Machine Learning‐Driven Optimization of Thermoelectric Materials Laser‐Printed on Flexible Substrate

ABSTRACT The development of flexible thermoelectric (TE) generators is essential for powering next‐generation wearable and Internet‐of‐Things (IoT) devices. Yet, conventional fabrication routes for benchmark Bi 2 Te 3 ‐based materials are restricted to rigid and small‐area devices. Emerging printing methods like laser powder bed fusion (LPBF) offer scalability, flexibility, and freeform shaping, but at the expense of jeopardizing performance. In this study, machine learning (ML) is applied to optimize the processing of LPBF‐fabricated Bi 0.5 Sb 1.5​ Te 3 ‐based materials printed on a flexible substrate. The developed ML tool identifies Sb concentration, printing atmosphere, and, to a lesser extent, laser energy, as the key processing parameters governing performance. The algorithm also prescribes the exact parameter values that maximize performance. The ML model not only supports the trend observed from the exhaustive traditional search of the experimental space, but it also enables a slight further improvement in average power factor, reaching a consistent ∼1280 µW m −1 K −2 value. A flexible 4 cm 2 ‐large, printed TE energy harvesting module produced with the optimized materials displays a power output of 44 µW at Δ T = 30 K. This work advances data‐driven TE materials manufacturing and enhances the understanding of LPBF processing‐property relationships, paving the way for next‐generation high‐performance, flexible TE generators.