Efficient Monte Carlo Simulations of Gas Molecules Inside Porous Materials

详细信息    查看全文

• 作者：Jihan Kim ; Berend Smit
• 刊名：Journal of Chemical Theory and Computation
• 出版年：2012
• 出版时间：July 10, 2012
• 年：2012
• 卷：8
• 期：7
• 页码：2336-2343
• 全文大小：446K
• 年卷期：v.8,no.7(July 10, 2012)
• ISSN：1549-9626

文摘

Monte Carlo (MC) simulations are commonly used to obtain adsorption properties of gas molecules inside porous materials. In this work, we discuss various optimization strategies that lead to faster MC simulations with CO₂ gas molecules inside host zeolite structures used as a test system. The reciprocal space contribution of the gas鈥揼as Ewald summation and both the direct and the reciprocal gas鈥揾ost potential energy interactions are stored inside energy grids to reduce the wall time in the MC simulations. Additional speedup can be obtained by selectively calling the routine that computes the gas鈥揼as Ewald summation, which does not impact the accuracy of the zeolite鈥檚 adsorption characteristics. We utilize two-level density-biased sampling technique in the grand canonical Monte Carlo (GCMC) algorithm to restrict CO₂ insertion moves into low-energy regions within the zeolite materials to accelerate convergence. Finally, we make use of the graphics processing units (GPUs) hardware to conduct multiple MC simulations in parallel via judiciously mapping the GPU threads to available workload. As a result, we can obtain a CO₂ adsorption isotherm curve with 14 pressure values (up to 10 atm) for a zeolite structure within a minute of total compute wall time.