Aqueous electrochemically-triggered atom transfer radical polymerization

Simplified electrochemical atom transfer radical polymerization (seATRP) using CuII–N-propyl pyridineimine complexes (CuII(NPPI)2) is reported for the first time. In aqueous solution, using oligo(ethylene glycol) methyl ether methacrylate (OEGMA), standard electrolysis conditions yield POEGMA with good control over molecular weight distribution (Đm < 1.35). Interestingly, the polymerizations are not under complete electrochemical control, as monomer conversion continues when electrolysis is halted. Alternatively, it is shown that the extent and rate of polymerization depends upon an initial period of electrolysis. Thus, it is proposed that seATRP using CuII(NPPI)2 follows an electrochemically-triggered, rather than electrochemically mediated, ATRP mechanism, which distinguishes them from other CuIIL complexes that have been previously reported in the literature.


Introduction
Electrochemical intervention in synthesis and catalysis has received renewed interest over the last 5-10 years. [1][2][3][4][5][6] From a synthetic point of view, the use of an applied potential/current enables accurate control over the thermodynamics and/or kinetics of electron transfer processes. 7,8 This can enhance the selectivity of chemical transformations and confer spatiotemporal control over synthetic and catalytic reactions of small and macromolecular organic molecules/polymers, amongst others.
In the context of reversible deactivation radical polymerization (RDRP) electrochemical intervention has been employed to regulate polymer synthesis through control of the dynamic equilibrium between dormant and active (radical) species which allows the overall radical concentration to be accurately controlled. [9][10][11]53 In atom transfer radical polymerization (ATRP) 12,13 the equilibrium (K ATRP ) is between a dormant alkyl (R-X) or macromolecular (P n -X) halide and propagating radicals (Rc/P n ) which undergo reversible redox reactions with transition metal complexes.
In 2011, Matyjaszewski and co-workers showed that the redox nature of the Cu-mediated ATRP mechanism could lend itself to electrochemical manipulation and control. 10 The active, yet oxidatively labile Cu I L complex was formed in situ when a reducing potential (E app ) was applied at the working electrode (WE) to induce a one electron reduction of an inactive Cu II L precursor. Activation of the dormant species (R-X/P n -X) in the reaction media then generated the radical species (Rc/P n ), and the Cu-complex in a higher oxidation state (X-Cu II L). Well controlled polymerization of methyl acrylate was reported suggesting that the deactivation step of the equilibrium, between the propagating radical ðP n Þ and X-Cu II L, reforming the dormant species (P n -X) and Cu I L respectively, was not perturbed by the electrochemical intervention. In fact, it was shown that by switching the E app at the WE to an oxidizing potential the polymerization could be completely switched off, conferring high delity on-off spatiotemporal control over polymer synthesis in solution.
In the 10 years since this discovery, eATRP has been employed for the synthesis of polymers with a variety of compositions and architectures including block copolymers, bioconjugates, star and gra (co)polymers. [14][15][16][17][18][19][20][21][22] It is compatible with aqueous 23,24 and organic 10 media whilst heterogeneous systems such mini-emulsion [25][26][27][28] and surface-initiated (si-eATRP) [29][30][31][32] polymerizations have also been reported. Furthermore, the complex reaction set-up, initially involving a 3-electrode divided electrochemical cell, has been simplied by the use of sacricial counter electrodes (typically Al-wire), enabling undivided cells to be used in either 3-electrode (potential controlled) or 2-electrode (current controlled) congurations giving rise to simplied electrochemical atom transfer radical polymerisaiton (seATRP). 33 This development is signicant as it enables the chemistry to be performed using commercial, standardized hardware. 24 The most widely studied systems for aqueous eATRP employ Cu II X salts with tetradentate ligands tris(2-(dimethylamino) ethyl)amine (Me 6 -Tren) 10,34,35 or tris(2-pyridylmethyl)amine (TPMA). 33,36,37 They form more active complexes, having high K ATRP values. 38 The ligands stabilize Cu II more than Cu I with cyclic voltammetry (CV) indicating that Cu I Me 6 -Tren and Cu I TPMA are strongly reducing complexes, leading to fast activation (k act ) of R-X/P n -X. 39 Cite this: Chem. Sci., 2022, 13, 5741 All publication charges for this article have been paid for by the Royal Society of Chemistry increase by orders of magnitude when aqueous media is employed, which in the absence of appropriate conditions and/ or external control of active catalyst generation, can result in high radical concentrations which has a detrimental effect on the polymerization. 