Controlling One- or Two-Electron Oxidation for Selective Amine Functionalization by Alternating Current Frequency

Abstract

Here, we report a unique electrosynthetic method that enables the selective one-electron oxidation of tertiary amines to generate α-amino radical intermediates over two-electron oxidation to iminium cations, providing easy access to arylation products by simply applying an optimal alternating current (AC) frequency. More importantly, we have discovered an electrochemical descriptor from cyclic voltammetry studies to predict the optimal AC frequency for various amine substrates, circumventing the time-consuming trial-and-error methods for optimizing reaction conditions. This new development in AC electrolysis provides an alternative strategy to solving challenging chemoselectivity problems in synthetic organic chemistry.

Graphical Abstract

INTRODUCTION

Controlling the degree of oxidation or reduction of substrates is critical to achieving the desired chemical transformation in organic synthesis. In the past decade, electroorganic synthesis has attracted significant attention because it allows for, in principle, straightforward and accurate control over the redox transformations of a substrate by selecting the proper electrode potentials.^1–5 Under typical direct electrolysis conditions, multielectron oxidation or reduction of substrates often takes place. For example, amine substrate I is prone to be oxidized by two electrons to N-acyl-iminium cation species II, which can be later trapped by nucleophiles such as MeOH and H₂O (Figure 1A). Mediated electrolysis is a common strategy to limit multielectron oxidation or reduction in organic electrosynthesis.^6–10 In mediated electrolysis, the electron transfer step is shifted to a homogeneous process that involves an electrochemically generated reagent that serves as a so-called “mediator”. The mediator engages in a reversible single-electron redox cycle initiated at the electrode, followed by single-electron transfer (SET) with the substrate (Figure 1B). This strategy provides access to the reactive intermediates formed by the SET events with substrates. For example, Ye and co-workers recently demonstrated using 2,2,6,6-tetramethylpi-peridinooxy (TEMPO) as a mediator to perform one-electron oxidation of I to III, providing access to amino radicals after deprotonation of III for radical–radical cross-coupling reaction (Figure 1B).⁷

Figure 1.

Existing and new strategies for controlling the degree of oxidation of amine substrates during electroorganic synthesis. (A) Direct electrolysis often leads to the two-electron oxidation of amine I to iminium cation II that can be captured by a nucleophile (Nu). PG stands for protecting group. (B) Mediated electrolysis enables the one-electron oxidation of I to cation radical III, providing access to amino radical IV after the deprotonation of III for radical–radical cross-coupling reaction. SET stands for single electron transfer. [Med_ox] and [Med_red] are the oxidized and reduced forms of the redox mediator, respectively. (C) Alternating current (AC) electrolysis controls one- or two-electron oxidation of amines by managing the redox environment during the deprotonation of III to IV.

More recently, AC electrolysis, where the flow of charge periodically changes direction, started to gain interest in the electroorganic synthesis community.^11–19 One of the experimentally observed properties of AC electrolysis is its ability to reduce overoxidation/reduction products without using any mediators relative to its direct current (DC) counterpart.¹⁶ For example, Hilt et al.¹³ demonstrated that AC electrolysis solved the problem of the overoxidation of disulfides to oxo species and over-reduction of disulfides to unidentified black precipitate in synthesizing unsymmetrical disulfides by an electrochemical sulfur–sulfur bond metathesis reaction. More recently, Baran et al.²⁰ presented chemoselective reduction of phthalimides to partially reduced hemiaminal or fully reduced lactam by applying different AC waveforms, which cannot be achieved by DC electrolysis, either. Despite these exciting findings, the mechanisms behind the controlled oxidation or reduction of substrates during AC electrolysis remain unclear. Due to the lack of mechanistic understanding, one must go through a time-consuming reiterative trial-and-error process to discover the optimal AC electrolysis conditions.

Here, we report a mechanism by which AC electrolysis controls the one- or two-electron oxidation of amines to amino radicals or iminium cations, enabling selective α-amine functionalization by simply adjusting the AC frequency. In our model reaction using tertiary amine and dicyanobenzene as the starting materials, the one- and two-electron oxidation processes of amines compete under paired electrolysis conditions. However, we found that such competition could be controlled by AC electrolysis. The mechanistic study reveals that the degree of amine oxidation is controlled by managing the redox environment where the amino radical cation III is deprotonated to amino radical IV, utilizing the alternating redox environment of AC electrolysis. When the deprotonation of III occurs in an oxidizing environment, IV is further oxidized to iminium cation II for nucleophilic addition; otherwise, IV remains available for radical-based reactions (Figure 1C). Such delicate reaction control requires precise synchronization of the dynamic redox environment and the deprotonation reaction kinetics at the “correct” AC frequency. However, because the deprotonation kinetics of substrates varies significantly, each substrate requires individual reaction optimization, making the empirical trial-and-error reaction optimization approach impractically time-consuming (several days per substrate). Therefore, we developed a convenient electroanalytical method to identify the optimal AC frequency in a few minutes to address it.

