Research

Photovoltaics

We design new earth-abundant materials for use in solar cells. We focus on defect-tolerant materials that can still show high efficiency even when using low-cost processing methods. Using a data-driven approach we can screen large libraries of crystals and we also generate our own datasets using high-throughput first-principles calculations.

---

Relevant publication: “The defect challenge of wide-bandgap semiconductors for photovoltaics and beyond” Nature Communications (2022)

Electron-phonon interactions

The electron–phonon interaction is a cornerstone of solid-state physics and controls the charge carrier mobility in semiconductors. We recently developed a new method to calculate electron–phonon and electron–defect interactions that is 500 times more efficient than existing methods. We’ve used it for the widescale screening of carrier lifetimes in 23,000 materials. We are currently applying it to discover new thermoelectric materials.

---

Relevant publication: “Efficient calculation of carrier scattering rates from first principles” Nature Communications (2021)

Machine learning for chemistry

Machine learning (ML) is a powerful tool for accelerating materials discovery. We build computational databases of materials properties and apply ML to extract chemical trends, predict the properties of unseen materials, and generate completely new crystal structures. We are particularly interested in applying ML to predict complex tensor properties of crystals such as dielectric and elastic properties.

---

Relevant publication: “Learned Force Fields Are Ready For Ground State Catalyst Discovery” arXiv:2209.12466