Machine learning for next-generation thermoelectrics

Thermoelectricity offers a ground-breaking solution for capturing waste heat and transforming it into valuable electricity. Despite its promise, the quest for high-performance materials faces challenges due to costly and time-intensive experimental processes. This review investigates the transformative role of machine learning in accelerating material discovery and optimization in thermoelectric research. Various machine learning algorithms employed in this domain are examined, alongside advancements in predicting lattice thermal conductivity and optimizing electronic properties such as power factor, Seebeck coefficient, and bandgap. Additionally, the manuscript explores machine learning applications in device optimization to enhance efficiency and power output. In conclusion, the review outlines the current challenges and prospects of machine learning in thermoelectric research. A comprehensive analysis of machine learning's impact on diverse thermoelectric properties promises to streamline material identification and refinement processes, paving the way for efficient energy conversion.