RESULTS AND DISCUSSION

In this study, our model reaction is the α-amino C–H arylation reaction, which was first discovered by MacMillan and co-workers using the strategy of accelerated serendipity. This reaction was initially accomplished by photoredox catalysis.²¹ We selected this model reaction for two reasons. First, it can be completed by electrochemically oxidizing amine 1 to amine radical cation 3 and reducing 1,4-dicyanobenzene 2 to the corresponding arene radical anion 6 at electrodes (rather than via SET with a photoredox catalyst). The C–H bonds adjacent to the nitrogen atom in 3 are weakened and undergo deprotonation by a base, B⁻ such as NaOAc, to give α-amino radical 4. A radical–radical coupling reaction between intermediates 4 and 6 yields the aromatized benzylic amine product 1a after the elimination of CN⁻ (one-electron oxidation (arylation) pathway in Figure 2). Second, unlikely photoredox catalysis, electrosynthesis produces a significant amount of cyanation byproduct 1b because of the overoxidation of 1 to iminium cation 5 (the two-electron oxidation (cyanation) pathway in Figure 2), enabling us to study how AC electrolysis alters the degree of amine oxidation and controls product selectivity.

Figure 2.

One- and two-electron oxidation pathways during electrochemical α-amino C–H functionalization reactions. One-electron oxidation pathway leads to the arylation product 1a, while the two-electron oxidation pathway results in the cyanation product 1b.

First, we carried out the model reaction using a DC electrolysis setup where 1 was oxidized at a carbon anode and 2 was reduced at a carbon cathode to initiate the reaction (Figure S1). Figure 3A shows the isolated yields of 1a and 1b at an applied voltage from 2 to 4 V. A mixture of arylation and cyanation products were observed in all cases. At 2.5 V, the best overall conversion of 70% was attained with 30% arylation and 40% cyanation products (1a/1b = 0.7). After lowering the voltage to 2.0 V, the product selectivity did not show any obvious changes (1a/1b = 0.8), suggesting that the lowered voltage does not favor the one-electron oxidation pathway over the two-electron one. At high voltages of 3.0 and 4.0 V, the product selectivity slightly shifted to the two-electron oxidation product 1b (1a/1b = ~0.4–0.5), and the overall conversion dropped to ~50% due to further oxidation of 1 by multiple electrons.

Figure 3.

(A) Isolated yields of arylation and cyanation products (blue, 1a; red, 1b) using DC electrolysis at different cell voltages. (B) Cyclic voltammograms of 1 in the presence (black) and absence (red) of the base, NaOAc, in N,N-dimethylacetamide (DMA) containing 0.1 M LiClO₄. Scan rate: 1 V/s.

The inevitable two-electron oxidation of amine 1 to iminium 5 and the consequent formation of cyanation product 1b during DC electrolysis can be explained by the cyclic voltammograms (CVs) of 1 collected in the absence and presence of the base, NaOAc, in Figure 3B. Without NaOAc, the CV of 1 (red curve) clearly shows two well-separated anodic peaks. The peak at E₁ = ~0.3 V vs Ag/Ag⁺ belongs to the reversible single-electron oxidation of amine 1 to its corresponding cation radical 3, and the other one at E₂ = ~0.9 V arises from the further oxidation of 3 to 5 by another electron. However, when NaOAc was added, the anodic peak at 0.9 V shifted negatively and merged with the 0.3 V peak (black curve), indicating that one- and two-electron oxidation of 1 can take place under similar potentials. The drastically lowered potential for the second-electron oxidation of 1 in the presence of NaOAc is because iminium cations are generated via the low-potential α-amine radical route (E₃ < E₂, Figure 2).²² As a result, the arylation product 1a is always accompanied by the cyanation product 1b under DC electrolysis, regardless of the applied voltage.

