Nanoindentation of explosive polymer composites to ... - Workspace

2 downloads 4000 Views 350KB Size Report
*Corresponding author, email [email protected]. © 2012 Institute of .... and cube corner (CC) tips have the same included angle ..... Science Campaign 2. JDY is ...
Nanoindentation of explosive polymer composites to simulate deformation and failure J. D. Yeager*1, K. J. Ramos1, S. Singh2, M. E. Rutherford1,3, J. Majewski2 and D. E. Hooks1 Crack initiation and propagation is a common concern for molecular composites such as plastic bonded explosives (PBXs) and pharmaceutical tablets. Under compressive stresses, cracks form at contacts between crystals and propagate along crystal-binder interfaces, causing composite failure. To investigate this process, crystal-binder interfaces have been characterised and their mechanical properties tested. Here, samples were created with interfaces representative of those in PBXs and characterised with surface energy measurements and neutron reflectometry (NR). Nanoindentation was performed to simulate the deformation and cracking that occurs at crystal– crystal contacts through the binder. NR revealed that use of a plasticiser disrupts typical crystal– binder intermixing and results in a mechanically weaker interface. During nanoindentation, a plasticised binder was observed by atomic force microscopy to delaminate around indentation impressions, whereas a non-plasticised binder did not. Differences in interfacial adhesion and incompatible strain, dictated by the elastic–plastic film compliances, were used to explain the contrasting delamination behaviours. Keywords: Nanoindentation, Neutron reflectometry, Surface energy, Microstructure, Composite, Explosives, Delamination

This paper is part of a special issue on ‘Hardness across the multi-scales of structure and loading rate’

Introduction Mechanical deformation and failure of plastic bonded explosives (PBX) is a critical area of research for understanding their response to off-normal stimuli and their subsequent safety and performance properties. Typical crack propagation under applied loading proceeds along the interface between explosive crystals and the polymer binder,1–3 as shown in Fig. 1. This delamination of crystals from the polymer is a potential concern for off-normal (e.g. unintentional) deflagration and detonation of the explosive.4,5 Deformation and delamination result in regions of local anisotropy which can change the detonative properties of the PBX6 and result in ‘hot spot’ formation under relatively mild mechanical or thermal insult.7–9 Hot spots are small regions of local inhomogeneity in the PBX which are thought to contribute to initiation of explosives, often formed by crystal cracking or delamination events.10 Thus, the interaction between crystals and the binder is not only responsible for mechanical stability during normal use 1

Shock and Detonation Physics, Los Alamos National Laboratory, Los Alamos, NM 87545, USA Lujan Neutron Scattering Center, Los Alamos National Laboratory, Los Alamos, NM 87545, USA 3 School of Chemistry, University of Edinburgh, Edinburgh EH9 3JJ, UK 2

*Corresponding author, email [email protected]

ß 2012 Institute of Materials, Minerals and Mining Published by Maney on behalf of the Institute Received 16 October 2011; accepted 15 January 2012 DOI 10.1179/1743284712Y.0000000011

but also for sensitivity during off-normal events. The interface between the two materials is of critical importance but has not been extensively examined. PBX 9501 was chosen as a model explosive in terms of performance and sensitivity as there is much existing data on its response and microstructure. PBX 9501 contains y200 mm diameter explosive crystals of cyclotetramethylene-tetranitramine (HMX), 95% by weight in a nitroplasticised poly(ester urethane) (Estane 5703) matrix. bHMX is the most stable polymorph at ambient conditions and crystallises in the monoclinic space group P21/n.11,12 PBX 9501 is formulated by mixing particles of b-HMX, suspended in water, with a polymer solution. The polymer solvent is methyl ethyl ketone (MEK), and the solution contains equal parts of Estane and nitroplasticiser (NP) by weight. The mixing process deposits nitroplasticised Estane (NP/Estane) on HMX particles to create millimetre scale agglomerates or prills, and these agglomerates can then be pressed to shape to create the composite. During this pressing process, cracks initiate within crystals or at crystal interfaces (Fig. 1) as faceted crystals are forced into contact with one another through the binder.13 Once cracks initiate, they typically propagate along crystal-binder interfaces, compromising the integrity of the PBX, and lead to mechanical failure. The mechanisms of crack initiation, both within the crystal and at the crystal interface, can be explored by indenting

Materials Science and Technology

2012

VOL

28

NO

9–10

1147

Yeager et al.