41,42 A great deal of discovery and optimization, of which eATRP is one example, has resulted in the development of efficient, well controlled aqueous ATRP reactions using these highly active complexes. 9,43 Prior to this, less active complexes composed of bidentate ligands such as bipyridine (bpy) and N-alkyl pyridine imines (NAPI) were more suitable for aqueous ATRP. [44][45][46][47][48][49] They stabilize Cu I more than Cu II , form less reducing Cu I complexes and have lower k act and K ATRP leading to lower radical concentrations. On one hand, this means that larger catalyst concentrations are required to mediate well controlled ATRP. On the other hand, it can also be benecial for polymerizations carried out in aqueous media wherein increased k act and K ATRP can lead to higher radical concentrations when more active complexes are employed. For example, 20 years ago, Haddleton and Perrier described in detail the efficient, well controlled polymerization of oligo(ethylene glycol) methyl ether methacrylate (OEGMA) using Cu I (NAPI) 2 complexes in water. [44][45][46] The rates of reaction and control over the polymerization were optimized with respect to [Cu I (NAPI) 2 ]/[Cu II (NAPI) 2 ] which was controlled from the outset by using known amounts of Cu I Br and Cu II Br 2 to form a mixed complex system. To accurately achieve the target [Cu I (NAPI) 2 ]/[Cu II (NAPI) 2 ] ratio's careful handling of oxidatively labile Cu I complexes and thoroughly deoxygenated reaction conditions were required. Looking back at this work, we considered the possibility of controlling [Cu I (NAPI) 2 ]/ [Cu II (NAPI) 2 ] electrochemically, thus avoiding the need to handle the oxidatively labile Cu I complexes. There are currently no reports of eATRP using Cu(NAPI) 2 complexes, in either organic or aqueous media in the literature. We were inspired to investigate these complexes with a view to mediate eATRP at less reducing potentials and currents. Long term, we hope this will help to overcome some of the initial limitations associated with oxygen reduction (at more reducing potentials) in our related work in scanning electrochemical probe directed eATRP. 50 To this end, herein we report for the rst time the use of the N-propyl pyridineimine (NPPI) ligand to form Cu II (NPPI) 2 complexes for eATRP of OEGMA 300 . Well controlled polymerization (Đ m z 1.30) is possible and initial investigations into the mechanism suggest that an alternative electrochemicallytriggered process is prevalent for these less-active copper complexes.

Results and discussion
Comparative CV of Cu II L complexes; Cu II TPMA, Cu II Me 6 Tren and Cu II (NPPI) 2 were initially performed in solutions of the reaction mixture (10% (v/v) OEGMA 300 in H 2 O) in the absence and presence of the initiator, hydroxyethyl-2-bromoisobutyrate (HEBiB) (Fig. S1-S3 †). In the absence of HEBiB, each complex exhibited the [Cu II L]/[Cu I L] redox process and as expected the standard reduction potential (E q z E 1/2 ¼ E pc + E pa /2) shied to less reducing potentials (vs. Ag/AgCl) going from Cu II Me 6 Tren (E 1/2 ¼ À0.40 V) to Cu II TPMA (E 1/2 ¼ À0.21 V) to Cu II (NPPI) 2 (E 1/2 ¼ +0.02 V) respectively. In the presence of HEBiB the voltammograms of the Cu II Me 6 Tren and Cu II TPMA complexes show a coupled increase in the cathodic current intensity (E pc ) and decrease in the anodic current intensity (E pa ). This is indicative of electrochemical reduction of Cu II L to Cu I L followed by fast activation of HEBiB by the Cu I L on the timescale of the CV (0.1 V s À1 ). In the case of Cu II (NPPI) 2 the coupled change in E pc and E pa was not observed. The currents decrease in both the cathodic and anodic scan suggesting that although the presence of HEBiB has an effect on the kinetics of electron transfer, the activation of HEBiB by Cu I (NPPI) 2 is slow on the timescale of the CV. These results are in agreement with the literature that suggests that with respect to k act , complexes with Me 6 Tren > TPMA [ NPPI. 35,38 Potentiostatic seATRP reactions using each complex were performed in undivided cells using an IKA ElectraSyn device. A commercial Pt-coated electrode (IKA) was employed as the cathode (WE), the anode (CE) was Al-wire and the reference electrode ( the resistance in the system was too high preventing the IKA Elec-traSyn from operating. However, when an overpotential of 60 mV was applied (E app ¼ À0.04 V) polymerization was complete within 2 h yielding POEGMA 300 with M n,SEC ¼ 9200 g mol À1 and Đ m ¼ 1.31 (Table 1, Table S1 †). This is likely due to the stoichiometry of Cu II Br 2 employed which equates to [Cu II Br 2 ] ¼ 8.8 mM. Although, this is suitable for the less active Cu II /NPPI system, it is much higher than is required for the so-called highly active complexes leading to higher than necessary [Cu I L] and [Rc/P n ] due to rapid over activation when E app ¼ E 1/2 which ultimately compromises the outcome of the polymerization.
