May 22, 2012 - F. Gao, R. Nigmatullin, S.Thompson. School of Science and Technology, Nottingham Trent University. Clifton Lane, Nottingham NG11 8NS, UK.
580
I. Preda et al.: Dielectric Response of Various Partially Cured Epoxy Nanocomposites
Dielectric Response of Various Partially Cured Epoxy Nanocomposites I. Preda, J. Castellon, S. Agnel Institut d’Électronique du Sud, Université Montpellier 2 Place Eugene Bataillon, 34095, Montpellier, Cedex 5, France
H. Couderc, M. Fréchette Institut de Recherche d’Hydro-Québec (IREQ) 1800 Lionel-Boulet Blvd., Varennes, Québec, J3X 1S1, Canada
F. Gao, R. Nigmatullin, S.Thompson School of Science and Technology, Nottingham Trent University Clifton Lane, Nottingham NG11 8NS, UK and A.-F. Vaessen Laborelec Belgium Rodestraat 125, B-1630 Linkebeek, Belgium
ABSTRACT Insufficient crosslinking and water uptake during fabrication or manipulation are known to affect the dielectric response of epoxies. Post thermal treatment may result in the completion of cross-linking, partial removal of water, and aging. In order to study the effect of manufacturing imprecision on dielectric response, several under-cured epoxybased nanocomposite samples with modified nanoclay fillers were investigated. In addition, the influence of silane coupling agents and the use of ultrasonic waves on the nanoclay intercalation were also studied. The structure of the samples and the extent of cross linking were characterized using X-ray Diffraction (XRD) and Differential Scanning Calorimetry (DSC) respectively. It was found that surface silanization lead to improved clay intercalation and higher extent of intercalation/exfoliation. The influence of post thermal treatment on the dielectric response of the materials was investigated using Broadband Dielectric Spectroscopy (BDS). Once the samples were in a stable dielectric state, relaxation maps were performed. It was found that the samples with silanized nanoclay have the lowest activation energy and they also proved to be the “strongest” vitreous materials. Index Terms — Dielectric properties, epoxy nanocomposites, nanoclay, X-ray diffraction, glass transition temperature, dielectric permittivity, dielectric loss.
1 INTRODUCTION LIQUID and solid are thermodynamically equilibrated states of matter. While cooling a liquid below its equilibrium temperature, its viscosity increases and the mobility of its molecules decreases. When reaching the crystallization temperature, which corresponds to the thermodynamic stable state of a crystal, the molecules need a specific time in order to rearrange themselves in the final crystal configuration. If the temperature decreases rapidly, the molecules will not have enough time to rearrange themselves in the crystal configuration Manuscript received on May 22, 2012, in final form December 17, 2012.
and the material will be in a supercooled liquid state. If the cooling continues, the viscosity will further increase and, at a certain temperature, known as the glass transition temperature Tg (which corresponds to a viscosity of approximately 1012 Pa·s [1]), the molecule movement will be slowed down and the material will be in a meta-stable, glassy state. At a temperature below Tg, a glassy material will continuously reorganize its structure in order to achieve a stable state. The return to the equilibrium state has several consequences such as a decrease in free volume or internal energy that determines the change in physical properties. This process is called physical ageing or structural relaxation. Unlike chemical ageing (which occurs when chemical bonds are broken), this is a
1070-9878/13/$25.00 © 2013 IEEE
IEEE Transactions on Dielectrics and Electrical Insulation
Vol. 20, No. 2; April 2013
reversible phenomenon. Physical ageing can be removed when the material is heated at a temperature above Tg. Several relaxation processes could be observed for amorphous polymers using Broadband Dielectric Spectroscopy (BDS). Each process has its average relaxation time τ [2]. The main relaxation, usually called α relaxation, is determined by the cooperative movements associated with the glass transition phenomenon. Other secondary relaxation phenomena (called β, γ …) can occur at lower temperatures and they are associated with the local movement of specific molecular groups. For inhomogeneous materials, an additional relaxation phenomenon, called the Maxwell-Wagner-Sillars (MWS) relaxation [3], can be observed. This phenomenon is due to a delay in charge transfer at the interface between components with different dielectric permittivity. Given that the variation of viscosity is a function of the average relaxation time [4], Angell suggested the representation of this variation as a function of the normalized temperature given by Tg/T [5]. This is an indicator of the sample’s behavior during its glass transition. Following this normalized scale, two extreme behaviors for vitreous materials could be identified; they could be either “strong” or “fragile”, depending on their temperature dependence. The fragility index m [6, 7] that characterizes the “fragile”/”strong” behavior can be calculated using the following equation:
