C(sp³)–F Adsorption Implicated for Reductive Activation at Ag Electrodes

Abstract

The reductive activation of C­(sp³)–F bonds enables the derivatization of fluorinated motifs with medicinal value as well as the recycling of fluorochemicals. The electrochemical reductive activation of C­(sp³)–Cl, C­(sp³)–Br, and C­(sp³)–I bonds at Ag electrodes is well-documented and commonly attributed to inner-sphere, surface-mediated pathways. In contrast, experimental support for surface-mediated reductive C­(sp³)–F activation at Ag electrodes remains limited. Here, we report evidence for the adsorption of organic fluorides at polarized Ag electrodes. We compare spectroscopic and voltammetric data of nitrile-tagged benzylic bromide, chloride, and fluoride substrate with their non-halogenated congener. Surface-sensitive and in situ spectroscopic data are consistent with their adsorption on polarized Ag surfaces, including the organic fluoride, prior to Faradaic charge transfer. Voltammetric data demonstrate enhanced reductive activation of organic halides at Ag, including the organic fluoride, compared to glassy carbon. Exposing organic fluoride adsorption at electrified surfaces motivates electrode designs that can exploit surface interactions for the development of electroorganic methods beyond outer-sphere modalities.

Introduction

Reductive C­(sp³)–F bond activation enables the derivatization of fluorinated motifs in drug discovery and development, − as well as the defluorination of pervasive and environmentally harmful fluorochemicals. , Among organofluorides, benzylic C­(sp³)–F bonds , exhibit reduction potentials that are accessible to conventional photoredox catalysts ,, and within the potential window of nonaqueous solvents used in electroorganic synthesis. , The ability to affect mesolytic cleavage of the benzylic C­(sp³)–F bond following electron transfer from photoredox catalysts has enabled the development of various reactions, including defluoroalkylation, − enantioselective aryldifluoromethyl-alkynylation of alkenes, hydrodefluorination, ,,,, defluorinative cross-coupling with terminal alkynes, and cyclization. In addition to photoredox methods, reductive dissociation of benzylic C­(sp³)–F bonds can be accessed via heterogeneous electron transfer from electrodes, offering the advantage of broader substrate scope as electrochemistry provides access to a wider range of reduction potentials. ,− Indeed, electrochemical methods have enabled access to defluorinative functionalization of carbonyls, hydrodefluorination, ,− defluorinative silylation, , and defluorinative carboxylation. − These reports demonstrate the synthetic utility of reductive benzylic C­(sp³)–F bond activation in both photoredox and electrochemical platforms. In both activation modalities, the reactivity is typically proposed to proceed through outer-sphere electron transfer, and thus, in electrochemical platforms, the reductively polarized electrode surface is assumed to serve as a source of electrons and otherwise not an active participant in the reaction.

In contrast to the hypothesized benzylic C­(sp³)–F bond activation proceeding via outer-sphere electron transfer, literature consensus implicates an inner-sphere mechanism for C­(sp³)–Cl, C­(sp³)–Br, and C­(sp³)–I bond activation, where substrate interactions with the transition-metal electrode enable the desired mesolytic cleavage. Among the diverse reported electrocatalytic reactivity of transition-metal electrodes, Ag has been shown to reductively activate benzylic C­(sp³)–Cl, C­(sp³)–Br, and C­(sp³)–I bonds − at potential values that are 400 to 800 mV more positive than carbon-based electrodes. This enhanced activation by Ag has been leveraged to enable electrocatalytic dehalogenation, carboxylation, − and more recently, fragment-based electrophile coupling from our group. For benzyl chlorides, the enhanced activation by Ag has been attributed to the weak pre-adsorption step of the substrate that pre-organizes it towards dissociative bond cleavage driven by the favorable adsorption of the benzyl radical and the chlorine. , In contrast to the proposed inner-sphere activation of these organic halide substrates by Ag, voltammetric studies on the reductive activation of C­(sp³)–X, where X is Cl, Br, or I, on glassy carbon electrodes have been primarily used to support the theory of concerted dissociative electron transfer based on a Morse curve model. In this model, the energy of the bond in the reactant and the repulsive interaction between the two fragments formed after electron transfer are approximated to be the same as the repulsive part of the reactant Morse curve. Furthermore, solvent reorganization is estimated using the Marcus-Hush-Levich model of outer-sphere electron transfer. Critically, this model describes bond cleavage in the absence of interactions between substrate, intermediate, or product with the electrode surface, and thus provides an explanation for the literature consensus that glassy carbon electrodes reductively activate C­(sp³)–X via outer-sphere electron transfer. Thus, taken together with the use of glassy carbon electrodes as an outer-sphere control, these studies provide the basis for the consensus hypothesis that Ag surfaces overcome the slow kinetics of OSET activation pathways by tuning the relative adsorption energetics of C­(sp³) organic chloride, bromide, and iodide substrates to their intermediates, leading to the earlier onset in the observed redox potential. ,,−

A surface-adsorption model for Ag-catalyzed inner-sphere reductive activation of organic halides raises the question of whether this mechanistic sequence is also possible with benzylic C­(sp³)–F bonds. This notion is further supported by the computed higher adsorption energetics of fluorine on low-index facets of Ag over chlorine, bromine, and iodine. Yet, to the best of our knowledge, an assessment of organic fluoride activation at Ag electrodes and simultaneous spectroscopic elucidation of surface-interacting species during C­(sp³)–F activation remains elusive. Here, we disclose the spectroscopic and electrochemical studies of benzylic C­(sp³)–F activation at Ag electrodes using a model substrate. We append a nitrile infrared-active tag to benzylic substrates to track and characterize their adsorption and activation using surface-enhanced infrared absorption spectroscopy (SEIRAS) in an attenuated total reflection (ATR) mode. In an analogy to the extensively studied activation of carbon–halogen bonds at Ag, namely C­(sp³)–Cl and C­(sp³)–Br, we investigated their activation at Ag electrode surfaces and compared it to a C­(sp³)–F congener. We find that, like organic chlorides and bromides, Ag electrodes can exhibit an earlier onset for C­(sp³)–F activation over carbon electrodes. Spectroscopic studies support the adsorption of each benzylic organic halide through the halogen. Our results demonstrate the adsorption of organic halides at Ag, including the organic fluoride, prior to reductive activation, implying that C­(sp³)–F activation is possible via an inner-sphere pathway at Ag electrodes.

