Scientists have uncovered thousands of previously overlooked chemical reactions that could significantly improve the conversion of carbon dioxide into valuable fuels and industrial chemicals. By combining machine learning with advanced computer simulations, researchers at the Indian Institute of Science identified 9,389 individual reactions occurring on a copper catalyst, revealing unexpected chemical pathways that conventional models had completely missed.
The findings, published in the scientific journal Nature Communications, could help researchers around the world develop more effective catalysts for transforming captured carbon dioxide into useful industrial products such as methanol. Beyond improving yield predictions, the research also uncovered a surprising way hydrogen can participate in these chemical reactions, challenging long-standing assumptions about how the industrial conversion process actually works at a molecular level.
One of the most promising approaches to recycling carbon dioxide and mitigating industrial emissions is hydrogenation. This chemical process uses a solid catalyst to help carbon dioxide react with hydrogen gas, ultimately forming stable chemicals and sustainable fuels such as methanol. However, the underlying chemistry involves an enormous and complex network of individual reactions taking place across the microscopic surface of the catalyst. Simulating every single possible step using fundamental quantum mechanics requires immense computing power. Because of this technical limitation, researchers have traditionally focused on a limited selection of reactions that they broadly believe to be the most important. Unfortunately, this simplification can leave hundreds or even thousands of potentially critical intermediate reactions entirely out of the picture.
"We began with a worry familiar to anyone who does mechanistic modeling: How do you know that your reaction network has not omitted the one step that matters?" says first author Anand Mohan Verma, who conducted the groundbreaking research as a CV Raman Postdoctoral Fellow at the Indian Institute of Science and is now serving as an Assistant Professor at the Motilal Nehru National Institute of Technology Allahabad.
AI Reveals Thousands of Overlooked Chemical Reactions
To overcome this persistent computational limitation and eliminate guesswork from the equation, the research team began by using quantum mechanics simulations to build a highly detailed foundational database containing 152 core chemical reactions. They then trained sophisticated machine learning models to rapidly and accurately estimate the energy barriers that additional, unmapped reactions would need to overcome to proceed.
Using automated computational tools designed for this scale of analysis, the scientists systematically examined the vast array of possible reactions involving 105 distinct chemical species present on the catalyst surface. They also determined which chemical transformations could realistically take place within a single reaction step. This innovative approach vastly expanded their original network from just 152 documented reactions to an astonishing 9,389 reactions.
The difference in predictive accuracy proved to be profoundly significant. When the researchers relied solely on the traditional, smaller network, their computer model incorrectly identified formic acid as the primary reaction product rather than the intended methanol. Furthermore, it predicted substantially less carbon dioxide conversion than physical laboratory experiments consistently indicated.
"When we modeled the process using the 152 reactions considered initially, the network wrongly predicted formic acid, not methanol, as the major product, and underestimated how much carbon dioxide gets converted," explains corresponding author Ananth Govind Rajan, an Associate Professor in the Department of Chemical Engineering at the Indian Institute of Science. "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."
Expanded Model Predicts 40 Times More CO2 Conversion
When the scientists incorporated the vastly larger reaction network into a comprehensive mathematical model tracking how quickly these chemical processes occur, the predicted amount of carbon dioxide conversion increased by approximately 40 times compared with the original, simplified model.
The newly expanded model also correctly identified methanol and carbon monoxide as the major products of the reaction, bringing its theoretical predictions into remarkable agreement with real-world experimental findings. This vital bridge between theory and practice helps validate the accuracy of the machine-learning-driven approach.
Experimental validation for the theoretical models was carried out by G Valavarasu and Santhosh Kotni at Hindustan Petroleum Corporation Limited’s Green Research and Development Centre, along with Amol Amrute and his research colleagues at the Agency for Science, Technology, and Research in Singapore. Additionally, Ambedkar Dukkipati, a professor in the Department of Computer Science and Automation at the Indian Institute of Science, contributed significantly to the development and refinement of the machine learning models.
A Surprising Hydrogen Pathway Challenges Conventional Chemistry
Beyond dramatically improving the predictive accuracy of carbon dioxide conversion rates, the much larger reaction network exposed a completely unexpected mechanism involving how hydrogen behaves on the catalyst surface.
In several key reaction steps identified by the algorithm, hydrogen was found to transfer to intermediate chemical compounds directly as intact molecular hydrogen, rather than strictly through the traditional mechanism of separate, dissociated hydrogen atoms. Additional, rigorous quantum mechanics calculations subsequently confirmed that this unique pathway can be especially favorable when hydrogen is transferred directly to chemical intermediates containing oxygen.
This unexpected discovery challenges a foundational assumption about how hydrogen participates in catalytic chemical reactions, potentially providing vital clues for the future design of superior industrial catalysts.
"The idea that hydrogen can transfer as an intact molecule, without first splitting into atoms, runs against what most of us were taught," says co-author Shivam Chaturvedi, a doctoral student in the Department of Chemical Engineering at the Indian Institute of Science. "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 molecular hydrogen could potentially enhance pathways leading to methanol."
The researchers believe that this unexpected molecular mechanism could guide the intentional development of novel catalysts that interact much more effectively with molecular hydrogen, ultimately improving the chemical pathways responsible for producing sustainable methanol from captured carbon dioxide emissions.
Furthermore, the broader computational framework developed by the team may prove useful well beyond the specific domain of carbon dioxide hydrogenation. By successfully integrating quantum mechanics simulations, machine learning algorithms, automated identification of chemical reactions, and dynamic models of reaction rates, the methodological approach could help researchers investigate a wide range of other important industrial chemical processes.
Potential future applications for the technique include carbon dioxide reduction using different types of catalytic materials, nitrogen reduction processes for fertilizer production, and advanced water splitting methods for clean hydrogen generation. By revealing chemical steps that simpler, traditional models inevitably miss, the advanced technique offers scientists a vastly more complete picture of complex catalytic reactions and opens up entirely new opportunities to optimize the industrial processes used to produce vital fuels and chemicals.
