Microstructure exploration and an artificial neural network approach for hardness prediction in AlCrFeMnNiWx High-Entropy Alloys

Abstract

The phase evolution of AlCrFeMnNiWx (X= 0, 0.05, 0.1, 0.5) High-Entropy Alloys (HEAs) during the solidification is understood by both thermodynamic simulation and experimental approach. The detailed structural and microstructural characterization of studied HEAs reveals the presence of BCC Fe–Cr–Mn rich (β1) primary phase and BCC Ni–Al-rich (β2) secondary dendritic phase. It is found that both primary and secondary BCC solid solution phases undergo spinodal decomposition, forming BCC_B2 (α1) and σ phases as well as BCC_B2 (α2) and BCC (γ) phases respectively. Interestingly, the hardness of the HEAs varies in the range 461–552.7 HV with alloying of W. The present investigation also reports the prediction of the hardness of AlCrFeMnNiWx (x = 0, 0.05, 0.1, 0.5) HEAs with the composition variation of tungsten by applying artificial neural network using various experimental data as input parameters. A back-propagation artificial neural network (ANN) model is used by taking the experimental data to understand the effect of alloying elements on the hardness. The ANN modeling results match well with experimental data with the accuracy of prediction 93.54% and error of the predicted value of 6.46%. © 2020 Elsevier B.V.