Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Materials Science and Engineering

Committee Chair/Advisor

Jianhua Tong

Committee Member

Feng Luo

Committee Member

Kyle Brinkman

Committee Member

Fei Peng

Committee Member

Rajendra Bordia


Hydrogen has significantly contributed to the global energy transition as a clean energy carrier. However, traditional hydrogen production involves energy-intensive processes with heavy carbon emissions. High-temperature water splitting to generate hydrogen has become a promising route for clean hydrogen production. Directly utilizing concentrated solar heat to perform solar thermochemical hydrogen (STCH) production is considered one of the most eco-friendly hydrogen production options. At the same time, high-temperature electrolysis in protonic ceramic electrolysis cells (PCECs) can efficiently utilize intermittent and cost-effective renewable electricity to produce hydrogen. Perovskite oxides with good redox capability and tolerance for oxygen deficiency are considered the most promising materials for STCH. Meanwhile, perovskite oxides are the most studied proton-conducting electrolyte materials for PCECs. The performance of STCH perovskite materials and proton-conducting perovskite electrolytes is closely correlated to the substitution of dopants. The investigations of STCH perovskite oxides or proton-conducting perovskite electrolytes still mainly rely on time-consuming trial-and-error experiments, limiting the search for high-performance perovskite materials due to the compositional complexity of perovskite oxides. The relationship between their performance and compositions has yet to be established. Machine learning (ML) is a powerful tool to accelerate the screening of high-performance perovskite materials for hydrogen production. To better discover potential high-performance perovskite oxides for STCH materials and proton-conducting electrolytes, my dissertation starts with the formability prediction of fractionally doped perovskite oxides (FDPOs) by ML and then investigates high-entropy perovskite oxides for STCH with the assistance of ML, and finally discusses ML-assisted discovery of high-performance perovskite electrolytes for PCECs.

For the formability prediction of FDPOs by ML, we collected 632 training data and defined 21 desirable features. Our gradient boosting classifier model achieved a high prediction accuracy of 95.4% and a high F1 score of 0.921. Furthermore, when verified on additional 36 experimental data from existing literature, the model showed a prediction accuracy of 94.4%. With the help of this ML approach, we identified and synthesized 11 new FDPO compositions, 7 of which are relevant for STCH.

With the assistance of the ML model in Chapter 2, a series of high-entropy perovskites were designed for STCH in Chapter 3. Compared with well-studied materials Sr0.4La0.6Mn0.6Al0.4O3 and BaCe0.25Mn0.75O3, our material has excellent redox reversibility and remarkable hydrogen yield at the oxidation temperature of 800 ℃. In contrast to BaCe0.25Mn0.75O3 material, our material exhibits superior phase stability in successive cycles due to the entropy-stabilized effect. Furthermore, high-entropy perovskite shows excellent chemical stability under 5 % steam.

In Chapter 4, ML models predict the total conductivity of proton-conducting perovskite electrolytes at an intermediate-temperature range (400-600℃). We collected 3797 experimental data and 44 features related to basic elemental properties, crystal structure, microstructure, and measurement conditions. Our XGB regressor achieved a high prediction performance with an R2 of 0.956. Compared with the 20 experimental total conductivities in the literature, the predicted results are close to the actual values.

Available for download on Saturday, May 31, 2025