Force field development

Accuracy of predictions made by molecular simulations is crucial for reliability of computational materials discovery. Reliable molecular simulation requires adequate descriptions of interaction parameters (i.e., force field), which is composed of molecule-molecule and framework-molecule interactions. Currently, force fields widely used in the literature are not developed in systematic approaches and can lack accuracy in the predictions of adsorption or transport phenomena. For this reason, our group attempts to develop systematic, efficient, and robust methodology for both the molecule-molecule and framework-molecule interactions.

1. Molecule-molecule interaction

Molecule-molecule interaction can be well-represented with an accurate molecular model of interest. Recently, we proposed a methodology to develop molecular models which not only involves all the possible model parameters in the optimization but also is computationally efficient.¹ (see Figure 1(a) for the flow diagram of methodology for the development of molecular models) With this, we developed several molecular models for gas molecules relevant to energy or environment issues we are facing today, including CO, CO₂, COS, H₂S, N₂, N₂O, and SO₂.² Our calculations show that our developed molecular models show improved accuracy in the adsorption predictions. For instance, for CO adsorption in nanoporous material, such as in Mg-MOF-74 as can be seen in Figure 1(b), our CO model is able to accurately reproduce the binding geometry predicted with quantum mechanical method (i.e., DFT), whereas previously reported CO models are not able to. We have named these molecular models as electrostatic potential optimized molecular models (ESPMMs) and are aiming to extend our methodology to other important gas molecules.

Figure 1. (a) Flow diagram of the methodology for molecular model development. (b) Binding geometry of CO adsorbed in Mg-MOF-74 predicted with DFT calculations by Lee et al.³ and by ESP-MM.

2. Framework-molecule interaction

We have proposed and developed a new methodology to parameterize framework-molecule interaction respectively for CO₂, N₂, and H₂O adsorbed in nanoporous materials such as Mg-MOF-74. (see Figure 2(a) for the flow diagram of methodology for parameterization of framework-molecule interaction)^4,5 Our results show significant improvements in the predictions of adsorption isotherm of gas molecules in Mg-MOF-74, as can be seen in Figure 2(b) for CO₂. Despite its high accuracy, our group strives to improve our proposed methodology in order to develop force fields that could be applicable to all types of porous materials and gas molecules.

Figure 2. (a) Flow diagram of force field development for framework-molecule interaction. (b) CO₂ adsorption isotherm in Mg-MOF-74 simulated with the DFT-derived force field in comparison with the experimental results. Simulation results using UFF is provided for comparison. (Adopted from Lin et al. (2014).⁵

References

(1) Cho, E. H.; Lin, L.-C. Systematic Molecular Model Development with Reliable Charge Distributions for Gaseous Adsorption in Nanoporous Materials. J. Mater. Chem. A 2018, 6 (33), 16029–16042.

(2) Cho, E. H.; Lin, L.-C. Electrostatic Potential Optimized Molecular Models for Molecular Simulations: CO, CO₂, COS, H₂S, N₂ , N₂O, and SO₂. J. Chem. Theory Comput. 2019, 15 (11), 6323–6332.

(3) Lee, K.; Howe, J. D.; Lin, L.-C.; Smit, B.; Neaton, J. B. Small-Molecule Adsorption in Open-Site Metal–Organic Frameworks: A Systematic Density Functional Theory Study for Rational Design. Chem. Mater. 2015, 27 (3), 668–678.

(4) Dzubak, A. L.; Lin, L.-C.; Kim, J.; Swisher, J. A.; Poloni, R.; Maximoff, S. N.; Smit, B.; Gagliardi, L. Ab Initio Carbon Capture in Open-Site Metal–Organic Frameworks. Nat. Chem. 2012, 4 (10), 810–816.

(5) Lin, L.-C.; Lee, K.; Gagliardi, L.; Neaton, J. B.; Smit, B. Force-Field Development from Electronic Structure Calculations with Periodic Boundary Conditions: Applications to Gaseous Adsorption and Transport in Metal–Organic Frameworks. J. Chem. Theory Comput. 2014, 10 (4), 1477–1488.