Learning Design Rules for Selective Oxidation Catalysts from High-Throughput Experimentation and Artificial...

Area of Science:

• Heterogeneous catalysis
• Materials science
• Artificial intelligence in chemistry

Background:

• Designing effective heterogeneous catalysts is complex due to numerous material possibilities and limited high-performing options.
• Understanding the interplay between material properties and catalytic performance is crucial for efficient catalyst design.

Purpose of the Study:

• To apply the subgroup-discovery (SGD) artificial intelligence approach to identify key physicochemical parameters governing outstanding catalytic performance.
• To establish "SG rules" that exclusively describe materials and reaction conditions leading to high catalytic activity.

Main Methods:

• Utilized a high-throughput experimentation approach to synthesize and test 120 SiO2-supported catalysts containing ruthenium, tungsten, and phosphorus.
• Employed artificial intelligence (subgroup discovery) on an experimental and theoretical dataset, considering temperature and 10 composition/chemical nature parameters.
• Tested catalysts in the catalytic oxidation of propylene.

Main Results:

• Identified temperature, phosphorus content, and composition-weighted electronegativity as critical parameters for high yields of acrolein and acrylic acid.
• The developed SG rules accurately describe materials and conditions associated with superior catalytic performance in propylene oxidation.
• Demonstrated the ability of SG rules to guide the design of more complex, multi-element catalysts.

Conclusions:

• The subgroup-discovery approach effectively extracts critical parameters and establishes predictive rules for heterogeneous catalyst design.
• This AI-driven methodology accelerates the discovery of advanced catalysts by focusing on key physicochemical constraints.
• The identified SG rules provide a foundation for designing novel, high-performance catalysts for value-added chemical production.