Research on control of thermal expansion of polymers has attracted significant attention, since polymers exhibit excellent mechanical and electronic properties, but suffer from high thermal expansion due to the thermal motion of their long molecular chains. Such problems can be addressed through formation of composites that contain an inorganic filler material. Filler materials reduce the thermal expansion of polymers through restriction of polymer chain motion. One particular area of interest is the introduction of negative thermal expansion (NTE) materials into polymer composites. The NTE property is expected to have an additional effect on the reduction of the coefficient of thermal expansion (CTE) of the composites. Several papers have demonstrated successful reduction of the CTE of polymer composites using cubic ZrW2O8, however, it is still unclear how much of this effect is caused by the NTE behavior, and how much is due to chain stiffening. To address whether the use of expensive NTE materials is justified, this project is designed to investigate the exact effects of NTE and chain stiffening on the reduction of thermal expansion of polymer composites. This objective was achieved through the preparation and testing of two sets of composites containing isomorphic particles with opposite thermal expansion (ZrW2O8 and ZrW2O7(OH)2¿2H2O), which possess identical chain stiffening effects.
The first goal of the project was to synthesize two different particles that have identical morphology but opposite thermal expansion, with cubic ZrW2O8 as the NTE material of choice. The initial idea was to use a-Al2O3 (corundum), which has a known positive CTE value, as the second material. This phase can be obtained through heat treatment of AlOOH at about 1100 °C. The synthesis of AlOOH with controlled morphologies based on choice of synthetic conditions has been reported. Attempts on the synthesis of AlOOH were made through two different routes. Neither of them delivered particles with similar size as cubic ZrW2O8. Additionally, it was found that the heat treatment at high temperature caused sintering of the particles, resulting in the formation of large particles. To circumvent this problem, the precursor of ZrW2O8, ZrW2O7(OH)2¿2H2O, was used as the counterpart for the comparison, since the topotactic transformation between the two phases results in unchanged morphology, giving rod-like shape for both materials.
The synthesis of ZrW2O7(OH)2¿2H2O was optimized to prepare particles with small size, high crystallinity, and good resistance to hydration after converting to the cubic NTE phase. The effects of acid concentration and reaction time were explored. The products were examined by powder X-ray diffraction (PXRD) and scanning electron microscopy, and the hydration rates were also estimated based on the PXRD patterns. Final reaction conditions were chosen as 6 M HCl at 230 °C for 7 d. The coefficient of thermal expansion was determined for ZrW2O7(OH)2¿2H2O using Pawley refinements of variable temperature PXRD data, and values of ¿a = 11 × 10-6 ± 1 × 10-6 K-1 and ¿c = 2.6 × 10-6 ± 0.3 × 10-6 K-1, respectively, were found. Rietveld refinements were carried out on PXRD patterns of both types of particles mixed with silicon to estimate their amorphous content. Results indicated that both particles were close to fully crystalline.
To improve the interaction between the particles and polymer, surface modification was carried out via in-situ polymerization in the presence of the particles using triphosgene and bisphenol A as monomers. Soxhlet extraction was used to purify the recovered particles. Thermogravimetric analysis was used to determine the surface coverage of the products and the presence of unbound polymer, and the required time for extraction was revealed to be 96 h based on the TGA results. Infrared spectroscopy was also used to examine the modified particles, which confirmed the presence of surface bound oligomers.
Optimization of synthetic conditions, including monomers ratio, reaction time and amount of particles, was carried out to obtain the highest possible coverage. It was found that the optimum ratio for the monomers is between 2.2 : 1 and 1.3 : 1. Leveling off was observed for the surface coverage after 21 h of reaction time. Smaller amounts of particles gave higher surface coverages, but resulted in very low quantities of recovered particles due to losses during recovery steps. To recover more particles from a single batch reaction, the particles were subjected to two consecutive modification steps, resulting in both high coverage and high recovered amounts. The precursor particles could be modified under the same optimum conditions found for NTE particles. The interaction between the particles and polymer was found to be improved after the modification.
Solution casting was used to prepare the composite films. A custom made glass vessel was created to provide an inert atmosphere with reduced pressure. This can lower the moisture level and increase the evaporation rate of the casting solvent, which can prevent moisture deposition and crystallization of the polymer. The interaction between the two phases was further enhanced through reprecipitation blending. Under optimized conditions, composite films loaded with bth types of particles were prepared with weight loadings ranging from 2 wt% to 25 wt%. Films with loadings above 12 wt% showed agglomeration on optical images. The homogeneity of the particle dispersion within the films was still acceptable based on combustion analysis.
Several properties of the composites were measured, including tensile properties, thermal stability, glass transition temperature and coefficient of thermal expansion. All films without agglomeration showed enhance thermal stability. On the other hand, most films with agglomeration exhibited slightly lower thermal stability. Similar trends were seen for the stress and strain at yield for both types of composites. The composites with lower thermal stability showed lower stress and strain at yield than pristine PC films, whereas the rest showed similar values for these two properties. The Young’s modulus of both types of composite films was found to slightly increase with the addition of the filler particles. All composites exhibited similar values as pristine PC. However, the local structure of the two types of the composites was revealed to be different by dynamic mechanical analysis. The films loaded with the precursor particles exhibited earlier softening than pristine PC, while a delay in softening was found for the ones loaded with NTE particles.
The coefficient of thermal expansion (CTE) was measured for the film samples at the University of Mulhouse. This instrument produced faulty numbers that required corrections for instrument contributions. The correction for instrument contributions was checked by comparing the corrected values of three selected film samples to values obtained through analysis at West Kentucky University. The composites blended with NTE particles showed consistently lower CTE values than pristine PC and decreased with increased particle loading, whereas the values of the other set of composites showed no clear trends. Overall, considering the errors associated with the CTE values, the difference caused by the NTE behavior of the particles may not be very significant. Additional samples with higher loadings need to be tested to obtain a clearer picture, and data should be collected on well calibrated instruments to reduce errors.