Computational Discovery of Salt Hydrates for Thermal Energy Storage

Abstract

Thermal energy storage (TES) has the potential to improve the efficiency of many applications, but has not been widely deployed. The viability of a TES system depends upon the performance and stability of its underlying storage material; improving the energy density of TES materials is an important step in accelerating the adoption of TES systems. Salt hydrates are a promising class of TES materials due to their relatively high energy densities and their reversibility. Despite their promise, relatively few salt hydrates have been characterized by a handful of experimental screening studies, presenting the possibility that new hydrate compositions with superior properties may exist. These new hydrates may be salt hydrates that have been observed experimentally, but have not been studied for TES, or they may be ‘hypothetical’ salt hydrates that have not yet been synthesized. Furthermore, current understanding of salt hydrate thermodynamics is limited to an approximation that thermodynamic properties of salt hydrates are additive and uniform. This study uses first-principles calculations, machine learning, and numerical system models to characterize new salt hydrates and identify TES materials that can out-perform known compounds, as well as develop deeper understanding about the features that affect the thermodynamics of salt hydration. Focusing first on experimentally known, but under-characterized salt hydrates, high-throughput density functional theory calculations were performed on metal halide hydrates and hydroxides mined from the Inorganic Crystal Structure Database. In total, 265 hydration reactions were characterized with respect to their thermodynamic properties, gravimetric and volumetric energy densities, and operating temperatures. Promising reactions were identified for three temperature ranges: low (< 100°C), medium (100–300°C), and high (> 300°C). Several high energy density reactions were identified, including the dehydration of CrF3•9H2O, a compound which appears to be unexplored for TES. Correlations linking TES performance with dozens of chemical features for salt hydrates were quantified using a Pearson correlation matrix. These analyses reveal property-performance relationships involving energy densities, but were less successful in explaining the thermodynamics of hydration. In salt hydrates, the energy densities depend strongly on the water capacity of the hydrate, although no strong correlations were found for the enthalpy of dehydration. Based on these correlations, design rules for hydration-based TES systems are proposed. Subsequently, the focus of this study turned to hypothetical salt hydrates. The energy densities, turning temperatures, and thermodynamic stabilities of 5292 hypothetical halide salt hydrates, 1779 hypothetical chalcogenide salt hydrates, and 5233 hypothetical complex anion salt hydrates are predicted using high-throughput density functional theory calculations. Several salt hydrates from each of these classes are identified as new, stable TES materials with class-leading energy densities and operating temperatures suitable for use in domestic heating and intermediate-temperature applications. The promising performance of these materials is demonstrated at the system level by parameterizing an operating model of a solar thermal TES system with data from the new hydrates and predicting system-level energy densities. Finally, machine learning models for salt hydrate thermodynamics are developed for each of the salt hydrate classes and used to identify design guidelines for maximizing energy density. In total, the new materials and design rules reported here are expected to foster the adoption of TES systems.