Machine Learning-Assisted Catalyst Discovery for Hydrogen Storage

A new report from Dongguan, People's Republic of China, presents data on machine learning-assisted catalyst discovery for hydrogen storage. Researchers at Dongguan University of Technology used computational screening and density functional theory (DFT) calculations to identify promising catalysts for the decomposition of formic acid (HCOOH), a potential solution for hydrogen storage. The study highlights the importance of the ensemble effect in HCOOH decomposition and showcases the effectiveness of a data-driven approach combined with DFT calculations and microkinetic modeling for fast catalyst discovery.

Key Takeaways:

• The study identified three promising catalysts: Au8Cu5IrPd21Pt13, Pd2Au2, and Pd2Au/Pd, which exhibit superior activity to conventional Pd catalysts by three orders of magnitude at 400 K while maintaining high H2 selectivity.
• DFT calculations and microkinetic modeling showed that the rate-determining steps in HCOOH decomposition are HCOO* and COOH* dehydrogenation on Au8Cu5IrPd21Pt13, Pd2Au2, and Pd2Au/Pd, respectively.
• The study emphasizes the importance of computational screening based on adsorption energies of key intermediates in catalyst discovery.
• The researchers used the Shared Informatics Space for Synthesis Optimization (SISSO) approach to predict adsorption energies on high-entropy alloy (HEA) surfaces.
• The study received financial support from the National Natural Science Foundation of China (NSFC), Natural Science Foundation of Hebei Province, and Guangdong Provincial Key Laboratory of Distributed Energy Systems.

Statistics:

• The researchers used DFT calculations to predict adsorption energies of CO on HEA AuCuIrPdPt(111) surfaces.
• The study identified 400 K as the optimal temperature for H2 selectivity in HCOOH decomposition.
• The three promising catalysts (Au8Cu5IrPd21Pt13, Pd2Au2, and Pd2Au/Pd) exhibit activity that is three orders of magnitude higher than conventional Pd catalysts.
• The study used a data-driven approach combined with DFT calculations and microkinetic modeling for fast catalyst discovery.

Sources:

• NewsRx. Investigators from Dongguan University of Technology Report New Data on Machine Learning (Machine Learning-assisted Computational Screening of High-entropy Alloy Catalysts for Hcooh Decomposition). Journal of Engineering. October 13, 2025; p 1636.
• Journal of Materials Chemistry A (2025).
• Royal Soc Chemistry, Thomas Graham House, Science Park, Milton Rd, Cambridge CB4 0WF, Cambs, England.