Thermodynamically Consistent ML Potential for Constant-Potential Simulation Predicts C–C Bond Barrier and Charge Transfer on Cu(100) Surface

1/ AI utilization method: training a thermodynamically consistent machine-learning potential based on energy and forces
2/ Application task: prediction of interatomic interactions and potential functions under constant-potential conditions
3/ Reported result: prediction of the voltage-dependent activation barrier and charge transfer for C–C bonding on the Cu(100) surface
4/ The PRL paper proposed a thermodynamically consistent machine-learning potential (TCCP-MLP) to handle interactions under constant-potential conditions.
5/ The authors trained it using only energy and forces, and reported that the work function is obtained as an exact derivative of a consistent energy function.
6/ From the perspective of materials research, this approach is directly relevant for atomic-scale treatment of electrochemical interfaces or surface reactions where potential matters.
7/ As a demonstration, the paper reported prediction of the voltage-dependent activation barrier and charge transfer for C–C bonding on the Cu(100) surface.
8/ However, these results are reported for a specific potential and demonstration system, and the same accuracy is not automatically guaranteed for other metal/electrolyte combinations or a broader potential range.
9/ In addition, because the training relies on energy and force data, extrapolation errors can occur under conditions not covered by the training data. [1][2][3]