Improving Monte-Carlo and Molecular Dynamics Simulation Outcomes Using Temperature-Dependent Interaction Parameters: The Case of Pure LJ Fluid

Abstract

The main proposition of this work is that introducing temperature-dependent interaction parameters (TDIP) instead of using temperature-independent interaction parameters (TIIP) may lead to improvement in the prediction of phase equilibrium properties such as vapor liquid equilibria, and transport properties e.g., self-diffusivity. Published second virial coefficient data was used to fit a simple two parameter temperature-dependent model for the collision diameter and well depth. This fitting procedure reduces the Root Mean Square Deviation (RMSD) between the experimental and predicted second virial coefficients by tenfold compared to the best TIIP and by 15 fold with the literature values. The vapor–liquid coexistence curve for argon was simulated in the NVT Gibbs ensemble in the temperature range: 110–148 K. The critical temperature and density were determined using the Ising-scaling model. The TDIP simulations produce, in general, a more accurate phase diagram compared to the diagram generated using TIIP. RMSD is reduced by 42.1% using TDIP. Also, there was no significant difference between the results obtained using TDIP and the highly accurate and computationally demanding phase diagrams based on three body contributions implementing Axillrod–Teller correction. Self-diffusivities of atomic argon were evaluated using the mean square displacement or the Einstein method using equilibrium molecular dynamics (MD) at a pressure of 13 bars and a temperature range from 90 K up to 135 K in the isobaric, isothermal NPT ensemble. TDIP, in general, produces more accurate self-diffusivities than the values computed by TIIP simulations. RMSD is reduced by about 64% using the temperature-dependent parameters.