Research by Columbia Faculty Reveals New Insights into Lithium-Ion Battery Chemistry

Research by Columbia University faculty in the Arts and Sciences and the School of Engineering and Applied Science has revealed a new insight that challenges long-held assumptions of the fundamental chemistry of lithium-ion batteries.

In a joint study published in Nature, the research team, co-led by Richard Friesner of the Department of Chemistry, and Alexander Urban, Lauren Marbella, and Daniel Steingart of Columbia Engineering, has shed new light on the atomic-scale structure of lithium-ion battery (LIB) electrolytes. These insights have the potential to produce higher-capacity, faster-charging, and longer-lasting batteries.

The research addresses a long-standing bottleneck in energy storage design. LIB electrolytes are crucial for battery performance and drive the formation of the solid electrolyte interphase, which is the critical protective film on the battery anode that impacts battery lifespan and safety. However, a detailed atomic-level understanding of how electrolyte structure influences these properties had been difficult to achieve – until now. 

Utilizing advanced machine learning force field technology provided by Schrödinger, Inc., with additional support from the New York Structural Biology Center, the team simulated LIB electrolytes at a much higher degree of accuracy and scale. The simulations showed that lithium ions frequently form clusters without the presence of anions, a fundamental structural phenomenon previously unrecognized.

“We discovered lithium ions in batteries actually group together into clusters in the absence of anions, and that these clusters are what drive the creation of the protective film on the battery's surface,” said Richard Friesner, William P. Schweitzer Professor of Chemistry, and a co-founder and member of the Board of Directors at Schrödinger.

“Predicting this unexpected chemistry computationally and confirming it experimentally demonstrates that we can accurately model the hidden atomic details of batteries,” he continued. “Furthermore, these results suggest that many, if not most, complex chemical systems are now amenable to predictive physics-based simulation, using machine learning force fields, to atomic resolution.”

Key Findings of the research include:

• Multi-Lithium Ion Clustering: In standard battery electrolytes, up to 40% of lithium ions group together into multi-ion clusters, with clustering becoming even more prevalent at higher salt concentrations. This finding upends the long-held assumption that lithium ions remain predominantly isolated in solution.
• Accurate Experimental Validation: The machine learning force field simulations demonstrate strong agreement with a wide range of condensed-phase experimental data, confirming the physical reality of the predicted multi-lithium clusters.
• Electrochemical Implications: Quantum chemical calculations indicate that these multi-lithium clusters have uniquely facile reduction potentials, suggesting they are key intermediates in the formation of the solid electrolyte interphase, which is critical for battery longevity and safety.
• Foundation for Rational Electrolyte Design: The insights gained provide a fundamental understanding that can guide the design of next-generation electrolytes with enhanced performance, stability, and charge rates.

Key contributions to the research included theoretical analysis by the Columbia and Schrödinger teams, and experimental work by Columbia’s School of Engineering and Applied Science.

“While the most novel aspect of the paper is the theoretical simulations, the experimental work confirming the results was essential and extremely difficult to execute,” added Professor Friesner.  “It’s a reminder – and an exemplary example – of how advances in AI-driven simulation need to be coupled closely to experiments.  Ultimately, that will allow us to accelerate our progress in solving problems in biology and materials science, including making better, longer-lasting batteries.”

This research was supported by funding from Gates Ventures.

The story was adapted from a press release from Schrödinger, Inc.