Combining Bayesian optimization and automation to simultaneously optimize reaction conditions and routes.

Area of Science:

• Artificial Intelligence in Chemistry
• Chemical Process Optimization
• Reaction Engineering

Background:

• Optimal reaction conditions are essential for high yields, reduced by-products, and sustainable chemistry.
• Traditional trial-and-error methods are time-consuming and inefficient.
• Artificial intelligence (AI) offers data-driven alternatives for chemical optimization.

Purpose of the Study:

• To showcase an integrated platform for automated chemical reaction optimization.
• To simultaneously optimize multiple substrates and reaction routes.
• To demonstrate the efficiency of AI in exploring chemical reaction spaces.

Main Methods:

• Utilized an integrated automation and Bayesian optimization platform.
• Performed simultaneous optimization of four different terminal alkynes and two reaction routes.
• Explored a fraction of the combinatorial space through a limited number of experiments.

Main Results:

• Achieved over 80% conversion rate for all four tested substrates.
• Optimization was completed within 23 experiments, representing approximately 0.2% of the combinatorial space.
• Identified the influence of various reaction parameters on outcomes.

Conclusions:

• The integrated AI platform significantly expedites reaction condition optimization.
• Demonstrates the potential for more efficient and sustainable chemical processes.
• Highlights the power of data-driven approaches in modern synthetic chemistry.