Advisor: Prof. Meilin Liu (MSE) and Prof. Xiaoming Huo (ISyE)

will propose a doctoral thesis entitled,

Accelerated Discovery of Oxygen Electrodes for Protonic Solid Oxide Cells via Data-driven Approach

On

Friday, July 25 at 11:30 a.m.

LOVE Room 183

Or

Virtually via Zoom Link

https://gatech.zoom.us/j/97575653573?pwd=cvzVrlmh4f9qjduZERmeKYEiOb93Y0.1

Committee

            Prof. Xiaoming Huo

            Prof. Angus P. Wilkinson

            Prof. Hamid Garmestani

            Prof. Matthew T. McDowell

            Prof. Preet Singh

Abstract

Protonic solid oxide cells (P-SOCs) offer high-efficient power generation and green hydrogen production, fostering a sustainable cycle of chemical-electrical energy conversion and paving the way toward a zero-emission future. However, the commercialization of P-SOCs is hindered by the absence of oxygen electrode materials that combine high electrocatalytic activity with long-term durability. To accelerate the discovery of such material – capable of operating under real-world conditions – including high humidity, exposed to contaminants, and reduced activity at lower temperatures – we conducted high-throughput density functional theory (DFT) calculations to systematically evaluate key descriptors relevant to oxygen electrode performance. These include the energy above hull (Ehull), vacancy formation energy (Ev), p-band center, d-band center, d-­p hybridization, and oxygen non-stoichiometry (δ), enabling rapid screening of candidate materials.

Building upon this foundation, we developed a supervised learning model to identify the dominant physical descriptors governing performance and to ensure transferability across diverse chemistries. These computational insights were integrated into an active learning-guided experimental loop, where the learned descriptors served as inputs to predict polarization resistance (Rp) – a critical multiphysics property indicative of electrocatalytic activity. By maximizing the expected information gain, the active learning framework strategically guided standardized experiments toward the most informative candidates. This approach enabled direct Rp prediction, and further analysis revealed that d-p­ hybridization and calcination resistance are key factors driving performance. Through large-scale screening of 6,940,032 compositions, we identified top-performing oxygen electrodes materials that achieved a peak power density of 2.68 W cm-2 at 600 °C exceeding benchmark materials by 35% and demonstrated stable operation for over 500 hours in a P-SOC configuration.