Application of Artificial Neural Network to Predict the Crystallite Size and Lattice Strain of CoCrFeMnNi High Entropy Alloy Prepared by Powder Metallurgy

Abstract

An equiatomic CoCrFeMnNi high entropy alloy (HEA) was prepared by the gas atomization process. In addition, high-energy milling was carried out to study the effects of milling parameters on the morphology and crystallographic properties of HEA powders. Phase identification and morphology of milled powders were observed by X-ray diffraction and scanning electron microscopy, respectively. Both the atomized and milled powders exhibited a single-phase face-centered cubic solid solution. The resultant crystallite size (CS) and lattice strain (LS) of milled HEAs were estimated using the Williamson Hall method and predicted using an artificial neural network (ANN) approach. With increasing the milling time from 0 to 240 min, the CS decreased from 39.7 to 6.56 nm and the LS increased from 0.25%–1.48%, respectively. Furthermore, the developed ANN modeling provides an excellent method for the prediction of the CS and LS with excellent accuracies of 96.25% and 93.43%, respectively. Graphical Abstract: [Figure not available: see fulltext.].