Incrementally increasing the overpotential by 40 mV had little effect on the rate of the polymerization with conversions remaining high (>95%), and control being retained with Đ m z 1.30 (Table 1, entries 2-4, Fig. S5-S7 †). The E app was then xed at À0.16 V and [OEGMA 300 ] was increased to 20% and 30% v/v respectively ( Table 1, entries 5-6). There was no signicant change in the control over the polymerization with low dispersities (Đ m < 1.30) obtained ( Fig. S8 and S9 †). Kinetic analysis showed that quantitative conversions were obtained with 2 h. However, the semi-log plot showed that whilst the pseudo rst order kinetics were observed at [OEGMA 300 ] ¼ 10% v/v, distinct deviations were apparent at the higher concentrations (Fig. S10 †). The observed increase in rate throughout the reaction is in agreement with Haddleton and Perrier who attributed it to water and monomer completing with the ligand for coordination at the Cu centre thus affecting the Cu II /Cu I equilibrium. 44 [7][8]. Increasing the length of the OEGMA monomer using OEGMA 500 and OEGMA 1100 had little effect on the control over the polymerization, with low dispersities retained (Đ m < 1.30, Fig. S11 and S12 †), though the rate of polymerization for OEGMA 1100 was slower than OEGMA 300/500 reaching 67% within 6.5 h (   [1] revealed that the apparent rate constant for propagation was k app p ¼ 0.0167 min À1 at E app ¼ À0.04 V (Fig. 1C). Initially, more reducing potentials (E app ¼ À0.08 V) resulted in a small increase in the rate of polymerization (k app p ¼ 0.0456 min À1 ). However, at higher overpotentials (E app ¼ À0.12 V, À0.16 V), the rate decreased back to k app p ¼ 0.0201 min À1 and 0.0174 min À1 respectively. At these potentials, E app is close to E pc at which point the reduction of Cu II /L to Cu I /L is not governed by the electrode potential and is limited by the rate of diffusion of accumulated Cu/L species to and from the electrode surface to and from the bulk. Irrespective of E app , a linear increase in M n,SEC as a function of conversion was observed with good agreement with the theoretical molecular weight (M n,th ) (Fig. 1D).
A hallmark of eATRP is the temporal control conferred by switching the potential/current on and off. At reducing potentials Cu II /L is reduced to Cu I /L leading to activation of dormant chains which can undergo propagation and subsequent deactivation events via the proposed ATRP mechanism. If the potential is switched off, or an oxidising potential is applied, reduction of Cu II /L nolonger occurs so activation of the dormant chains stops and the polymerization is halted.  (Fig. 2). The potential was then switched off (E app ¼ 0 V) and stirring was continued for a further 20 min, aer which conversion unexpectedly increased to >80%. The polymerization reached >90% conversion through an additional 'on' (20 min) and 'off' (20 min) cycle, indicating that Cu II (NPPI) 2 lacked the temporal control associated with eATRP. The lack of temporal control with this less active catalyst system is in agreement with reported differences in temporal control related to catalyst activity observed in photo-ATRP. 51 Unlike in eATRP reactions using Cu II (Me 6 Tren) and Cu II (TPMA) complexes, it was observed that the reaction solutions containing Cu II (NPPI) 2 changed colour, from green to brown, during electrolysis (Fig. S14 †). The brown colour resembled that reported in early aqueous ATRP using Cu I (NAPI) 2 complexes. [44][45][46] Thus, it was hypothesized that the overpotentials applied and the less activating nature of the Cu I (NPPI) 2 complex, resulted in accumulation of stable Cu I (NPPI) 2 in the reaction media which was capable of continuing to mediated the polymerization of OEGMA when E app was removed.
