An on-line mass spectrometry method has been developed to monitor, model and optimise continuous flow reactions. This method makes use of dual-piston pumps, tubular reactor block, Vici sample actuator, an Advion Compact Mass Spectrometer (CMS) and other analytical systems to investigate a variety of chemical systems based on their need for process improvement. Full reaction automation employed MATLAB, the Snobfit algorithm, along with Modde DoE software.
On-line mass spectrometry has advantages over other analytical techniques as it has shorter acquisition times (2-60 s), low chemical sensitivity (~108 mol%) and chemical identity as well as the potential to provide quantitative information. In this work, reaction quantitation has been explored using four chemical systems, where each of them was monitored by a variety of analytical techniques, with the overall aim being to examine if on-line mass spectrometry can be used for quantitative analysis. For all cases investigated, process improvements were made whilst also determining optimal operating conditions to improve conversions, yields or selectivities as well as looking at reaction waste reduction. Flow chemistry and the work conducted has shown how waste can be reduced for certain reactions when compared to more traditional approaches. This method relies on machine learning, full process automation and quick process analytical technology to determine optimum conditions as well as build large reaction data sets. Large data sets were created using a hybrid DoE-kinetic composite circumscribed orthogonal design.
Mass spectrometry provided valuable reaction information and has the potential for reaction quantitation depending on the required application, reaction system and ionisation settings. Compound thermal stability can be problematic in APCI+ mode whilst ion suppression is problematic in ESI+ mode. Still a versatile analytical tool, on-line mass spectrometry was found to be inherently quantitative. The continuous-flow-on-line-MS-self-optimisation platform was used to investigate a variety of different reactions to show versatility of the MS system. These reactions are summarized below.
1) An N-Boc deprotection of AZD5634 for optimisation and process scale-up, with achieved conversions >95% and scale-up to pilot and commercial scale using on-line mass spectrometry).
2) An N-Boc deprotection reaction using a hybrid DoE-kinetic model for optimisation and large data set generation, with achieved conversion >90%.
3) An SNAr reaction of AZD4547 for product selectivity and yield improvement, with achieved conversion of ~38% and DP yield of ~30%.
4) The synthesis and optimisation of Fe-N-heterocyclic carbene complexes using an electrochemical method for use in a C-H hydroxylation reaction. Optimum electrochemical conditions of either 7 V and 4 minutes residence time, or 2.5 V and 15 minutes residence were achieved.
