Lepidine and cyclohexane were selected as model substrates. Using TFA (2 equiv), Bu4NPF6 as supporting electrolyte (SE, 1 equiv) in a mixture of acetonitrile and 1,1,1,3,3,3-hexafluoroisopropanol (HFIP, 8:2, 0.03 M) the reaction was run with a Glassy Carbon Electrode (GCE) as anode and Nickel as cathode, at 5 mA (total charge of 3 F/mol) at room temperature. Under these conditions, the desired product was obtained in low yield (Table 1, entry 1). Following our previous work, we decided to test different pyridine N-oxides as HAT (Table S1), obtaining a higher yield in the presence of 30 mol% of 2,4,6-collidine N-oxide (HAT3), but still insufficient (entry 2). Different electrodes were tested as anode and cathode (Table S2), which provided poorer yields. We decided to explore heterogeneous catalysis by anchoring 2,6-dimethylpyridine on the GCE. The oxidation of the anchored pyridine was conducted by submerging the electrode in an MCPBA solution overnight, mimicking the structure of HAT3. The reaction using this GCE-HAT3 anchored electrode provided a slightly better yield (entry 3). Surprisingly, the control reaction performing the overnight MCPBA-mediated oxidation of the bare GCE in the absence of the anchored pyridine-N-oxide gave a significantly higher reaction yield, obtaining 80% of the expected product (entry 4) and complete conversion by increasing the total charge (entry 5), suggesting that the oxidated area of the GCE is a better HAT mediator than HAT3. We tested a more environmentally friendly oxidation of the GCE using a phosphate buffer (NaH2PO4 and Na2HPO4 (pH 7, 0.1 M)).23 The number of scans needed to activate the electrode was further optimized (Table S4). Using the optimal activation method, the modified electrode (AGCE) yielded 85% of product 1 under standard reaction conditions, which rose to 94% when the amount of cyclohexane was increased (entry 7). Importantly, the control reaction without current gave complete recovery of the lepidine (entry 8).
Table 1. Optimization of the model reaction
Reactions were performed on a scale of 0.25 mmol of lepidine. aGC yield based on remaining SM without calibration. b HAT3: 2,4,6-collidine N-oxide. cGCE-HAT3: glassy carbon electrode anchored with 2,6-lutidine. d Electrode activated with MCPBA overnight. eAGCE: activated glassy carbon electrode using phosphate buffer.
Substrate scope
Having found the optimal conditions, we approached the substrate scope by examining different azaarenes in the reaction with cyclohexane as the C(sp3)-H coupling partner (Fig. 2). Quinolines with substituents placed at C4 reacted better than the C2 substituted, obtaining C2 or C4 cyclohexyl derivatives (products 1–5) from moderate to good yields (36–82%). Unsubstituted quinoline and 3methylquinoline were converted into inseparable mixtures of C2- and di-alkylated products at C2/C4 (6) and C4- and C2- alkylated products at C2/C4 (7), respectively. Halides groups (-Cl,-Br) in different positions of quinolines gave the desired products (8–11) in moderate yields. We also explored the reaction with a nitro group as an electron-withdrawing group, obtaining the corresponding product (12) in good yield. Other azaarenes, such as 2,2'-biquinoline and phenanthridine, were suitable substrates for the reaction, giving moderate yields of the corresponding alkylated products (13, 14). Interestingly, our protocol was applied to functionalize the natural cinchonine, giving 15 in a 35% yield. Different C-H partners were also evaluated. Cycloalkanes reacted to provide the desired products (16–19) from moderate to good yields. The bridged alkane norbornane was successfully converted into the desired exo product 20 with excellent endo/exo selectivity. Interestingly, dioxane reacted smoothly, giving the desired product 21 in 38% yield, and the reaction using pivalaldehyde provided the alkylated product (22) after decarbonylation. Finally, adamantane and other derivatives were also tested, giving the desired products (23–25) in moderate yields. Interestingly, in all three cases, the stronger C-H bond of the adamantane (BDE(C1-H) 99 kcal/mol vs. BDE(C2-H) 96 kcal/mol)29 was alkylated preferentially. This results from an enhanced polar effect at C1-H due to the positively charged surface of the anode. Presumably, there is an increased charge-transfer character in the transition state since a positive charge is developed at the C of the adamantane through the HAT process, a charge that is more stable at the tertiary C1 than at the secondary C2. The same phenomena can explain the high selectivity observed in product 25, where the weaker secondary and tertiary Csp3-H bonds of the side chain were untouched.
While performing the scope of the Minsici reaction, we observed by GC-MS the formation of a lighter product in large amounts in some cases. For instance, when adamantane was used, apart from the formation of the Minisci product, a new peak with high intensity and with m/q of 193 was detected. It is worth noting that an excess of adamantane was needed for the Minisci reaction, so this was not affected by the formation of the byproduct. After carefully analyzing the fragmentation pattern, we realized it matched N-(adamant-1-yl)acetamide. Indeed, when we performed the reaction without lepidine using adamantane as a limiting reagent, complete conversion to the N-(adamant-1-yl)acetamide 26 was observed. Interestingly, high selectivity was again found towards C1 due to the enhanced polar effect previously commented on. This Ritter-type reaction demonstrated that reactions that proceed through a different mechanistic pathway were also possible using the AGCE. In literature, Ritter-type reactions have been performed under electrochemical conditions.20,30,31 However, to our knowledge, this is the first one functionalizing an unactivated C(sp3)-H bond directly using the electrode and without a HAT mediator. Thus, we became interested in this new opportunity and tried to reoptimize the reaction conditions for this reaction (Table S3). After optimization, using 2.5 F/mol at 10 mA provided full conversion in shorter reaction times, and using a simpler SE such as LiClO4 facilitated the isolation of the different products.
