Research in Nguyen group focuses on understanding chemical processes and applying that knowledge to improve them for industrial applications. The current main research areas in the group are:
1. Machine Learning methodologies to generate intepretable prediciton models chemical properties, reactivity and reaction outcomes, including impurities and workup.
2. High throughput computational and analytical techniques to generate high quality data for AI/Machine Learning in chemical sciences.
3. Theoretical and practical framework for discovery and development of 'on water' reactions as green processes in High Value Chemical Manufacture.
Our research tools include AI/Machine Learning and computational chemistry, structure-reactivity relationships, spectroscopic studies, kinetic analysis and models, and electrochemical chemistry. We are also part of the iPRD (https://www.iprd.leeds.ac.uk), and have access to a wide range of equipment and expertise in batch and flow reactors from lab scale to pilot scale.
Some of our recent projects are described below:
Solubility prediction remains a critical challenge in drug development, synthetic route and chemical process design, extraction and crystallisation. We developed a successful approach to solubility prediction in organic solvents and water using a combination of machine learning and computational chemistry (doi:10.1038/s41467-020-19594-z). Rational interpretation of dissolution process into a numerical problem led to a small set of selected descriptors and subsequent predictions which are independent of the applied machine learning method. These models gave significantly more accurate predictions compared to benchmarked open-access and commercial tools, achieving accuracy close to the expected level of noise in training data (LogS±0.7). Finally, they reproduced physicochemical relationships between solubility and molecular properties in different solvents, which led to rational approaches to improve the accuracy of each models.
The copper-catalysed Ullmann-Goldberg C-N and C-O coupling reactions are potentially highly important reactions in High Value Chemical Manufacture, with the cheap and readily available copper catalysts replacing their expensive palladium counterparts. However, these reactions are notoriously unreliable and difficult to scale up, with poor reproducibility often cited as the main problem in industrial context.
Our mechanistic investigation of these reactions (doi: 10.1039/c7sc02859h) using soluble and partially soluble bases led to the identification of various pathways for catalyst deactivation through (i) product inhibition with amine products, (ii) by-product inhibition with inorganic halide salts, and (iii) ligand exchange by soluble carboxylate bases. The reactions using partially soluble inorganic bases showed variable induction periods, which are responsible for the reproducibility issues in these reactions. Surprisingly, more finely milled Cs2CO3 resulted in a longer induction period due to the higher concentration of the deprotonated amine/amide, leading to suppressed catalytic activity. These results have significant implications on future ligand development for the Ullmann–Goldberg reaction and on the solid form of the inorganic base as an important variable with mechanistic ramifications in many catalytic reactions.
Through collaborations with other research groups in fluid dynamics and inorganic chemistry, we developed an extremely efficient and scalable electrochemical flow-cell which can effect the conversion of imidazoliums and a copper electrode to Cu-NHC catalysts (NHC = N-heterocyclic carbene) in very high yields (DOI: 10.1021/ja512868a). The purity of the Cu-NHC catalysts is sufficient to allow direct 'dispensing' of the catalyst solution to catalytic reactions, when required, with no detectable change in catalytic activity and selectivity compared to using the purified catalysts. This will enable rapid catalytic screening, as well as catalyst top-up (coupled with computer controlled reactors) in the near future.
The technology is applicable to a wide range of ligands beyond NHCs.
Research in the group investigated the 'activation' of CO2 using guanidine/amidines catalysts and the subsequent catalytic reactions between CO2 and propargylamines. Factors controlling the complexation between amines and CO2 and catalytic activity have been carefully examined using a range of physical organic chemical techniques. This had led to a 10 times increase in the performance of the catalytic system at low loading using a much cheaper catalyst (doi: 10.1039/C4CY00480A and doi: 10.1021/acscatal.7b04108).