Nanoindentation to simulate failure in explosive composites

1 a micrograph of PBX 9501 showing crack initiation by particle–particle contact and b schematic of simulation of crack initiation with indentation. Once cracks are initiated, they proceed primarily along c crystal-binder interfaces. Micrographs are from Skidmore et al.13 (a) and Rae et al.1 (c) and are used with permission

single crystals of HMX that have been coated by the binder through monitoring the load–depth response and post-characterisation of the indent and delamination of the binder from the crystal. Nanoindentation is a well characterised technique for probing the mechanical properties of materials on the micro- to nanoscale.14 Indentation has been applied to interfacial characterisation in traditional composites15 and elastically mismatched film systems16–18 as well as for elastic modulus and hardness information in complex anisotropic materials like the explosive cyclotrimethylene-trinitramine.19 Figure 1 schematically describes the principal benefit of simulating PBX deformation by indenting the explosive polymer samples in the configuration used. The hard indenter tip pressing through the thin polymer and into the HMX crystal is an analogue for crystal–polymer–crystal interaction, which occurs in explosive composites under compressive loading during PBX formation and under compressive stresses during use. Additionally, the micrometre scale sharpness of the indenter tip may facilitate distinguishing the interphase regions, which were observed with neutron reflectometry (NR), by variations in load–depth response. While indentation can provide high quality property measurements, a detailed understanding of the PBX mechanical behaviour requires processing–structure– property relationships. In order to link the change in processing (i.e. inclusion of the plasticiser) to the observed mechanical properties, NR was used to identify the structure of the composite at the nanoscale. We have previously used reflectometry to observe the solubility and intermixing behaviour of acetaminophen and the PBX 9501 binder, finding that the NP/Estane–acetaminophen composites had a sharp interface while the Estane–acetaminophen samples had a diffuse interface.20 The presence of the plasticiser appeared to limit the dissolution of the crystal during the dip coating process, presumably by lowering the solubility of acetaminophen in MEK. Here, we use similar reflectometry measurements to learn whether HMX binder samples exhibit the same type of structure. Because neutron penetration is limited to y500 nm, HMX thin films were used in place of bulk crystals. Comparison between the indentation and structure measurements presented here and their applicability to real PBX formulations relies on the premise that the materials are similar in chemical structure at the crystalbinder interface. Here, we simulate the PBX 9501 formulation with a dip coating process to deposit the 9501 binder on single crystals of b-HMX. The dip coating involves immersing and then slowly removing a

1148

Materials Science and Technology

2012

VOL

28

NO

9–10

crystal or film of HMX from the 9501 binder solution. This mimics the PBX 9501 formulation directly by creating a multiphase system consisting of HMX, Estane, NP, and MEK for a brief amount of time. When the HMX is removed from the solution, the MEK evaporates and leaves behind a polymer coating. This process replicates the parameters of PBX 9501 formulation that we hypothesise are most important, in particular HMX– MEK interaction, as described in greater detail in our previous work.20

Experimental Sample preparation Preparation of b-HMX single crystals and films began by dissolving HMX powder in acetone (Fisher Chemical, Pittsburgh, PA, USA). The films were prepared by dip coating silicon wafers (Silicon Sense Inc., Nashua, NH, USA) from 10 wt-% HMX–acetone solutions. b-HMX single crystals were grown from acetone solutions by evaporation and prepared for nanoindentation as described previously.19,21 Crystal morphology and size were typical for these materials, generally producing semirectangular prisms approximately 3 mm tall, 1 mm wide and 10 mm long. Flatness and roughness were evaluated by optical microscopy and atomic force microscopy (AFM). Crystals that appeared optically flat and smooth under 61000 magnification were selected as indentation candidates, and representative crystal surfaces were found by AFM to have slopes of ,2u and root mean square roughness of ,5 nm. The crystals and films were then coated with a polymer by dip coating the samples into concentrated polymer solutions. The poly(ester urethane) Estane 5703 (B. F. Goodrich, Jacksonville, FL, USA), with and without NP, was used as polymer binder. Estane 5703 is a statistical copolymer with a molecular weight average of 45 kDa and has been characterised previously with neutron scattering techniques.22 Non-plasticised polymer solutions were made by dissolving Estane pellets in MEK (Fisher Chemical, Pittsburgh, PA, USA) to make a 7 wt-% solution. Nitroplasticiser/Estane solutions were formulated by mixing equal amounts of NP and Estane pellets in MEK at 7 wt-%. The NP was manufactured in-house and is a eutectic mixture of bis-2,2-dinitropropyl-acetal (BDNPA) and bis-2,2-dinitropropyl-formal (BDNPF), usually referred to as BDNPA/F. Dip coating involved immersing and removing the HMX samples from the appropriate solution at a constant rate (50 mm min21 in and out of solution) within an environmentally regulated box. The total time in which crystals were in contact with

Yeager et al.

