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

Abstract

It is a common practice in ab initio molecular dynamics (AIMD) simulations of water to use an elevated temperature to overcome the over-structuring and slow diffusion predicted by most current density functional theory (DFT) models. The simulation results obtained in this distinct thermodynamic ensemble are then compared with experimental data at ambient temperature based on the rationale that a higher temperature effectively recovers nuclear quantum effects (NQEs) that are missing in the classical AIMD simulations. In this work, we systematically examine the foundation of this assumption for several DFT models as well as for the many-body MB-pol model. We find for the cases studied that a higher temperature does not correctly mimic NQEs at room temperature, which is especially manifest in significantly different three-body correlations as well as dynamics. In many of these cases, the effects of NQEs are exactly the opposite of the effects of carrying out the simulations at an elevated temperature.

Quantum Dynamics of Water from Møller-Plesset Perturbation Theory via a Neural Network Potential

“…In addition, recent work from Voth and co-workers found that higher temperatures do not accurately replicate NQEs at room temperature, which is evident in different three-body correlations as well as dynamics. [36] Furthermore, classical molecular dynamics overestimates the vibrational frequencies of the O-H stretch region and the H-O-H bend region compared to results that take the NQEs into account. [29] Regarding the reactivity of water, "classical" water is more basic than "quantum" water (i.e., water in nature) with a pH of 8.6, which is about a 30-fold change in the ionization constant.…”

Condensed Phase Water Molecular Multipole Moments from Deep Neural Network Models Trained on Ab Initio Simulation Data

“…54 Recent work has pointed out limitations in the usage of the higher temperatures alone in AIMD classical simulations aimed at modeling the detailed impact of nuclear quantum effects. 55,56 Electrostatic interactions under periodic boundary conditions are treated with the Ewald method. The simulations are run for 30 ps with a time step of 0.5 fs in the NVT ensemble, generating 60 000 configurations along with the total potential energy E and force on each atom F i , where i is the atom index.…”

“…We use the revised Perdew, Burke, and Ernzerhof (revPBE) functional , together with the Grimme D3 dispersion correction. , A Nosé–Hoover thermostat chain of length 3 is used to maintain a temperature of 330 K to increase the fluidity and produce an oxygen–oxygen radial distribution function that agrees better with experiment . Recent work has pointed out limitations in the usage of the higher temperatures alone in AIMD classical simulations aimed at modeling the detailed impact of nuclear quantum effects. , …”

Liquid water simulation using hydrogen bond corrected SCAN and neural network potentials.

“…(NQEs) This is not a good way to account for these effects however. 9 Additionally, reproducing the sensitive experimental properties of aqueous solutions is obviously going to be difficult if there is a 30 degree discrepancy in temperature.…”