Explainable AI supports improved nickel catalyst design for converting carbon dioxide to methane

The conversion of carbon dioxide into clean fuel is considered an important route to carbon neutrality. co₂ In particular, methanation is of increased interest due to its favourable thermodynamic properties and environmental benefits. However, large-scale deployments continue to face challenges such as insufficient catalytic activity at low temperatures and vulnerability to carbon deposition.

Researchers are currently applying an explanatory machine learning (ML) framework to support the rational design of nickel-based catalysts in CO.₂ Methanation.

This study is published in the journal ACS Sustainable Chemistry & Engineering.

Instead of relying on traditional trial and error methods, this study introduces a systematic approach to data processing, cross-validation, and construction of ensemble learning models. Among the methods tested, the catboost model achieved R² Co's value of 0.77₂ Convert, 0.75 for ch₄ Selectivity.

By analyzing the main descriptors, this study identified the optimal reaction conditions: temperatures between 250 and 350 °C, gas hourly rates below 15,000 cm³ g^-1 h^-1bet surface area of ​​50-200m² g^-1and the nickel content is above 5%.

These insights demonstrate how data-driven methods can guide catalyst optimization and shorten the pathway from laboratory research to industrial applications.

“This work shows how machine learning can help you better understand the key factors affecting CO.₂ Methanation performance,” said Hao Li, a well-known professor at Tohoku University's Institute of Advanced Materials (WPI-IAMR).

“By being able to explain the model, we not only predict outcomes, but also provide knowledge as to why certain conditions are important.”

In the future, the research team will integrate density functional theory calculations and high-throughput experimental data to build a multi-scale prediction model. It also conducts systematic experimental verifications to improve catalyst design.

“Our goal is to establish a platform that combines computational chemistry, machine learning and catalytic engineering,” explained Li. “In doing so, we want to provide practical solutions for the efficient use of carbon recycling and renewable energy.”

This study provides a perspective on how explanatory machine learning can be applied to catalytic research, supporting both the development of cleaner fuels and the broader transition to sustainable energy systems.