Having found the optimal conditions, we approached the substrate scope by first examining different adamantanes with acetonitrile as a nitrile source (Fig. 3). Adamantane and 1,3-dimethyladamantane gave good yields (products 26 and 27, 84% and 67%, respectively). We also explored the reaction with the isopentyl-adamantane-1-carboxylate, obtaining the corresponding product (28) in good yield (76%). Interestingly, this reaction was again completely selective to the C1 of the adamantane, even in the presence of weaker C-H bonds present at the open side chain, such as the secondary methylene in αposition to the oxygen atom or the tertiary methine. Furthermore, our protocol was successfully applied when using adamantane in the presence of isobutyronitrile as a nitrile source, giving the desired product (29) in moderate yield. Cycloheptane and cyclooctane were also suitable substrates for the reaction in the presence of acetonitrile, giving products 30 and 31 in moderate yields (57% and 60%). Finally, indane was tested, providing the desired acetamide 32 in 44% yield (Fig. 3).
AGCE analysis
To better understand the mechanism allowing for this alkane activation in the absence of any HAT mediator, a series of analyses were performed on the surface of the AGCE, and these results were compared with the surface of polished GCE. The activation of the GCE to obtain AGCE was monitored by CV, observing the appearance of a new oxidation and reduction peak couple, which is attributed to the generation of quinone/hydroquinone structure on the surface (Figure S2). A wettability test was also conducted, adding 10 µL of water to the GCE and the AGCE (Fig. 4A) and observing that the water slips easier on the AGCE than on the GCE, indicating the presence of a more hydrophilic surface on the AGCE. IR analysis (Figure S4) revealed a broad signal at > 3000 cm− 1, suggesting the presence of hydroxyl groups on the surface of the AGCE. Furthermore, XPS analysis of the surface of the AGCE revealed a higher content of O 1s on its surface compared to the GCE surface, where the only significant peak belongs to the C 1s (Figs. 4B and S5). Indeed, analyzing the distribution of the C 1s in the AGCE, a substantial increase of the C = O and C-O groups is detected compared to the GCE, combined with a decrease in the C-C and C = C groups (Figs. 4C, 4D, and S6). Raman spectroscopy performed on both surfaces (Fig. 4E) showed a calculated I(D/G) ratio of 1.39 and 1.36 for the GCE and the AGCE, respectively, which indicates the dominance of sp2-bonded carbons (see Figure S7 for further details). Finally, SEM images and EDX analysis were performed (Figures S8-S10). The GCE showed a 100% composition of C atoms (Figs. 4FG), while the AGCE showed an average of 28% of O atoms on the surface (Figs. 4H-I). All these results indicate that the buffer treatment can generate oxygenated functional groups on the surface of the GCE, activating the surface toward possible HAT processes.
Mechanistic insights and reaction mechanism proposal
Mechanistic studies were conducted to gain a better understanding of the mechanistic pathways. To confirm the formation of the alkyl radical during the reaction, β-pinene was added to the reaction of lepidine with cyclohexane, observing the formation of the β-pinene-cyclohexane adduct and confirming the formation of the cyclohexyl radical (Figs. 5A and S13). The addition of TEMPO or 1,1-diphenylethylene (radical scavengers) gave no formation of the corresponding Minisci product, as well as the addition of CuCl2, a well-known single electron scavenger (Figs. 5A, S11 and S2).
The electrode analysis and the mechanistic studies allowed us to draw plausible mechanisms for these reactions (Fig. 5B). The formation of an oxygen-centered radical on the surface of the AGCE after electrochemical oxidation promotes the HAT from the alkane, producing the corresponding alkyl radical. In the Minisci reaction, this alkyl radical is trapped by the protonated azaarene which, after deprotonation, gives the corresponding radical intermediate that suffers a second oxidation on the surface of the electrode, giving the corresponding product after a second deprotonation. On the other hand, the alkyl radical can also be further oxidized on the surface of the AGCE, forming the corresponding alkyl cation. A nucleophilic media, such as MeCN, promotes the Ritter amidation, furnishing the corresponding acetamides after hydration. It is worth noting that for the Minisci reaction, a smaller current intensity was necessary to avoid overoxidation of the formed radical. In contrast, a higher intensity is allowed to perform the Ritter reaction since the alkyl radical must be further oxidized on the surface of the electrode.
To further demonstrate the utility of this procedure, a scale-up of the Minisci-type reaction was performed at 1.5 mmol-scale, giving an even better yield than when 0.25 mmol-scale was used (68% vs 48%, Figs. 6A and S14). Moreover, since this is an electrocatalytic procedure, we tested both the Minisci and the Ritter-type reactions using a series of 1.5 V batteries (3 X 1.5 V were measured to have a voltage of ~ 4.0 V; see Figures S15-16 for further details), obtaining comparable results to the yields achieved using controlled current intensity. This experiment highlights the robustness and simplicity of this reaction, which can be performed using a regular battery.