In this blog, I discuss the importance of electrolytes for device performance and how (sometimes) electrolyte-based computational descriptors can be enough to predict device performance, bypassing the need to investigate highly complex electrolyte-electrode interfaces.
What are electrolytes anyway?
Electrolytes are more than the sum of their parts (individual solvents, salts, and additives).
Why do we care about descriptors in the first place?
(Perhaps) it all started with the need to bypass expensive and time-consuming modeling of electrode-electrolyte interfaces in the 1990s when there was shortage of high-performance computing resources. The general idea was to replace the need for explicitly modeling the complex electrode-electrolyte interfaces with computationally cheap electrolyte-only properties, often termed as descriptors, that can predict the electrochemical device properties. The first successful attempt was the proposal of hydrogen binding free energy on catalyst surfaces as descriptors for hydrogen evolution reaction (HER) activity.
Why we can’t rely on traditional approaches alone?
What does AI bring to the table?
Examples of AI in battery research
So where does it leave us?
Note: These are my personal views with the limited experience I have on working in this field. I would be happy to hear your thoughts and suggestions.
References
- R. J. Gomes, R. Kumar, H. Fejzic, B. Sarkar, I. Roy, C. V. Amanchukwu. “Modulating Water Hydrogen Bonding within a Nonaqueous Environment Controls its Reactivity in Electrochemical Transformations.” Nat. Catal. 7, 689–701 (2024). DOI