Systematic searches for new inorganic materials assisted by materials informatics

Area of Science:

• Materials Science
• Computational Chemistry
• Data Science

Background:

• Exploring novel inorganic functional materials is crucial for technological advancement.
• Traditional methods for material discovery are often time-consuming and resource-intensive.
• Integrating machine learning and computational chemistry offers a promising avenue for accelerated discovery.

Purpose of the Study:

• To introduce proprietary Materials Informatics (MI) technologies and a chemistry-oriented methodology for discovering new inorganic functional materials.
• To leverage machine learning and advanced computational techniques for predicting material properties and designing novel structures.
• To facilitate the collection and analysis of large-scale experimental data for material selection and validation.

Main Methods:

• Development of 'Element Reactivity Maps' using machine learning on crystal structure databases for 80x80x80 elements.
• Application of Delaunay tetrahedral decomposition to analyze atomic coordinates, establishing the concept of 'Delaunay Chemistry'.
• Creation of the 'Crystal Cluster Simulator' and 'Starrydata2' web systems for crystal structure design and large-scale experimental data collection from academic papers.
• Utilizing proprietary ion diffusion control technologies and sodium metal in synthesis for discovering new compound types.

Main Results:

• Discovery of numerous new material phases, including solid solutions in novel element combinations, through large-scale synthesis experiments (>7,000 samples).
• Identification of new cage-like and intercalation compounds using advanced synthesis techniques.
• Successful application of Element Reactivity Maps for selecting barrier metals for device electrodes.
• Selection of candidate materials for new thermoelectric applications through data analysis of collected experimental data.

Conclusions:

• The developed Materials Informatics technologies and chemistry-oriented methodology significantly accelerate the discovery of novel inorganic functional materials.
• The integration of machine learning, computational chemistry, and large-scale data analysis provides a powerful framework for materials innovation.
• The discovered materials and methodologies hold potential for applications in thermoelectrics, device electrodes, and other advanced technologies.