(342l) Workflow Development for First-Principles Molecular Simulations in CP2K

Reproducibility in molecular simulations has become a prominent issue in recent years [1,2]. With the growing complexity of systems being studied, it is becoming more challenging for researchers to disseminate their computational details and methods in traditional forms. Moreover, modern simulation codes have a large number of control parameters beyond the underlying force field, and these parameters can affect the outcome of the simulations in a qualitative manner near a phase transition or in a quantitative manner when high data precision is desired. To tackle these problems, the Molecular Simulation Design Framework (MoSDeF) is being developed to increase the reproducibility and transparency of the simulations [2,3]. With MoSDeF, simulation workflows for complicated molecular simulations can be developed, which streamlines the process of setting up and performing the simulations. The workflows can also be used to automatically perform multiple simulations at different state points, decreasing the possibility of human error. Once the workflow is made available (e.g., as supporting information of a research article), it can be used by other researchers in the field to set up their simulations. Therefore, MoSDeF simulation workflows help create TRUE (Transparent, Reproducible, Usable by others, and Extensible) simulations.

We developed workflows for first-principles molecular dynamics (FPMD) and Monte Carlo (FPMC) simulations in CP2K, an open-source quantum chemistry package that can be used to perform simulations with the interactions described by Kohn-Sham density functional theory (KS-DFT). Our workflows make the process for setting-up, running, and analyzing certain types of first principles simulations convenient and straightforward.

In this poster, we present MoSDeF simulation workflows for two scientific applications.

Molecular dynamics simulations to probe the liquid-phase structure of dihalogens

Using the workflow developed for FPMD simulations in CP2K, we conducted MD simulations in the canonical ensemble using two different generalized gradient approximation exchange-correlation functionals to investigate the structure of liquid Cl₂, F₂, and ClF. The structure factor calculated from simulations for Cl₂ and F₂ shows good agreement with the experimental results. We developed a consistent and transferable set of criteria to detect halogen bonds that is exclusively based on the nuclear positions, making its application to force field-based simulations straightforward.

Gibbs ensemble Monte Carlo simulations for vapor–liquid equilibria and gas adsorption in BCR-704 zeolite

Argon and nitrogen adsorption isotherms are commonly used to characterize the pore size distribution in nanoporous materials. However, for certain materials, the argon and nitrogen adsorption isotherms do not yield interchangeable results. BCR-704 is a calcium aluminosilicate zeolite that is used as a reference material for zeolites with strong adsorption sites. From the argon, and nitrogen adsorption isotherm experiments for BCR-704, experimental measurements show that the relative pressure at which the onset of the steep increase in loading occurs for nitrogen is much lower than that for argon [4]. Reproducing the argon and nitrogen adsorption isotherms in BCR-704 with transferable force fields proved challenging [5]. FPMC simulations in the Gibbs ensemble were conducted to obtain an estimate for the saturated vapor pressures of argon and nitrogen and to investigate the adsorption of these gases in BCR-704. In addition, the spatial distributions of adsorbate molecules are compared to force-field-based simulations.

References

1. Baker M. 1500 scientists lift the lid on reproducibility. Nature. 2016;533(7604):452.

2. Thompson MW, Gilmer JB, Matsumoto RA, Quach CD, Shamaprasad P, Yang AH, Iacovella CR, McCabe C, Cummings PT. Towards molecular simulations that are transparent, reproducible, usable by others, and extensible (TRUE). Mol. Phy. 2020;118(9-10):e1742938.

3. Cummings PT, McCabe C, Iacovella CR, Ledeczi A, Jankowski E, Jayaraman A, PalmerJC, Maginn EJ, Glotzer SC, Anderson JA, Ilja Siepmann J, Potoff J, Matsumoto RA,Gilmer JB, DeFever RS, Singh R, Crawford B. Open-source molecular modeling software in chemical engineering focusing on the Molecular Simulation Design Framework. AIChE Journal. 2021;67(3):e17206.

4. Ross RB, Brennan JK, Frankel KA, Moore JD, Moore JD, Mountain RD, Ahmad R, Thommes M, Shen VK, Schultz NE, Siderius DW, Smith KD. Perfluorohexane adsorption in BCR-704 Faujasite zeolite benchmark studies for the seventh industrial fluid properties simulation challenge. Fluid Phase Equilib. 2014;366:141–145.

5. Bai P, Ghosh P, Sung JC, Kohen D, Siepmann JI, Snurr RQ. A computational study of the adsorption of n-perfluorohexane in zeolite BCR-704. Fluid Phase Equilib. 2014;366:146–151.