the specimen may be subjected either to constant stress amplitude or to the mean stress, respectively defined as [Moet, 1986]:
σa = σmax – σmin; σmean = (σmax + σmin) / 2
(173)
Construction of the S-N curve is time consuming since for statistically valid results ca. 10 samples have to be tested at each stress level. In Figure 137 a schematic S-N plot for PS is shown [Sauer et al., 1976]. The specimens were tested at a frequency of 1600 cycles/min at constant maximum stress, σmax = 17.25 MPa, systematically varying sa from about 4.75 to 17.25 MPa. There is a substantial difference in the material behaviour when cycling is done at the same level of stress amplitude, but either fully in tension, e.g., σmin = 0 and σmax = 17.25 MPa, or in tension and compression, e.g., σmin = -8.62 and σmax = +8.62 MPa
Figure 137 Schematic representation of the S-N plot for PS at 500 Hz. After [Sauer et al., 1976].
429
Clay-Containing Polymeric Nanocomposites – the corresponding lifetime is N = 100 and 30 k-cycles, respectively. Thus, stress reversal has a deleterious effect on the specimen endurance. As shown in Figure 137, at σa ≅ 6 MPa there is a change in the slope of the S-N plot, suggesting two different mechanisms of fracture below and above this limit. Since the usual requirement is that the material survives 107 to 108 cycles, here the ‘endurance limit’ is about 4.4 MPa. The main difference in fatigue behaviour between metals and polymers is the sensitivity to test frequency. While metals are insensitive to frequency, polymers with high energy dissipation and low thermal conductivity are. Depending on the material characteristics (e.g., internal friction coefficient and thermal conductivity), frequency, ν, and stress amplitude, σa, two mechanisms have been identified. The boundary between these two is determined by the rate of energy input: (∂E/∂t) ∝ν × σa. At low rates (∂E/∂t) < (∂E/∂t)crit the material is able to dissipate the energy of deformation, thus the temperature rises only to a specific, dynamic level. Under these conditions, the fatigue fracture proceeds by the conventional fatigue crack initiation and propagation (FCP) mechanism. However, when the rate of energy input exceeds the critical value, there is an unbounded temperature increase. Under these circumstances FCP cannot proceed and the specimen fails by the thermal fatigue failure (TFF) mechanism [Crawford and Benham, 1975]. These authors developed an empirical relation between the critical value of the stress amplitude, σa, crit, and frequency, νcrit:
(∂E / ∂t ) = (∂E / ∂t )crit ⇒ σ a, crit = (c1 − c2 log ν crit )
A/V
(174)
where ci are material parameters, while A and V are specimen surface area and volume, respectively. The data schematically shown in Figure 138 are based on the reversed load fatigue tests of polyoxymethylene (POM or acetal) at different levels of frequency: ν = 0.167 to 10 Hz and stress amplitude, σa. At low frequency and low σa the conventional FCP fracture mechanism was observed and the data followed the FCP-type dependence, shown in Figure 137. However, when the frequency and stress amplitude exceeded the critical values for POM, the number of cycles to failure, N, dramatically decreases forming TFF-branches, each for a specific frequency. Evidently, the two mechanisms lead to significantly different morphology of the fractured surfaces. Stress concentration caused by a notch reduces the critical value of the energy rate, thus it increases the probability of the TFF mechanism. The FCP mechanism often starts by craze formation which, depending on the material and test conditions, leads either to cracks or shear banding, which in turn proceed to the crack propagation stage. The crack opens either in a single craze or shear band. The initiation is best described in terms of the fracture mechanics stress intensity factor, σth ∝ l / r where l is the crack length and r its radius of curvature. Growth of the crack length with the number of cycles during the fatigue test was found to be proportional to the amplitude of the stress concentration factors in the tensile mode, σKI = σI, max – σI, min:
∂l / ∂N = co Δσ lm
(175)
where co and m are characteristic constants of this ‘Paris law’ equation. Continuum theories that assume that the crack growth rate is proportional to the crack opening displacement predict the exponent value: m = 2. Nguyen et al. [2001] have
430
Mechanical Behaviour
Figure 138 S-N plot for two mechanisms of failure in POM: thermal fatigue failure (TFF) and fatigue crack propagation (FCP). After [Crawford and Benham, 1975].
