Model reaction and batch experiments
The electrochemical methoxylation of 4-methylanisole was selected as model reaction for this study. This well-known transformation is carried out on a large scale at BASF using graphite disk electrodes in a narrow-gap arrangement and recirculation of the electrolyte [11, 13]. It is very often utilized as model for the evaluation of electrochemical flow cells [20, 31,32,33,34]. A 2-electron anodic oxidation of the benzylic group in methanol provokes its acetalization (Scheme 1). Formation of the desired diacetal 3 occurs after a second 2-electron oxidation step. The process therefore requires a theoretical charge of 4 F/mol for completion. Acidic workup of the diacetal yields the desired aldehyde 3′. If too much charge is applied to the reaction mixture, compound 4 is formed. Its acidic workup produces the corresponding methyl ester 4′. Reduction of protons from the methanol on the cathode produces hydrogen gas as byproduct.
Scheme 1Anodic methoxylation of 4-methylanisole
Batch experiments were carried out in a commercial reactor (IKA ElectraSyn 2.0) using graphite as the anode material and graphite or stainless steel as the cathode. In a typical experiment, a standard 5-mL vial was loaded with a solution of 4-methylanisole (0.1 M) and a variable amount of Et4NBF4 as supporting electrolyte in methanol (3 mL). A current of 20 mA (current density ca. 13 mA/cm2) was applied until a charge of 1–6 F/mol had been passed through the solution (Table 1). Graphite and stainless steel as cathode material provided similar results. The conversion was slightly higher with stainless steel as a cathode when charge amounts up to 4 F/mol were applied (entries 1–4). After an excess of charge had been passed, the reaction using stainless steel as a cathode material showed a better performance. Thus, higher conversion, current efficiency, and GC yield for compound 2 were observed (Table 1, entry 5). Notably, the current efficiency remained essentially constant with a value of ca. 60% independently of the amount of charge passed. This nearly constant value could indicate that the efficiency is independent of the concentration of the substrate (1), at least at conversions of up to 90%.
Table 1 Electrochemical methoxylation of 1 in a batch reactora Continuous flow setup
Continuous flow experiments were carried out using a commercial parallel plates flow cell (Vapourtec Ion Reactor) [35]. The reactor consisted of two electrodes (5 × 5 cm) separated by PTFE or FEP spacers of variable thickness. The surface area of the electrode exposed to the reaction mixture was 2 × 12 cm2. Graphite was utilized as the anode material. For the cathode, graphite or 304 stainless steel was utilized. The reaction mixture was introduced into the reactor using a 1.5-mL sample loop and a 6-port valve (Fig. 2). The reactor was pressurized in some cases (vide infra) using an adjustable back-pressure regulator (BPR). A programmable laboratory power supply was utilized to electrify the flow cell. The current and/or voltage of the cell was simultaneously monitored using a digital multimeter (see the “Experimental section” for details)
Galvanostatic vs potentiostatic operation in continuous flow
Electrochemical reactions in galvanostatic mode are carried out under constant current. The voltage of the cell is variable and adapts to the reaction conditions to achieve the set current. The amount of charge applied to the reaction mixture can be easily calculated from the value of the current using Faraday’s law (Eq. 1), where I is the current, V̇ is the flow rate in milliliter per minute, c the concentration of the substrate in solution in molar, F the Faraday constant (1608.0889 min A/mol), and Q the amount of charge in F/mol. The same equation can be utilized to calculate the flow rate/current ratio for a given amount of charge and substrate concentration. During potentiostatic operation, the cell voltage is set to a constant value and the cell current is variable.
$$ Q=\frac{I}{\dot{V}\times c\times F} $$
(1)
Constant current is the most common mode of operation of continuous flow cells [9, 12]. More robust results are typically achieved under constant current, as the current density is known and reproducible and the desired amount of charge being applied to the reaction mixture can be achieved by adjusting the current [13]. Constant voltage operation can be more problematic, especially if a two-electrode configuration is utilized, because the cell voltage can be influenced by many parameters, including temperature, small amounts of impurities in the reaction mixture, or even external factors such as contact issues between the cell and the power supply [13]. Moreover, small variations in the cell voltage can lead to dramatic changes in the current and ultimately in the amount of charge applied to the mixture. This is especially relevant for single-pass continuous flow reactions.
During lab scale optimization of continuous flow reactions, it is commonplace to use small amounts of reagents, in some instances introducing the solutions into the reactor with the aid of sample injectors. It should be noted that, under galvanostatic mode, current should only be applied to the flow cell under steady-state conditions (i.e., when the cell is filled with solution and its concentration is constant). If current is applied too early or in a later stage when the amount of substrate in the flow cell is too low, high amounts of undesired overoxidation products may occur, which in some cases could provoke further issues such as clogging of the cell or electrode fouling.
