The resurgence of electrochemistry in medicinal chemistry laboratories has created the demand for the rapid delivery of up to a few kilograms of material by the same methods in order to quickly evaluate whether the candidate molecule is a viable clinical candidate or not. Electrochemical processes have been successfully scaled up to commercial scale using parallel plate reactors, but usually this requires a time-consuming complete redevelopment of the reaction conditions, including the electrode materials. In addition, mass transfer is of critical importance to electrochemistry given that reaction only occurs within very close proximity of the electrode surface, and in a plate reactor mass transfer is flow rate dependent. One solution is to rapidly recirculate the process mixture through the electrode plates, but this limits overall throughput relative to the reactor footprint, so a better solution would be to decouple mass transfer from flow rate by using mechanical mixing. Interestingly, two flow electrochemical reactor concepts designed for rapid scale up to kilogram scale using different mechanical mixing principles have been recently reported in the literature.
The first system based on the concept of using two cylindrical electrodes separated by a narrow gap with the central electrode rotating to create excellent mass transfer has been developed by the team at the University of Graz in Austria (DOI: 10.1021/acs.oprd.3c00255). This type of reactor, known as Taylor-Coutte system, has been independently reported by a team at the University of Nottingham in the UK and shown to deliver up to hundreds of grams of product (Org. Process Res. Dev. 2022, 26, 2674). What was of particular interest to me was that the team in Graz has designed reactors and a workflow to facilitate rapid scale-up from the IKA ElectraSyn 2.0 system which is popular among medicinal chemists. To accomplish this, they built a family of three different sized reactors which maintaining the same 5 mm gap between the electrodes as the IKA ElectraSyn 2.0 system. The small reactor has a working volume of approximately 50 mL, the middle reactor 250 mL, and the large reactor 1250 L, so there is a 5-fold scale increase from one to the other. All three reactors can be operated in batch and flow modes which is important for conserving materials during the screening of reaction conditions.
In order to demonstrate the rapid scale-up workflow the team in Graz used a Shono methoxylation reaction that had been first developed in an IKA ElectraSyn 2.0 system. The reaction conditions were then transferred without modification to the medium scale reactor in batch mode and the current increased from 10 A incrementally to 60 A in order to find the best balance between throughput, conversion and selectivity. This was found to occur at 40 A, so a reaction was performed on a 1 L scale in recirculation mode to convert 113 g of material into 129 g of product (93% yield).
Next, the Graz team examined whether a flow cascade operating in single-pass mode could be used to improve the throughput relative to operating in recirculation mode. For this experiment they transferred conditions for the anodic decarboxylative methoxylation of diphenyl acetic acid identified in batch mode in a 3 mL IKA ElectraSyn 2.0 system to a cascade of three rotating electrode reactors (3 x 250 mL, 3 x 470 cm2 electrode surface area) at a flow rate of 20-30 mL/min and a total current of 49.2 A to enable full conversion in a single pass. They found that the best balance of conversion and selectivity occurred at 24.5 mL/min (corresponding to 2.5 F/mol) to afford 711 g of product (97% isolated yield) over the course of a 5 h run. This corresponds to 146 g/h or 3.5 kg day, which is an excellent throughput for a system that can be readily housed in a laboratory fumehood.
Finally, the Graz team show that their system is resistant to clogging by electrolysis of cortisone to cleave the side chain and afford adrenosterone. This reaction was performed as a slurry at 0.5 M which is approximately 10-fold more concentrated than the solubility limit of cortisone in the electrolyte. The reaction was performed on a 1 L scale (180 g of cortisone) using the medium sized reactor (250 mL) in recirculation mode (200 mL/min) with wide bore tubing and fittings to minimize the risk of clogging, using conditions identified in batch mode on a 3 mL scale. No clogging was observed, and 141 g of pure product was obtained in 94% yield. They also demonstrated that this reaction could be performed in batch mode in all three sizes of reactor at constant linear velocity of the electrode to give consistent results across the different scales, so this rotating electrode system does appear to be a reactor design that could enable rapid scale-up of electrochemical reactions.
