MIT broke a battery trade-off by shrinking the electrolyte molecule, not redesigning it

Sodium-metal batteries are attractive because sodium is roughly 1,000 times more abundant than lithium and about a hundredth the cost per pound, but sodium metal is highly reactive, making it hard to get both long-term stability and fast charge/discharge in the same battery. The electrolyte — the component that ferries charged ions between electrodes — is usually the bottleneck: most electrolytes react with the electrodes in ways that build up insoluble compounds, block ion transport, and eventually kill the battery.

In 2021, the same MIT group had identified DMTMSA, a sulfonamide solvent that stayed stable at both electrodes in lithium batteries — described by group leader Ju Li as 'magically stable.' The open question for sodium batteries was whether a related molecule could keep that stability while also being small enough to move sodium ions quickly, since bulkier solvent molecules slow ion transport the way a large suitcase slows someone moving through a crowd.

Rather than searching broadly, the team constrained the search to 'congeneric' molecules — structurally and electronically similar to DMTMSA — reasoning that family membership was more likely to preserve stability. PhD student Chia-Wei Hsu built an AI-guided algorithm that generated 100,000 candidate molecules in 24 hours, narrowed them to 200 by shape and electronic-property criteria, then selected 27 representative candidates spanning that range for lab testing under matched conditions.

DMFSA, the smallest of the 27, won on both axes: stability and fast ion transport. The group is now repeating the same congeneric-search process starting from DMFSA itself, hoping to find an even better solvent, and frames the underlying method — using molecular size and family similarity as the search variables — as a general electrolyte-design principle applicable beyond sodium batteries.

Search only molecules structurally similar to a known-stable solvent, then pick the smallest one

Restricting candidates to molecules structurally related to a proven-stable solvent meant the team wasn't gambling stability against speed from scratch — similarity acted as a stability prior.

Generating 100,000 candidates computationally and filtering to 27 meant the expensive step, synthesizing and testing electrolytes head-to-head, only had to happen on a short list.

Testing all 27 finalists under identical conditions made the comparison a fair, direct read of which trade-off point won, rather than results confounded by different setups.

An independent researcher (Jinhyuk Lee, McGill University) called the size-and-similarity approach generalizable to other energy-storage chemistries, not just this one system.

When two properties trade off, search molecules related to a known-good example, not the full space — holding family fixed while optimizing size turns an open search into a testable short list.

The findings were published in the journal Joule by 15 members of the MIT team. The group has moved on to a second round of congeneric search using DMFSA as the new reference point rather than DMTMSA, aiming to find a solvent that improves further on both stability and charge speed. The work was supported in part by a National Research Foundation of Korea grant and a U.S. National Science Foundation graduate research fellowship, with characterization equipment provided by MIT.nano.