Autonomous and dynamic precursor selection for solid-state materials synthesis.

Area of Science:

• Materials Science
• Chemical Engineering
• Computational Chemistry

Background:

• Solid-state synthesis is crucial for new materials development.
• Current methods often involve extensive, trial-and-error precursor and condition screening.
• Advanced characterization and computation aid understanding but don't fully automate discovery.

Purpose of the Study:

• To develop an automated algorithm for selecting optimal precursors in solid-state synthesis.
• To reduce the number of experiments needed for new material discovery.
• To integrate domain knowledge into optimization for materials synthesis.

Main Methods:

• Introduction of the ARROWS³ (Active REaction path-finding With Optimized Selection) algorithm.
• Algorithm actively learns from experimental outcomes to identify unfavorable reaction pathways and stable intermediates.
• ARROWS³ proposes new experiments by predicting precursors that avoid these intermediates, maximizing thermodynamic driving force.

Main Results:

• Validation on three experimental datasets with over 200 synthesis procedures.
• ARROWS³ significantly reduces experimental iterations compared to black-box optimization.
• Demonstrated ability to identify effective precursor sets for target materials.

Conclusions:

• ARROWS³ successfully automates precursor selection for solid-state synthesis.
• The algorithm's active learning approach accelerates materials discovery.
• Integrating domain knowledge into optimization is key for autonomous research platforms in materials science.