the dipping solution was y5 s. The dip coating procedure appeared to induce some etching of certain regions in the HMX, but defects on the crystal surface were apparent under optical microscopy and easily avoided during characterisation. The films varied in thickness and smoothness towards the edges of the sample due to the uneven drying following the dip coating, but large regions in the centre of the sample were smooth and of constant thickness. Ellipsometry was used to measure the film thickness on the samples, as in our previous work.20 Ellipsometry of the crystal–polymer samples proved difficult due to the small sample size, so the film thicknesses were assumed to be the same as identical films prepared on clean silicon wafers. Indentation of the HMX samples and silicon wafers verified this assumption as substrates were sensed at similar depths and were in agreement with the ellipsometry measurements. Film thicknesses for both Estane and NP/Estane indentation samples were y470 nm. This thickness was chosen so that the indentation data would ideally show effects from both the polymer and the crystal. The films for the reflectometry samples were required to be thinner in order to achieve useful signal to noise for the data analysis, and so for these measurements, both types of sample were y110 nm thick. The films were allowed to dry for 4 days between dip coating and indentation or reflectometry measurements.

Interfacial chemistry and structure measurements The interfacial properties of the composite samples were characterised through measurement of the surface energies and intermixing behaviour of the constituents. The surface energy was measured with a contact angle meter (ChemInstruments, Mentor, OH, USA), as in our previous study on polymer binders.23 Briefly, the total surface energy of the polymer cs can be considered as a sum of dispersive cd and polar cp components. The contact angle formed between a liquid and the polymer is a function of these components. Measuring the contact angles from several standard liquids allows for solving the unknown polar and dispersive components of the polymer surface energy with a geometric method.24 Thermodynamic work of adhesion Wa between the polymers and HMX can then be calculated using published surface energy values for HMX.25 The standard liquids used were deionised water, diethylene glycol and hexadecane. Neutron reflectometry measurements were performed at the Surface Profile Analysis Reflectometer beamline at the Los Alamos Lujan Neutron Scattering Center. The technical details of the measurement have been presented elsewhere,20,26 but the key principles are presented here. A neutron beam from a spallation Table 1 Theoretical SLDs for materials used in present study Material

SLD/61026 A˚22

Silicon Silicon dioxide Estane 5703 BDNPA/F NP/Estane HMX

2.0727 2.4627 1.3027 2.6827 1.9020 4.5828

Nanoindentation to simulate failure in explosive composites

source is focused onto the HMX–polymer sample at a very low angle, and the reflected beam intensity is measured with a neutron detector. The reflectivity R is the ratio of intensity of the incident beam to the reflected beam and is a function of the momentum transfer vector Qz, where Qz54psin (h)l21. h is the angle of incidence of the beam, and l is the wavelength of the neutrons. l is determined by measuring the time of flight for the neutrons. Analysis of the reflection of neutrons with various wavelengths enables modelling of the scattering length density (SLD) of the sample. The SLD is a function of chemistry and density of the material and is modelled as a function of depth into the sample, revealing chemical and structural information at high spatial resolution (,1 nm) through the thickness of the sample. The theoretical SLDs of the materials used in the present study are shown in Table 1.

Nanoindentation Nanoindentation was performed using a Hysitron TribonIndenter (Minneapolis, MN, USA) equipped with a low load QSM/DMA transducer and an included AFM. The AFM was manufactured by Quesant Instrument Corporation, and the combined system provides AFM to tip calibration, which enabled imaging of pre- and postindent areas. This capability was especially useful for indentifying partially recovered impressions in the compliant films. The AFM used a non-contact silicon cantilever in tapping mode, operated at the resonant frequency of 177 kHz. Several tips having the same included angle were used to study the effects of tip sharpness on the deformation of the binder/crystal interfaces. Ninety degree conical and cube corner (CC) tips have the same included angle so the impressions are comparable if indentation is performed at similar depths that are deep enough to ensure self-similar indentation [A(hc) and a(hc)!included angle]. Achieving this was difficult because of the difference in compliance between the films and the difficulty sensing first contact with the films. As a compromise, the final unloaded depth rather than the maximum was used to distinguish comparable indents.

Results Interfacial structure Contact angle measurements were taken on thin films of Estane and NP/Estane in order to determine the surface energies of each. In general, the contact angles of all three standard liquids used were quite similar between the two polymers, resulting in similar surface energies (Table 2). The Wa for each polymer to HMX is also quite similar, and indeed, the difference in adhesion is smaller than the error in the measurements. From a thermodynamic standpoint, any differences in the mechanical Table 2 Dispersive, polar and total surface energy of Estane and NP/Estane along with calculated work of adhesion to HMX: values are listed in dyn cm21 along with standard deviations Polymer

cd

cp

cS

Wa to HMX

Estane NP/Estane

28.1¡2.3 29.3¡5.1

6.8¡2.3 6.8¡2.2

34.9¡3.7 36.1¡5.7

79.6¡3.7 80.9¡5.7

Materials Science and Technology

2012

VOL

28

NO

9–10

1149

Yeager et al.