Next, we tested AC electrolysis for its effect on product selectivity. During AC electrolysis, the oxidation of 1 and the reduction of 2 occurred sequentially at the same electrode driven by the alternating voltage polarity. Other reaction conditions, including electrodes, electrolytes, and chemical reagents, were identical to the DC electrolysis conditions (Figure S2). We applied sine waveforms with a frequency from 1 to 30 Hz at a fixed AC peak amplitude of 2.5 V. This amplitude was calculated from the potential difference between the oxidation peak of 1 (0.28 V) and the reduction peak of 2 (−2.22 V) in the CVs after iR drop correction (Figures S4–S6). Figure 4A shows the product yields and selectivity for DC and AC electrolysis. All AC experiments exhibited a higher selectivity toward the one-electron oxidation pathway (arylation) product 1a than DC electrolysis, meaning a reduced degree of amine oxidation. At 10 Hz, we observed the maximum yield of 68% for 1a and the biggest 6-fold selectivity improvement toward 1a over DC electrolysis (1a/1b = 4.3 vs 0.7). Under an optimal frequency of 10 Hz, the faradaic efficiency for synthesizing 1a is 4.4% (see the Supporting Information (SI)).

Figure 4.

(A) Isolated yields and selectivity of arylation and cyanation products (1a and 1b) at different AC frequencies from 1 to 30 Hz. (B) Cyclic voltammograms of 1 in the presence and absence of NaOAc in DMA containing 0.1 M LiClO₄ at different scan rates and their equivalent AC frequencies. (C) Proposed mechanism for the reduced degree of amine oxidation during AC electrolysis.

To understand the AC frequency-dependent product selectivity, we conducted the following experiments to examine the time-resolved amine oxidation behavior. The CVs of 1 were acquired in the presence (black) and absence (red) of NaOAc at different scan rates from 0.02 to 20 V/s (Figure 4B). As previously discussed, 1 undergoes a reversible one-electron redox cycle in the absence of NaOAc, so its anodic peak current (i₀) is an ideal reference of a one-electron oxidation process. After introducing NaOAc to the reaction, the two-electron oxidation of 1 becomes possible at low potentials (<0.6 V). If two-electron oxidation does happen, the anodic peak current (i₁) would be larger than i₀; otherwise, i₁ = i₀. By varying the scan rate, we can control the oxidation time (t_ox): the higher scan rate, the shorter t_ox. In this study, t_ox is defined as the time to scan the electrode potential from the onset potential where the oxidation of 1 starts to the anodic peak potential. For example, at a scan rate of 0.02 V/s, t_ox is ~8 s and i₁ is approximately twice of i₀ (Figure 4B), suggesting that two-electron oxidation of 1 is dominant for t_ox = 8 s. Interestingly, the difference between i₁ and i₀ decreases with increasing scan rate, suggesting that the two-electron oxidation of 1 is suppressed as t_ox is shortened. At 5 V/s or t_ox = ~45 ms, i₁ became equal to i₀ and part of the CV curve overlapped with the reference CV (red curve), meaning that only one-electron transfer occurred during the 45 ms long anodic potential scan. Recall that the second-electron oxidation of 1 at low potentials is achievable only after deprotonation of 3 to 4. Therefore, when t_ox is short, the second-electron oxidation is hindered possibly due to the slow deprotonation of 3, resulting in such t_ox-dependent oxidation behavior.

On the basis of the CV results above, we propose the following mechanism for the product selectivity change during AC electrolysis: due to the slow deprotonation of 3, a portion of 3 formed in the positive half-cycle of AC waveform cannot be immediately deprotonated and thus stays intact until the subsequent negative half-cycle. In the reducing environment of the negative half-cycle, deprotonation continues but the further oxidation of the deprotonated product 4 to 5 is prohibited, thereby shifting the product selectivity toward 1a (Figure 4C). According to the CV data, 3 can be hardly deprotonated to 4 and then oxidized to 5 within ~45 ms. Thus, the AC frequency that provides t_ox of 45 ms (i.e., f_pred = 11 Hz, see Figure S7) should result in the highest selectivity toward 1a. The f_pred of 11 Hz from the CV measurements is in excellent agreement with the experimentally observed optimal frequency (f_exp) of 10 Hz. At even higher frequencies such as 20 or 30 Hz, the deprotonation of 3 might not finish in one AC period and thus is carried over to the following AC period, resulting in deprotonation in an oxidizing environment again and an increased yield for two-electron oxidation product 1b.

The intriguing agreement between the CV-derived frequency (f_pred) and f_exp led us to think if we can predict f_exp for other amine substrates from their CVs. It is critical to be able to predict f_exp because it takes several days to determine the optimal frequency for one substrate using the trial-and-error method as every reaction requires a separate power supply and takes more than 24 h to complete. To test this idea, we collected the CVs of a variety of amines, including N-aryl pyrrolidines bearing electron-donating and withdrawing groups (7–18), N-aryl piperidine (19), N-aryl tetrahydroisoquinoline (20), and aliphatic amine (32). These amines showed similar scan-rate dependent CVs as 1 (Figures S16–S29 and S41): the ratio between i₁ and i₀ (|i₁/i₀|) generally decreases with increasing scan rate, suggesting the slow deprotonation of amine cation radicals by NaOAc is true for all tertiary amines. We determined f_pred from the scan rate that produced |i₁/i₀| that is most close to one (Figures S7 and S9C).