To explore this hypothesis, the temporal control experiment was repeated using [ [1.25]. Aer the electrolysis period (E app ¼ À0.08 V; t E app ¼ 30 min), the reaction solution was brown, indicative of Cu I (NPPI) 2 accumulation, and conversion had reached 33% (Fig. S15A †). Concurrently, electrolysis was stopped and sparging with compressed air was commenced to rapidly introduce O 2 into the reaction solution to stop the polymerization. The solution quickly changed colour from brown to green, indicative of oxidation of Cu I (NPPI) 2 to Cu II (NPPI) 2 and no further conversion was observed (Fig. S15B †). In an attempt to re-initiate the polymerization, the reaction solution was sparged for second time, this time with N 2 to displace the O 2 previously added to the solution, prior to a second period of electrolysis. Pleasingly, re-initiation was observed (E app ¼ À0.08 V; t E app ¼ 30 min) with the polymerization reaching 64% conversion (Fig. S15C †), yielding POEGMA 300 with M n,SEC ¼ 9800 g mol À1 and Đ m ¼ 1.25 (Fig. S16 †) [1.25]) were electrolyzed at constant potential (E app ¼ À0.08 V) for increasing periods of time (t E app ¼ 5, 10, 20, 30 min) before the potential was removed (E app ¼ 0 V). Samples were taken for analysis aer electrolysis and at regular intervals aer the potential was removed. Increasing the initial electrolysis times led to an increase in initial conversion from 2% (E app ¼ À0.08 V; t E app ¼ 5 min) to 22% (E app ¼ À0.08 V; t E app ¼ 30 min). In all experiments monomer conversion continued upon removal of E app (Fig. 3A). At shorter electrolysis times, conversion continued up to a total reaction time of 60 min resulting in nal conversions of 18% and 35% when t E app ¼ 5 and 10 min respectively. Increasing the initial electrolysis time to 20 min yielded initial conversions of 11% with monomer conversion continuing thereaer to reach a nal conversion of 56% aer a total reaction time of 70 min. When the reaction solution was electrolyzed for 30 min monomer conversion continued for a total reaction time of 90 min, reaching 93% conversion. Kinetic analysis of these reactions revealed that the rate of the reaction also increased from k app p ¼ 0.0028 min À1 when t E app ¼ 5 min to k app p ¼ 0.0425 min À1 when t E app ¼ 30 min (Fig. 3B). An overlay of the SEC chromatograms shows that the polymerization continues aer the initial electrolysis period with the molecular weight distributions shiing to higher molecular weights as a function of time (Fig. 3C). The nal polymer obtained (E app ¼ À0.08 V; t E app ¼ 30 min) was comparable to the polymer obtained by uninterrupted electrolysis (Table 1; entry 2) with M n,SEC ¼ 9300 g mol À1 and Đ m ¼ 1.33.
