Moving away from fossil fuels requires environmentally friendly and economically viable alternative energy sources. A wide adoption of new technologies for energy production and storage depends on better performing materials. Computational methods, such as electronic structure calculations and machine learning, hold the promise to work in conjunction with traditional experimentation to accelerate the discovery of materials needed to make those technologies more efficient. This thesis presents a first-principles methodology in the context of perovskite discovery for hydrogen fuel production via solar thermochemical water splitting. We calculate the properties of an exhaustive list of compounds to search for the ideal materials for water splitting, some that have never been experimentally synthesized. In addition, we use this large dataset of materials to benchmark current machine learning techniques to further reduce the number of expensive calculations required to discover new materials. Finally, we look at the entropy of reduction of cerium to explain the good performance of ceria for water splitting.

Last modified

• 10/22/2018

Creator

• Antoine Alexandre Emery

DOI

• https://doi.org/10.21985/N2KF21

Subject

• Materials Science and Engineering

Keyword

• Perovskites
• Hydrogen production
• Water splitting
• Density functional theory
• Materials discovery
• Machine learning

Date created

• 2017-01-01

Resource type

• Dissertation

Rights statement

• In Copyright