Nanoindentation to simulate failure in explosive composites

2 a reflectivity versus momentum transfer vector Q and b SLD as function of depth into sample for HMX coated with Estane. Data points in a are shown by circles with error bars indicating one standard deviation, and model fit is shown by solid line. Scale in b is abbreviated for brevity

properties of the composites are unlikely to be caused by differences in secondary bonding at the interface. The NR results, however, strongly indicate a difference in interfacial structure for the two composites. Figure 2 shows the reflectivity data along with a model fit (a) and the modelled SLD profile (b) for HMX– Estane. The SLD profile shows the silicon wafer, with a small native oxide layer, coated by a thin layer of HMX. The pure HMX has a depth of ,4 nm. A relatively large region of varying chemistry is then observed between the pure HMX and the bulk polymer, indicating intermixing of both components. As with our previous study on acetaminophen composites, this is a result of HMX– MEK interaction during the dip coating process. The HMX is soluble in a wide variety of polar solvents,29 such as MEK, though it is less soluble in general than acetaminophen. It is therefore unsurprising to observe a ˚) similar but smaller interphase in Estane–HMX (y60 A ˚ ). than in Estane–acetaminophen (y400 A While qualitative observation of the interface in Fig. 2b is instructive, quantitative analysis of the SLD

values provides insight as well. The fitted SLD of the pure HMX layer was significantly lower than the theoretical value shown in Table 1. This is similar to our previous study on acetaminophen composites and is most likely a result of the dip coating process producing a low density, polycrystalline film. In contrast, Estane ˚ 22, which is much higher had an SLD of y1?861026 A than the theoretical SLD. This is evidence of further HMX intermixing in the bulk polymer away from the interface, again due to the presence of the MEK solvent during dip coating. The dissolved HMX raises the SLD of the polymer by either increasing the density of the polymer or contributing to the neutron reflection by a rule of mixtures effect. Further study of the polymer should resolve the differences in possible explanations. Though the Estane–HMX samples showed extensive intermixing, little to no intermixing was observed in the NP/Estane–HMX samples (Fig. 3). The SLD profile in Fig. 3b shows a high spike in the SLD between the HMX and the polymer, indicating some NP has migrated from the polymer to the interface during the dip coating process.

3 a reflectivity versus momentum transfer vector Q and b SLD as function of depth for HMX coated with NP/Estane. Data points in a are shown by circles with error bars indicating one standard deviation, and model fit is shown by solid line. Scale in b is abbreviated for brevity

1150

Materials Science and Technology

2012

VOL

28

NO

9–10

Yeager et al.

The reflectometry showed a clear difference in interfacial structure between the two samples. In the NP/Estane–HMX film system, the dissolution of the HMX is limited by the NP, which has migrated to the interface. No gradually changing SLD region is observed between the HMX and the polymer, unlike in Estane–HMX. The NP/Estane segregation has been observed previously,27 though its effect on the PBX chemical inhomogeneity has only been speculated.20 Given the negligible differences in surface energies between the two polymer systems, the observed differences in interfacial structure/interphase must be a result of solvent mediated interdiffusion of the components. Though the difference in interphase is dramatic, it is unlikely that the NP completely prevents dissolution of HMX during the formulation processes. Interdiffusion of HMX and NP/Estane has been inferred previously during heating experiments30,31 but is shown here to be less preferable than interdiffusion without the plasticiser, at least at room temperature.

Nanoindentation and AFM A testing schedule was developed, as described in Table 3, to produce comparable indents with a 90u conical and a CC tip so that the effects of radii of curvature on the deformation of the interface could be studied. Note that micrometre scale indentation depths were achieved with the nanoindenter. Representative load–depth curves for all the samples and tips are shown in Fig. 4. In the Estane–HMX samples, the effect of Estane is obvious in the displacement profiles. With the conical tip, Estane exhibited ideal plastic flow, plastically flowing under indentation and retaining its deformed shape, whereas NP/Estane behaved as a rubber, stretching elastically under indentation in a very compliant manner. In all indentations with the CC tip, the effect of the polymer was not seen directly in the initial loading, indicating that the tip was piercing the compliant polymer rather than elastically deforming it. Note that in all cases, the final unloaded depth was much larger than the thickness of the polymer film, indicating that the HMX had flowed plastically along with the film. Thus, the PBX material failure scenario outlined in Fig. 1 was successfully replicated with these HMX composites. The post-indentation AFM revealed delamination in the NP/Estane samples from the conical tip, while the Estane samples showed little plastic deformation away from the indent (Fig. 5). The maximum scan size on the AFM is 40 mm, and the delamination blister appeared to extend beyond the scan. Note in particular the scan heights in the AFM images: NP/Estane which had deformed under the indenter tip recovered almost completely (height range of 198 nm) in addition to the delamination, while Estane did not recover from the deep indent (height range of 1?9 mm). Less elastic

