Mapping inorganic crystal chemical space

Area of Science:

• Materials Science
• Computational Chemistry
• Inorganic Chemistry

Background:

• The vast chemical space of elemental combinations is largely unexplored.
• Discovering new inorganic materials is crucial for technological advancement.
• Current methods for exploring chemical space are limited in scale.

Purpose of the Study:

• To create an extensive library of stoichiometric inorganic compositions.
• To develop a comprehensive map of inorganic crystal chemical space.
• To facilitate the discovery of novel, synthesizable inorganic materials.

Main Methods:

• Enumeration of binary, ternary, and quaternary element and species combinations.
• Vectorization of unique compositions using machine learning-derived embedding vectors.
• Application of dimensionality-reduction techniques for visualization.

Main Results:

• Generation of a library exceeding 10^10 unique inorganic compositions.
• Creation of a two-dimensional representation of inorganic crystal chemical space.
• Labeling of compositions based on chemical filter compliance and database presence.

Conclusions:

• The study provides a scalable framework for exploring inorganic chemical space.
• The generated map aids in identifying promising candidates for new material discovery.
• This approach accelerates the search for functional inorganic materials.