SEED: System for Electrolyte Exploration and Discovery

Team: Venkat Viswanathan, Aditya Parameswaran

The energy density of batteries is a critical bottleneck for shrinking dimensions of portable electronics and electric vehicles. Unfortunately, the inverse coupling between energy density & safety limits the practicality of several advanced battery concepts. Safety problems stem from various characteristics of the electrolyte. The goal is to develop a tool, SEED, to enable battery scientists to quickly gain insights from large collections of electrolyte simulation data, and automatically-inferred and manually-specified domain knowledge. For datafication, we collect or generate a large quantity of electrolyte data spanning molecular, micro- and meso- scales with spatial and temporal heterogeneity. This data is then exposed via novel declarative discovery mechanisms (with visual and textual input) aimed at facilitating the seamless identification of trends, patterns, and anomalies. These discovery mechanisms will utilize knowledge representations to encode domain knowledge, physical/empirical laws, and desirable characteristics. We expect SEED to have an enormous impact on research in battery science enabling rapid search and retrieval of desired electrolyte patterns, trends and anomalies, as well as shaping the face of electrolyte design via the discovery of entirely new electrolyte formulations. Based on interest for our initial prototype–with 10+ organizations either currently using it or intending to use a future version–we expect SEED to not just facilitate electrolyte discovery, but also spur the development of the next generation of batteries for a range of current and as-yet-undiscovered use cases.


A. Khetan, H. Pitsch, V. Viswanathan, J. Phys. Chem. Lett. 5, 2419, J. Phys. Chem. Lett. 5, 1318 (2014)
A. Parameswaran, N. Polyzotis, and H. Garcia-Molina, SeeDB: Visualizing Database Queries Efficiently (Vision Paper), VLDB, 2014

Upcoming datasets

Minnesota Solvation Database
NIST IL Thermo