m
d log Tg d T T T
(1)
g
The fragility index m varies from 16, for very strong glassforming liquids, to 250, for very fragile glass-forming liquids. Strong glass forming liquids are made by high directional bonds (van der Waals bonds). In this case the molecules resists to stress shearing motions associated with plastic deformation [6] and they are particularly brittle. For the fragile glass forming liquids, the inter-atomic or inter-molecular bonds are non-directional (covalent bonds) and they permit an easy plastic deformation [6], as planes of molecules can “slide” without disrupting their interactions. If the vitreous material is “strong”, the variation of the viscosity as a function of temperature has an Arrhenian variation; otherwise, it has a non-Arrhenian variation. In this work, we are interested in several partially cured epoxy-based materials. As the curing reaction between the base epoxy resin and a curing agent occurs under the influence of a thermo treatment [8, 9], the resin goes through several stages, from liquid to gelation (when the material will exhibit viscoelastic properties) and ends with rigidification (when the chemical network is hardened and the expected mechanical rigidity is obtained). Since high curing temperature could damage the material, low temperature is usually applied for the curing (the temperature within the material will increase as well since curing reactions are highly exothermic).
581
Afterwards, a post curing at higher temperature is needed to complete cross-linking reaction [10]. It has been demonstrated that clay/polymer nanocomposite technology can be used to improve the dielectric properties of epoxy [11-13]. In our case, the chosen organoclay, Cloisite 30B (C30B), is a natural montmorillonite (MMT) modified with a quaternary ammonium ion (MT2etOH - methyl, tallow, bis-2-hydroxyethyl, quaternary ammonium). MMT is a layered structured clay mineral comprising a central alumina octahedral sheet sandwiched between two tetrahedral silica sheets. In its natural state, a clay particle consists of a stack of thousands of layers. The layers are bounded by the van der Waals force and filled with cations, such as sodium or magnesium, plus water. Partial exchange of sodium ions with organophilic cationic surfactants can occur in order to improve the compatibility between the clay and the epoxy. The improved compatibility ensures a better exfoliation of the organoclay in the polymer. Replacing small ions such as Na+ with large structures such as MT2etOH ions increases the gallery spacing so that polymer chains can be more easily intercalated between the silicate layers [14]. Given the partial exchange, it has been reported that the organoclay produced by the ion exchange treatment can still absorb water molecules [15], since both clay surface and the surfactant have the ability to absorb water [16]. Also, the length of the organic chain will influence the hydrophobicity of the treated surface. Many studies have shown that the polymer-particle interface is a potential location for water [15-19]. Since dramatic increase of the polymer/clay interface area follows the intercalation [20], these materials are particularly vulnerable to the water uptake. Post thermo treatment was proven to have a substantial effect on the dielectric stability of the material. This is mainly associated with the contribution at the end of the curing reaction, partially removing the moisture contained in the material [16, 21-23]. As previously stated, in addition to improving the hydrophobicity, the surface treatment of the clay could also improve clay intercalation in a polymer matrix [24]. In order to further control and quantify the filler dispersion, an ultrasonic wave and a centrifugal force were found to further influence the sample’s dielectric properties [25]. In this paper, partially cured organoclay/epoxy nanocomposite samples were characterized using several methods. Clay surface functionalization was also used for two of the samples. Clay intercalation, the calorimetric properties, the effects of the extent of curing and the influence of the clay surface treatment on the dielectric response were investigated.
