Improved Thermoelectric Properties of Hot-Extruded Bi-Te-Se Bulk Materials with Cu Doping and Property Predictions via Machine Learning

Cu-doped Bi2Te2.85Se0.15 bulk thermoelectric materials are fabricated using a hot-extrusion technique. The Cu atoms are found to intercalate into interstitial sites between the Te(1)-Te(1) layers, which results in a reduction in the carrier concentration, and thus increases in the related Seebeck coefficient and electrical resistivity and a decrease in the carrier thermal conductivity. A resulting ZT(max) value of 0.86 is obtained in the Cu0.05Bi2Te2.85Se0.15 sample, which is 83% higher than that of the Cu-free sample. As data-driven materials science is becoming increasingly important for materials design, quantitative information on the processing, microstructure, and properties of the hot-extruded materials is also estimated using a machine learning approach, where property predictions are conducted using an artificial neural network model. Moreover, an inverse exploration of the potential best material properties and their corresponding microstructure and processing is further attempted by using a Bayesian optimization algorithm. This study is expected to provide a new route to fabrication of high-performance Bi2Te3-based thermoelectric materials with a focus on data-driven materials design.