High-throughput design and performance validation of superior latent heat eutectic salt materials

Eutectic salts are widely utilized as latent heat storage materials in various thermal energy storage and management applications. However, developing efficient and accurate methods to predict the thermophysical properties of molten salts and guide their design remains a significant challenge. This study focuses on thermal properties, constructing an element gene feature database, and introducing an integrated machine-learning strategy for the design, screening, and evaluation of eutectic salts. Marginal contribution analysis reveals that the melting point is primarily influenced by features related to structural stability, while the fusion enthalpy is strongly associated with features affecting atomic bonding strength. Based on these insights, five novel eutectic salts are designed and experimentally validated. Among them, NaCl-NaF (66.0–34.0 mol%) exhibits a melting point of 683.7 °C and a fusion enthalpy of 588.8 J/g. High-temperature testing demonstrates that the prediction accuracies for the melting point and fusion enthalpy of the five salts are 94.49 % and 95.01 %, respectively. This study presents an effective and efficient approach for the precise targeted design of novel eutectic salts, offering expanded possibilities and options for the efficient utilization of renewable energy.