Machine Learning Prediction of the Entropy Forming Ability for Synthesizability in High Entropy Borides

High-entropy materials have a plethora of useful properties that make them a topic of interest in many areas of research. A major descriptor in synthesizing these materials is the entropy-forming ability (EFA). This describes how likely a high-entropy material is to be synthesized in a single-phase form. It is time-consuming to calculate the EFA, so we used machine learning models to predict the EFA of high-entropy carbides (HEC) and borides (HEB). The HEC compounds already have literature, so they were used to confirm what we already knew and to ensure the validity of our models. We found that HEB and HEC compounds containing Chromium have the same positive correlation between the EFA and the mean ionic character, while those without Chromium have the same negative correlation. Random forest and XGBoost models were successfully built to predict the EFA values of the HEC and HEB compounds.