Modeling Equilibrium Solid-Liquid Interfaces under Effective Constant Chemical Potential Using Machine Learning...

Area of Science:

• Computational chemistry
• Materials science
• Physical chemistry

Background:

• Chemical potential (μ) is crucial for interfacial processes.
• Conventional molecular dynamics (MD) struggles to maintain constant chemical potential, especially under dynamic conditions.
• Solute migration to interfaces can alter bulk concentrations, limiting simulation accuracy.

Purpose of the Study:

• Introduce iterative quasi-constant chemical potential molecular dynamics (iqCμMD) for simulating targeted molar concentrations.
• Enable MD simulations of interfaces under bulk conditions comparable to experimental settings.
• Provide a computationally efficient method to overcome limitations of fixed-composition MD.

Main Methods:

• Developed and implemented the iqCμMD simulation approach.
• Applied iqCμMD with machine learning interatomic potentials (MLIPs) for Na2SO4(aq)-graphene interface.
• Validated transferability using classical force fields for NaCl(aq)-air and NaCl(aq)-graphite interfaces.

Main Results:

• iqCμMD efficiently achieves target bulk ion concentrations within two iterations.
• MLIP-driven iqCμMD enables converged results with smaller-scale simulations.
• Results are comparable to previous CμMD simulations and DFT-level accuracy.

Conclusions:

• iqCμMD offers a robust and simple computational framework for constant chemical potential simulations.
• The method requires only the ability to converge interfacial simulations with a measurable bulk region.
• Combines iqCμMD with MLIPs for accurate modeling of solid-liquid interfaces under effective constant chemical potential.