Results and Discussion

To gain insight into the adsorption behavior of organic fluorides at Ag electrodes, we chose the model substrate, m-(fluoromethyl)­benzonitrile (mCNBzF), its organic chloride (mCNBzCl) and bromide (mCNBzBr) analogs, and the dehalogenated m-tolunitrile variant (mCNBzH). These substrates were chosen because: (1) the nitrile (CN) group enables us to track all compounds via in situ infrared (IR) spectroscopy; (2) the CN group is sufficiently close to the halogen that through-space electric field effects may be observable, and (3) CN stretching frequencies inform on the reduction of aryl nitriles to radical anions and radicals. We note that other spectroscopically observable features in this study are primarily attributed to the interfacial dynamics of the solvent, dimethylformamide (DMF), as previously observed and noted. This observation further highlights the utility of nitrile-based vibrational tags because other spectral regions are primarily obscured by competitive DMF dynamics (Figure S1 and Tables S1, S2). We note that the organic iodide congener, mCNBzI, was omitted due to competitive decomposition occurring on the time scale of targeted electrochemical and spectroscopic measurements at room temperature. To track and compare the interaction between Ag and mCNBzF, mCNBzCl, mCNBzBr, and mCNBzH, we utilized in situ SEIRAS in an ATR mode. , This technique leverages surface nanostructure to amplify IR absorption of molecules with normal modes perpendicular to the surface. SEIRAS is ideally suited to characterize the interfacial structure because the technique is (1) sensitive to distances within around 8 nm of the polarized surface with enhancements expected as a function r ^–6, where r is the perpendicular distance from the surface; (2) independent of the solution background; and (3) compatible with select transition-metal surfaces, including Ag.

The addition of organic halides, regardless of the halogen identity, gives rise to spectroscopic features commensurate with their adsorption to polarized Ag surfaces prior to the observation of Faradaic charge transfer. Figure a,c,e depict the linear sweep voltammogram (LSV) traces, and Figure b,d,f depict the associated SEIRA spectra collected for mCNBzF, mCNBzCl, and mCNBzBr, respectively. For all experiments, a reference spectrum was collected at −1.01 V vs Fc/Fc⁺ (all potentials quoted versus Fc/Fc⁺) in the presence of DMF and 0.1 M tetrabutylammonium perchlorate (TBAClO₄). Following, the organic halide was injected at −1.01 V, and distinct absorption features at 2230 ± 1, 2230 ± 3, and 2229 ± 1 cm^–1 were observed for mCNBzF, mCNBzCl, and mCNBzBr, respectively (Figure b,d,f). Figure S2 show the data presented in Figure , but with expanded wavenumber ranges, and Tables S3-S5 list the IR peaks observed. Data collected with mCNBzCl are consistent with our previous work, assigning the 2231 cm^–1 feature to the nitrile stretching mode (νCN) of surface-bound adsorbed species. For the organic fluoride and bromide, in situ SEIRAS experiments conducted using independently synthesized and isotope-labeled m ¹³CNBzF and m ¹³CNBzBr show a 49 and 51 cm^–1 redshift in this feature at −1.01 V, respectively, consistent with an assignment to the nitrile stretching mode, νCN, of these organic halides (Figure S3 and Tables S6, S7). Our previous study of interfacial catalytic halogen atom transfer reaction of alkyl bromides by polarized Ag surfaces suggests that the bromide anion and/or organic bromides can remain on negatively charged Ag electrode surface up to −2.92 V. When the SEIRA spectra are similarly collected at −1.01 V using the substrate that does not contain a halogen, mCNBzH, the spectra are featureless in this wavenumber regime, Figure g,h, demonstrating the critical role of the halogen in giving rise to an observable νCN, which, importantly, includes the organic fluoride. In contrast, the bulk ATR-IR spectrum for mCNBzH collected at greater than 20-fold higher bulk concentration gives rise to a feature at 2229 cm^–1, assigned to the νCN (Figure S4).

1.

Linear sweep voltammetry of 10 mM of mCNBzF (a), mCNBzCl (c), mCNBzBr (e), and mCNBzH (g) in DMF containing 0.1 M TBAClO₄ recorded on SEIRAS-active Ag film electrodes. Data collected at 2 mV s^–1 with a negative direction of scan. The onset of reductive current (∼−200 μA cm^–2) attributed to C­(sp³)–X bond activation is indicated by the dashed line. The open circles indicated on the voltammogram represent the potential at which the IR spectrum was obtained. Simultaneously collected SEIRA spectra for mCNBzF (b), mCNBzCl (d), mCNBzBr (f), and mCNBzH (h). Background spectra collected at −1.01 V vs Fc/Fc⁺ in the absence of substrates.

