This work aims to translate a multi-step and multi-phasic cascade reaction into a one-step and one-phase reaction via a continuous flow process, exemplified by the synthesis of Rufinamide (1 ) and its analogue 1-(2,6-difluorobenzyl)-4-phenyl-1H -1,2,3-triazole (2 ) (Scheme 1).
Scheme 1. (a) Reaction scheme for the 2-step synthesis of Rufinamide (1 ) and analogue 1-(2,6-difluorobenzyl)-4-phenyl-1H -1,2,3-triazole (2 ). (b), (c) Proposed ONE-FLOW processes for the preparation of product1 and 2 .

Results and Discussion

As for the current production of 1 and 2 , multiple steps are required to obtain the purified product and most of the published processes are energy intensive and time consuming; meanwhile, sustainability and circularity issues such as reagent and catalyst recovery are in most published papers still not addressed.10-12 For instance, in previous work from our group, the first step involving the synthesis of the azide, a 40 min residence time at 160 °C and 7 bar (yield 93%) was selected as optimized reaction condition in flow,10 following the use of the ‘novel process windows’ (NPW) concept to boost flow chemistry reactions.13 In this work, we used a packed-bed resin-N3 microreactor to achieve possibly full conversion under moderate reaction conditions. In the previous study,10also a solvent-free three-step Rufinamide flow process was developed, as another NPW variant, based on the use of undiluted reactants (omission of the solvent) with resulting sustainability benefits.14 Yet the need for three consecutive steps, each one with a long reaction time, requires a start-up time for the whole synthesis of a few hours,15 while in this work starting up the reaction takes only a few minutes, in combination with compression of two steps into one and favorable reaction rates. We have designed a system, which is meant to simplify the whole cascade of reactors and separators traditionally used in a ONE-FLOW operation leading directly to the purified solid product. This is achieved by the selection of a designed solvent, screened by COSMO-RS,16 which enables a selective separation of the product simply by temperature variation; and a designed catalytic nanoreactor based on cross-linked polymer vesicles, or polymersomes,17 which facilitates catalyst recycling.
For the reaction solvent selection, we used a COSMO-RS based selection to screen out the reaction solvent candidates for the Rufinamide (1 ) synthesis, as shown in Table 1. Firstly, the solubilities of the two reactants (benzyl chloride R1, propiolamide R2) and product (rufinamide P2) at two temperatures (T1: 25 °C, T2: 65 °C) in common solvents were calculated separately by an auxiliary batch-processing program in COSMOthermX.\(\text{Log}_{10}\left(x\right)\) (\(\log S\), Equation S1, in supporting information) describes the optimized mole fraction of solute in one solvent and was chosen as the key parameter to characterize solubility. The maximal value of \(\log S\) is 0, meaning total dissolution in the solvent; as the values of \(\log S\) decrease, the solubility tends to be smaller.18 At 25 °C, the potential solvent should comprise high values for\(\log S_{R}\), while keeping low \(\log S_{P}\) and\(\log S_{P-R}\). The target was to find a candidate solvent which can dissolve reactants R1, R2 at 25 °C but in which product P2 is insoluble. As the key scenario constraint, the cut-off values of the following four parameters were chosen as \(\log S_{R1}>-1\) (freely soluble),19\(\ \log S_{R2}>-1\), \(\log S_{P2}<-1\),\(\log S_{P2-R1}<-1\), \(\log S_{P2-R2}<-1\). It decreased the total solvent candidates from 11957 to 1833. Next the difference in product solubility at different temperatures was used as the second main constraint. The aim was to find a solvent in which the product dissolves at higher reaction temperature 65 °C to achieve homogeneous conditions and precipitates automatically while cooling down to 25 °C. The chosen values were \(\log S_{P2}^{65}>-1.5\) sparingly soluble),19\(\log S_{P2}^{65}-\log S_{P2}^{25}>-0.5\). With the above screening step, the number of solvents that satisfied this condition was reduced to 356. To improve on the current synthetic processes, three commonly used solvents for this reaction were used as benchmark solvents and only solvents with more favorable characteristics (37 candidates, Table S1, in supporting information) were considered for the next step.20,21To narrow the screening space further, other chemical and physical properties and the economic effect of the solvents were considered. Finally, acetonitrile (ACN) was recognized as the top solvent and its suitability was validated with solubility tests.
Table 1. Number of solvents in different screening steps.