Low stress electroforming of Ni and Ni–Co alloys enabled by machine learning

Internal stress control is critical for the electroforming of high-precision components such as X-ray focusing mirrors, which typically require stress levels below 0.1 MPa to maintain shape accuracy. Meanwhile, the complex nonlinear interactions among multiple deposition parameters make stress optimization challenging using traditional methods. We propose a machine learning-assisted framework to systematically investigate and optimize the low-stress (< 0.1 MPa) electroforming of nickel and nickel-cobalt alloys from sulfamate baths. Among various machine learning algorithms, random forest is identified as the optimal predictive model. Based on this, a processing map correlating current density and cobalt concentration with low-stress conditions is established. SHAP analysis reveals that current density is the dominant factor, followed by cobalt concentration. Combined EBSD, XRD, and DFT calculations elucidate the underlying physical mechanisms. Current density induces grain refinement, while cobalt incorporation promotes solid solution formation, both of which lead to lattice expansion in the nickel structure and consequent tensile stress generation. Finally, finite element simulations guide the optimization of engineering solutions, leading to the successful fabrication of X-ray focusing mirrors with exceptional thickness uniformity (coefficient of variation < 2%)