Automated Flow Synthesis: In-Line Reaction Monitoring and Machine Learning for the Optimisation of Continuous Flow Photocatalytic Reactions

自动流动合成：用于优化连续流动光催化反应的在线反应监测和机器学习

基本信息

批准号：
2894726
负责人：
金额：
--
依托单位：
Heriot-Watt University
依托单位国家：
英国
项目类别：
Studentship
财政年份：
2023
资助国家：
英国
起止时间：
2023 至无数据
项目状态：
未结题

来源：
https://gtr.ukri.org/projects?ref=studentship-2894726
关键词：
Automated Flow Synthesis Line Reaction

项目摘要

This is a PhD research project in Chemistry. We propose a new and ambitious project to utilise in-line benchtop NMR and/or UV-vis monitoring of continuous flow photocatalytic radical decarboxylation reactions, exploiting real time data analytics in a digital feedback-loop to optimise several reaction parameters. Photocatalytic direct C-H Minisci functionalisations of heterocycles will be developed and optimised in this way, as these scaffolds are highly relevant to drug discovery programmes. Our system will enable technology for more efficient photocatalytic reactions including scale up processes compared to batch reactions. The project aims to provide proof-of-principle towards an ultimate goal of automated development and optimisation, using control theory and machine learning techniques, of continuous flow photocatalytic reactions. Stepwise objectives are therefore: 1. Optimise in continuous flow the photochemical Minisci transformations 2. Integrate in-line/on-line monitoring systems (NMR and UV-Vis) with the flow system as a unified platform for the generation of training data 3. Model the input/output processes of the continuous flow and design control/reinforcement learning algorithms to maximise meaningful metrics (e.g., selectivities, conversions and yields)

这是化学博士研究项目。我们提出了一个新的雄心勃勃的项目，以利用连续流光催化自由基脱羧反应的在线台式NMR和/或UV-VIS监测，从而利用数字反馈环中的实时数据分析来优化几个反应参数。杂环的光催化直接C-H Minisci功能将以这种方式进行开发和优化，因为这些脚手架与药物发现计划高度相关。我们的系统将启用技术，以提供更有效的光催化反应，包括与批处理反应相比，包括扩展过程。该项目旨在使用控制理论和机器学习技术来实现自动化发展和优化的最终目标，并提供连续流动光催化反应。 Stepwise objectives are therefore: 1. Optimise in continuous flow the photochemical Minisci transformations 2. Integrate in-line/on-line monitoring systems (NMR and UV-Vis) with the flow system as a unified platform for the generation of training data 3. Model the input/output processes of the continuous flow and design control/reinforcement learning algorithms to maximise meaningful metrics (e.g., selectivities, conversions and yields)