Title: Unveiling a New Homologous Series of Barium Compounds
A remarkable breakthrough by Mercouri Kanatzidis and his team at Northwestern University reveals a novel homologous series of barium compounds that offers an endless array of interconnected structures with predictable unit cells. This progress has the potential to substantially improve the ability of machine learning models to identify new inorganic materials by comprehending structure–composition relationships.
In chemistry, homologous series are typically linked with organic compounds, defined by a sequence that includes a repeating unit and a universal formula. Illustrations of this include straight-chain alkanes and alkenes. In contrast, inorganic homologous series are less common but can be found in forms such as non-stoichiometric titanium oxides and two-dimensional halide perovskites, which are significant in solar cell applications.
Kanatzidis’s team concentrated on barium antimony telluride (BaSbTe₃), partially replacing tellurium with sulfur, another element from group 16. Logically, one might expect sulfur and tellurium atoms to distribute randomly within the anionic sites, leading to a solid solution. However, sulfur atoms, being more electronegative, favor electron-rich positions, altering the electron density of telluride anions. This anomaly inhibits the formation of a solid solution, thereby creating a new type of material order, which Kanatzidis describes as remarkable.
This investigation produced ten members of the series that progressively escalated in structural complexity. The final compound, BaSbSTe₂, showcases an electronic instability known as a charge density wave. Charge density wave materials can demonstrate superconductivity at elevated temperatures or under low pressures, hinting at promising new predictions for superconductors.
The team highlighted that while machine learning has advanced in chemical design, it generally flourishes within established structure types, achieving greater success in organic chemistry owing to a broader foundation of recognized chemical principles. Discoveries such as this new phase homology could provide vital training data to enhance the accuracy of machine learning in innovating inorganic chemistry.
Leslie Schoop from Princeton University praises the robustness of the study, encouraging researchers to examine these new materials for noteworthy properties. Schoop further supports the idea of integrating such relationship insights into AI to advance material discovery initiatives in inorganic chemistry.