We examine the possibility that changes in the oscillator strength of mCNBzF over mCNBzH could lead to the enhanced νCN signal. To probe this hypothesis, we compared the integrated band intensity of the IR peak assigned to νCN (2245–2210 cm^–1) for spectra collected as a function of the bulk concentration of the mCNBzF and mCNBzH using either SEIRAS-active Ag surfaces polarized at −1.01 V or undoped Si, without any SEIRAS-active Ag film on the surface of the Si ATR prism. This experiment enables us to assess the influence of the polarized Ag surface on affecting νCN intensities, as well as the comparison of the νCN signal at the same bulk concentration between mCNBzH and mCNBzF, where interaction with the underlying substrate can be ruled out on Si surfaces. Figure a depicts the resulting titration data. On bare Si surfaces, we observe that mCNBzF gives rise to comparable integrated νCN IR band intensities compared to mCNBzH over the concentration regime of 2–64 mM at −1.01 V (Figure a, red versus black), demonstrating that the molar absorption coefficient, and thus the oscillator strength for νCN, between the two molecules does not significantly change in DMF containing TBAClO₄. Consistent with this hypothesis, bulk transmission Fourier Transform Infrared (FT-IR) analysis gives rise to molar absorption coefficient values of 22 M^–1 cm^–1 for both mCNBzF and mCNBzH (Figure S5). In contrast, on Si surfaces coated with a SEIRAS-active Ag layer polarized at −1.01 V, we observe that mCNBzF gives rise to increased integrated νCN IR band intensities compared to mCNBzH over the concentration regime of 2–64 mM at −1.01 V (Figure a, blue versus green), inconsistent with similar adsorption motifs for mCNBzF and mCNBzH on the SEIRAS-active Ag surface. We note the higher intensities of mCNBzH on the surface of Ag relative to bare Si at the same concentration, in line with the expected SEIRAS properties of the underlying and nanostructured metal. The surface selection rule of SEIRAS , precludes observation of νCN modes parallel to the surface, suggesting that the increased νCN intensity observed on the Ag surface for mCNBzF versus mCNBzH is consistent with a higher interfacial population of the organic fluoride at Ag with νCN aligned with the vibrational vector perpendicular to the interface through the interaction between the fluorine atom and Ag surface compared to the dehalogenated congener. If mCNBzF is oriented relative to the Ag surface with minimal net directionality, then similar integrated νCN band intensities would be expected for both molecules at every concentration probed. Plotting the νCN intensity of mCNBzF tracked by SEIRAS against the bulk concentration gives rise to a curve that can be well described by a Langmuir adsorption isotherm fit (Figure S6). Taken together, the observation of νCN only for halogen-functionalized nitriles, including the fluorinated derivative, at low concentrations by Ag SEIRAS, is consistent with a structural model in which organic fluorides, like organic chlorides and bromides, , adsorb to electrified Ag surfaces with the νCN vector positioned vertically to the surface (Figure a, blue). ,

2.

(a) Titration experiment with mCNBzF and mCNBzH on Ag film and bare Si surface in DMF containing 0.1 M TBAClO₄. Using the Ag film, SEIRA spectra were collected at −1.01 V vs Fc/Fc⁺. Background spectra were collected at −1.01 V vs Fc/Fc⁺ in the absence of mCNBzF and mCNBzH. Integrated area of nitrile stretching bands, νCN, was used. The non-zero integration documented when νCN is not observed (0.0001) is attributed to consistent fluctuations in the background. Error bars correspond to the standard deviation from data collected on two independently prepared Ag films and trials. The structural models to the right are simplified; solvent and electrolyte are omitted, and the hypothesized adsorption of mCNBzF to the Ag surface is presented in model 3. We note that the parallel orientations of the substrates shown in models 1, 2, and 4 are only shown to highlight the difference between the perpendicular orientation hypothesized for model 3. The orientation of substrates in models 1, 2, and 4 may be predominantly random in aggregate. Chronoamperometry traces recorded in the presence of 60 mM mCNBzH (b) and simultaneously collected SEIRA spectra (c). Background spectra collected at −1.01 V vs Fc/Fc⁺ in the absence of substrate.

The application of reductive potential gives rise to changes in the νCN band when the halogen is present. Switching the applied potential from −1.01 to −2.89 V, we observe enhanced reductive current for mCNBzF, mCNBzCl, and mCNBzBr beginning at −2.43, −1.97, and −1.21 V, respectively (Figure a,c,e). For mCNBzF, as the potential is swept to more negative values from −1.01 to −2.40 V, prior to the observation of Faradaic charge transfer, we observe the rise in the νCN peak assigned to adsorbed mCNBzF at 2229 cm^–1. We do not observe a significant Stark slope for this peak within experimental error (1 ± 1 cm^–1 V^–1, average and standard deviation from data collected on three independently prepared Ag films). As we observe the passage of reductive current beginning at −2.43 V, we find that this band is replaced by a new feature at 2072 cm^–1 at −2.89 V (Figure b). Data are consistent with the assignment of the peak at 2072 cm^–1 to a reduced aryl nitrile species. It is well-documented that the reduction of aryl nitrile species to nitrile radical anions or neutral radicals results in a redshift of the νCN. , Furthermore, this reduced aryl nitrile species was not detected with mCNBzH, supporting the hypothesis that the interaction between halogen and the Ag surface increases the interfacial concentration of the substrate (Figure h). The accumulated halogenated substrates, mCNBzX (X = F, Cl, Br), could lead to a higher surface concentration of reduced aryl nitrile species (i.e., mCNBz^• and/or mCNBzX^•–) via outer-sphere electron transfer near the Ag electrode surface. To independently probe the νCN feature of a reduced aryl nitrile species, we conducted in situ ATR-IR of mCNBzH at six-fold higher concentrations (60 mM), where we expect to see the bulk signal from mCNBzH (Figure a). We chose to conduct this experiment at −3.01 V because the reversible reduction of mCNBzH to the corresponding radical anion (mCNBzH^•–) has a reduction potential of −2.83 V (E _1/2 = −2.78 V, Figure S7). Indeed, at −3.01 V we observe the rise of a new feature at 2064 cm^–1 (Figure b). Furthermore, this peak at 2064 cm^–1 is shifted to higher wavenumbers upon changes in the substituent pattern. When changing methyl to methyl ester or ether substituents, we observe that the peak assigned to νCN of the radical anion shifts to 2073 and 2077 cm^–1, respectively (Figure S8). Thus, while we cannot rule out the assignment of the peak to a neutral radical aryl nitrile species that may form following reductive C–F bond cleavage (mCNBz^•), the observation of a peak at 2072 cm^–1 at −2.89 V in the presence of low concentration of mCNBzF is consistent with the presence of mCNBz^• or radical anion (mCNBzF^•–) at the Ag interface, and not mCNBzH^•–, as its bulk species would appear at 2064 cm^–1. We note that an adsorbed benzyl radical anion intermediate at polarized Ag surfaces has been hypothesized to explain the catalytic effects of benzyl chloride activation by Ag electrodes. If mCNBz^• is formed, it may interact with the Ag surface and form a carbene that follows α-elimination. Carbenes have been reported on transition metal surfaces, and are well-documented following α-elimination for homogeneous transition metal complexes. We disfavor this assignment because we do not observe any features in the region where Ag-bound hydride has been theorized (1660–1930 cm^–1).