developed a finite element model to simulate the fatigue behaviour of a specimen under plane strain. The material was characterised by a continuum plasticity model, with a combination of isotropic and kinematic hardening, a cohesive law that progressively softens and has loading-reloading hysteresis, which results in the crack tip blunting. The numerical calculations of the fatigue crack growth rate predicted m ≈ 3, confirmed with the experimental data for aluminium. The damage accumulation models predict the exponent value, m = 4, frequently observed for polymeric systems. The effect of exfoliated clay on fracture and fatigue behaviour has seldom been studied. For example, PA-6 and CPNC of PA-6 containing 2 wt% of organoclay (Ube resins: PA-1015B and -1015C2, respectively) were dried and injection moulded [Bureau et al., 2001; Gloaguen and Lefebvre, 2001]. The former authors measured the tensile and fracture toughness using either dried or conditioned at 50 RH specimens. The tensile properties were determined in the machine direction, while the fracture tests with a notch were carried out in the transverse direction. The tensile modulus, E, tensile strength, σy, elongation at break, εb, stress intensity factor in mode I, KIc, and the fracture energy parameter (composed of two parts, elastic and plastic: Jc = Jel + Jpl) were determined. The parameter Jc is a measure of an internal energy change per unit crack length. It characterises the fracture toughness prior to the onset of stable crack extension. Its value is independent of the in-plane dimensions. Selected results from Bureau et al. [2001] are listed in Table 62. Note that the mechanical properties of semicrystalline reinforced polymers depend on crystallinity and reinforcement. Addition of clay to PA-6 induces formation of the γ-crystalline form, which during annealing may convert to the more stable α-form (more information on the topic is in Section 3.4.2). Furthermore, injection moulded PANC specimens have skin dominated by the α-form and core by the 431
Clay-Containing Polymeric Nanocomposites
γ-form. In these nanocomposites the level of crystallinity is also higher than that in neat PA-6, which in part accounts for the increased values of E and σy as well as for the reduction of εb. The data in Table 62 show that addition of 2 wt% of organoclay (equivalent to 0.64 vol% of MMT) has a large effect on the tensile (Young´s) modulus, E, especially when the matrix is plasticised by the moisture. There is also a significant increase of the yield (or ultimate) strength, σy. However, the elongation at break, εb, is reduced – the dry CPNC specimens are particularly brittle. From the fracture test data the stress intensity factor, KIc, and the fracture energy, Jc, were computed. For CPNC the values of these parameters are also below those for neat resin. Noteworthy is the difference in the moisture effect on the fracture mechanics of PA and CPNC – while for PA the humidity changes linear-elastic behaviour into an elastoplastic one; for CPNC linear-elastic behaviour is observed for both, dry and conditioned specimens. These results are in good agreement with data reported by Gloaguen and Lefebvre [2001], who also studied the Ube material. The authors carried out tensile tests at constant strain-rate of 10-3 s-1, recording the stress, linear strain, and volume strain (a specific video-extensometer was used). CPNC showed brittle behaviour at room temperature (εb = 5-7%). At T = 80 °C (above the glass transition temperature, Tg) the elongation at break for PANC was higher than that for PA-6 (εb = 900 versus 800%). The authors also reported large volume expansion during tensile tests; about twice as large for CPNC as that for the matrix polymers, PA-6 or PP. The cavitation in CPNC was large and orientationdependent, similar to that observed in HIPS. In the authors’ opinion, the cavitation originates in local damage at the polymer/clay interface, and thus in an unspecified nanoscale defect (crystalline organisation in the vicinity of the clay platelet, localised necking and fibrillation, nanoscale interfacial cavitation, etc.). The recent publication by Bellemare et al. [2002] describes fatigue test results for the same two Ube resins (PA-1015B and PA-1015C2). The measurements
Table 62 Mechanical properties at room temperature of PA-6 and its CPNC with 2 wt% organoclay (Ube Ind. PA-1015B and 1015C2, respectively). Data [Bureau et al., 2001] Property
PA
CPNC
CPNC/PA
Symbol
Units
Dry
Cond.1
Dry
Cond.
Dry
Cond.
E
GPa
2.7
0.8
4.1
1.8
1.5
2.25
σy
MPa
74
39
100
52
1.35
1.33
εb
%
175
700
