Combinatorial synthesis for AI-driven materials discovery
Combinatorial synthesis of solid-state materials comprises the use of automation or parallelization to systematically vary synthesis parameters. This approach to materials synthesis is a natural fit for accelerated mapping of composition–structure–property relationships, a central tenet of materials research. By considering combinatorial synthesis in the context of experimental workflows, we envision a future for accelerated materials science promoted by the co-development of combinatorial synthesis and artificial intelligence (AI) techniques. To evaluate the suitability of a synthesis technique for a given experimental workflow, we establish a collection of ten metrics spanning speed, scalability, scope and quality of synthesis. We summarize select combinatorial synthesis techniques in the context of these metrics, elucidating opportunities for further development. These opportunities span initial deployment in high-throughput experimentation through to seminal demonstrations of automated decision-making using AI. Historical analysis of combinatorial synthesis in the context of the Gartner hype cycle establishes a recent rise in productivity, indicating that the field is poised to realize accelerated materials science workflows that transform materials discovery and development.