Virtual VenkatAI

The work in brief

Venkat's work spans battery science from atomistic mechanisms to vehicle-level requirements. Early papers (2011–2016) used first-principles calculations to study nonaqueous Li–O2 batteries: charge transport through Li2O2, solvent stability descriptors, and how solvating additives and electrolyte anions drive solution-mediated discharge and toroid growth. From around 2016 the focus moved to lithium metal anodes, where it developed stability criteria for electrodeposition at solid–solid interfaces, machine-learning screening of solid electrolytes for dendrite suppression, phase-field simulations of dendrite growth, and studies of voids and chemomechanics in solid-state batteries. Alongside this ran DFT phase studies of graphite and NMC cathodes with uncertainty quantification, and electrode kinetics beyond Butler–Volmer, such as Marcus–Hush–Chidsey models implemented in open-source tools.

A second strand assesses what batteries must deliver for electrified transport, including performance metrics for electric semi-trucks, VTOL and commercial aircraft, safety and degradation of aircraft batteries, an eVTOL cycling dataset, and battery state estimation and observers. Since about 2019 the work has increasingly combined machine learning, automation and differentiable programming. This includes robotic electrolyte optimization with Bayesian planning (Clio, Dragonfly), the differentiable geometric deep learning model DiffMix, machine-learning interatomic potentials for lithium, self-driving labs for salt solubility, and a proposed "differentiable electrochemistry" paradigm. Recent papers also examine SEI formation during cell formation and design targets for solid-state batteries.

AI-written from the paper titles and abstracts; Venkat hasn't reviewed it.