In parallel, we experimentally identified f_exp via conventional reaction optimization (i.e., running reactions at different AC frequencies and determining the selectivity from crude NMR, Figures S9–S11). Figure 5A shows the excellent agreement between f_exp and f_pred for various amines, confirming the predictive power of the simple electrochemical descriptor (i.e., |i₁/i₀|) in finding the optimal frequency for selective arylation over cyanation. We also attempted to understand the dispersed f_exp values (by 2 orders of magnitude) for different amines from the calculated pK_a values of their corresponding cation radicals (Figures S12 and S13, Table S1). However, we did not find any obvious correlation between f_exp and pK_a (Figure S13), suggesting that the deprotonation rate is not solely determined by the acidity of these cation radical intermediates. We further varied the cyanoaromatic coupling partners, including 1,2-dicyanobenzene (35), ethyl 4-cyanobenzoate (36), 4,4′-biphenylcarbinitrile (37), and 2,5-dicyanotoluene (38) and bases, including LiOAc, CsOAc, NaOMe, NaOH, Na₂CO₃, and NaHCO₃ to examine their effects on f_exp. Cyanoaromatics 35–38 show similar reversible redox behaviors with a reduction potential of ~−2 V vs Ag/Ag⁺ as 2 (Figure S44) and have little impact on f_exp because they are not involved in the deprotonation of 3. In contrast, Figure 5B shows that base strongly influences f_exp and the CV-based method accurately predicts f_exp (Figures S45–S50), further supporting our proposed mechanism that the deprotonation step plays an essential role in controlling the degree of amine oxidation and product selectivity during AC electrolysis.

Figure 5.

(A) Plot of experimentally determined optimal frequency (f_exp) vs predicted optimal frequency (f_pred) for various tertiary amines and cyanoaromatics. (B) Plot of f_exp vs f_pred for 1 with different bases.

To confirm the one-electron oxidation of amines to α-amine radicals at the optimal frequency, we performed the radical trapping experiment on amine 1 and 20. Five equiv of TEMPO was added to the reaction mixture under the optimal AC electrolysis conditions. We did not observe the formation of arylation product 1a or 20a. However, the formation of the TEMPO-adduct of 1 and 20 (Figures S14 and S15) indicates that this reaction proceeds via α-amine radical pathway.

Finally, we evaluated the substrate scope and compared the product selectivity between AC and DC electrolysis in Figure 6. For all substrates, we performed AC electrolysis under f_exp and DC electrolysis. For N-aryl pyrrolidine bearing electron-donating groups such as methyl, tertiary butyl, 2,6-dimethyl, indane, dioxole, ethyl acetate, and phenyl, the arylation product yields were significantly improved from 5 to 30% under DC electrolysis to 43–68% under AC electrolysis (1a, 7a-14a) while the cyanation product yields dropped from 11 to 40 and 0–38% (1b, 7b–14b). The product selectivity shifted from the cyanation products under DC electrolysis to the arylation products under AC electrolysis. The N-aryl pyrrolidine substituted with electron-withdrawing groups such as F, Cl, and Br also afforded one-electron transfer products (15a–17a) with greater selectivity and yields under AC electrolysis than the DC counterpart. In the case of N-aryl pyrrolidine without any substituent on ring (18) afforded poor yield (<20%) and selectivity under both electrolysis conditions (18a and 18b). The reaction of six-membered cyclic amine, N-aryl piperidine 19, with 2 under both AC and DC electrolysis conditions provided poor yield for 19a and 19b, possibly because the impaired overlap between the lone pair of amine and the unpaired electron at the α-carbon significantly destabilizes the α-amino radical intermediate.⁷

Figure 6.

Substrate scope and product selectivity comparison between AC and DC electrolysis. Note: 18b and 19b were characterized using high-resolution mass spectrometry. Yields of 30a and 30b were determined using ¹H NMR of the purified inseparable compounds. Regioisomers of 38a were formed in 1:0.6 ratio. The experimental frequency for each substrate is provided in parentheses. All yields were calculated considering 0.5 mmol of 1,2-dicyanobenzene as the limiting reagent.

