Field Effects Govern the Decarboxylation of Copper(II)-Benzoates: Kinetic and Mechanistic Studies

Abstract

An experimental mechanistic study of the decarboxylation reactivity of well-defined 1,10-phenanthroline-ligated copper(I)- and copper(II)-benzoate complexes is reported. This work demonstrates decarboxylation to occur from both copper(I)- and copper(II)-benzoates with the (phen)Cu^II(2-nitrobenzoate)₂ (2) proceeding through an oxidative decarboxylation pathway to generate the homocoupled biaryl product while decarboxylation from (phen)CuI(2-nitrobenzoate) (1) generates primarily the arene via redox-neutral protodecarboxylation (phen = 1,10-phenanthroline). The time course kinetics and additive experiments demonstrate the decarboxylation of copper(II)-benzoates to both generate and be catalyzed by copper(I). We also disclose the characterization and reactivity of an off-pathway bis(phenanthroline)copper(I)-benzoate species [(phen)₂CuI][2-nitrobenzoate] (3) as well as the interconversion of these species under typical reaction conditions. These studies reveal an unexpected inhibitory effect of phen on the decarboxylation of copper(I) species with the unligated species CuI(2-nitrobenzoate) (4) showing rapid decarboxylation. Finally, synthesis, characterization and reaction kinetics of the decarboxylation of a series of differently substituted copper(II)-benzoate complexes reveal the field effect (F) to be the dominant factor governing the rate of decarboxylation in these systems.

Graphical Abstract

INTRODUCTION

Decarboxylative coupling reactions, which leverage carboxylic acids as coupling partners, have garnered attention as efficient and step-economic routes to substituted arenes and heteroarenes.¹ (Hetero)aromatic acids are available in great structural diversity from natural and synthetic sources, and they are easy to store and handle making them ideal starting materials. Additionally, the metal-mediated extrusion of CO₂ from a metal-carboxylate results in a metal-aryl intermediate that affords functionalization selectively at the ipso-carbon, thereby overcoming the inherent selectivity challenges associated with C-H functionalization reactions.

Copper catalysts in particular have demonstrated an innate ability to facilitate a wide array of decarboxylative coupling reactions.² Early work from Shepard and coworkers demonstrated the protodecarboxylation of furoic acid derivatives by copper bronze,³ while later studies by Nilsson and coworkers established the first copper-mediated decarboxylative cross-coupling of 2-nitrobenzoic acids and iodobenzene.⁴ Building on these pioneering examples, Goossen and coworkers developed the catalytic decarboxylative coupling of benzoic acids with aryl bromides by employing cocatalytic Pd and Cu (Scheme 1a).⁵ Later advances have extended this strategy to couplings with less-activated aryl halides⁶ and pseudohalides^5–9 as well as the development of copper-only catalyst systems (Scheme 1b).¹⁰

Scheme 1.

Copper-Catalyzed Redox-Neutral Decarboxylative Coupling Reactions and the Corresponding Mechanistic Proposals and Support.

In each of these reactions, the decarboxylation is believed to occur at a copper(I)-benzoate to access an active copper(I)-aryl intermediate (Scheme 1c). The nucleophilic copper(I)-aryl then undergoes transmetalation with palladium or direct reaction with the electrophilic coupling partner. These proposals evolved from foundational studies of the copper(I)-aryl species (Scheme 1d). In particular, Sheppard and coworkers isolated the copper(I)-aryl intermediate in the stoichiometric decarboxylation of cuprous pentafluorobenzoate in quinoline solvent¹¹ and independently demonstrated the reactivity of this species with aryl halides.¹² Similarly, Nilsson and coworker prepared the copper(I)-phenyl species and demonstrated its coupling reactivity with iodobenzenes.^4a,13 Later studies by Goossen and coworkers provided computational support for copper(I)-aryl intermediates in both protodecarboxylation¹⁴ and decarboxylative coupling reactions.¹⁵

In contrast, the recent development of oxidative decarboxylative coupling reactions¹⁶ has enabled coupling of the purported copper-aryl intermediate with a wide array of nucleophilic coupling partners to afford new C-C,¹⁷ C-N,¹⁸ C-S,¹⁹ C-O²⁰ and C-halogen²¹ bonds (Scheme 2). Under these oxidative conditions a variety of decarboxylation pathways are possible due, in part, to the multiple oxidation states available to copper.²² Some proposals include decarboxylation of a copper(I)-benzoate intermediate to generate a copper(I)-aryl species which then undergoes oxidation and coupling to form product.^17c,19b,23 Alternative proposals suggest oxidation precedes decarboxylation which then occurs at a copper(II)-^{17f,17g,18a,21b,24} or copper(III)-benzoate intermediate.^17d However, there have been no mechanistic studies to establish the operative pathway in these reactions.

Scheme 2.

Copper-Catalyzed Oxidative Decarboxylative Coupling Reactions.

