UvA researchers discover new salt hydrates for sustainable energy storage

Salt hydrates can store enormous amounts of heat and release it later. This makes them interesting for sustainable applications: capturing solar heat, utilizing industrial waste heat, thermochemical heat storage.

The problem: scientists estimate there could be more than 100 million possible hydration reactions worth investigating. Only about 6,000 of those have been studied experimentally or computationally so far. It’s simply not feasible to test everything one by one.

Chemical composition as starting point

Dr. Alberto Pérez de Alba Ortíz led a team of chemists, physicists, and computational scientists who developed a machine-learning system. The key: the system only needs to know a material’s chemical composition to predict its thermodynamic properties.

This is crucial because it means researchers don’t need detailed information about atomic structure. Many other approaches do. This makes the machine-learning model ideal for screening where structural data is scarce.

The system learns from elements and their ratios which heat-storage properties to expect.

Prediction and laboratory validation

The UvA team used the model to search through unexplored chemical reactions and identified new candidate salt hydrates that looked particularly promising for heat storage.

In collaboration with researchers from Radboud University Nijmegen, they subsequently synthesized some of these materials in the lab and tested them. The experimental results matched the predictions remarkably well.

This shows that machine learning can reliably guide the search for new salt hydrates—without running a million experiments first.

From screening to sustainable applications

This research shows that AI-driven screening can fundamentally accelerate the materials development process. Instead of going through a million possibilities step by step, scientists can now use predictions to identify the most promising candidates.

The researchers expect the same strategy can be applied to other advanced materials and technologies. Especially for sustainability applications, where rapid innovation is critical.

Source: website UvA