For mCNBzCl, as the potential is swept from −1.01 to −2.01 V, prior to the observation of Faradaic charge transfer, we observe the rise in the νCN peak assigned to the adsorbed mCNBzCl and its corresponding redshift to 2225 cm^–1 at −2.01 V (Figure d). However, we refrain from assigning this redshift to a Stark shift as the redshift is within the error of the measurement (4 ± 4 cm^–1 V^–1, average and standard deviation from data collected on three independently prepared Ag films). As the Faradaic charge is passed beginning at −2.01 V, in contrast to mCNBzF, we observe the peak assigned to the adsorbed neutral organic halide persists, however, like mCNBzF, a new feature at 2082 cm^–1 is observed at highly reducing potentials, i.e., −2.89 V. This 2082 cm^–1 value is within 5 cm^–1 compared to that observed for data collected for mCNBzF, suggestive of an assignment to a common reduced aryl nitrile species. Finally, for mCNBzBr, as the potential is swept to −1.21 V, we immediately observe the onset of Faradaic charge transfer accompanied by the rise in the νCN peak assigned to the adsorbed mCNBzBr at 2229 cm^–1 with minimal Stark slope of 1 ± 1 cm^–1 V^–1 within error (average and standard deviation from data collected on three independently prepared Ag films, Figure f). Like mCNBzCl, we observe the persistence of the band at 2229 cm^–1 at highly reducing potentials, i.e., −2.89 V, and we observe a new feature at 2074 cm^–1. This 2074 cm^–1 value is in a similar region compared to that observed for data collected for mCNBzF, suggestive of an assignment to a common reduced aryl nitrile species. We note that the Stark tuning of νCN has been shown to be highly dependent on the angle of the nitrile motif relative to a polarized surface; lower slopes (approximately 1–4 cm^–1 V^–1) have been predicted for smaller angles, particularly when continuous conjugation from the electrode through the nitrile group is interrupted, as is the case here. Therefore, the negligible Stark tuning effect observed for the halogenated substrates within experimental error does not exclude their adsorption at the surface, even though through-space electric field effects, expected for the nitrile motif in an electric field environment within 8.4 Å, have been spectroscopically and computationally shown to give rise to Stark shifts. Importantly, the SEIRA spectra of mCNBzH remain featureless in the spectral region corresponding to the nitrile at equivalent low substrate concentrations (Figure g,h), indicating that both the neutral and reduced aryl nitrile species observed by SEIRAS require the presence of a halogen substituent, including fluorine, albeit at varying applied potential values. This result is consistent with prior surface-enhanced Raman studies of benzyl chloride reductive activation by Ag electrode surfaces, where toluene was not detected and attributed to its weak interaction with the surface. Together, these SEIRAS observations demonstrate the critical role of the halogen in enabling detection of both neutral and reduced aryl nitrile species at Ag.

