Abstract

The paper presents Atomic Cluster Expansion (ACE) and MACE interatomic potentials trained on a new DFT dataset for indium phosphide (InP), targeting dislocation mobility. The dataset includes bulk, strained bulk, point defects, stacking faults, and dislocation quadrupoles. The models are validated against RSCAN DFT and compared with literature potentials, showing substantially improved accuracy for partial dislocation formation energies while remaining computationally efficient.

Abstract

The paper presents Atomic Cluster Expansion (ACE) and MACE interatomic potentials trained on a new DFT dataset for indium phosphide (InP), targeting dislocation mobility. The dataset includes bulk, strained bulk, point defects, stacking faults, and dislocation quadrupoles. The models are validated against RSCAN DFT and compared with literature potentials, showing substantially improved accuracy for partial dislocation formation energies while remaining computationally efficient.

Abstract

The paper presents Atomic Cluster Expansion (ACE) and MACE interatomic potentials trained on a new DFT dataset for indium phosphide (InP), targeting dislocation mobility. The dataset includes bulk, strained bulk, point defects, stacking faults, and dislocation quadrupoles. The models are validated against RSCAN DFT and compared with literature potentials, showing substantially improved accuracy for partial dislocation formation energies while remaining computationally efficient.

Abstract

The paper presents Atomic Cluster Expansion (ACE) and MACE interatomic potentials trained on a new DFT dataset for indium phosphide (InP), targeting dislocation mobility. The dataset includes bulk, strained bulk, point defects, stacking faults, and dislocation quadrupoles. The models are validated against RSCAN DFT and compared with literature potentials, showing substantially improved accuracy for partial dislocation formation energies while remaining computationally efficient.