Nanoindentation to simulate failure in explosive composites

4 Load–displacement profiles for HMX–polymer samples, where triangles represent NP/Estane, circles represent Estane, filled icons specify indentation using conical tip and open icons specify indentation using CC tip: in all cases, solid line is real data

recovery by the NP/Estane was evident from CC indentation (Fig. 6). The NP/Estane sample recovered enough from the indent to leave a residual circular impression (Fig. 6a), while the Estane sample exhibited perfect flow and retained the CC impression (Fig. 6b). With both tips, NP/Estane–HMX exhibited elastic recovery beyond that of Estane–HMX. Before testing, the degree of importance of the radius of curvature (‘sharpness’) of the tip on delamination was not anticipated. Figure 7 illustrates schematically the proposed deformation/delamination process. During indentation with the conical tip, NP/Estane stretched compliantly, causing incompatible shear strain at the interface and delamination surrounding the impression. Once unloaded, NP/Estane elastically recovered over the impression in HMX, and a delamination blister formed around the impression. During indentation with the CC tip, the sharper tip perforated the NP/Estane, limiting stretching and prohibiting incompatible shear strain at the interface so that delamination did not occur. In all cases, Estane was nearly perfectly plastic, deforming irreversibly. The nanoindentation provides evidence that the Estane/HMX interface was tougher than the NP/ Estane/HMX interface given the similar load–depth testing schedule shown in Table 3. This is certainly a result of the differences in the interfaces in the two composites. The smaller interface with concentrated plasticiser limiting interdiffusion was weak in comparison to the larger interface with extensive intermixing. While this type of relationship between interfacial structure and adhesion or toughness is typically seen in traditional composites, it has not been demonstrated

Table 3 Nanoindentation strategy for polymer–crystal samples

NP/Estane Estane

Load/mN

Depth with conical tip/nm

Depth with cube corner tip/nm

10 4. 25 10 2.75

1200 … 900–1150 …

2050 1250 1950 1000

Materials Science and Technology

2012

VOL

28

NO

9–10

1151

Yeager et al.

Nanoindentation to simulate failure in explosive composites

5 Images (AFM) of post-indent impressions from conical tip showing a elastically recovered NP/Estane–HMX sample and b its impression profile and c retained indent of Estane–HMX sample: height range of AFM image in a is 198 nm, while range in c is order of magnitude larger, 1?88 mm

previously for molecular composites, and its effect on PBX safety and performance may be substantial.

Discussion Taken together, these results have important implications for the interpretation of how PBX microstructures contribute to the deformation and failure phenomena. In particular, the crystal–crystal cracking and crystal– binder delamination previously observed in PBX 9501 (Fig. 1) have been partially explained by these nanoindentation results on simulated systems. The crack

propagation between the binder and the crystals in the real formulation is likely to be facilitated by the weakened interface as a result of the NP. It is clear that compressive stress in this material is supported almost entirely by particle–particle contacts between crystals. The NP/Estane is extremely compliant and can support negligible stress by comparison to the much stiffer HMX crystals. Whenever there is incompatible shear strain, as a result of either plastic deformation or shifting at crystal contacts, delamination occurs readily. This is entirely consistent with the observation of failure in PBX 9501 predominately by delamination along the NP/

6 Images (AFM) of post-indent impressions from CC tip showing partially recovered indent in a NP/Estane–HMX sample and b retained indent of Estane–HMX sample. Note round impression in NP/Estane compared to expected CC shape in Estane. Height range of AFM image in a is 620 nm, while range in b is 648 nm

1152

Materials Science and Technology

2012

VOL

28

NO

9–10

Yeager et al.

