IISc researchers use machine learning and computation to unveil mechanisms for CO2-to-fuel conversion

Researchers at the Indian Institute of Science (IISc) have developed a data-driven computational framework that maps nearly 10,000 chemical reactions involved in converting carbon dioxide (CO₂) hydrogenation into fuels and chemicals on a copper catalyst.

According to IISc, scientists are increasingly exploring CO₂ hydrogenation, where CO₂ reacts with hydrogen over a catalyst to give rise to products like methanol and convert it into useful chemicals and fuels. However, this process involves thousands of tiny chemical steps happening on the catalyst surface.

“To model these steps computationally, researchers traditionally select a relatively small number of likely reactions, as modelling every possible reaction using quantum mechanics is prohibitively expensive. But this means that hundreds of important reactions may be missed,” IISc said.

New approach

However, this new approach by the institute to develop a data-driven computational framework could help scientists better understand and eventually design effective catalysts for turning CO₂ into chemicals and fuels.

IISc said that to develop this framework, the researchers first generated a carefully curated, extensive database of 152 reactions using quantum-mechanical simulations. Then, they trained machine learning models to rapidly predict activation energy barriers for additional reactions. In addition, they used automated tools to predict all possible reactions involving 105 surface species and to predict which of these would occur as single-step reactions. This allowed the researchers to expand the original network to 9,389 elementary reactions.

“When we modelled the process using the 152 reactions considered initially, the network wrongly predicted formic acid, not methanol, as the major product, and underestimated how much CO₂ gets converted. Only when we expanded the network to include thousands of additional, previously overlooked reactions did the predictions fall in line with what we and others see experimentally,” said corresponding author Ananth Govind Rajan, associate professor in the Department of Chemical Engineering, IISc.

Kinetic model

When incorporated into a kinetic model, the expanded network predicted an approximately 40-fold increase in CO₂ conversion and correctly identified methanol and carbon monoxide as the major products, consistent with experimental observations.

It further added that the larger network also revealed that, in several key steps, hydrogen can be transferred to reaction intermediates directly as molecular H₂, rather than only through individual hydrogen atoms.

“Quantum-mechanical calculations confirmed that this pathway can be particularly favourable for hydrogen transfer to oxygen-containing intermediates,” it added.

“The idea that hydrogen can transfer as an intact molecule, without first splitting into atoms, runs against what most of us were taught. This surfaced only because the network was large enough to allow for it, and the observation held up when we went back and computed those steps explicitly. This also suggests that catalysts that interact more strongly with H₂ could potentially enhance pathways leading to methanol,” said co-author Shivam Chaturvedi, a PhD student in the Department of Chemical Engineering, IISc.