Machine-Learning-Assisted Development of Carbon Steel With Superior Strength and Ductility Manufactured by Electron Beam Powder Bed Fusion

Abstract

AbstractIn this study, based on a novel support vector machine optimization method, a wide processing window for manufacturing defect-free S25C carbon steel by electron beam powder bed fusion (EB-PBF) was identified. Samples with same energy densities exhibited similar microstructures and mechanical properties. One sample showed an optimum strength and elongation combination of 459.3 MPa and 57.6 pct. The pearlite region with irregular cementite particles was the first to crack during deformation, and the cracks gradually expanded into the surrounding area. Ferrite, cellular structures, and pearlite with parallel and straight cementite particles could effectively modulate the deformation by slip and enhance the plasticity of the S25C parts. After quenching, the strength improved to an unprecedented value of 1722.5 MPa owing to the presence of martensite and dislocation entanglements, with an elongation of 16.8 pct. The strength decreased after further tempering, and the plasticity evidently increased, with an optimum strength and elongation combination of 722.7 MPa and 44.2 pct, respectively. The microstructure of tempered sample contained lath martensite, cementite particles, and sparse dislocation lines. These results demonstrate that the current method can serve as a powerful tool for effectively optimizing the high-dimensional parameters of the EB-PBF process to produce carbon steel with excellent mechanical properties.