OCPs are measured without passing current or applying an external voltage, and we initiated our study by measuring OCPs for zinc and manganese relative to a reference electrode (Ag/AgNO₃) under a variety of conditions relevant to Ni-catalyzed XEC reactions (Fig. 2A). A three-electrode set-up was used to enable determination of the Fc/Fc⁺ redox potential (used for referencing the OCP) in the same solution.^37,43 Eight solvents commonly used in Ni-catalyzed XEC reactions were selected for OCP measurement to probe the influence of solvent on the Zn and Mn reduction potentials (Fig. 2B): dimethylformamide (DMF), dimethylacetamide (DMA), N-methyl-2-pyrrolidone (NMP), N,N′-dimethylpropyleneurea (DMPU), 1,3-dimethyl-2-imidazolidinone (DMI), dimethyl sulfoxide (DMSO), acetonitrile (MeCN), and tetrahydrofuran (THF). Zn²⁺ or Mn²⁺ salts were included in the solution (10 mM) to ensure well-defined thermodynamic conditions for the redox process of interest (i.e., M ⇌ M²⁺ + 2e⁻). The OCPs measured in the different solvents were converted to formal thermodynamic potentials versus Fc/Fc⁺ (E°’) by correcting for the non-standard-state concentration of the M²⁺ ions (see Section 2 of SI for details). The redox potential of [Ni(bpy)₃]Cl₂ was also measured in each of these solvents to provide a benchmark for Ni-based redox potentials (black data, Fig. 2B).

The Zn and Mn reduction potentials show modest variation in a series of amide and urea solvents (DMF, DMA, NMP, DMPU, and DMI): ΔE°’ ≤ 50 mV for Zn, ΔE°’ ≤ 80 mV for Mn. The Mn potentials in DMSO and THF are similar to the corresponding potentials observed in amide solvents. The Zn potential is more negative in DMSO and more positive in THF, relative to the corresponding potentials in amide solvents. The Zn and Mn potentials change substantially in MeCN, increasing by up to 260–370 mV relative to those measured in DMSO and amide solvents. The redox potentials of Zn and Mn are more positive than those of [Ni(bpy)₃]Cl₂, although the difference is rather small for Mn in DMSO and amide solvents (ΔE ≤ 90 mV). A somewhat larger difference is evident for Zn in these solvents (ΔE ≤ 270 mV).

The unique behavior of MeCN has important implications for Ni-catalyzed XEC reactions. The high reduction potentials observed for Zn and Mn in MeCN likely explain why XEC reactions are rarely conducted with these reductants in MeCN.^17,24 For example, Weix and coworkers evaluated a series of solvents, including MeCN, for the cross-coupling of benzyl chloride with phenyl iodide. Efficient cross-coupling was observed in DMA with Zn as the reductant, affording 82% yield of diphenylmethane, while only 15% yield was observed in MeCN (Fig. 2C).²⁴ This result could be rationalized by unfavorable thermodynamics for Zn reduction of the Ni catalyst in MeCN. Support for this hypothesis was obtained by conducting constant-potential electrolysis experiments in DMA and MeCN, applying potentials associated with the reduction potential of Zn in each solvent. At an applied potential of −1.07 V, the Zn reduction potential in MeCN, negligible current was observed in DMA with >90% unreacted starting materials remaining after 36 h, while a 23% yield of product was observed in MeCN. At the more reducing potential of −1.35 V, the Zn reduction potential in DMA, sustained current and moderate-to-good product yields were observed in both solvents (50% in DMA, 78% in MeCN; see full screening data in Supplementary Table 4). These observations show how solvent effects on the reduction potential of metallic reductants can have a significant influence on the outcome of reductive coupling reactions.

DMF is one of the most common solvents used for XEC reactions. Additives, such as LiCl, LiBr, among others, are often included in the reaction mixture.⁸ OCP measurements for Zn and Mn in the presence of various additives (200 mM) in DMF show that the Zn and Mn reduction potentials can change by nearly 300 mV (Fig. 2D), depending on the additive identity. Halide ions, especially chloride salts, shift the Zn and Mn reduction potentials to more negative values relative to those recorded in the presence of weakly coordinating anions. The influence of halide ions was further probed in MeCN and THF, and a similar phenomenon was observed. One exception is a shift in the Mn potential to a slightly more positive value in the presence of LiI relative to that measured in the presence of ⁿBu₄NPF₆. As the quantity of these additives is commonly screened in the development of XEC reactions,^44,45 OCP measurements were performed with different concentrations of chloride salts. The results show that Zn becomes more reducing at higher chloride concentration (ΔE = 360 mV from [Cl⁻] = 0 – 1.2 M; see Supplementary Fig. 3). These observations show that halide ions could have a significant influence on XEC reactions by changing the potential of the metal reductant. For example, the significantly lower reduction potentials of Zn and Mn in the presence of chloride salts in MeCN and THF could provide new opportunities to develop XEC reactions in these non-amide solvents. In another aspect, LiCl has been reported to accelerate Ni-catalyzed XEC reactions, and this behavior was attributed to the kinetic influence of chloride and/or lithium ions on Ni^II reduction by Zn metal.⁴⁶ The results in Fig. 2D, however, show that Zn is a stronger reductant in the presence of LiCl, highlighting an important, but previously unrecognized, thermodynamic contribution to this reaction. It is worth noting that the methodology outlined herein can be readily extended to the study of other conditions, such as those using mixed solvents (e.g., DMF/THF) or other additives (e.g., trimethylsilyl chloride) (see Supplementary Tables 1 and 2 for full data).