To illustrate the importance of applying constant current only under steady-state conditions, we carried out the model reaction using the continuous flow setup depicted in Fig. 2 and an interelectrode gap of 0.1 mm. The sample loop was loaded with a solution of 1 (0.1 M) and Et4NBF4 (0.1 M) in MeOH. A flow rate of 400 μL/min and a constant current of 257 mA (4 F/mol) was applied during the whole process. To avoid major issues with undesired oxidations (e.g., solvent oxidation or even electrode damage) a maximum voltage of 5 V was set in the power supply. Figure 3 shows the voltage profile recorded during the process. As the reaction mixture entered the flow cell, the voltage dropped to a value of 3.2 V. When no electrolyte was present in the cell, the power supply reached the maximum of 5 V and switched to constant voltage. During the transient period until steady state is reached, all the current is applied to a small fraction of reaction mixture, leading to overoxidation. This was demonstrated by GC monitoring of the output stream before and after steady state. Notably, high amounts of side-products (42%) were observed.
Under potentiostatic operation, the reaction selectivity is not affected by the reactant concentration. Thus, steady-state conditions are not required to apply electricity to the cell, which may be advantageous when only small amounts of starting material are available. When the reaction mixture enters the flow cell, the current will increase with the amount of electrolyte existing in the cell. Potentiostatic operations can lead to improved selectivities in some instances [36], as the potential can be tuned below the redox potential of sensitive functionalities. However, to achieve robust and reproducible results under potentiostatic conditions, a three-electrode configuration is typically required. As mentioned above, the cell voltage in a two-electrode arrangement can be affected by external factors such as temperature or small amounts of impurities on the electrode surface. Moreover, small changes in the cell voltage can lead to significant alterations in the cell current (Fig. 4a). This effect, less relevant in flow experiments with recirculation of the solution, is very important for single-pass continuous flow experiments in which a particular amount of charge to be applied to the reaction mixture is targeted.
To exemplify these potential issues under potentiostatic operation using a two-electrode configuration setup, the model reaction was carried out under constant voltage. A potential of 3.2 V was initially selected, which corresponded to the potential observed in a previous experiment under constant current (cf. Fig. 3). The same reaction conditions were utilized (0.1 M 1 and 0.1 M Et4NBF4 in MeOH). Notably, the observed current (Fig. 4b) was significantly lower than expected. A value of 130–140 mA under steady-state conditions was recorded. In this case, as the current is not constant, the amount of charge transferred to the reaction mixture can be calculated by integrating the current value over the time period (see Eq. 2 in Fig. 4b). This can be easily done by numerical integration using any spreadsheet software (e.g., MS Excel) with the current vs time data. The amount of charge, in this case, 2.07 F/mol was insufficient. Increasing the cell potential by 0.5 to 3.7 V essentially doubled the observed cell current and consequently the amount of charge (Fig. 4b). In this case, a small excess of current over the theoretical amount required (4.04 F/mol) was applied.
Importance of the interelectrode distance and current density
The interelectrode gap has a significant influence on the performance of a flow cell. In general terms, it can be stated that narrower gaps lead to lower cell resistance, typically resulting in smaller amounts of supporting electrolyte being required to run the electrochemical reaction [9, 27]. In addition, narrower gap cells exhibit larger electrode area to cell volume ratios. This ratio has a direct effect on the rate with which the electrolysis can be carried out [9].
Current and current density (current relative to the electrode area) are the parameters that define the cell productivity and one of the most important factors that characterize an electrochemical reaction, both in batch and flow mode. Electrochemical reactions typically perform better in terms of current efficiency and selectivity at low current densities. The development of flow cells and other scale-up strategies should therefore focus on the maximization of the current density that is possible to be applied to the reaction, in addition to the utilization of large electrode surface areas. Importantly, the interelectrode gap influences the current density values that can be applied to the cell: narrow gaps enable higher current densities (vide infra). This effect results in higher productivities for lower volume cells for a given electrode surface area (a narrower interelectrode distance provides a lower cell volume), which may sound counter-intuitive to the synthetic organic chemist.
We carried out the model flow electrochemical methoxylation of 1 using variable concentrations of supporting electrolyte and interelectrode spacers of different thickness (Fig. 5). As expected, the cell voltage was lower for the narrower interelectrode gaps even with a 0.1 M concentration of Et4NBF4. As the concentration of supporting electrolyte was decreased, the cell voltage significantly incremented in the batch reactor (5 mm gap). The increase was less pronounced in the flow reactor with a 0.5-mm interelectrode spacer. Notably, with a 0.1-mm gap the supporting electrolyte concentration could be decreased down to 0.02 M without any detectable increase in the cell voltage (Fig. 5). A 5-fold decrease in the amount of a relatively costly substance could be achieved by transferring the electrochemical reaction to flow and reducing the interelectrode distance.
We next turned our attention to the effect of both the interelectrode distance and the current density on the reaction conversion and selectivity. Four different interelectrode distances and flow rates were evaluated (Table 2). The cell current was set according to the flow rates to keep the amount of charge constant at 4 F/mol. As expected, the conversion and yield decreased with increasing current densities for all interelectrode spacers. With the thinnest spacer (0.1 mm), a decrease in the conversion from 96 to 88% and in the GC yield of 3 from 87 to 64% was observed when the current density was gradually increased from 11 to 43 mA/cm2. The drop in the reaction performance was more pronounced with wider interelectrode gaps. Thus, with a 1-mm gap, the conversion decreased from 89 to 44%, and the amount of 3 from 65 to 29% (Table 2).
