Researchers Use AI and Clean Electricity to Identify Breakthrough CO₂-Dissolving Solvents for Climate Technology
Researchers at Stony Brook University have developed a new data-driven approach that could accelerate climate change mitigation by identifying optimal solvents for capturing and converting carbon dioxide (CO₂) using clean electricity.
The study, led by doctoral researcher Kuldeepsinh Raj and Professor Nav Nidhi Rajput from the Department of Materials Science and Chemical Engineering, focuses on improving CO₂ electroreduction, an emerging technology that uses electricity to transform captured carbon dioxide into useful chemicals and fuels such as carbon monoxide, ethylene and ethanol.
The research, published in Cell Reports Physical Science, addresses a major bottleneck in carbon conversion systems: the difficulty of identifying suitable liquid electrolytes that control how CO₂ dissolves, reacts and converts into valuable products. Because millions of possible molecular combinations exist, traditional experimental testing would take decades to complete.
To overcome this, the research team developed a computational framework combining physics-based simulations, chemistry modelling and machine learning techniques. Using this system, they screened around 1.3 million candidate molecules to identify the most promising solvent structures for efficient CO₂ processing.
From this large dataset, the researchers identified six previously untested solvents—five cyclic ethers and one nitrile—that demonstrated strong CO₂ solubility and favourable transport properties. These characteristics are essential for improving reaction efficiency and scaling up electrochemical carbon conversion technologies.
The study also uncovered fundamental molecular design principles that explain why certain solvents perform better than others. These insights provide what the researchers describe as “design rules” for developing next-generation electrolytes tailored for carbon utilisation systems.
Importantly, the data and models have been made openly available through an online database called COSMIC (CO₂ Solvent Materials Informatics Collection), allowing researchers worldwide to build on the findings and accelerate innovation in carbon conversion technologies.
Professor Dilip Gersappe, Chair of the Department of Materials Science and Chemical Engineering, said the work demonstrates how advanced computational tools can be combined with fundamental science to address urgent environmental challenges. He highlighted that making the models openly accessible will support global collaboration in clean energy research.
The breakthrough adds to growing international efforts to develop scalable carbon capture and utilisation technologies powered by renewable electricity, offering potential pathways to both reduce atmospheric CO₂ and produce valuable industrial chemicals.