Quantication of the end group delity using conventional 1 H NMR analysis was not possible as poly(methacrylates) do not contain an u-methine proton to integrate against signals at the a-chain end. To exemplify end group delity, a chain extension experiment was performed.  (Fig. 4). The molecular weight of the nal POEGMA 300 obtained (M n,SEC ¼ 16 600 g mol À1 ) was in reasonable agreement to the theoretical molecular weight (M n,th ¼ 12 200 g mol À1 ) relative to the homopolymerizations performed. Although this result exemplies good chain-end delity, there is scope for optimization based on a gradual increase in tailing to low molecular weight, which increased during the course of the reaction resulting in a gradual increase in dispersity (Đ m ¼ 1.27-1.45, Table 2). The electrochemically triggered reaction conditions were also compatible with polymerizations of OEGMA 300 targeting higher molecular weights. When DP n,th ¼ 200, the reaction solution was electrolyzed for 30 min (E app ¼ À0.08 V) leading to 13% conversion. The reaction continued in the absence of electrolysis for an additional 90 min, reaching 89% conversion (Fig. S18 †) furnishing POEGMA 300 with relatively low dispersity (Đ m ¼ 1.32, Fig. S19 †). To expand the monomer scope, 2-Nmorpholinoethyl methacrylate (DP n,th ¼ 200) was electrolyzed for 30 min (E app ¼ À0.08 V) resulting in 40% conversion. The reaction was allowed to continue in the absence of electrolysis for an additional 90 min, reaching 65% conversion ( Fig. S20 2 to Cu I (NPPI) 2 followed by UV-vis spectroscopy (Fig. 5). Prior to electrolysis the reaction solution was green and the characteristic Cu II (NPPI) 2 absorbance band was present at l ¼ 670 nm,  assigned to the d-d transitions of the Cu II centre. Aer electrolysis the reaction solution was brown in colour, qualitatively conrming the reduction of Cu II (NPPI) 2 to Cu I (NPPI) 2 . UV-vis of the reaction solution immediately aer electrolysis showed disappearance of the absorbance band at l ¼ 670 nm and appearance of a new, strong absorbance band at l ¼ 465 nm, conrming reduction of Cu II (NPPI) 2 to Cu I (NPPI) 2 . The absorbance at l ¼ 465 nm was assigned to MLCT between the Cu I centre and the p* of the surrounding NPPI ligands, as reported for other bipyridyl and/or diimine based complexes of Cu I . 52 In order to quantify the concentration of Cu I (NPPI) 2 present aer electrolysis, a calibration plot of Cu I (NPPI) 2 was used to determine the molar extinction coefficient of Cu I (NPPI) 2 (3 ¼ 1359 M À1 cm À1 , Fig. S22 †). Prior to electrolysis, the concentration of Cu II (NPPI) 2 in the reaction solution was 8.8 mM. Aer electrolysis for 30 min, conversion reached 24% (Fig. 5A) and [Cu I (NPPI) 2 ] was measured and found to be 4.94 mM (Fig. 5B). The reaction was again allowed to continue in the absence of an applied potential (E app ¼ 0 V). Though the polymerization continued, the colour of the reaction solution gradually changed from brown back to green over the course of the reaction. Further UV-vis analysis of the reaction solution allowed [Cu I (NPPI) 2 ] to be followed, revealing a steady decrease over time eventually reaching 1.37 mM aer 30 min at which point the reaction had reached 72% conversion.
Identical analyses were performed during polymerization of OEGMA 300 using Cu II TPMA and Cu II Me 6 -Tren. Due to the tetradentate nature of TPMA and ME 6 Fig. S1 and S2 †) and electrolysis was initially performed for 30 min before the potential was removed and stirring continued at room temperature. The progress of the reactions was followed by 1 H NMR revealing $5% conversion aer the initial period of electrolysis. UV-vis analysis showed very little change in the absorbance spectra of each complex (Fig. 5C) and unlike the Cu II (NPPI) 2 system, no further conversion of monomer to polymer was observed when the reaction was continued for 30 min at E app ¼ 0 V (Fig. 5A). This is perhaps unsurprising considering the relative activity of the Cu I TPMA and Cu I Me 6 -Tren complexes relative to Cu I (NPPI) 2 . Thus we repeated the reaction using bipyridine (bipy) to form Cu I (bipy) 2 in situ which has intermediate activity relative to the highly active complexes derived from Me 6 -Tren/TPMA and the less active complex derived from NPPI. Similar to Cu II (NPPI) 2 , initial conversion in the presence of Cu II (bipy) 2 increased with increasing electrolysis time (E app ¼ À0.08 V; t E app ¼ 10-30 min). However, in the absence of electrolysis polymerization was only maintained for 10-20 min reaching only moderate nal conversions (< 65%, E app ¼ À0.08 V; t E app ¼ 30 min, Fig. S23 †). This suggests that the ability to conduct electrochemically triggered eATRP is directly related to the activity (k act and K ATRP ) of the Cu-complex employed.