Nanoindentation to simulate failure in explosive composites

7 Illustration of proposed deformation/delamination process for indentation of NP/Estane–HMX samples

Estane-crystal interfaces. Similar pathways for fracture initiation and propagation may be present in other PBX materials with poor crystal–binder adhesion, whether due to a plasticiser or residual components from the formulation. Applying the results to material design, there are two obvious ways to make PBX 9501 tougher. A less compliant, more ideally irreversible plastic binder could be used or the adhesion between NP/Estane and HMX could be improved through formulation techniques or interface modifying additives. With a more plastic binder, the material will support compressive stresses and plastically flow to fill in regions where plastic deformation occurs between crystal contacts. In that case, failure of the PBX would be controlled by plastic deformation of the constituents rather than delamination at the interfaces. However, it is unclear whether a formulation with greater toughness is desirable. PBX 9501 is known to be a ‘safer’ formulation of HMX because of the high compliance of the binder composition as compared to similar formulations without plasticisers.32–34 What these results may imply is that it is not simply the compliance of the binder that matters; rather, it is the combination of compliance transferring load to crystal–crystal contacts and the prompt delamination of the binder that prevents the transfer of shear stresses to the crystals. The threshold of stresses that induce this delamination can be studied in more detail by varying the geometry of the indenter. We hypothesise that increased delamination of the NP/Estane-HMX interface can be induced using a blunter tip with a larger included angle. At similar depths, the contact radius for a tip with a larger included angle will be larger and cause greater stretching, resulting in incompatible shear strain at the interface. This approach may be useful for demonstrating a relationship between typical facet radii for crystals and the desensitisation observed when crystal facets are rounded.35 Increasing or decreasing the plasticiser content and formulation procedure will change the mechanical strength and sensitivity to impact of the PBX

through modification of the binder/crystal interface strength. The contribution of morphology of the crystals to sensitivity through both crystal-crystal contacts and crystal-binder delamination can also be studied by varying the intender tip geometry. Thus, the indentation technique can be used to both infer the safety advantages of crystal shape modification and provide criteria for idealised crystal shape and size in reduced sensitivity crystalline powder products. Such experiments are underway and will be complemented by larger scale investigations of the mechanistic manifestations of failure in varied formulations. The downside of prompt delamination as a safety feature is the fact that it opens up porosity in the PBX, leading to a material that may be more sensitive to impact than pristine material by virtue of the fact that there are an increased number of ‘hot spots’. A selfhealing interface might solve this problem, but adding materials that might enable self-healing might also have unintended consequences for the overall stability of the material over time. The methodology presented in the present research can be applied to processing–structure–property relationships in PBXs and other molecular composites. For example, pharmaceutical tablets often have poor compaction and require the use of filler material to form durable pills.36 Mechanical failure of the tablets has implications for shelf life and drug release rate.37,38 The chemistry at the drug-binder interface may be studied with reflectometry and related to mechanical properties with nanoindentation. Additionally, the molecular crystal in many molecular composites exhibits poor phase stability, often changing phase under low stresses that may be encountered during processing. Development of the nanoindentation strategy presented here could lead to fundamental understanding of binder delamination, stress distributions and therefore phase change and defect generation with the crystals and the effect of crystal morphology on crack initiation.

Materials Science and Technology

2012

VOL

28

NO

9–10

1153

Yeager et al.

Nanoindentation to simulate failure in explosive composites

Conclusions The interfacial structure and mechanical behaviour in PBXs and other molecular composites are of importance for predicting and modelling performance as theory and modelling efforts advance to capture physics at the grain scale. The methodology presented in the present work provides key insight into fundamental deformation processes and relates those processes to the properties of the interface. Reflectometry measurements showed that the Estane-HMX interface is altered by the presence of the plasticiser. The plasticiser is concentrated at the interface and results in less HMX distributed into the binder material at the interface. The NP/Estane-HMX interface is weaker due to less intermixing than in the nonplasticised Estane, facilitating delamination of the binder from the crystal at relatively low stresses during nanoindentation. In addition to the very high compliance of the plasticised binder, this prompt delamination prevents transfer of shear stress to the crystal, providing another possible explanation for why the PBX 9501 formulation has favourable safety characteristics as compared to similar HMX formulations without plasticising agents. The techniques presented and the processing–structure–property relationships determined here are applicable to the design of improved explosives and other related materials, including pharmaceutical compacts.

Acknowledgements Los Alamos National Laboratory, an affirmative action equal opportunity employer, is operated by Los Alamos National Security, LLC, for the National Nuclear Security Administration of the US Department of Energy under contract no. DE-AC52-06NA25396. The present work was performed, in part, at the Center for Integrated Nanotechnologies, a US Department of Energy, Office of Basic Energy Sciences user facility and the Lujan Neutron Scattering Center at LANSCE funded by the DOE Office of Basic Energy Sciences. Funding for the present work was provided by the DOE/ DoD Joint Munitions Project and the Explosives Science project of the National Nuclear Security Administration Science Campaign 2. JDY is supported by the Agnew National Security Fellowship. The authors particularly thank D. Bahr of Washington State University for helpful discussions on the present work.