2 MATERIALS AND SAMPLE PREPARATION The epoxy resin used in the study was diglycidyl ether of bisphenol A based resin, commercially known as DER332. The selected hardener was poly (propylene glycol) bis (2aminopropyl ether) (PPG-b-AP) and (3-Glycidyloxypropyl) triethoxysilane (GOPTES) was used as a coupling agent. As mentioned before, the organoclay was Cloisite 30B.
582
I. Preda et al.: Dielectric Response of Various Partially Cured Epoxy Nanocomposites
To further improve the compatibility between the clay and epoxy, the clays for two of the samples were treated using silane GOPTES. In this process, 15 g of clay were dispersed in 600 ml of distilled water, using magnetic stirring for 3 h. Afterwards, 100 g of 10 wt% GOPTES solution in distilled water were stirred for 1.5 h to initiate silane hydrolysis reaction at room temperature. The solution was added to the clay suspension and the resulting mixture was stirred for 48 h at room temperature to obtain surface modified clay. The modified clays were separated from the solution by filtration and dried using freeze drying. The stoichiometric ratio of the epoxy/NH2 groups suggests a content of 34 g of PPG-b-AP for 100 g of DER332. In order to produce under-cured samples, all the samples were produced using 64 g of PPG-b-AP for 100 g of DER332. The required quantities of hardener and organoclay were mixed using a glass rod until an uniform, semi-transparent, gel-like mixture has been produced. Pot-life of the DER332/hardener mixture was estimated using viscometry and was determined to be around 90 minutes. Therefore final stages of sample preparation namely mixing, degassing and mould filling were completed in 60 minutes. A required amount of DER332 resin was introduced into prepared hardener/organoclay mixture and stirred at 333 K for 20 minutes using magnetic stirring. The prepared liquid formulation was degassed for 30 minutes in a vacuum desiccator. Two glass plates separated with a 1mm thick Teflon spacer were used as a mould for the preparation of sample plates. The mould was sealed with a bead of thermally resistant silicon sealant. Glass surfaces were coated with a mould release agent QZ13 (Robnor Resins Ltd., UK). After loading of degassed formulations into the moulds, the samples were cured at 373-378 K for 4 h. For one particular sample containing Cloisite 30B, the preparation procedure was altered in order to investigate the influence of ultrasonic treatment (sonication). The required amounts of hardener and C30B were premixed using a glass rod and sonicated for 1 min at full amplitude using an S-4000 Sonicator ultrasonic processor (Misonix, USA) equipped with a microtip. The samples were kept at room temperature in a controlled dry and UV-proof environment. Table 1. Available samples and codes. Sample
Sample code
Neat epoxy resin
NE
Epoxy resin filled with 2 wt% organoclay
EC
Epoxy resin filled with 2 wt% silanized organoclay
ECS
Epoxy resin filled with 2 wt% sonicated and silanized organoclay
ECSS
3 EXPERIMENTAL METHODS To investigate the intercalation of clay, X-Ray Diffraction (XRD) measurements were performed on treated and untreated C30B, and afterwards on composite specimens in
order to determine the changes in the interlayer distance. XRD characterization was carried out using a Philips XPert Pro XRD with a CuKaα radiation source at 40 kV and 30 mA. The samples were scanned by applying 1°/min rate with a step size of 0.008°. The X-ray wavelength was 1.540598 Å. Using the (001) diffraction peak, the interlayer distance of clay d001 was calculated using Bragg's equation:
2d sin nd
(2)