The preferred reductant in Ni-catalyzed XEC reactions can vary between Zn and Mn. The reduction potentials measured for these metals together with redox potentials of the Ni catalysts used in the reactions provide a foundation for understanding the origin of the preferred reductant. We elected to explore this approach using two reactions reported by Shu and coworkers depicted in Fig. 1B. The cross-coupling of alkenyl acetates and alkyl bromides proceeds effectively with Zn, but not with Mn, as the reductant (Fig. 3A).²⁹ This reaction was proposed to be initiated by reduction of Ni^II to Ni^I, followed by coordination and oxidative addition of the alkenyl acetate to form an alkenyl-Ni^III species. One-electron reduction of this species to give an alkenyl-Ni^II species and reaction with an alkyl radical was then proposed to generate the desired product. The cross-coupling of benzyl alcohols with aryl triflates uses dimethyl oxalate (DMO) to convert the alcohol to an oxalic ester electrophile, and this reaction proceeds effectively with Mn, but not with Zn, as the reductant (Fig. 3B).³⁰ The reaction was proposed to be initiated by reduction of Ni^II to Ni⁰, in this case, followed by reaction of Ni⁰ with the benzyl oxalate to afford a benzyl-Ni^II species. One-electron reduction of this species to give a benzyl-Ni^I and oxidative addition of aryl triflate was then proposed to afford the desired product.

Analysis of these reactions was initiated by recording CVs of the Ni catalysts in their corresponding reaction solvents (DMA, DMF). CVs were obtained in the presence and absence of alkenyl acetate or benzyl oxalate, the substrates proposed to initiate reaction with their respective catalysts (Figs. 3C and 3D; see Supplementary Figs. 9 and 10 for CVs involving both coupling partners). For the first reaction, the CV of the Ni catalyst alone reveals two quasi-reversible redox couples, reflecting sequential one-electron reduction of Ni^II. In the presence of 10 equiv alkenyl acetate, a new reduction peak is evident at −1.29 V, more positive than the original Ni^II/I feature. An increased current is observed for the peak at −1.46 V and, even more significantly, at −1.73 V. The reduction potentials of Zn and Mn, determined from OCP measurements under the same conditions, are −1.31 V (Zn) and −1.55 V (Mn). These data suggest that the preference for Zn as the reductant for this reaction reflects the ability of Zn to reduce Ni^II to Ni^I without significant generation of Ni⁰. The reduction potential of Mn is sufficiently close to Ni^I/0 potential to generate Ni⁰, which was shown to promote alkyl-alkyl and vinyl-vinyl homocoupling in addition to the formation of other side products.²⁹

A similar approach was adopted to analyze the cross-coupling of benzyl alcohol and aryl triflate (Fig. 3D), using a benzyl oxalate directly as the coupling partner to facilitate the CV analysis. This reaction features two different ancillary ligands, dppf or phen (dppf = 1,1′-bis(diphenylphosphino)ferrocene; phen = 1,10-phenanthroline), and the CV trace of the Ni catalyst shows multiple reduction peaks. The first peak at −1.37 V is attributed to one-electron reduction of Ni^II to Ni^I, while assignments of the peaks at lower potentials, −1.56 V and −1.86 V, are complicated by the mixture of ligands present. Addition of benzyl oxalate (20 equiv) to the solution leads to a small increase in current at potentials lower than approximately −1.60 V, relative to the Ni-only CV. The reduction potentials of Zn (−1.36 V) and Mn (−1.55 V), determined by OCP measurements under these conditions, suggest that the preference for Mn as the reductant for this reaction arises from the need to access the lower-potential Ni species in this reaction, unlike the first reaction, which leads to deleterious reactivity at lower potentials. These results, for the first time, show how redox potentials of metal reductants in organic solvents correlate with catalyst redox potentials in XEC reactions.

