A Flexible and Generalized Constant-Potential Framework in i-PI.

Area of Science:

• Computational chemistry
• Materials science
• Electrochemistry

Background:

• Realistic simulations of electrochemical interfaces under operating conditions require constant-potential molecular dynamics.
• Existing frameworks often lack portability and extensibility due to tight coupling with specific electronic-structure codes.
• Simulations are further complicated by integer-electron constraints in some codes.

Purpose of the Study:

• To present a flexible, portable, and extensible constant-potential molecular dynamics framework.
• To enable simulations interfacing with multiple density functional theory (DFT) engines and machine-learning potentials.
• To overcome integer-electron limitations in DFT codes.

Main Methods:

• Implemented a constant-potential framework within the i-PI driver, introducing an explicit electronic degree of freedom and a potentiostat module.
• Employed a mixed-Hamiltonian interpolation scheme, running two adjacent integer-charge clients in parallel and linearly interpolating their properties.
• Validated the method on a 1D double-well model and an Al(111) surface.

Main Results:

• Demonstrated stable potential control and well-behaved charge fluctuations on model systems.
• Successfully coupled constant-potential ab initio molecular dynamics (AIMD) with enhanced sampling.
• Enabled efficient characterization of potential-dependent reactivity and free-energy landscapes for CO2 reduction on NiN4-doped graphene.

Conclusions:

• The developed framework offers a portable and scalable platform for constant-potential simulations.
• It supports diverse electronic-structure clients and has potential for machine-learning potentials.
• This work facilitates rigorous simulations of electrochemical interfaces under operating conditions.