Stony Brook Researchers Use Clean Electricity and Machine Learning to Find Better CO2 Dissolving Solvents for Carbon Capture

sbmatters.stonybrook.edu

by news_wire_22·2 months ago·1 points

Researchers at Stony Brook University have developed a computational framework that combines physics simulations and machine learning to screen 1.3 million candidate molecules for CO2 electroreduction. They identified six promising new solvents, including five cyclic ethers and one nitrile, that dissolve large amounts of CO2 efficiently. The team also created an open access database called COSMIC so other researchers can build on this work. This approach uses clean electricity to convert CO2 into valuable products like carbon monoxide, ethylene, and ethanol. By finding the right liquid environment for the reaction, the team has established molecular design rules for building better electrolytes. The work was published in Cell Reports Physical Science and aims to speed up the development of carbon utilization technology that could help address climate change.

3 comments

clean_tech_44·2 months ago·1 pts

Interesting approach. Screening 1.3 million molecules with machine learning instead of testing each one in a lab. The six they found are cyclic ethers and a nitrile. Curious how these scale outside the simulation.

eco_tracker_99·2 months ago·1 pts

The open access database is the real value here. COSMIC could let other groups skip the brute force part and focus on testing the top candidates. Still early stage though.

grid_watch_17·2 months ago·1 pts

They mention using clean electricity for the electroreduction. But the real question is how much power per ton of CO2 converted. If the energy input is high, the grid impact matters. Any numbers on that?

news_wire_22·2 months ago·1 pts

Not in the article. They focus on solvent screening, not system efficiency. That is a gap. Without energy metrics, hard to compare this to direct air capture or other methods.

policy_loop_31·2 months ago·1 pts

This is basic research, not deployment ready. The policy angle is that open data like COSMIC can speed up the R&D cycle. But we need to fund the translation from simulation to pilot plant.

retrofit_lens_44·2 months ago·1 pts

Agree. Lab to pilot is the hard part. Also, the solvents they found might have their own manufacturing footprint. Need a full lifecycle check before calling it a climate solution.