Since we observed that mCNBzF exhibits evidence for adsorption on Ag electrodes, we next examined whether its reductive activation is enhanced relative to carbon-based electrodes. The reductive activation of all organic halides examined, including mCNBzF, occurs at more positive potentials on Ag compared to glassy carbon electrodes. Figure a–c depict the difference in the CVs of mCNBzF, mCNBzCl, and mCNBzBr reduction on planar Ag and glassy carbon electrodes. We have confirmed that the SEIRAS-active Ag film, probed above, gives rise to similar systematic shifts in the electrochemical reduction of mCNBzF, mCNBzCl, and mCNBzBr as we observe on planar Ag electrode surfaces. The onset potentials observed for reductive current (defined at −200 μA cm^–2) using Ag planar electrodes at −2.30, −1.71, and −1.18 V (Figure a–c) for mCNBzF, mCNBzCl, and mCNBzBr, respectively, similarly shift to more positive values with increasing halogen size for data collected at a 100 mV/s scan rate. We note that, unlike the data collected for SEIRAS, see above, where slow scan rates enabled observation of potential-dependent spectroscopic features that report on the steady-state interfacial structure, CV data shown here are conducted at scan rates greater than or equal to 100 mV s^–1 to mitigate voltammetric analysis that reports on competitive convection effects. Slow scan rate data collected on planar Ag electrodes show a similar potential onset values for substrate reduction compared to SEIRAS-active Ag surfaces (Figure S9). We note observed sharp reductive peaks for mCNBzCl, and mCNBzBr at −2.24 and −1.77 V, respectively, consistent with prior work assigning similar features to the reduction of adsorbed species. ,− Importantly, as shown in Figure , we discover that, for all three substrates, we observe a more positive reductive peak potential value for the first irreversible wave on Ag (E _p,Ag) compared to glassy carbon (E _p,GC). For mCNBzF, the E _p,Ag value is 50 mV more positive than E _p,GC. For mCNBzCl, the E _p,Ag value is 220 mV more positive than E _p,GC. This shift is consistent with literature reports on Ag-catalyzed C­(sp³)–Cl reductive activation, attributed to the pre-adsorption of the organic chloride to the Ag surface, lowering the barrier for reduction, as well as the favorable thermodynamics associated with the liberated halide adsorption to the surface. ,, Finally, for mCNBzBr, the E _p,Ag value is 360 mV more positive than E _p,GC. In contrast, at more negative potentials where reversible reduction of the aryl nitrile is accessed for mCNBzF, mCNBzCl, and mCNBzBr, the half-wave potential values (E _1/2) are indiscriminate between data collected on Ag and glassy carbon under the conditions examined in this study (Figure S10). The observed E _1/2 value is consistent with the E _1/2 values observed for the reduction of m-tolunitrile and 3,3′-(ethane-1,2-diyl)­dibenzonitrile (dimerization product) (Figure S7). The quasi-reversible wave of the dimerization product supports the assignment of the reversible wave of halogenated compounds in Figure S10 to tolunitrile or to an unreacted neutral radical (mCNBz^•) from reductive carbon–halogen bond cleavage. We note that similar behavior has been reported for 4-(trifluoromethyl)­benzonitrile and 4-(chloromethyl)­benzonitrile, , where the reversible wave following the irreversible reduction was assigned to the protodehalogenated species. The positively shifted E _p values for the irreversible wave persist at varying scan rates (Figure d–f). The peak currents at E _p,Ag, and E _p,GC scale linearly with the square root of the scan rate (Figure S11), consistent with a diffusion-controlled Faradaic process and in line with previous observations for benzyl chloride reduction on Ag. The voltammetric features at E _p,Ag, and E _p,GC both remain irreversible over the scan rates probed, in line with prior electroanalytical studies on related benzyl fluoride substrates using glassy carbon electrodes. Data are inconsistent with the nitrile functional group serving as the primary driver of the enhanced reduction. Comparison of the CVs of 1-(fluoromethyl)-3-methoxybenzene (mMeOBzF) collected using Ag and glassy carbon electrodes also reveal an E _p,Ag value that is 110-mV more positive than that of E _p,GC (Figure S12). Together, these data show an earlier onset in the first irreversible reductive peak for all organic halides, including the organic fluoride, on Ag compared to glassy carbon.

3.

Cyclic voltammograms of 10 mM of mCNBzF (a), mCNBzCl (b), and mCNBzBr (c) in DMF containing 0.1 M TBAClO₄ using Ag electrodes (solid lines) and GC electrodes (dashed lines). Data collected at 100 mV s^–1 with a negative direction of scan. Scan rate dependence of the peak potential for data collected using 2 mM of mCNBzF (d), mCNBzCl (e), and mCNBzBr (f) in DMF containing 0.1 M TBAClO₄ using Ag and glassy carbon electrodes. Error bars correspond to the standard deviation from three independently collected data sets.

Bulk electrolysis is consistent with the assignment of the peak at E _p,Ag to the irreversible reduction of the carbon–halogen bond for each substrate. Bulk electrolysis experiments of both mCNBzF and mCNBzCl at E _p,Ag resulted primarily in the formation of dehalogenated mCNBzH product (Figure S13a,b) after the passage of 1 F mol^–1 of substrate. Although we note the variability in the observation of a dimerized product previously observed for the organic chloride, this result is consistent with our previous work implicating the intermediacy of a neutral benzyl radical following activation that can be terminated via hydrogen atom transfer from DMF. XPS results further support the substrate adsorption through the halogen. The Ag surface before and after the bulk electrolysis was characterized by XPS (Figure S14). Peaks corresponding to Ag 3d_5/2 and Ag 3d_3/2 transitions, at 374.4 and 368.4 eV, respectively, were observed before and after electrolysis. However, a peak at 685.0 eV, consistent with an assignment to a F 1s transition of metal fluorides , and inconsistent with an assignment to organofluorine compounds, is only detected on Ag surfaces following electrolysis. Bulk electrolysis of mCNBzBr at E _p,Ag resulted primarily in the formation of the dimerized product (Figure S13c) after the passage of 1 F mol^–1 of substrate. This result is consistent with the assignment of the irreversible reductive feature to carbon–halogen bond cleavage that generates a neutral benzyl radical, which can either dimerize or undergo a second electron transfer to form a carbanion that reacts with the starting material to form the dimer. We note that we cannot rule out the direct reduction of the nitrile to produce an amine, however, we do not see evidence for this product in our gas chromatography analyses. While the bulk electrolysis data do not inform on the elementary (electro)­chemical steps that follow reductive cleavage of the examined substrates, the results show that E _p,Ag of each substrate reports on the reductive activation of each C­(sp³)–X bond.

