Intelligent automated reactor drastically cuts time it takes to optimise cross-coupling reactions
Identifying the ideal conditions for a catalytic reaction can be a laborious process: selecting the most efficient catalyst system in addition to assessing continuous variables like temperature, time and reagent concentrations can easily take hundreds of separate experiments.
Now, a team of chemists and chemical engineers led by Stephen Buchwald and Klavs Jensen from the Massachusetts Institute of Technology has developed a system that drastically cuts the time researchers spend optimising reactions. The hands-free approach identified optimum conditions within 96 experiments, each taking less than ten minutes, for a series of Suzuki-Miyaura cross-couplings – important carbon–carbon bond forming reactions that are widely used in industry.
‘The robotic system we present here conducts its own experiments, processes the data, and uses the information it collects to select new experiments to optimise each reaction, similar to how a chemist would,’ explains Brandon Reizman, who worked on the project.
The system uses an automated liquid handler to produce microlitre scale reaction droplets. These are pushed through a Teflon tube reactor and the resulting mixtures separated by high-performance liquid chromatography and analysed by UV-Vis spectroscopy and mass spectrometry. A computer algorithm then processes the results and feeds back instructions for the liquid handler to prepare the next reaction droplet.
‘The algorithm looks at the shared relationships among both the discrete variables and continuous variables, like temperature and time, and then using those relationships it converges very quickly to where it thinks the best results might be,’ explains Reizman.
Ryan Hartman, an expert in flow chemistry and microsystems based at New York University, US, lauds the research as a ‘vital contribution that advances both synthetic chemistry and chemical engineering.’ Commenting on the significance of the new system, he says that ‘the automation of laboratory-scale reactors is one of the most important, yet challenging chemical engineering feats that could revolutionise the way we discover science in the 21st century’. He also notes that the research is ‘an extraordinary benchmark of why scientists and engineers must work together’.
Looking to the future, Jensen explains that the group is keen to develop the system further so that it can be used for a wider range of complex reactions. ‘We are trying to build algorithms that are very robust, because ideally what we would like is a method that anyone could walk up to and use,’ he says.
This article is open access
B J Reizman et al, React. Chem. Eng., 2016, DOI: 10.1039/C6RE00153J