High-Entropy Alloy Design for Extreme Environment Applications: Synthesis, Characterisation, and Property Prediction Using Machine Learning Guided by Nigerian Materials Research Infrastructure

This study develops a machine learning-guided design methodology for high-entropy alloys (HEAs) targeted at extreme service environments including high temperature, radiation, and corrosion, and demonstrates the experimental synthesis and characterisation workflow feasible within Nigerian university research infrastructure. HEAs represent a paradigm shift from conventional alloy design, and developing indigenous design and fabrication capability is a strategic priority for Nigeria's advanced materials research community. A database of 1,450 HEA compositions with published experimental mechanical, thermal, and oxidation properties is assembled from the literature. Machine learning models including random forest, gradient boosting, and artificial neural network algorithms are trained on the dataset with input features derived from elemental property descriptors and CALPHAD-calculated mixing entropy, enthalpy, and valence electron concentration. The best-performing model is used to screen a compositional design space of CrMnFeCoNi-based alloys with systematic substitutions, predicting Vickers hardness, yield strength, and phase stability. Three predicted high-performance compositions are synthesised by vacuum arc melting at the study institution, homogenised, and characterised by XRD, SEM, EDS, and tensile testing. Corrosion resistance in 3.5 percent NaCl and simulated marine atmosphere is evaluated. Model predictions for the synthesised compositions show mean absolute errors of 18 HV for hardness and 24 MPa for yield strength. Keywords: high-entropy alloy, machine learning, alloy design, extreme environment, Nigeria.