A slightly different approach to electrochemical scale-up has been reported by the Process Research and Development team at Abbvie (https://doi.org/10.26434/chemrxiv-2023-mqzn0). Their system is based on a design that they have previously reported for the use in rapidly alternating polarity electrochemical reactions on up to a 1 kg scale in batch mode that uses a more conventional laboratory glass jacketed reactor and overhead stirring with interchangeable rod electrodes. In this case mass transfer limitations are overcome using an impeller system to agitate the reactor mixture, and that by constant spacing of the electrode rods so that they can produce a range of reactor sizes that should provide similar performance across scale. They mention in their introduction that they have used this system for 3 years and have observed the advantages of this system compared to a number of parallel plate and microchannel reactor designs.
For this paper the Abbvie team developed a CSTR cascade of four reactors operating in single pass mode in order to scale-up with sufficient throughput from lab scale to commercial scale. They too used a Shono oxidation as a model reaction to demonstrate their system, albeit on a different substrate from the team in Graz. A major reason for selecting a Shono oxidation is that it is prone to over-oxidation, especially as the reaction mixture reaches close to full conversion, so this provides a good test of how to tune the system parameters to simultaneously deliver high conversion and selectivity.
One key difference between the two systems is that the Graz team operated their reactors at constant current, whereas the Abbvie team operated at constant potential. Running at constant current has the advantage that the reaction rate is constant throughout, but this can lead to over-reaction and therefore lower selectivity. Running at constant potential in batch mode will lead to slower reaction, especially at the end of the reaction when the substrate is depleted. To overcome this limitation the Abbvie team took advantage that the potential can be set to different levels in each reactor in a CSTR cascade in order to drive a high reaction rate at the start of the cascade and high degree of selectivity at the end.
The Abbvie team also used an IKA ElectraSyn 2.0 system to rapidly identify the initial reaction conditions, which they then scaled-up in batch mode to evaluate the parameters that would maximize the reaction rate. These studies showed and that the rate increased up to 6.0V but then plateaued because of mass transfer limitations due to the maximum stirrer rate being 1000 rpm. Monitoring the reaction profile over time showed that reducing the reaction temperature from 25 C to 15 C significantly decreased over-reaction at higher conversion, but at lower temperature they needed to modify the electrolyte to to prevent it from crystallizing from the solution.
Using the detailed reaction profile information that they had acquired in batch mode they were able to design the four reactor CSTR cascade. Reactors 1 and 2 were set to 6.0 V cell potential to maximize the initial rate of conversion. Batch kinetics showed that at 6.0 V problematic over-oxidation occurred above 72% conversion which was obtained in 44 minutes, so the flow rate of the CSTR system was set to have a mean residence time of 22 minutes in each reactor. Reactor 4 was set at 3.5 V to maximize the selectivity for the desired product during the final 10% of the reaction when over-reaction is most likely. Reactor 3 was set at the intermediate value of 4.0 V to achieve the remaining 18% conversion. An initial 1 kg run over 9.3 hours gave the desired product in 85% yield and held the over-oxidation by-product below 2%. No electrode degradation was observed over the run, and despite the temperature in Reactors 1 and 2 exceeding the set temperature due to Ohmic heating, the authors extrapolate that this system that fits inside an R&D fumehood could operate at 2.6 kg/day.
Finally, the Abbvie team report a 10-fold scale-up reactor based on electrode surface area (6 L volume) designed for commercial scale production. A reaction performed in batch mode on a 400 g scale achieved 95% conversion in 5.3 h and afforded the desired product in 80% isolated yield, but the observed current was only ~40% of that expected from the same applied potential at smaller scale. The authors attribute this to lower electrode surface area to volume in the larger reactor and differences in the electrode spatial arrangement. By applying a higher potential to a 700 g batch they were able to achieve the expected rate and reaction profile, and by extrapolating these results to a 4-CSTR cascade they extrapolate that >30 kg/day throughput should be possible.
Reading both papers in together I found it interesting that both sets of authors gave similar motivating factors for their work and recognized the challenges that plate reactors pose to rapid scale up. To me, the Abbvie system appears to be the more expedient approach that could be more readily built in-house provided the company has sufficient resources to do so. In contrast, the Graz system looks more like an advanced prototype that a commercial equipment manufacturer could develop into a portfolio of reactors for an off-the-shelf implementation by scientists with limited electrochemistry expertise, but whether that will happen is unknown given that rotating Taylor-Couette reactors have been known for many years but have yet to find widespread use for rapid scale-up. Hopefully, in the near future scale up electrochemists will have the tools they need to rapidly deliver the same novel chemistry that is emerging from medicinal chemistry labs.