What drives property prediction for solid-state hydrogen storage? Data or smart features?

Metal hydrides play a pivotal role in a wide range of applications, including hydrogen storage, compression, heat management, and catalysis, making them a central focus of interdisciplinary research spanning chemistry, materials science, and engineering. The performance of the metal hydride based systems is strongly governed by the thermodynamics of metal-hydrogen interactions. Among key thermodynamic properties, the equilibrium plateau pressure (Peq ) is particularly critical, as it defines the conditions under which hydrogen absorption and desorption occur, directly influencing operating conditions of the system. Traditionally, determination of P eq requires extensive experimental measurements, limiting the pace of materials discovery. In this work, we introduce EquiP, a machine learning framework designed to predict P eq as a function of temperature. Beyond single-point predictions, EquiP generates complete Van’t Hoff plots (Peq vs. 1/T), enabling rapid determination of enthalpy (∆H) and entropy (∆S) of hydride formation. We further show that incorporating structural descriptors derived from X- ray diffraction (XRD) data enhances performance of the model, with sparse training datasets. Comparative analyses demonstrate that the extended feature set improves predictive accuracy and generalizes well to unseen hydride compositions. This work demonstrates that with lim- ited data, intelligent feature design grounded in domain knowledge is the key to improving predictions of complex material properties.