The dissociative electron transfer of mCNBzF, mCNBzCl, and mCNBzBr were examined using voltammetric analysis. For outer-sphere dissociative electron transfer, the dissociation can occur either in a concerted or stepwise fashion. In the stepwise case, diagnostic criteria for the relationship between the variation of the scan rate with the peak potential (i.e., ∂E _p/∂log ν) have been derived. If the initial electron transfer acts as a pre-equilibrium, followed by kinetic control of the cleavage step, then ∂E _p/∂log ν equals 29.6 mV at room temperature. If the overall reaction is, instead, under kinetic control of the electron transfer step, such that the cleavage step is fast compared to electron transfer, then ∂E _p/∂log ν equals 29.6/α mV at room temperature, where α is the transfer coefficient. If the potential at which bond cleavage occurs is near the standard potential for the reductive activation, then α is close to 0.5, and thus ∂E _p/∂log ν equals 59.2 mV. When performing this analysis for voltammetry collected on mCNBzF on glassy carbon electrodes, we obtain a value of ∂E _p/∂log ν equaling 43 ± 2 mV with α value equaling 0.68 (Figure S15a), consistent with stepwise dissociation under kinetic control of the electron transfer step. This mechanism is consistent with prior proposals for electrochemically driven defluorination of benzylic fluorides by outer-sphere electron transfer, including 4-(fluoromethyl)­benzonitrile, 4-(difluoromethyl)­benzonitrile, 4-(trifluoromethyl)­benzonitrile, 3-(trifluoromethyl)­benzonitrile, and 2-(trifluoromethyl)­benzonitrile. In the concerted dissociative electron transfer case, ∂E _p/∂log ν similarly equals 29.6/α mV at room temperature, however the energy barrier for the reaction contains a contribution of bond-breaking, thus giving rise to α values smaller than 0.5. When performing this analysis for voltammetry collected on mCNBzCl and mCNBzBr on glassy carbon electrodes, we obtain a value of ∂E _p/∂log ν equaling 77 ± 2 and 84 ± 3 mV, respectively (Figure S15b,c). These slope values correspond to α values equaling 0.38 and 0.35 for mCNBzCl and mCNBzBr, respectively, consistent with concerted dissociative electron transfer. This mechanism is consistent with that proposed for electrochemically driven dehalogenation of benzylic chlorides and bromides by outer-sphere electron transfer, including benzyl chloride, benzyl bromide, 2-(bromomethyl)­benzonitrile, and 4-(bromomethyl)­benzonitrile. Compared with electron transfer reactions occurring via outer-sphere mechanisms, we note that surface-mediated electron transfer involving surface-adsorbed reactants, such as mCNBzF, mCNBzCl, and mCNBzBr invoked here, precludes the utilization of these diagnostic criteria. Prior work has circumvented this limitation by utilizing Ag microelectrodes, revealing two voltammetric features on benzyl chloride at fast scan rates; one corresponding to the reduction of the adsorbed substrate by inner-sphere electron transfer, and the other corresponding to the direct reduction of the substrate by outer-sphere electron transfer. Our attempts to translate these results to Ag microelectrodes at fast scan rates (>10 V s^–1) rendered varying results, however slower scan rates (0.1 V s^–1) shown in Figure S16 preserve similar onset for reduction at −2.40, −1.86, and −1.45 V for mCNBzF, mCNBzCl, and mCNBzBr, respectively, as those observed for planar Ag disks and SEIRAS-active Ag films, further suggesting that a variability in the Ohmic drop between glassy carbon and Ag is not the primary role for the enhanced mCNBzF activation we observe herein. Taken together, the spectroscopic and voltammetric results implicate that halogenated substrates interact with the polarized Ag surface before the Faradaic charge transfer. We propose that the more positive E _p values observed for mCNBzF activation on Ag, like mCNBzCl and mCNBzBr, are associated with the weak interfacial adsorption of the organic halides prior to electron transfer, which can lower the apparent activation barrier relative to glassy carbon. ,, We note that our data lie in contrast to prior work on electrochemical C­(sp³)–F bond activation reported at other transition metals, including Au, , Ni, Pb, Pt, Fe-based alloy electrodes, that implicate an outer-sphere electron transfer process. Instead, our work builds on prior studies of the catalytic effects of Ag on the activation of organic chlorides, bromides, and iodides ,,,− ,− by demonstrating that organic fluorides can interact with Ag surfaces to alter the reduction potential required to trigger reductive carbon–halogen activation.

Conclusions

Our combined studies provide a starting point for defining the mechanistic role of Ag surfaces in facilitating reductive C­(sp³)–F activation. The in situ surface-enhanced infrared absorption spectroscopic data are consistent with the adsorption of nitrile-tagged benzyl bromide, benzyl chloride, and benzyl fluoride. The reduction of all three substrates occurs at more positive potentials than glassy carbon. While we note that these studies focus on a singular organic fluoride substrate, the spectroscopic observation of the pre-adsorbed substrate and its enhanced reductive activation motivates further study of a broader class of organofluorides and electrode materials that may be capable of inner-sphere activation for electroorganic synthesis reactions. More broadly, these studies emphasize the need to benchmark electrodes previously assumed to be inert in electroorganic synthesis.

Experimental Methods

Materials and Chemicals

Reagents were purchased from commercial sources and used without further purification unless otherwise noted. Standard Schlenk line techniques were utilized for synthetic manipulations unless otherwise noted. Dimethylformamide (DMF), tetrahydrofuran (THF), and acetonitrile (MeCN) were dried and purified using a solvent purification system (PureSolv MD 5, INERT Corporation) under N₂. Thin-layer chromatography (TLC) analyses of reaction mixtures were performed using Merck silica gel 60 F254 TLC plates and visualized using iodine, KMnO₄ stain, or ultraviolet (UV) light. Column chromatography and preparative TLC were carried out using Mackerey-Nagel Silica Gel 60 Å, 0.04 × 0.063 mm or Merck silica gel 60 F254 TLC plates, respectively. As an electrolyte, anhydrous 0.1 M TBAClO₄ electrolyte in DMF was prepared by drying the recrystallized TBAClO₄ salt under vacuum overnight at 60 °C, and dissolving the salt in anhydrous DMF (Sigma-Aldrich, 99.8%), and then storing over freshly activated molecular sieves (3 Å, Supelco) in a N₂-filled glovebox. Immediately before use, the electrolyte was filtered through a 0.22 μm frit.