References 1. P. J. Rae, H. T. Goldrein, S. J. P. Palmer, J. E. Field and A. L. Lewis: ‘Quasi-static studies of the deformation and failure of bHMX based polymer bonded explosives’, Philos. Trans. R. Soc. Lond. A, 2002, 458A, (2019), 743–762. 2. S. J. P. Palmer, D. M. Williamson and W. G. Proud: ‘Adhesion studies between HMX and EDC37 binder system’, AIP Conf. Proc., 2006, 845, (1), 917–920. 3. S. J. P. Palmer, J. E. Field and J. M. Huntley: ‘Deformation, strengths and strains to failure of polymer bonded explosives’, Proc. R. Soc. Lond. A, 1993, 440A, (1909), 399–419. 4. A. M. Mellor, D. A. Wiegand and K. B. Isom: ‘Hot spot histories in energetic materials’, Combust. Flame, 1995, 101, (1–2), 26–28. 5. H. L. Berghout, S. F. Son and B. W. Asay: ‘Convective burning in gaps of PBX 9501’, Proc. Combust. Inst., 2000, 28, (1), 911–917. 6. H. L. Berghout, S. F. Son, C. B. Skidmore, D. J. Idar and B. W. Asay: ‘Combustion of damaged PBX 9501 explosive’, Thermochim. Acta, 2002, 384, (1–2), 261–277.

1154

Materials Science and Technology

2012

VOL

28

NO

9–10

7. W. C. Davis and L. G. Hill: ‘Joints, cracks, holes, and gaps in detonating explosives’, Proc. 12th Int. Detonation Symp., San Diego, CA, USA, August 2002, Office of Naval Research. 220–229. 8. F. P. Bowden and A. D. Yoffe: ‘Initiation and growth of explosion in liquids and solids’, 104; 1952, Cambridge, Cambridge University Press. 9. S. M. Walley, J. E. Field and M. W. Greenaway: ‘Crystal sensitivities of energetic materials’, Mater. Sci. Technol., 2006, 22, (4), 402–413. 10. J. K. Dienes, Q. H. Zuo and J. D. Kershner: ‘Impact initiation of explosives and propellants via statistical crack mechanics’, J. Mech. Phys. Solids, 2006, 54, (6), 1237–1275. 11. H. H. Cady, A. C. Larson and D. T. Cromer: ‘The crystal structure of a-HMX and a refinement of the structure of b-HMX’, Acta Crystallogr., 1963, 16, (7), 617–623. 12. C. S. Choi and H. P. Boutin: ‘A study of the crystal structure of bcyclotetramethylene tetranitramine by neutron diffraction’, Acta Crystallogr. B, 1970, 26B, (9), 1235–1240. 13. C. B. Skidmore, D. S. Phillips, P. M. Howe, J. T. Mang and J. A. Romero: ‘The evolution of microstructural changes in pressed HMX explosives’, Proc. 11th Int. Detonation Symp., Snowmass Village, CO, USA, August–September 1998, Office of Naval Research, 556–564. 14. W. C. Oliver and G. M. Pharr: ‘Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and refinements to methodology’, J. Mater. Res., 2004, 19, (1), 3–20. 15. S.-L. Gao and E. Ma¨der: ‘Characterisation of interphase nanoscale property variations in glass fibre reinforced polypropylene and epoxy resin composites’, Composites A, 2002, 33A, (4), 559–576. 16. M. J. Cordill, N. R. Moody and D. F. Bahr: ‘The effects of plasticity on adhesion of hard films on ductile interlayers’, Acta Mater., 2005, 53, (9), 2555–2562. 17. D. F. Bahr, J. W. Hoehn, N. R. Moody and W. W. Gerberich: ‘Adhesion and acoustic emission analysis of failures in nitride films with a metal interlayer’, Acta Mater., 1997, 45, (12), 5163–5175. 18. T. D. Nguyen, J. D. Yeager, D. F. Bahr, D. P. Adams and N. R. Moody: ‘Nanoindentation of compliant substrate systems: effects of geometry and compliance’, J. Eng. Mater. Technol., 2010, 132, (2), 021001–021007. 19. K. J. Ramos, D. E. Hooks and D. F. Bahr: ‘Direct observation of plasticity and quantitative hardness measurements in single crystal cyclotrimethylene trinitramine by nanoindentation’, Philos. Mag., 2009, 89, (27), 2381–2402. 20. J. D. Yeager, M. Dubey, M. J. Wolverton, M. S. Jablin, J. Majewski, D. F. Bahr and D. E. Hooks: ‘Examining chemical structure at the interface between a polymer binder and a pharmaceutical crystal with neutron reflectometry’, Polymer, 2011, 52, (17), 3762–3768. 21. D. E. Hooks: ‘Isentropic compression of cyclotetramethylene tetranitramine (HMX) single crystals to 50 GPa’, J. Appl. Phys., 2006, 99, (12), 124901. 22. J. T. Mang, R. P. Hjelm, E. B. Orler and D. A. Wrobleski: ‘Smallangle neutron scattering of a solvent-swollen segmented polyurethane as a probe of solvent distribution and polymer domain composition’, Macromolecules, 2008, 41, (12), 4358–4370. 23. J. D. Yeager, A. M. Dattelbaum, E. B. Orler, D. F. Bahr and D. M. Dattelbaum: ‘Adhesive properties of some fluoropolymer binders with the insensitive explosive 1,3,5-triamino-2,4,6-trinitrobenzene (TATB)’, J. Colloid Interface Sci., 2010, 352, (2), 535–541. 24. D. K. Owens and R. C. Wendt: ‘Estimation of the surface free energy of polymers’, J. Appl. Polym. Sci., 1969, 13, (8), 1741–1747. 25. R. Y. Yee, A. Adicoff and E. J. Dibble: ‘Surface properties of HMX crystal’, 1981-0125cc, Naval Weapons Center, China Lake, CA, USA, 1981. 26. J. Majewski, T. L. Kuhl, J. Y. Wong and G. S. Smith: ‘X-ray and neutron surface scattering for studying lipid/polymer assemblies at the air–liquid and solid–liquid interfaces’, Rev. Mol. Biotechnol., 2000, 74, (3), 207–231. 27. G. S. Smith, C. B. Skidmore, P. M. Howe and J. Majewski: ‘Diffusion, evaporation, and surface enrichment of a plasticizing additive in an annealed polymer thin film’, J. Polym. Sci. B, 2004, 42B, (17), 3258–3266. 28. NIST: ‘Scattering length density calculator’, http://www.ncnr.nist. gov/resources/sldcalc.html (accessed 19 September 2011). 29. B. Singh, L. K. Chaturvedi and P. N. Gadhikar: ‘A survey on the cyclotetramethylene telranitramine (HMX)’, Defence Sci. J., 1978, 28, (1), 41–50.