Encouraged by the broad generality of our approach on five-membered cyclic amines, we tested it on another class of tertiary amines, N-aryl tetrahydroisoquinoline. Tetrahydroisoquinoline alkaloids are an important class of bioactive natural products and display a myriad of biological activities.^23,24 N-aryl ring bearing electron-donating and electron-withdrawing groups such as methoxy, fluoro, dioxine, dimethyl, trifluoromethyl (20–26), and N-benzyl (27) all afforded arylated and cyanation products (20a–27a and 20b–27b). Under DC electrolysis, the formation of cyanation product was predominantly observed in good yield 36–63% because the benzylic position is prone to overoxidation to form iminium cation.^25–27 Under AC electrolysis, arylation product formation significantly improved to good/moderate yields of 26–51%. However, the cyanation product yields were comparable to the DC electrolysis results, even though the CV data in Figures S29–S36 predicts that the two-electron oxidation of N-aryl tetrahydroisoquinolines should be largely reduced at f_exp, like the N-aryl pyrrolidine oxidation. A possible explanation for the discrepancy is that there are alternative pathways to generate iminium cations that does not require NaOAc, for example, via deprotonation by in situ generated CN⁻²⁷ or dicyanobenzene 2.²⁸ This explanation is in part supported by the formation of arylation and cyanation products at moderate yields for tetrahydroisoquinoline 20 even in the absence of NaOAc (Table S2).

We have also identified the tertiary amine-containing heteroaromatic rings as a suitable class of substrates (28–31). Various N-benzyl/alkyl thienopyridines containing susceptible functional groups such as ester and allyl were tested. The reaction of 28 with benzonitrile 2 under DC electrolysis provided poor yield and selectivity for arylation product 28a relative to cyanation product 28b. In contrast, the yield of 28a improved from 17 to 54% under AC conditions. Notably, the arylation occurred regioselectively at cyclic amine and the benzylic methylene group remained intact. In the presence of ester (29), DC electrolysis did not yield any products. This situation was altered when AC was applied, delivering both products at a yield of 31% for 29a and 37% for 29b. In the case of N-allyl group, significant enhancement in yield and selectivity was not observed (30a and 30b). For late-stage functionalization of antiplatelet drug Ticlopidine 31, the arylation and cyanation product yields (31a and 31b) slightly increased under AC electrolysis.

The tertiary aliphatic amines were also investigated. DC electrolysis did not furnish any arylation or cyanation products. In contrast, one-electron oxidation arylation product exclusively formed under AC conditions in moderate yields 28–39% (32a–34a). Dealkylation of aliphatic amines under electrooxidative conditions is the possible reason for their moderate yields.^29,30 Tertiary amines containing strong electron-withdrawing functional groups such as ester and nitro (39–40), pyrimidinyl (41), and benzenesulfonyl (42) did not result in any products (Figure 6). The unavailability of nitrogen lone pair electrons due to strong electron-withdrawing effect makes it difficult to oxidize these amines to stable radical cation intermediates.

Moreover, the substrate scope for aromatic partner 2 was also explored. Cyanoaromatics 35–38 successfully underwent coupling with 1 to furnish arylated products in moderate to good yields (35a–38a). In contrast, the product involving two-electron transfer was predominantly obtained in most cases in moderate to low yields (1b) under DC electrolysis. The arylated product 38a was a 1:0.6 mixture of two isomers due to a lack of regioselectivity in elimination of CN⁻ group in reaction. The coupling of 1 with electron-deficient hetero-aromatics such as 4-cyanopyridine (43), 2-chlorobenzthiozole (44), and 2-chlorobenzoxazole (45) were not successful using AC or DC electrolysis (Figure 6), possibly because of the poor stability of their anion radicals. Compared with the TEMPO-mediated electrolysis method developed by Ye and co-workers,⁷ our AC electrolysis method exhibits similar substrate scope and yields, which is not surprising because both methods rely on limiting the oxidation level of amines to generate the critical intermediate of α-amine radicals but via different approaches.

CONCLUSIONS

In conclusion, we have reported a mechanism by which AC electrolysis controls the one- or two-electron oxidation of amines to amino radicals or iminium cations, enabling selective α-amine functionalization by adjusting the AC frequency. We found that the selective one-electron oxidation of amines to amino radicals was achieved by providing a reducing environment for the deprotonation of amino radical cations. We have also developed a convenient electroanalytical method to identify the optimal AC frequency that maximizes the arylation product of the one-electron oxidation pathway, eliminating the time-consuming trial-and-error approach in conventional AC electrolysis condition optimization. We envision that this highly efficient and unique AC electrolysis protocol can be applied to achieve unique reactivities in synthetic and medicinal chemistry.