Written by: Basant El-Faramawy
A research team led by Tohoku University, in collaboration with the University of Tokyo, has successfully developed the Digital Hydrogen Platform (DigHyd), an advanced database aimed at solving technical dilemmas related to hydrogen storage and release, paving the way for a cleaner and more sustainable future.
The new platform includes over 30,000 data entries extracted from more than 4,000 experimental publications. Using these resources, the team was able to create a physical map that correlates material properties with hydrogen storage capacity and equilibrium pressure, helping to identify the optimal conditions under which materials absorb and release hydrogen.
Data Integration and Artificial Intelligence in Scientific Research
Building this reliable database relied on integrating AI-assisted literature mining with precise human verification to ensure data quality. The researchers focused specifically on metal hydrides, organizing 4,643 measurements of storage capacity and 652 pairs of enthalpy and entropy measurements, which are thermodynamic quantities that help predict how hydrogen storage behavior changes with temperature.
The team used the GoodRegressor methodology, a method based on symbolic regression of artificial intelligence, to identify mathematical relationships between the data. Tests showed that this method achieves predictive performance comparable to complex models, but with the added advantage of providing mathematical equations that scientists can interpret to understand the underlying physical mechanisms.
The Relationship Between Material Properties and Storage Efficiency
The resulting map revealed an interconnected hierarchy of material properties that affect hydrogen storage. The results showed that lighter host atoms generally enhance hydrogen storage capacity by weight. It was also found that the mechanical properties of the material play a crucial role; when hydrogen enters the metal lattice, the structure must deform and expand, and harder materials resist this deformation, increasing the energy required and affecting hydride stability and equilibrium pressure.
These findings indicate the necessity of considering storage capacity and equilibrium pressure as interconnected elements when designing new materials, rather than studying each property in isolation. The team is currently seeking to expand the database and identify promising material combinations for experimental testing, with the goal of accelerating access to practical and efficient hydrogen storage solutions.
