Automated Library Generation and Serendipity Quantification Enables Diverse Discovery in Coordination Chemistry

Area of Science:

• Coordination chemistry
• Automated experimentation
• Chemical library generation

Background:

• High-quality chemical libraries are crucial for discovering new materials, methods, and models.
• Generating such libraries efficiently and effectively remains a significant challenge in chemical research.

Purpose of the Study:

• To demonstrate the automation of library generation workflows using automated hardware.
• To develop a novel workflow for prospecting library formation by combining Bayesian optimization with novelty quantification.
• To explore chemical and process spaces for targeted synthesis and library generation.

Main Methods:

• Utilized automated hardware, including the Chemputer, for library generation experiments.
• Employed target-oriented synthesis of coordination complexes to validate synthetic operations.
• Implemented Bayesian optimization with a Gaussian process surrogate model for directed exploration.
• Quantified novelty using mass spectrometry data to guide the search for new chemical entities.

Main Results:

• Successfully automated key workflows for library generation in coordination chemistry.
• Generated focused libraries within defined chemical and process spaces.
• Developed and validated a new workflow for prospecting library formation, demonstrating directed exploration.
• Observed the formation of unexpected Co(III) anhydride complexes, showcasing "engineered serendipity".

Conclusions:

• The developed automated workflow effectively directs the exploration of chemical process spaces.
• This approach facilitates the discovery of novel compounds by "engineering serendipity" into the search process.
• The methodology provides a powerful tool for accelerating materials discovery through automated and intelligent library generation.