Overall, these results consolidate the hypothesis that the less activating nature of the Cu I (NPPI) 2 complex, and its stability in water results in its accumulation in the reaction media. The accumulated Cu I (NPPI) 2 is then capable of mediating the polymerization of OEGMA when E app was removed. We therefore propose that Cu complexes containing pyridine-imine ligands (Cu II (NAPI) 2 ) follow an electrochemically-triggered, rather than electrochemically mediated, ATRP mechanism wherein E app is only required in order to generate the required [Cu I (NAPI) 2 ] to initiate and maintain the polymerization reaction.
Finally, to simplify the reaction set up further, current vs. time (I vs. t) graphs obtained from reactions performed under potentiostatic conditions (Fig. 6A) were used to design a step-wise current prole to enable the electrochemically triggered polymerizations to be performed using a 2-electrode current controlled conguration.
Using  [1.25], a 3-step current prole was initially applied over 30 min (I app ¼ À3.5 mA, 8 min; À1.9 mA, 7 min; À0.5 mA, 15 min) resulting in 12% conversion. At this point the reaction continued in the absence of electrolysis for a further 120 min reaching 80% conversion (Fig. S24 †) with comparable control (M n,SEC ¼ 11 600 g mol À1 ; Đ m ¼ 1.25, Fig. 6B) to the polymerizations performed with a potentiostatic trigger. Considering future translation to ow electrolysis, it would be benecial to trigger these reactions using a single current, truly galvanostatic reaction conguration. With this in mind the reaction was repeated with I app ¼ À2.0 mA leading to 16% conversion aer the 30 min electrolysis period reaching 95% aer a further 120 min stirring in the absence of electrolysis (Fig. S25 †). This is very promising for intensication to ow electrolysis, though it should be noted that under these conditions, a higher overall charge is passed during the reaction which has an effect on the outcome of polymerization. Whilst the control respect to the dispersity is retained (Đ m ¼ 1.28, Fig. S26 †), the M n,SEC (14 300 g mol À1 ) and M n,th (6211 g mol À1 ) diverge relative the potentiostatic and step-wise current prole triggered reactions, leaving scope for optimisation in future. We attribute the divergence in the M n,SEC and M n,th to the gradual increase in the potential (required to maintain I app ) during the initial electrolysis period. This leads to an increase in [Cu I (NPPI) 2 ] and subsequently ½P n ; leading to increased termination and reduced initiator efficiency, relative to the potentiostatic and stepwise current prole triggered polymerizations.

Conclusions
In summary, seATRP using Cu(NPPI) 2 complexes in aqueous solution has been reported for the rst time. Typical electrolysis conditions require less reducing potentials (E app ¼ À0.08 V) than complexes derived from Me 6 -Tren and TPMA. Using OEGMA 300 as monomer, a range of molecular weights have been targeted with the polymerizations typically complete within 2 h, yielding POEGMA 300 with good control over the molecular weight distribution (Đ m < 1.35). However, the dening 'on-off' control experiment revealed that the polymerizations were not under complete electrochemical control, as monomer conversion continued in the absence of E app . This is contrary to previous reports using more active Cu II L complexes. Through electrochemically triggered control experiments and UV-vis spectroscopy we have been able to propose that these less activating complexes, that stabilize Cu I more than Cu II , follow an alternative, previously unreported, electrochemicallytriggered polymerization pathway. The polymerizations proceed with good control enabling a range of molecular weights to be targeted (DP n,th ¼ 20-200). In situ chain extension is also possible alluding to potential application to the synthesis of block copolymers. The reaction set-up can also be further simplied to a 2-electrode, galvanostatic conguration which is promising for future intensication through translation to ow electrolysis. However, though suitable for eATRP at reduced catalyst loadings, more active ligands such as Me 6 -Tren, TPMA and bipy do not support the electrochemically-triggered polymerization pathway. Indeed, the ability to conduct electrochemically triggered eATRP seems to be directly related to the activity of the Cu-complex and can be related to the k act and K ATRP of the complexes employed. Thus, it is possible that other ligands that stabilize Cu I over Cu II (e.g. other substituted NAPI and 1,4-diazabutadiene ligands) could also follow or favour this electrochemically triggered pathway.

Data availability
Experimental procedures and data supporting the research, not presented in the main manuscript, is included in the ESI. † Raw data les are available from the Warwick Research Archive