Automated small molecule synthesis for accelerated discovery of photo- and electroactive organic materials

Yi, Seungjoo

Permalink

https://hdl.handle.net/2142/132725 Copy

Description

Title

Automated small molecule synthesis for accelerated discovery of photo- and electroactive organic materials

Author(s)

Yi, Seungjoo

Issue Date

2025-08-14

Director of Research (if dissertation) or Advisor (if thesis)

Schroeder, Charles M

Doctoral Committee Chair(s)

Schroeder, Charles M

Committee Member(s)

• Evans, Christopher M
• Wang, Hua
• Jackson, Nicholas E

Department of Study

Materials Science & Engineerng

Discipline

Materials Science & Engr

Degree Granting Institution

University of Illinois Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Automated small molecule synthesis
• Modular synthesis
• Closed-loop
• Organic materials
• Triplet state quencher
• Photostability
• Light-harvesting small molecules
• Transparent-to-black electrochromism
• Electrochromic
• Viologens
• Spectroelectrochemistry
• Artificial Intelligence
• Single-molecule electronics
• Au-C anchor
• Time-dependent density functional theory
• Single-anchor
• Charge transport
• Scanning tunneling microscope break junction

Abstract

Recent advances in artificial intelligence (AI)-guided closed-loop experimentation have transformed materials discovery, enabling optimization of functional properties and determination of fundamental principles in molecular design. A key innovation is the closed-loop transfer (CLT) method, which integrates closed-loop workflows with physics-based feature selection and supervised learning to uncover new insights for objective function optimization. In this thesis, CLT is used to optimize the photostability of light-harvesting organic molecules with donor–acceptor structures. CLT identified critical mechanistic factors such as the role of high-energy triplet state manifolds via automated modular synthesis and characterization of only ~1.5% of the theoretical molecular space. This approach yielded a physics-based model for photostability that was validated across diverse test sets and enhanced by modulating triplet-state energies in solvent media, surpassing initial saturation of the objective function during optimization. Extending CLT as a general human-in-the-loop framework addresses limitations in prior black-box AI methods by combining Bayesian optimization (BO)-guided exploration, experimental validation of predictive machine learning (ML) models, and hypothesis-driven molecular design. CLT was validated in several different case studies including the photostability of organic light-harvesting molecules, organic laser dyes, and stereoselective aluminum catalysis. In this way, CLT accelerates the discovery of interpretable structure–property insights, offering a blueprint for integrating synthesis, characterization, and ML for the development of new functional molecules. In a second project, an integrated computational-experimental framework was developed for discovery of viologen pairs for transparent-to-black/grey electrochromism. In this way, time-dependent density functional theory (TDDFT) -guided method was used to design new viologen pairs to overcome limitations in using single viologen molecules for broadband visible absorption. A curated library of 950 viologens from 17 bipyridine cores and 41 pendant groups—selected via k-means clustering with DFT and RDKit descriptors for steric, electronic, and hydrophobic diversity—provided a large chemical library for synthesis using modular anhydrous Suzuki-Miyaura coupling and Menshutkin quaternization. TDDFT simulations were used to predict dication and radical cation absorption spectra, guiding virtual pairwise blending for complementary spectral profiles. Spectroelectrochemical validation was used to validate TDDFT predictions, with the recommended viologen pairs exhibiting near-uniform 380–780 nm absorption and aligned reduction potentials (~-0.5 V). This workflow introduces quantitative greyness metrics and paves the way for scalable electrochromic devices, with potential ML integration for closed-loop optimization. A third project focused on understanding the role of terminal anchor groups on the electron transport properties of molecular junctions. Using automated chemical synthesis, single-molecule junctions based on p-terphenyl derivatives with one pre-installed anchor were studied using single-molecule electronics experiments, molecular dynamics simulations, electrochemistry, spectroscopy, and non-equilibrium Green’s function-DFT calculations. Results from single-molecule electronics experiments for junctions with only one pre-installed anchor showed conductance features similar to junctions with two pre-installed terminal anchors. In particular, 4-amino-p-terphenyl exhibited a high conductance state that is absent in molecular analogs lacking amine anchors or substitutions at the terminal para position in the terphenyl derivatives. Results from experiments and simulations showed that the low conductance state arises due to π-π stacking interactions and intermolecular electron transport, whereas the high conductance state arises due to Au-C bond formation via single-electron oxidation and radical substitution at the gold electrode surface. A series of control experiments was used to understand the role of the primary amine in Au-C bond covalent bond formation, offering insights for molecular electronic device design and junction formation mechanisms. Taken together, this research demonstrates the synergy of using AI-guided methods, closed-loop approaches, and modular synthesis for accelerating the discovery of new functional materials for applications in light-harvesting organics, electrochromism, and molecular electronics.

Graduation Semester

2025-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/132725

Copyright and License Information

Copyright 2025 Seungjoo Yi

Owning Collections