The effect of moisture on the mechanical and thermophysical properties of the crosslinked network of the SU-8 photoresist.

A. Goldberg, A.R. Browning, T. Morisato, T. Vadicherla, M.D. Halls
United States

Keywords: molecular dynamics simulations, SU-8 photoresist, crosslinking, elastic constants, glass transition temperature, grand canonical Monte Carlo


With the fast growth in computer power, molecular modeling has great potential for developing new materials for different industries. Schrödinger Materials Science Suite software stands in the best position for these developments. This study is devoted to the SU-8 photoresist epoxy-based material that, when crosslinked, exhibits excellent mechanical and thermophysical properties. An efficient crosslinking algorithm that is based on the chemical reaction when two bonds are broken, and two new bonds are formed leads to the construction of the SU-8 epoxy network with a high crosslinking degree. After every formation of new bonds, an equilibration process derived from molecular dynamics at elevated temperatures is applied. This procedure establishes the accurate density of the crosslinked system. Following up on the construction of the SU-8 epoxy network, molecular dynamics simulations were applied to investigate the moisture effect on the mechanical and thermophysical properties of the generated systems. To account for the different ways crosslinking can occur, different models are elucidated. In the simulation, the density (), Young’s modulus (E), some other important elastic constants, and glass transition temperature (Tg) are determined. These properties are compared with the experimentally available data, and a discussion on the differences between dry and wet as well as cured and uncured systems is presented. The obtained results show good agreement with the experiment. The dry crosslinking network is much more rigid compared to the initial unlinked system. Although the density of the cured network is slightly smaller compared to the initial system, the elastic constants obtained using different models show considerably higher values. The density shows a monotonic increase with increasing moisture content. The Young’s modulus (E) displays an initial increase followed by a decrease for 3 models out of 4. To estimate the maximum water loading into the generated network, a grand canonical Monte Carlo method is applied to simulate the amount of water that penetrates the system vacancies. The cured and non-cured water loading systems are compared. The Tg values of cured and non-cured systems are also discussed. This work will be beneficial for the design of the epoxy-based molecule networks in predicting their structural stability, especially when the moisture effect is of concern.