Data mining approaches to high-throughput crystal structure and compound prediction

Area of Science:

• Materials Science
• Computational Chemistry
• Data Mining

Background:

• High-throughput computational materials design requires efficient prediction of unknown inorganic compounds and their crystal structures.
• Data mining and machine learning offer promising approaches for accelerating materials discovery.
• Understanding phase stability is crucial for predicting new compounds.

Purpose of the Study:

• To present data mining algorithms for inorganic compound prediction.
• To apply these algorithms to identify novel compounds and their crystal structures.
• To demonstrate the experimental validation of computationally predicted materials.

Main Methods:

• Statistical learning of patterns governing phase stability from inorganic compound databases.
• Development of probabilistic or regression models for compound prediction.
• Assessment of predicted compound stability using ab initio techniques.

Main Results:

• Identification of novel inorganic compounds and their crystal structures through data mining.
• Statistical models successfully captured correlations governing phase stability.
• Computationally predicted compounds were experimentally synthesized and confirmed.

Conclusions:

• Data mining driven predictions are effective for discovering new inorganic materials.
• Integrating computational prediction with experimental synthesis accelerates materials discovery.
• Machine learning approaches hold significant potential for future materials design.