University of Amsterdam Drastically Accelerates Heat Storage Material Discovery with Machine Learning, Rapidly Identifying Promising Salt Hydrates for Thermochemical Storage

September 6, 2026

University of Amsterdam Netherlands

Overview

University of Amsterdam scientists have developed a new machine learning approach that significantly accelerates the discovery of heat storage materials. This method requires only the chemical composition of potential storage materials to evaluate and select suitable salt hydrates for thermochemical heat storage. By predicting the most promising candidates with AI instead of testing millions of possibilities one by one, new materials can be discovered more rapidly and efficiently, enabling breakthroughs in energy storage technology.

In Depth

Key Findings

Scientists at the University of Amsterdam have developed a novel approach that dramatically accelerates the discovery process for heat storage materials by leveraging machine learning (ML). Their research, published in ‘Nature Communications Materials,’ demonstrates the effectiveness of an ML method that uses only chemical composition information to efficiently evaluate and select salt hydrates suitable for thermochemical heat storage systems. This groundbreaking technique enables a shift from the time-consuming process of testing millions of candidates individually to a more efficient material discovery where AI rapidly predicts the most promising candidates.

Technical / Clinical Details

The developed machine learning approach learns complex relationships between composition and performance from existing databases of thermochemical heat storage materials. Specifically, this model can predict critical thermochemical properties such as hydration heat, phase transition temperature, stability, and cycle life from the basic elemental composition and structural descriptors of materials. Using this ML model, the research team explored a vast space of unknown salt hydrate candidates, identifying a small number of promising materials most suitable for thermochemical heat storage from millions of virtual candidates. This process enables the discovery of materials with specific performance targets (e.g., high energy storage density, specific operating temperature range) in dramatically shorter times and at lower costs compared to traditional experimental-driven exploration. It resolves development bottlenecks by significantly reducing the number of candidates sent for experimental validation.

Background & Context

Efficient energy storage is a critical challenge in building sustainable energy systems. Technologies that can store surplus heat energy, such as solar thermal energy and industrial waste heat, and utilize it when needed, contribute to improving energy efficiency and reducing CO2 emissions. Thermochemical heat storage is a promising technology with higher energy density compared to sensible or latent heat storage, but the discovery of suitable storage materials (especially salt hydrates) has been a major factor impeding its practical application. The introduction of machine learning provides a powerful means to overcome this material search challenge and accelerate the development of new heat storage solutions.

Strategic Significance & Outlook

This research by the University of Amsterdam demonstrates the applicability of AI not only in the field of thermochemical heat storage materials but also across a wide range of materials science, including other energy storage technologies, catalysts, and high-performance composite materials. This machine learning approach will shorten the “trial-and-error” period of material development, promoting faster innovation. In the future, AI is expected to predict even more advanced material properties and optimize the entire synthesis process, dramatically improving the cost and performance of energy storage systems, and providing essential technologies for the realization of a sustainable energy society.

Source: https://hims.uva.nl/content/news/2026/09/machine-learning-successfully-predicts-new-materials-for-storing-heat.html?origin=QHAZwjzdTzivRzgW9Q2Rng

Get our weekly technology intelligence — free

Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.

Subscribe Free — Weekly Tech Intelligence

By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.