Machine Learning‐Guided Design of L1 0 ‐PtCo Intermetallic Catalysts: Zn‐Mediated Atomic Ordering

ABSTRACT Promotion of atomic ordering in Pt‐based intermetallic compounds (IMCs) is a proven strategy to enhance catalytic activity and durability, for the cathode catalysts in proton exchange membrane fuel cells (PEMFCs). However, achieving higher atomic ordering typically requires elevated temperature annealing, which induces nanoparticles (NPs) sintering and surface area loss, resulting in a challenge for catalyst design. Here, we demonstrate that Zn incorporation in L1 0 ‐PtCo IMCs promotes the ordering, endowing the enhanced stability and activity. Machine learning interatomic potential (MLIP) simulations reveal that Zn lowers vacancy formation energies and modifies atomic migration, thereby accelerating ordering during annealing. These results are validated experimentally by X‐ray‐based analyses. Electrochemical measurements show that L1 0 ‐Zn‐PtCo/ZnNC achieves a mass activity (MA) of 1.76 A mg Pt −1 at 0.9 V RHE , outperforming Pt/C (0.24 A mg Pt −1 ). In single‐cell tests, it delivers 438 mA cm −2 at 0.7 V, surpassing Pt/C (293 mA cm −2 ). After 30 000 cycles, it retains 89.7% initial current density, compared with only 54.6% retention for Pt/C. By integrating ML‐guided design with experimental validation, this work establishes a rational strategy to engineer atomically ordered Pt‐based IMCs under practical conditions, advancing the development of efficient electrocatalysts.