Electrode Materials and Preparation

Polycrystalline Ag working electrodes were purchased from CH Instruments (2 mm diameter, 0.0314 cm² area). Ag electrodes were polished mechanically on a MicroCloth polishing pad (Buehler) using 1 μm alumina slurry with Milli-Q (Millipore Type 1, 18.2 MΩ) water and rinsed with Milli-Q water and acetone and dried under vacuum. Glassy carbon (GC) working electrodes were purchased from CH Instruments (3 mm diameter, 0.0707 cm² area). GC electrodes were polished mechanically on a MicroCloth polishing pad (Buehler) using 0.5 μm MetaDi monocrystalline diamond suspension (Buehler) and rinsed with Milli-Q water, acetone, and hexanes and dried under vacuum. The Ag microelectrode was prepared with Ag wire (diameter: 25 μm). The Ag wire was sealed in glass tube and polished on a MicroCloth polishing pad (Buehler) using 1 μm alumina slurry with Milli-Q water. After polishing, the Ag microelectrode was rinsed with Milli-Q and acetone and dried under vacuum.

Electrochemical Methods

All CV experiments with Ag and GC working electrodes were conducted in an N₂-filled glovebox. All CV experiments with Ag microelectrodes were conducted under N₂ flow. Electrochemical experiments were conducted using a Biologic VSP Potentiostat, CHI 760E potentiostat, or Gamry Reference 1010B potentiostat, an Ag/Ag⁺ reference electrode (BASi, 10 mM AgBF₄ in 0.1 M TBAClO₄ in DMF, Fc/Fc⁺ at 0.01 V versus Ag/Ag⁺) or an Ag/AgCl reference electrode (eDAQ, leakless, 0.23 V vs SHE), a high surface area Pt-mesh counter electrode (Alfa Aesar, 99.997%), and a glassy carbon (BASi, 3 mm diameter, 0.0707 cm² area) or Ag (BASi, 2 mm diameter, 0.0314 cm² area) working electrodes. The Ag/Ag⁺ electrode was stored in DMF between measurements and was periodically checked relative to Fc/Fc⁺ to ensure against potential drift. Electrode potentials were converted to the Fc/Fc⁺ standard using the conversion, E _Fc/Fc ⁺ = E _Ag/Ag ⁺ – 0.01 V. For each experiment using the Biologic VSP Potentiostat, an initial starting point for the uncompensated Ohmic loss was measured using the Precise Function in the EC-Lab Biologic software workstation v.11.60 (roughly 100–300 Ω). This initial value was further corrected by collecting sequential cyclic voltammograms with increasing R _u values until oscillations corresponding to total compensation were observed. This corrected R _u value (R _corr) was approximately 100–500 Ω higher than the initial value measured using the R _u test function, that is, 300–500 Ω. This R _corr was input manually into the iR compensation menu in the EC-Lab Biologic software workstation v.11.60. Current density values are reported relative to the geometric surface area of the working electrode. A two-compartment electrochemical cell was used, holding 6 mL of electrolyte in the working compartment and counter compartments (James Glass Inc.). The compartments were separated by an ultra-fine glass frit (P5, Adams and Chittenden). For the data collected in Figures and S7, S9, S11, S12, S15, and S10, the first CVs were reported on a freshly polished Ag electrode. For electrodeposition, the Ag/AgCl electrode was stored in Milli-Q water between measurements and was periodically checked relative to a pristine Ag/AgCl to ensure against potential drift. All experiments were performed at 24 ± 1 °C.

Bulk Electrolysis Procedures

Bulk electrolysis experiments were performed in an N₂-filled glovebox. Experiments were conducted using a Biologic VSP Potentiostat, a non-aqueous Ag/Ag⁺ reference electrode (BASi, 10 mM AgBF₄ in 0.1 M TBAClO₄ in DMF, Fc/Fc⁺ at 0.01 V versus Ag/Ag⁺), a high surface area Pt-mesh counter electrode (Alfa Aesar, 99.997%) and an Ag mesh working electrode (1 cm² geometric surface area, 99.9%, Alfa Aesar), attached to a stainless-steel rod by threading through Ag wire of 99.9% purity. For every experiment, the Ag mesh electrode was cleaned by immersing the electrode for 30 min in H₂SO₄, sonication in Milli-Q water (30 min), followed by sequential washing with Milli-Q water and acetone. mCNBzX (40 mM, X = F, Cl, and Br) were added to the cathode side containing 0.1 M TBAClO₄ in 6 ml of DMF. Before each experiment, a CV of each substrate was taken using an Ag working electrode to verify the potential of Ag-catalyzed substrate reduction. The working electrode was switched to the Ag mesh, and a chronoamperometry trace was initiated at a potential value around 50 mV more negative than the peak potential of the catalyzed reduction determined by CV. During the bulk electrolysis, the working and counter compartments were continuously stirred (roughly 600 rpm). After the current value reached the background value (roughly −0.1 mA), the chronoamperometry experiment was terminated. Consumption of the substrates was confirmed by CV with an Ag working electrode. For determining yields by gas chromatography mass spectrometry, 40 mM of decane in ethyl acetate (EA) internal standard solution was added before the liquid injection at a 1:1 ratio with the reaction solution. All experiments were performed at 24 ± 1 °C.

