Thermoelectric ConversionEutectogels for Highly SensitiveSelf-Powered Sensors and Machine Learning-Assisted Temperature Identification

Endowing flexible sensors with self-powering capabilities is of significant importance. However, the thermoelectric conversion gels reported so far suffer from the limitations of insufficient flexibility, signal distortion under repetitive deformation, and insufficient comprehensive performance, which seriously hinder their wide application. In this work, we designed and prepared eutectogels by an ionic liquid and a polymerizable deep eutectic solvent (PDES), which exhibit good mechanical properties, adhesion, and excellent thermoelectric conversion and thermoelectric response performance. The Seebeck coefficient (<i>S</i>_i) can reach 30.38 mV K^–1 at a temperature difference of 10 K. To amplify the self-powered performance of individual gel units, we assembled them into arrays and further prepared temperature sensors. The combination of the K-means clustering algorithm of machine learning can filter out the noise of traditional thermoelectric sensors and improve the consistency of signals, thereby enabling the prediction of absolute temperature under the conditions of 10 or 20 K temperature difference. This study also demonstrates potential application of these eutectogels in thermoelectric self-powered sensing.