Metallic Zn and Mn are effective reductants for many reductive cross-coupling reactions, but their individual redox potentials are not necessarily optimal for every reaction, while electrochemical potentials can be tuned continuously over a wide range.⁴⁷ The OCP measurements outlined herein enable the first direct comparison of chemical and electrochemical Ni-catalyzed XEC reactions at reduction potentials of Zn and Mn under the reaction conditions, and these results may be compared with electrochemical reactions conducted at other applied potentials. Reactions with N-alkyl-2,4,6-triphenylpyridinium reagents (Katritzky salts)^32,48–51 represent a compelling target for exploration of these issues because they exhibit variable performance in Ni-catalyzed XEC reactions when using homogenous organic reductants with different reduction potentials.²⁶

Two Katritzky salts, one with a 1° alkyl and one with a 2° alkyl substituent (1 and 2, respectively), were investigated in XEC reactions with six different (hetero)aryl bromides (Fig. 4). Previously reported thermochemical conditions using Mn as the reductant⁴⁹ were used as the starting point for development of electrochemical conditions for the coupling of 1 and 2 with ethyl 4-bromobenzoate (Conditions A and B, Fig. 4; see Supplementary Table 5 for evaluation of different cell configurations, electrode materials, and electrolytes). Good results were observed under constant potential conditions in an undivided cell with a Ni foam cathode and sacrificial Fe anode. Each pair of substrates was then compared under four reaction conditions: two thermochemical conditions using metallic Zn and Mn powder as the reductant, and two electrochemical conditions conducted at applied potentials corresponding to E°’_Zn (−1.31 V) and E°’_Mn (−1.55 V) determined from OCP measurements with Zn and Mn under the reaction conditions. A fifth condition was then used for each substrate pair, based on optimization of the applied potential (see Supplementary Figs. 5 and 6 for optimization details).

The data in Fig. 4 show how thermochemical and electrochemical reduction methods perform at the same potential (E°’_Zn and E°’_Mn). For example, coupling product 1a was obtained in 72% yield with Zn and a 62% yield under electrochemical conditions at the Zn potential (−1.31 V). Significantly lower yields were observed with Mn (27%) and under electrochemical conditions at the Mn potential (0%; −1.55 V). Quantitative differences between electrochemical and thermochemical conditions can arise from various factors, for example, differences in the mass transport behavior of dissolved species interacting with suspended metal particles versus a solid-state electrode, and differences in the redox-active surface area of the metal powders and electrodes, that could alter the outcome. Additionally, the composition of reaction mixtures will evolve as the reaction progresses through the release of metal salts that alter redox equilibria. Thus, the electrochemical reduction at E°’_Zn and E°’_Mn represents only a first-order approximation of the potential supplied by the stoichiometric metal reductant. Nonetheless, the results in Fig. 4 show good qualitative agreement between thermochemical and electrochemical results obtained at the Zn potential of substrates 1a–f and 2f, and at the Mn potential for all substrates. Notable differences were observed for many of the XEC reactions of 2 at the Zn potential: negligible quantities of 2a–e were observed under thermochemical conditions, while the corresponding electrochemical conditions (at E°’_Zn) led to moderate-to-good product yields. Further analysis of this behavior showed that soluble Zn^II salts, generated in the reactions that used metallic Zn, convert the aryl bromide into the proto-dehalogenation byproduct, likely via the formation of protolytically sensitive arylzinc species (see Supplementary Table 6).^45,52

The reactions of each substrate pair were then evaluated at different applied potentials, and the results showed that higher yields could be obtained at potentials other than E°’_Zn or E°’_Mn in all cases. For example, 1a was generated in 81% yield at −1.25 V, a redox potential higher than E°’_Zn and E°’_Mn. A survey of the results shows that the optimal potential changes for each reaction, and even the position of the optimal potentials relative to the Zn and Mn potentials changes. For the six reactions performed with the 1° Katritzky salt 1, only 1a performs best at a potential higher than E°’_Zn. Four products (1b, 1c, 1e, 1f) exhibit an optimal yield at a potential between Zn and Mn, and one (1d) maximizes at a potential more negative than E°’_Mn (Fig. 4). With the 2° Katritzky salt 2, three of the products afford the highest yields at a potential higher than E°’_Zn (2a, 2e, 2f), while three are optimal at potentials between Zn and Mn (2b, 2c, 2d).

