High-throughput design and machine-learning optimization of SLM-fabricated high-entropy alloy coatings on magnesium alloys

Magnesium (Mg) alloys are limited by rapid corrosion in chloride-containing environments and the poor deformation compatibility of conventional ceramic coatings. Herein, AlCrFeCoNi high-entropy alloy (HEA) coatings were fabricated on an Mg-6Zn-2Dy substrate by high-throughput selective laser melting (SLM). A process-property database containing 157 samples was constructed using laser power and scan speed as inputs, and coating hardness and thickness as outputs. A back-propagation neural network (BPNN) model was developed to predict coating hardness and thickness, with the highest R 2 reaching 0.97, and was coupled with non-dominated sorting genetic algorithm II (NSGA-II) for multi-objective optimization. The optimized parameters were identified as approximately 298 W and 919 mm s −1 , producing a dense FCC/BCC HEA coating with a hardness of approximately 450 HV and a thickness of approximately 330 µm. The coating formed a continuous metallurgical interface with local elemental redistribution and a gradual mechanical transition, which contributed to improved yield strength, tensile strength and elongation. Electrochemical tests showed that the coating shifted Ecorr from −1.270 V to −0.681 V and reduced Icorr from 4.73 × 10 −5 to 3.64 × 10 −6 A cm −2 . After 24 h immersion in 3.5 wt.% NaCl solution, no obvious pitting or Cl − enrichment was observed. These results indicate that the high-throughput SLM-BPNN-NSGA-II framework enables efficient optimization of mechanically compatible and corrosion-resistant HEA coatings on Mg alloys.