Efficient parametrization and deployment of constant potential models based on automatic differentiation: application...

Area of Science:

• Computational Chemistry
• Materials Science
• Electrochemistry

Background:

• Electrochemical interfaces are crucial for catalysis, but their simulation is complex.
• Accurate modeling requires efficient methods for parameter optimization and property prediction.

Purpose of the Study:

• To introduce a differentiable charge equilibration (Qeq) approach within a JAX-based framework for molecular force fields.
• To enable efficient optimization of force field parameters for electrochemical systems.
• To investigate CO2 adsorption on M-C-N catalysts in aqueous electrolytes.

Main Methods:

• Integration of a differentiable Qeq method into the JAX-based Differentiable Molecular Force Field (DMFF) package.
• Automatic differentiation of properties with respect to atomic Qeq parameters.
• Simulations of CO2 adsorption on M-C-N catalysts using explicit solvent models.

Main Results:

• The surface charge of M-C-N catalysts significantly influences interfacial water molecule orientation and hydration layers.
• A negatively charged electrode optimizes CO2 adsorption by balancing electrostatic forces and water competition.
• Distinct hydration layers are observed, impacting catalytic activity.

Conclusions:

• The developed differentiable Qeq approach provides a versatile framework for optimizing force fields in electrochemical catalysis.
• Understanding interfacial water behavior is critical for designing efficient CO2 reduction catalysts.
• Explicit solvent simulations are essential for accurately modeling electrochemical interfaces.