Machine learning-assisted defect control in LPBF CuCrZr: Unifying process, structure, and performance

The Laser Powder Bed Fusion (LPBF) of copper alloys, particularly CuCrZr, is critical for high-performance thermal and electrical applications but remains challenging due to the material’s high reflectivity and thermal conductivity. These properties frequently induce lack of fusion defects and porosity, severely compromising mechanical and functional performance. This study aims to overcome these metallurgical constraints and maximize the densification, electrical conductivity, and corrosion resistance of CuCrZr. By employing a data-driven optimization framework coupling the ExtraTrees algorithm with a Genetic Algorithm, an optimal processing window for defect minimization was identified. Subsequently, the model-recommended process window was experimentally validated using tiered builds, contrasting suboptimal density targets (97–99%) against an ML-optimized condition. Archimedes' measurements, optical metallography, SEM, and EBSD analysis confirmed that the optimized parameters yielded a near-full relative density of 99.8%, accompanied by reduced internal defects, fewer defect-interrupted regions, and improved crystallographic continuity. Furthermore, 3D optical profilometry revealed improved surface integrity under the ML-optimized parameters, evidenced by lower areal roughness and reduced peak–valley excursions compared with the suboptimal reference. Microhardness measurements further confirmed enhanced mechanical integrity, with the optimized condition exhibiting the highest hardness and improved uniformity. Beyond structural integrity, the optimized CuCrZr demonstrated enhanced electrical performance, reaching 24.1% IACS with an electrical resistivity of 7.16 × 10⁻⁶ Ω·cm, together with improved electrochemical stability, as evidenced by a charge-transfer resistance of 4.81 × 10⁴ Ω·cm² measured by electrochemical impedance spectroscopy. These findings elucidate the critical process–structure–property relationships in CuCrZr and establish a robust pathway for fabricating dense, multifunctional copper components using conventional infrared LPBF systems.