Research

Machine-learned force fields (MLFFs), density functional theory, and graph neural networks are combined to screen and study energy-relevant materials. The work ranges from accelerating molecular dynamics with learned potentials to improving their accuracy and transferability, with active learning used to decide which expensive quantum-chemistry calculations are worth running.

Today's commercial lithium-ion batteries rely on unsustainable materials and are surprisingly short-lived. A typical smartphone battery loses noticeable performance after only 1.5 to 2 years of use. Surface and bulk engineering strategies target the failure modes of lithium-ion electrodes, especially the cathode, including cracking, gas evolution, metal dissolution, and surface fouling, with the goal of longer lifetimes and more stable cycling.

Work here covers CO₂ capture and electrochemical CO₂ reduction, the oxygen evolution reaction (OER), and hydrogen evolution. Candidate catalysts are screened computationally, including how their surfaces reconstruct under reaction conditions, which strongly affects stability and activity. The most promising ones are then synthesized and tested in the lab.

This area covers the growth and synthesis of colloidal quantum dots, halide perovskites (CsPbX₃) for optoelectronic devices, and materials for blue OLEDs. A central question is how synthesis controls the structure and stability of these materials, and how to translate that understanding into efficient, durable light-emitting devices.

This area spans two threads. The origins-of-life work combines experiment and computation: recreating plausible prebiotic conditions in the lab, and computationally exploring the inorganic chemistry that may have preceded biology. A second thread focuses on myalgic encephalomyelitis / chronic fatigue syndrome (ME/CFS) and Long COVID, working toward biomarkers that could help detect them, including through machine-learning analysis of biological data.