The results in Fig. 4 have important implications for Ni-catalyzed XEC reactions. Large-scale applications of these reactions with chemical reductants are already complicated by challenges in using metal-powder reagents (non-uniform particle properties/reactivity, difficulty in suspending dense powders in reactors⁵³) or air-sensitive organic reductants²⁸. The data in Fig. 4 show that the quantized/non-tunable redox potentials of chemical reagents will limit opportunities to optimize the reaction performance. Each of these challenges may be addressed by using electrochemistry to supply the electrons needed in the reaction.

The results outlined above document thermodynamic potentials of Zn and Mn reductants in organic solutions for the first time, revealing the influence of different solvents and additives on the reduction potentials. These data are ideally paired with cyclic voltammetry studies to gain fundamental insights into the relationship between the redox potential of the reductant and critical redox processes in the reaction. The approach used here is readily adapted to other reduction and reductive coupling reactions, including those using different metal reductants, solvents, and reaction conditions. Access to the Zn and Mn reduction potentials also provided the first opportunity to directly compare the influence of chemical versus electrochemical reduction methods on the reaction outcome. Good qualitative agreement is observed from reactions conducted at the same potential (e.g., at E°’_Zn or E°’_Mn), but the optimization studies show that the best performance is often observed at reduction potentials different from the specific potentials accessible from the chemical reductants. These results highlight the importance of tuning the reductant redox potential when optimizing Ni-catalyzed XEC, and presumably other reductive coupling, reactions. The tunability of electrochemistry will offer significant advantages in the future development of these reactions.

In a solution containing ZnBr₂ (11.2 mg, 10 mM), ⁿBu₄NPF₆ (387 mg, 0.2 M), and anhydrous DMF (5 mL), a Zn plate (polished using a smooth file before measurement) was used as the working electrode for OCP measurements, with a Pt wire counter electrode and a Ag/AgNO₃ reference electrode. The solution was rigorously purged with nitrogen and stirred for 15 min before the measurement. While stirring at 600 rpm, the open-circuit potential was measured in a homemade CV cell (Fig. 2A) and recorded after it remained stable (< 5 mV fluctuation over 1 min). The measured potential versus Ag/AgNO₃ was then adjusted to be referenced versus Fc/Fc⁺ and converted to the formal thermodynamic potential (E°’) using the Nernst equation.

In a nitrogen-filled glove box, NiCl₂(dme)/NiBr₂(dme), ligand, and anhydrous solvent (5 mL) were added to a 2-dram vial. This solution was stirred at 800 rpm in the glove box for 30 min to allow complexation of Ni with the ligand. Under air, to an undivided cell fitted with a cross-shaped stir bar was added KPF₆ and the electrophiles. The cell was transferred into the glove box, followed by addition of the catalyst solution and installation of a Ni foam cathode, a Ag/AgNO₃ reference electrode, and an Fe rod anode. The cell was then sealed with a rubber septum before removing from the glove box. A thin Teflon tube was introduced immediately into the cell to allow continuous nitrogen bubbling. The reaction mixture was stirred at 1000 rpm for 10 min to allow full dissolution of all reagents and exclusion of adventitious oxygen. After that, the Teflon tube was removed, and the top of the rubber septum was sealed with silicone gel. The reaction mixture was electrolyzed at a constant potential (−1.07 to −1.65 V vs. Fc/Fc⁺) at the desired temperature (25 to 80 °C) for 10 to 24 h (2.5 to 4.0 F/mol) with stirring (800 rpm).

In a nitrogen-filled glove box, NiCl₂(dme)/NiBr₂(dme), ligand, and anhydrous solvent (1 mL) were added to a 1-dram vial. This solution was stirred at 800 rpm in the glove box for 30 min to allow complexation of Ni with the ligand, followed by sequential addition of the two electrophiles and Zn or Mn powder (2.0 equiv with respect to the limiting substrate). The vial was capped with a screw cap fitted with a PTFE-faced silicone septum, then removed from the glove box and stirred at 1000 rpm in a sand bath at the desired temperature (25 to 80 °C) for 17 h.

All CV experiments were performed using a Pine WaveNow PGstat without IR compensation under an N₂ atmosphere at the desired temperature (25 to 80 °C). The experiments were carried out in a three-electrode cell configuration with a glassy carbon (GC) working electrode, a platinum wire counter electrode, and a Ag/AgNO₃ non-aqueous reference electrode (internal solution, 0.1 M ⁿBu₄NPF₆ and 0.01 M AgNO₃ in DMF). The redox potential of ferrocene/ferrocenium (Fc/Fc⁺) was measured (under same experimental conditions) and used to provide an internal reference. The potential values recorded versus Ag/AgNO₃ were then adjusted relative to Fc/Fc⁺, and electrochemical studies in organic solvents were reported accordingly. The GC working electrode was polished with alumina powder (5 μm) before each experiment. The resultant voltammograms were plotted following IUPAC convention.