Agile Self-Optimisation for High Pressure Flow Chemistry

点击翻译按钮获取中文摘要

In flow chemistry, optimising the parameters (concentration, temperature, pressure etc.) is important in ensuring a high-quality product. The standard procedure to do this is manual optimisation in which adjustment to one parameter at a time is done. This approach can be very time consuming; it can deplete resources very quickly and produces considerable amounts of waste which can be a major issue with expensive starting materials.The concept of self-optimisation involves machine learning algorithms and process analytical technology (PAT) working together to reduce the consumption of reagents needed whilst simultaneously accelerating the optimisation process. Self-optimisation facilitates process development as it allows more precise control over reaction parameters and is becoming increasingly popular for flow chemistry. However its application to high pressure continuous flow reactions has been rather more limited. High pressure is important because, for example, it permits the use of supercritical solvents which have particular advantages for reactions involving gaseous reagents such as hydrogenation with hydrogen or oxidation with molecular oxygen both of which can lead to highly atom efficient and more sustainable reactions. Optimising such reactions can be complicated because many of the reactor parameters are highly correlated - changing one can affect several others. This project is supervised by an interdisciplinary team involving computational chemistry, mathematics, spectroscopy and process chemistry. Initially a flexible high pressure flow reactor will be constructed with facilities for using a variety of process analytical techniques for monitoring thermal catalytic reactions. Then this equipment will be applied to reaction self-optimisation. The aim is to build up a library of algorithms that can manipulate data from the reaction monitoring to identify the optimum reaction parameters. Understanding the advantages and disadvantages of each algorithm will enable reactions to be optimised as efficiently as possible, thereby minimizing the use of chemicals during process development and delivering sustainable processes.