Surface-Enhanced Infrared Absorption Spectroscopy (SEIRAS)

SEIRA spectra were recorded in an attenuated total reflection (ATR) configuration using a Nicolet iS50 Fourier transform IR spectrometer equipped with a HgCdTe (MCT) detector (Ref./2× Gain Setting) and a PIKE VeeMax III accessory (incident angle of 60°). The Nicolet spectrometer was operated in the Series mode with an optical velocity of 1.8988 cm s^–1. Spectra were sequentially acquired with a spectral resolution of 4 cm^–1 at intervals of every 19.690 s. A single-beam spectrum collected at the starting potential in the absence of any substrates (that is, in the presence of 0.1 M TBAClO₄ in DMF) was used as the reference spectrum. All ATR-SEIRA spectra are reported in absorbance units defined as A = −log (I/I ₀), where I and I₀ stand for the sample and reference single-beam spectra, respectively. Data analyses were conducted using the OMNIC v.9.12.928 software. The PIKE Jackfish J2 spectroelectrochemical cell was utilized and held 14.5 mL of electrolyte. The spectroelectrochemical cells were equipped with a SEIRAS-active Ag film as the working electrode, a Pt wire counter electrode, and a non-aqueous Ag/Ag⁺ reference electrode. They were assembled inside an N₂-filled glovebox, sealed after the addition of 14 ml of DMF containing 0.1 M TBAClO₄, and then taken out before the SEIRAS measurements. Reference spectra (Tables S1, S3–S5, S8, and Figures S1 and S4) collected in the absence of an Ag film on top of the Si, i.e., only using a bare Si prism, were obtained using the same spectroscopic setup.

Preparation of Ag Films for SEIRAS

Ag films were prepared as previously detailed and characterized in our group. Au films were prepared on undoped Si prisms (PIKE) using the “double deposition method” as previously detailed. − The geometric surface area of the SEIRAS film exposed to the electrolyte was 0.71 cm². Following the deposition, the Au-coated Si prism was assembled into the PIKE Jackfish J2 spectroelectrochemical cell and cleaned using an electrochemical procedure. The Au film was cycled in 0.1 M H₂SO₄ from 0.2 V versus Ag/AgCl to 1.5 V versus Ag/AgCl at 50 mV s^–1 for ten continuous cycles without pause. Following the Au deposition, the 0.1 M H₂SO₄ was replaced with 0.1 M H₂SO₄ containing 2 mM Ag₂SO₄ and degassed for 10 min with continuous N₂ sparging. Then, chronoamperometry traces were conducted using the following sequence. First, 0.3 V versus Ag/AgCl was applied for 300 s twice (for a total of 600 s). Second, 0.25 V versus Ag/AgCl was applied for 300 s once (for a total of 300 s). This deposition procedure resulted in a film that looked like an Ag layer by eye. Following, the film was washed with Milli-Q water, followed by ethanol. Then, the film was dried for 30 min under a continuous stream of N₂ and cycled into the glovebox. The background spectra for these four independent films in the absence of substrates but in the presence of DMF and 0.1 M TBAP are reported in Figure S1. These data are collected after at least one 2 mV s^–1 scan cycle between −1.01 and −2.89 V versus Fc/Fc⁺ and subsequent background collection at −1.01 V versus Fc/Fc⁺. During a constant applied potential of −1.01 V versus Fc/Fc⁺, concentrated solutions of mCNBzX (X = F, Cl, and Br) were injected sequentially.

X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy (XPS) measurements were collected on an as-prepared Ag foil before and after bulk electrolysis at −2.5 V, E _p,Ag, with 40 mM mCNBzF in 0.1 M TBAClO₄ in DMF (Figure S14). The Ag foil was removed from the electrochemical cell and rinsed with DMF and acetone twice, respectively, in a N₂-filled glovebox and protected by a N₂-filled bag until the XPS measurement. The X-ray photoelectron spectra were collected using an AXIS Nova spectrometer (Kratos Analytical) equipped with a monochromatic Al Kα X-ray source. The Al anode was powered at 10 mA and 15 kV. The instrument work function was calibrated to give an Au 4f_7/2 binding energy of 83.95 eV. The instrument base pressure was ca. 1 × 10^–9 Torr. The analysis area size was 0.3 × 0.7 mm². The reported binding energies were referenced to the C 1s peak at 284.8 eV. The high-resolution spectra were collected with a pass energy of 40 and a 0.1 eV step size.

General Methods

¹H and ¹³C Nuclear Magnetic Resonance (NMR) experiments were conducted using Bruker Ultrashield 400 Plus instruments. Peaks were referenced using residual solvent peaks, e.g., for ¹H NMR CDCl₃ (δ = 7.26 ppm) and ¹³C NMR in CDCl₃ (δ = 77.16 ppm). High-resolution mass spectra (ESI/APCI ion source) were obtained on an Agilent 6224 Time-of-flight Mass Spectrometry (TOF-MS) spectrometer instrument and were reported for molecular ions [M]⁺, [M + Na]⁺, or [M + H]⁺, with chloroform, and MeCN as the solvents. Gas chromatography with mass spectrometry (Electron Ionization, Agilent Tech HP-5Å MS, 30 m × 0.25 mm × 0.25 μm column) was conducted using an Agilent 5975 Series MSD, Hewlett Packard 7683 Series injector, and Agilent 6890 Series GC instrument. DMF and dichloromethane (DCM) were used as the solvent. Liquid IR analysis was conducted using a sandwich configuration KBr liquid IR cell (BECKMAN) with a 0.1 mm optical path length.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acselectrochem.6c00101.

• Experimental details and methods are listed in the SI. Synthetic procedures, electrochemical data, in situ IR spectroscopic data, NMR data, and HRMS data (PDF)

The authors declare no competing financial interest.