Scientists from Kyushu University, Osaka University, and the Fine Ceramics Center in Japan have developed a new framework that employs machine learning to expedite the discovery of materials for green energy technology. By leveraging this approach, the researchers were able to identify and successfully synthesize two candidate materials for use in solid oxide fuel cells, which generate energy without emitting carbon dioxide. Their findings, published in the journal Advanced Energy Materials, have the potential to not only enhance the power-generating efficiency of hydrogen fuel cells but also accelerate the search for innovative materials in various sectors beyond energy.
The urgency to develop alternative energy sources that do not rely on fossil fuels is driven by the need to combat climate change. One promising avenue towards achieving carbon neutrality is the creation of a hydrogen society. However, in addition to optimizing hydrogen production, storage, and transportation, it is crucial to improve the power-generating efficiency of hydrogen fuel cells.
To generate an electric current, solid oxide fuel cells require the efficient conduction of hydrogen ions, or protons, through a solid material called an electrolyte. The current research on new electrolyte materials has primarily focused on oxides with specific crystal arrangements, known as perovskite structures.
While the first proton-conducting oxide discovered was in a perovskite structure, we aim to expand the discovery of solid electrolytes to non-perovskite oxides that also exhibit excellent proton conductivity, explains Professor Yoshihiro Yamazaki from Kyushu University’s Q-PIT Department of Materials Science and Technology.
However, the traditional method of discovering proton-conducting materials through trial and error has its limitations. To enhance proton conductivity, trace amounts of another substance, called a dopant, must be added to the base material. With numerous base and dopant candidates, each possessing different atomic and electronic properties, finding the optimal combination becomes time-consuming and challenging.
Instead of relying on traditional methods, the team of researchers employed machine learning to calculate the properties of various oxides and dopants. Through data analysis, they identified the factors that influence proton conductivity and made predictions about potential combinations.
Guided by these factors, the researchers successfully synthesized two promising materials with unique crystal structures and examined their proton conductivity. Remarkably, both materials exhibited proton conductivity after just a single experiment.
One of the materials showcased a sillenite crystal structure, making it the first-known proton conductor with this arrangement. The other material, which possesses a eulytite structure, demonstrated a high-speed proton conduction path distinct from perovskites. Although their performance as electrolytes is currently low, the research team believes that further exploration and refinement can enhance their conductivity.
Professor Yamazaki concludes, Our framework has the potential to significantly expand the search space for proton-conducting oxides and accelerate advancements in solid oxide fuel cells, bringing us closer to realizing a hydrogen society. With minor modifications, this framework could also be applied to other fields of materials science, potentially expediting the development of various innovative materials.
In summary, the researchers from Kyushu University, Osaka University, and the Fine Ceramics Center in Japan have harnessed the power of machine learning to expedite the discovery of materials for green energy technology. Through their innovative approach, they successfully identified and synthesized two new candidate materials for solid oxide fuel cells. Their findings not only contribute to the advancement of hydrogen fuel cells but also have the potential to accelerate the search for innovative materials across different sectors.
