Transfer Learning from Homogeneous to Heterogeneous: Fine-Tuning a Pretrained Interatomic Potential for Multicomponent Mo Alloys with Localized Substitutional Alloying

Machine learning interatomic potentials (MLIPs) are typically constructed for homogeneous crystalline systems that exhibit only minimal local deviations from equilibrium configurations. However, substitutional alloying elements in multicomponent engineering alloys are often distributed in a locally heterogeneous form. To address this, we develop a fine tuned MLIP based on the MACE foundation model, specifically tailored for Mo based dilute alloys containing one or two out of 20 substitutional elements: Cr, Fe, Mn, Nb, Re, Ta, Ti, V, W, Y, Zr, Al, Zn, Cu, Ag, Au, Hg, Co, Ni, and Hf. The model is trained on more than 7,000 non equilibrium structures derived from first principles density functional theory (DFT) calculations. The optimized large scale fine tuned model attains state of the art accuracy, with mean absolute error (MAE) and root mean square error (RMSE) of 2.27 meV/atom and 3.79 meV/atom for energy predictions, and 13.83 meV/Å and 24.26 meV/Å for force predictions, respectively. Systematic evaluation of model transferability to unseen alloying elements under different data splitting protocols demonstrates that incorporating even a modest set of new element DFT data during refinement reduces the energy MAE below ~20 meV/atom. The fine tuned models reduce the MAE by approximately 7–10 times compared to models trained from scratch, and by 10–20 times relative to zero shot foundation models. This performance gain remains consistent across varying dataset sizes (equilibrium vs. non equilibrium structures) and model scales. Our work illustrates the efficacy of transfer learning from globally homogeneous systems to locally heterogeneous multi element alloy environments, delivering a robust MLIP tool for the accelerated design of multicomponent alloys.