Table 2 Effect of the interelectrode gap distance and the current density on the conversion and selectivity for the continuous electrochemical methoxylation of 1a The results collected in Table 2 also illustrate the effect of the interelectrode distance on the current density that could be applied to a flow cell. For example, analogous conversion and selectivity was achieved with a current density of 11 mA/cm2 when the 1-mm gap was used (entry 1) and with 43 mA/cm2 when the 0.1-mm spacer was utilized instead (entry 4) (ca. 89% conversion and 65% yield). The current increase enabled by the narrower interelectrode gap permitted to increment the flow rate and therefore the cell productivity from 200 to 800 μL/min. Thus, as mentioned above, flow cells display larger productivities for smaller reactor volumes on equal electrode surface area, due to the higher performance of cells with narrower interelectrode gaps.
Flow reactor back-pressure in reactions with gaseous byproducts
Most anodic oxidations utilize a proton source (e.g., water or an alcohol) as a convenient electron acceptor for the concurrent cathodic reduction. This generates stoichiometric amounts (1 equiv for a 2-electron process) of H2 gas as byproduct. While in batch the gas simply moves to the headspace, in continuous flow reactors, with no head space available, the generated H2 affects the cell performance. The behavior of the gas phase and its effect on the flow electrolysis has been studied in detail [37,38,39]. Gas generation typically results in gas/liquid segments in a flow reactor. As they do not conduct electricity, the presence of gas segments increases local current densities in the liquid phase, which has a negative effect on the cell performance. On the other hand, gas segments moving through the reactor channel influence the mass transfer. Depending on the flow regime, this might have a positive effect by promoting turbulence. Increasing the operating pressure of the flow cell has been suggested as a possible solution to minimize the issues associated with the generation of gas by reducing the size of the segments [37,38,39]. High pressure could also have a negative effect on the current efficiency by incrementing the amount of H2 dissolved in the liquid phase, aiding its oxidation on the anode [40].
The electrochemical methoxylation of 1, a 4-electron process, generates two equivalents of H2 gas. For our model reaction conditions (0.1 M concentration of 1 in MeOH), approximately 2.2 mL of gas are generated per milliliter of reaction mixture. Using an adjustable BPR, the continuous flow reaction was evaluated under back-pressures ranging from 0 to 5 bar. A 0.5-mm interelectrode gap distance was utilized in this case. As expected, the average cell potential decreased at higher pressure due to the smaller size of the gas segments that provoke increased cell resistance (Table 3). Interestingly, the reaction conversion, yield of 3, and the current efficiency gradually decreased with increasing pressures. This effect could be ascribed to large amount of H2 dissolved in the liquid phase, resulting in undesired oxidation of the gas to protons. This pressure effect has been described by Attour et al. [40], although in their study higher pressure had to be applied in order to observe this decrease in cell efficiency. Another interesting effect was observed in the cell voltage when back-pressure was applied (Fig. 6). Gas and liquid slugs typically produce different resistance for passing through a BPR. Gas segments can pass more easily than liquid segments, which provokes pressure oscillations in the system. These oscillations translate to the cell potential, likely due to expansion-compression of the gas segments and detachment of small gas bubbles from the electrode surface.
Table 3 Effect of the flow cell back-pressure on the conversion, selectivity, and current efficiency for the continuous electrochemical methoxylation of 1a Batch vs flow cell efficiency
Even though the study described herein is not representative of large-scale operation (only 0.3-mmol scale reactions were carried out in batch), important differences between the reaction performance in batch and flow mode could be observed.
The batch reactions were carried out using a current density of 13 mA/cm2 and a 5-mm interelectrode distance. The relatively small amount of reaction mixture processed (3 mL, 0.3 mmol) required 96 min of electrolysis time for an amount of charge of 4 F/mol. Yet, only moderate conversions were achieved with the theoretical amount, and a 50% excess (6 F/mol) was required to obtain good conversions over a 144-min time period (Table 1). The current efficiency for the batch reaction was ca. 60%. Using a similar current density (Table 2, entry 1) in continuous flow with a 0.1-mm interelectrode gap distance resulted in excellent conversion and yield using a charge of 4 F/mol. In this case, the current efficiency was nearly quantitative. The time required to process the same amount of reaction mixture (3 mL) in flow under these conditions was 15 min. Importantly, the narrow interelectrode gap enabled higher current densities in the flow cell. A 4-fold increase of current density resulted in very good conversion also using 4 F/mol. In this case, 3 mL of reaction mixture could be processed in 3.75 min, with a productivity ca. 40 times compared with the batch reactor. The comparison between the efficiency of the batch and flow processes is summarized in Table 4.
Table 4 Comparison of the current efficiency and cell productivity obtained for the batch and flow electrochemical methoxylation of 1