Yeager et al.

30. L. Smilowitz: ‘On the nucleation mechanism of the b–d phase transition in the energetic nitramine octahydro-1,3,5,7-tetranitro1,3,5,7-tetrazocine’, J. Chem. Phys., 2004, 121, (11), 5550. 31. L. Smilowitz, B. Henson, B. Asay and P. Dickson: ‘The b–d phase transition in the energetic nitramine-octahydro-1,3,5,7-tetranitro1,3,5,7-tetrazocine: kinetics’, J. Chem. Phys., 2002, 117, (8), 3789. 32. K. S. Vandersall, C. M. Tarver, F. Garcia, S. K. Chidester, P. A. Urtiew and J. W. Forbes: ‘Low amplitude single and multiple schock initiation experiments and modeling of LX-04’, Proc. 13th Int. Detonation Symp., Norfolk, VA, USA, July 2006, Office of Naval Research, 904–913. 33. K. S. Vandersall, C. M. Tarver, F. Garcia, P. A. Urtiew and S. K. Chidester: ‘Shock initiation experiments on the HMX based explosive LX-10 with associated ignition and growth modeling’, Proc. APS Topical Conf. on ‘Shock compression of condensed matter’, Waikoloa, HI, American Institute of Physics. USA, June– July 2007, 1010–1013.

Nanoindentation to simulate failure in explosive composites

34. R. L. Gustavsen, S. A. Sheffield, R. R. Alcon and L. G. Hill: ‘Shock initiation of new and aged PBX 9501’, Proc. 12th Int. Detonation Symp., San Diego, CA, USA, August 2002, Office of Naval Research, 530–537. 35. A. E. D. M. van der Heijden, R. H. B. Bouma and A. C. van der Steen: ‘Physicochemical parameters of nitramines influencing shock sensitivity’, Propellants Explos. Pyrotech., 2004, 29, (5), 304–313. 36. C. Bacher, P. M. Olsen, P. Bertelsen and J. M. Sonnergaard: ‘Compressibility and compactibility of granules produced by wet and dry granulation’, Int. J. Pharm., 2008, 358, (1–2), 69–74. 37. L.-F. Huang and W.-Q. Tong: ‘Impact of solid state properties on developability assessment of drug candidates’, Adv. Drug Deliv. Rev., 2004, 56, (3), 321–334. 38. M. L. Hamad, K. Bowman, N. Smith, X. Sheng and K. R. Morris: ‘Multi-scale pharmaceutical process understanding: from particle to powder to dosage form’, Chem. Eng. Sci., 2010, 65, (21), 5625– 5638.

Materials Science and Technology

2012

VOL

28

NO

9–10

1155