where d is the average basal distance of layered silicate, θ the diffraction angle, nd the diffraction order and λ the X-ray wavelength. Mass losses with temperature were investigated by applying the Thermogravimetric Analysis (TGA) technique. This was carried out using a Pyris Dyamond TG/DTA produced by Perkin-Elmer. The apparatus was calibrated using the Indium and Zinc fusion temperatures. All the experiments were conducted first under inert nitrogen atmosphere, between 323 K and 873 K, using a heating rate of 10 K/min. After a 5-minute isotherm at 873 K, for measurements performed at temperatures ranging from 873K to 1073 K (with the same heating rate), dry air was used in order to oxidize the residual carbon. Differential Scanning Calorimeter (DSC) measurements were conducted using a Thermal Analysis Q20 heat flow calorimeter in order to obtain qualitative and quantitative information about thermal phenomena that occur during the material’s phase transition or during its structural evolution. The apparatus was calibrated using the Indium fusion temperature and enthalpy. All of the experiments were conducted under inert nitrogen atmosphere at a heating/cooling rate of 10 K/min, with the following experimental procedure being applied for all the samples: an initial heating run, from 223 K to 523K, continued with an isotherm for 10 minutes at 523 K in order to remove thermal history, followed by a cooling run down to 223 K and ending with a second heating run to 523 K. Dielectric relaxation phenomena were investigated using Broadband Dielectric Spectroscopy (BDS). The equipment used was a Novocontrol Alpha-A Analyzer with applied frequency from 10-1 to 106 Hz. Samples were produced in disc shapes, with thickness of approximately 1 mm and a radius of around 20 mm. They were studied in a parallel plate configuration. In addition, 50 µm thick silver foil electrodes were used to ensure a good contact between the sample and the brass electrodes of the instrument. A sinusoidal voltage of 1Vrms was applied. For each point, the final values for the obtained parameters represent the average value of 10 consecutive measurements. All the measurements were carried out under nitrogen with a ±0.01 K deviation in temperature. The following measurement protocol was used for all the available samples: an initial characterization was performed at 293 K, followed by 38 h of post thermo treatment at 373 K (when the permittivity was measured each hour), and finally a dielectric relaxation map of the sample was performed from 223 to 373 K, with a 5 K step. Using BDS measurements, the real and imaginary parts of the complex dielectric permittivity were obtained. The
IEEE Transactions on Dielectrics and Electrical Insulation
Vol. 20, No. 2; April 2013
complex dielectric function that connects the two parts is given by the following formula:
* ( f ) ( f ) i ( f )
(3)
where f is the frequency of the applied voltage, ε' the real or conservative part and ε'' the imaginary or loss part. The data was analyzed by fitting the relaxation peaks with the Havriliak and Negami equation [26].
* ( )
(1 (i ) HN ) HN
(4)
where ε∞ is the high frequency permittivity, Δε the dielectric relaxation strength, τ the relaxation time , αHN and βHN the Havriliak-Negami parameters that describe the symmetric and asymmetric broadening of the relaxation time distribution function (0< αHN 0) and ω the angular frequency (ω=2π·f). The following function was considered for the frequency-dependent conductivity contribution [27]:
* ( ) B(i ) n 1
Ea k BT
(6)
where τ0β is a pre-exponential factor, Ea the activation energy of the β relaxation process and kB the Boltzmann’s constant. At high temperature, the α relaxation process is dominant. In this case, the variations of the molecular mobility as a function of temperature can be described using the Vogel – Tamman – Fulcher (VTF) empiric law [28]:
0 exp
B T TK
error was calculated using (8):
error %
log(reali ) log( fittedi ) 1 n 100 n i 1 log(reali )
(8)
where n represents the number of measured frequencies for the imaginary permittivity at a specific temperature, reali represents the measured value for a specific frequency and fittedi represents its value using the fitted function.
4 EXPERIMENTAL RESULTS AND DISCUSSION 4.1 X-RAY DIFFRACTION RESULTS
(5)
where B represents the intensity of the conductivity phenomenon and n the frequency dependence (0
