Closed-loop iterative molecular discovery engines

Despite advances in robotics and artificial intelligence, the synthesis of new small molecules is primarily performed manually by expert chemists, and remains the rate determining step in the discovery of new molecules. With a pressing need to accelerate the pace of molecular innovation in order to solve society’s emergent challenges in health, energy, and sustainability, new strategies to discover molecules and understand molecular function are required. The generalized and modular automation of small molecule synthesis has the potential to shift the bottleneck of molecular discovery from synthesis to prediction and analysis, enabling the creation of engines for molecular discovery. Molecular discovery engines would incorporate fully automated experimental feedback loops of making, testing, and analyzing new molecules to understand and discover new molecular function. If realized, this technology has great potential to shift the paradigm of molecular discovery from structural design to functional understanding. The purpose of this dissertation is to validate the concept of molecular discovery engines by building them for the first time and testing them in impactful scientific contexts. I first present practical advances to the Burke group’s automated modular small molecule synthesis platform and demonstrate its utility in the study of fundamental small molecule charge-transport behavior, in Chapter 2. Then, I present three case studies where closed-loop iterative molecular discovery engines were built and tested to solve frontier problems. In Chapter 3, a molecular discovery engine was built which discovered globally optimized, maximally general reaction conditions. In Chapter 4, a molecular discovery engine was built which discovered increasingly photostable light-harvesting small molecules. In Chapter 5, a molecular discovery engine was built which discovered high gain solid-state organic lasers. In each of these cases, practical advances in emerging areas of research were found from otherwise intractably large molecule search spaces.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Chemistry

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2023

Author dc:creator

• Angello, Nicholas Henry

Contributors dc:contributor

• Burke, Martin D.
• Denmark, Scott E.
• Moore, Jeffrey S.
• Schroeder, Charles M.

• Closed-loop
• Automated Small Molecule Synthesis
• Iterative Cross-coupling
• Inverse Design
• Democratization
• Suzuki Coupling
• Single Molecule Charge Transport
• Generality
• Datamining
• Machine-learning
• General Reaction Conditions
• Optimization
• Bayesian
• Organic Photovoltaics
• Organic Lasers

Statement dc:rights

• © 2023 Nicholas H. Angello

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/121195

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F121195