First-principles molten salt phase diagrams through thermodynamic integration

Tanooj Shah,Kamron Fazel,Jie Lian,Liping Huang,Yunfeng Shi,Ravishankar Sundararaman

2023-06-05

Abstract:Precise prediction of phase diagrams in molecular dynamics (MD) simulations is challenging due to the simultaneous need for long time scales, large length scales and accurate interatomic potentials. We show that thermodynamic integration (TI) from low-cost force fields to neural network potentials (NNPs) trained using density-functional theory (DFT) enables rapid first-principles prediction of the solid-liquid phase boundary in the model salt NaCl. We use this technique to compare the accuracy of several DFT exchange-correlation functionals for predicting the NaCl phase boundary, and find that the inclusion of dispersion interactions is critical to obtain good agreement with experiment. Importantly, our approach introduces a method to predict solid-liquid phase boundaries for any material at an ab-initio level of accuracy, with the majority of the computational cost at the level of classical potentials.

Materials Science,Chemical Physics

What problem does this paper attempt to address?

The main objective of this paper is to develop a method to accurately predict the solid-liquid phase boundary of molten salts (such as NaCl) using first-principles calculations combined with thermodynamic integration techniques. Specifically, the paper attempts to address the following key issues: 1. **Improving the accuracy of phase diagram predictions**: Traditional methods face challenges in predicting phase diagrams in molecular dynamics simulations because they require long time scales, large spatial scales, and precise descriptions of interatomic potentials. The proposed method uses thermodynamic integration (TI) from low-cost force fields to neural network potentials (NNPs) trained on density functional theory (DFT) to quickly predict the solid-liquid phase boundary under first-principles. 2. **Evaluating the accuracy of different DFT exchange-correlation functionals**: The paper uses the proposed TI method to compare the accuracy of several different DFT exchange-correlation functionals (such as PBE, PBESol, etc.) in predicting the phase boundary of NaCl and finds that including dispersion interactions is crucial for obtaining good predictions consistent with experimental results. 3. **Developing a general method to predict the solid-liquid phase boundary of materials**: The study proposes a new method that can predict the solid-liquid phase boundary of any material with first-principles level accuracy, while most of the computational cost remains at the classical potential level. 4. **Combining low-cost force fields with high-accuracy neural network potentials**: The paper introduces a method that uses low-cost classical force fields (e.g., Fumi-Tosi potentials) to perform most of the complex path transformations, and then performs additional bulk phase transformations from classical potentials to neural network potentials to obtain the melting point of the neural network potentials. In summary, the paper aims to provide an efficient and accurate computational framework for predicting the solid-liquid phase boundary of molten salts (particularly NaCl), which will aid in the design of new nuclear reactors and thermal energy storage systems.

First-principles molten salt phase diagrams through thermodynamic integration

Modeling Solvation Thermodynamics in Molten Salts with Quasichemical Theory and Ab Initio-Accurate Deep Learning-Accelerated Simulations

Interfacial heat and mass transfer at silica/binary molten salt interface from deep potential molecular dynamics

A DFT accurate machine learning description of molten ZnCl2 and its mixtures: 1. Potential development and properties prediction of molten ZnCl2

Liquid-Vapor Phase Equilibrium in Molten Aluminum Chloride (AlCl3) Enabled by Machine Learning Interatomic Potentials

Atomistic Simulation Framework for Molten Salt Vapor-Liquid Equilibrium Prediction and its Application to NaCl

Best Practices for Fitting Machine Learning Interatomic Potentials for Molten Salts: A Case Study Using NaCl-MgCl2

$\textit{Ab initio}$ construction of full phase diagram of MgO-CaO eutectic system using neural network interatomic potentials

Transferable Deep Learning Potential Reveals Intermediate-Range Ordering Effects in LiF-NaF-ZrF4 Molten Salt

Scalable molecular simulation of electrolyte solutions with quantum chemical accuracy

Powerful predictability of FPMD simulations for the phase transition behavior of NaCl-MgCl2 eutectic salt

Thermodynamic Description of the K, Be//F, Cl Salt System with First-Principles Calculations

A Molecular Approach for Predicting Phase Diagrams of Ternary Aqueous Saline Solutions

Elucidating the Local Structure and Properties of Molten Na2CO3-K2CO3 Salts Using Machine Learning-Driven Molecular Dynamics

Machine Learning Aided High-Throughput First-Principles Calculations to Predict the Formation Enthalpy of Σ Phase

Molecular Dynamics Simulations of Molten Magnesium Chloride Using Machine‐Learning‐Based Deep Potential

A DFT Accurate Machine Learning Description of Molten ZnCl2 and Its Mixtures: 2. Potential Development and Properties Prediction of ZnCl2-NaCl-KCl Ternary Salt for CSP

A deep potential molecular dynamics study on the ionic structure and transport properties of NaCl-CaCl2 molten salt

Molecular dynamics simulations on AlCl3-LiCl molten salt with deep learning potential

Actinide Molten Salts: A Machine-Learning Potential Molecular Dynamics Study

Thermodynamic and Transport Properties of LiF and FLiBe Molten Salts with Deep Learning Potentials