Minimal Experimental Bias on the Hydrogen Bond Greatly Improves Ab Initio Molecular Dynamics Simulations of Water

Abstract

Experiment Directed Simulations (EDS) is a method within a class of techniques seeking to improve molecular simulations by minimally biasing the system Hamiltonian to reproduce certain experimental observables. In a previous application of EDS to ab initio molecular dynamics (AIMD) simulation based on electronic density functional theory (DFT), the AIMD simulations of water were biased to reproduce its experimentally derived solvation structure. In particular, by solely biasing the O-O pair correlation functions, other structural and dynamical properties that were not biased were improved. In this work, the hypothesis is tested that directly biasing the O-H pair correlation (and hence the O-H…H hydrogen bonding), will provide an even better improvement of DFT-based water properties in AIMD simulations. The logic behind this hypothesis is that for most electronic DFT descriptions of water the hydrogen bonding is known to be deficient due to anomalous charge transfer and over polarization in the DFT. Using recent advances to the EDS learning algorithm, we thus train a minimal bias on AIMD water that reproduces the O-H radial distribution function derived from the highly accurate MB-pol model of water. It is then confirmed that biasing the O-H pair correlation alone can lead to improved AIMD water properties, with structural and dynamical properties in even closer to experiment than the previous EDS-AIMD model.

A systematic comparison of the structural and dynamic properties of commonly used water models for molecular dynamics simulations

“…[156][157][158] While the careful testing of newer water models with older force fields for biomaterials and other molecules is one solution, another current solution is to develop new force fields for compatibility with the more recently developed water models. 4,159,160 Finally, while the accuracy limit may indeed have been reached for 3-point and 4-point non-polarizable models, other models for water are continually in development, such as the 7-point TIP7P, 161 the reactive force field ReaxFF, 162 the FFLUX polarizable and multipolar model, 163 as well as others based on DFT and machine learning, [164][165][166] to name a few. 85,167 As computing power increases over the coming years, we expect that the ability to predict and reproduce the observed properties of water molecules will also continue to improve, which in turn will improve our understanding of this remarkable solvent.…”

Classical Ab Initio Molecular Dynamics Run at an Elevated Temperature is Not a Good Model for the Nuclear Quantum Effects in Water at Ambient Temperature

“…31,32 The EDS correction was added via a modified version of PLUMED2 33 coupled to CP2K, with the EDS parameters taken from Ref. 22 , while the simulations with the SCAN exchange-correlation functional were carried out by linking the Libxc library 34 (DPMD) 36 . The BLYP-D3 energies and forces were calculated with CP2K, the forces for the DP potential were computed using LAMMPS 37 coupled to DeepMD-kit, the EDS correction was computed using PLUMED2, and the MD simulations were performed using the i-Pi force engine.…”

Using Constrained Density Functional Theory to Track Proton Transfers and to Sample Their Associated Free Energy Surface

“…It was found that excess proton and water diffusion better match experimental values after including the EDS correction. This EDS method for excess protons in water, which follows earlier work for pure water, 36 is briefly summarized in the Supporting Information , and more details will be published in the future.…”