Composition-Only Prediction of Density and Bulk Modulus for Solid/Solid PCM Design

Solid/solid phase-change materials (PCMs) provide compact and durable thermal energy storage by using reversible structural transitions within the solid state. They avoid the typical limitations of solid/liquid PCMs such as leakage, corrosion, and large volume changes. To identify suitable candidates, a thermodynamic screening based on the CALPHAD method, using FactSage databases, was applied to locate materials with high transformation enthalpy and good thermal stability at elevated temperatures. Since CALPHAD calculations describe enthalpy and stability per mass rather than per volume, connecting these predictions to real material performance requires accurate density data, which motivates the use of data driven modelling.
In this work, data driven workflows are developed to predict key physical properties that are central to PCM design, including transformation temperature, enthalpy, density, bulk modulus, and other thermo-mechanical parameters, using only chemical composition as input. The models rely on open materials databases such as the Materials Project, OQMD, and JARVIS, and use uncertainty aware regression techniques trained on well established compositional descriptors.
The study mainly presents results for density and bulk modulus. Composition only models reproduce general chemical trends and enable rapid screening of potential PCMs according to volumetric energy density and mechanical stability. Bulk modulus values are derived both from elastic tensors and from energy volume equations of state to ensure robustness.
The illustrated case studies show how these predictions help narrow the materials search space for high-temperature cycling by supporting down-selection of chemistries with adequate stiffness and manageable density for compact reactor designs. The outcome is an extensible pipeline that provides fast property priors to PCM materials selection, reactor engineering and techno-economic studies, aligning with SuCiMaT’s emphasis on computational methods that accelerate sustainable, circular materials development.
Future work will extend these composition-based models to include temperature dependence and physical correlations among key properties, providing an internally consistent and validated multi property prediction framework.

Abstract

Erwerben Sie einen Zugang, um dieses Dokument anzusehen.

Dr. Irina Roslyakova