Massively Parallel GPU-Accelerated String Method for Fast and Accurate Prediction of Molecular Diffusivity in Nanoporous Materials

The diffusivity of guest molecules in nanoporous materials is instrumental for practical applications ranging from gas separation to catalysis and energy storage. Conventional methods to predict diffusion coefficients are computationally demanding, in particular for polyatomic molecules with small diffusivity in nanoporous materials. In this work, we have implemented a massively parallel graphic processing unit (GPU)-accelerated string method to calculate the minimum energy path for the diffusion of polyatomic molecules in nanoporous materials. The GPU parallelization enables fast prediction of molecular diffusivity in nanoporous materials, speeding up the computation by a factor of over 500 in comparison with serial CPU calculations. The massively parallel GPU-accelerated string method yields diffusion coefficients in excellent agreement with results from molecular dynamics while reducing the computational cost by several orders of magnitude. It will thus open up opportunities for high-throughput screening and inverse design of nanoporous materials.

Automated summary

The study introduces a GPU-accelerated method for predicting the diffusion paths of polyatomic molecules in nanoporous materials, significantly reducing computational costs and improving efficiency. The results show excellent agreement with molecular dynamics, opening up opportunities for high-throughput screening and inverse design of nanoporous materials.