Furthermore, both catalytic^16–21 and stoichiometric decarboxylations^11–13,25 have demonstrated superior reactivity of ortho-nitrobenzoates under copper-mediated conditions. In particular, Cohen and coworkers demonstrated enhanced rates of decarboxylation of ortho-nitrobenzoates relative to other substituted benzoates in the decarboxylation of well-defined copper(I)-carboxylates.^25b Similarly, our past work has revealed related trends in the decarboxylation of well-defined silver(I)-benzoate complexes.²⁶ In these studies, we uncovered the field effect of the substituent to be a controlling feature in the silver-mediated decarboxylation.²⁷ This substituent feature may also be significant in copper-mediated decarboxylations. There are, however, no studies addressing the substituent influences in the decarboxylation of copper(II)-benzoates.

In this study, we compare the decarboxylation reactivity of well-defined 1,10-phenanthroline-ligated copper(I)- and copper(II)-2-nitrobenzoate complexes (1 and 2, respectively) and demonstrate autocatalytic behavior in the decarboxylation of copper(II)-benzoates. We also disclose the characterization and reactivity of an off-pathway bis(phenanthroline)copper(I)-2-nitrobenzoate species (3). Our findings demonstrate that decarboxylation occurs from all isolated complexes and each forms a unique product distribution characteristic of the copper oxidation state (Scheme 3). The copper(I)-2-nitrobenzoate species (1), which is in equilibrium with (3), proceeds primarily through a redox-neutral decarboxylation pathway forming arene, while the copper(II)-bis(2-nitrobenzoate) (2) follows an oxidative coupling pathway to generate the corresponding biaryl product. We also describe the interconversion of these species under typical reaction conditions as well as the influence of benzoate substitution on the rates of decarboxylation from copper(II). Here we demonstrate that both the rate of decarboxylation and the duration of the lag phase are strongly correlated with the field effect of the benzoate substituents.

Scheme 3.

Decarboxylation of Well-Defined Copper-Benzoate Complexes.

RESULTS AND DISCUSSION

Decarboxylation of Copper(I)- and Copper(II)-Benzoates.

Under copper-catalyzed oxidative decarboxylative coupling conditions, both copper(II)- and copper(I)-benzoates may undergo the key decarboxylation step. To compare the propensity of each species toward decarboxylation, we first measured the rates of decarboxylation from the well-defined copper(I)- and copper(II)-(2-nitrobenzoate) complexes 1 and 2 (Figure 1a).²⁴ The 2-nitrobenzoate complexes were chosen for initial study because this carboxylic acid is a privileged substrate in decarboxylative coupling reactions^{11–13,25,26} and 1,10-phenanthroline (phen) is a common ligand employed under copper-catalyzed conditions.^16–21

Figure 1.

(a) The stoichiometric decarboxylation and (b) corresponding CO₂ evolution time course data of complexes 1 (red circles ●, 100 μmol, 50 mM) and 2 (blue squares ■, 100 μmol, 50 mM) in DMSO (2 mL) at 140 °C under N₂. Each time course is an average of two runs and the error bars are given as (2(stddev))/√2.

While NMR spectroscopy is suitable for monitoring the reactions of copper(I) complexes and UV-visible spectroscopy is appropriate for copper(II) species, these techniques require dramatically different concentrations and therefore hinder the direct comparison of reaction rates. Methods involving the quantification of the organic products alone are not suitable because the rate of product formation may not correlate with the rate of decarboxylation. Additionally, the stoichiometric decarboxylation reactions of complexes 1 and 2 lead to different product distributions of nitrobenzene and biaryl (30% and 48% for 1; 17% and 73% for 2, Figure 1a). Thus, we sought a technique that would enable the direct measurement and comparison of decarboxylation rates for both copper(I) and copper(II) species.

The use of a ReactIR 15 system fitted with a gas cell enabled quantification of the CO₂ produced over the course of the reactions (see SI for full details). The decarboxylation reactions of complexes 1 and 2 were conducted in DMSO at 140 °C and the headspace was monitored by IR spectroscopy revealing dramatically different reaction profiles for the two complexes (Figure 1b). Decarboxylation of the copper(I) complex 1 begins immediately and proceeds with an initial rate of 4.4 mM min⁻¹ (for [1]_o = 50 mM, Figure S6). In contrast, decarboxylation of the copper(II) complex 2 gives rise to a sigmoidal time course with CO₂ formation beginning after ~8 minutes and reaching a maximum rate of 13.6 mM min⁻¹ (for [2]_o = 50 mM, Figure S7). The appearance of this reaction profile is reminiscent of the kinetics previously measured for the copper-catalyzed aerobic oxidative decarboxylative thiolation reaction,²³ in which a copper(II)-carboxylate complex underwent in situ disulfide-mediated reduction to copper(I) during an initial lag phase. In the absence of disulfide, however, a different pathway for reduction must be operative.

In Situ Reduction of Copper(II)-Carboxylate Complexes.

The visual appearance of the decarboxylation reactions of 2 are also instructive. During the initial lag phase of the reaction, the mixture changes from a blue-green color typical of (phen)copper(II) to a deep red color characteristic of (phen)copper(I), consistent with reduction to copper(I). Reduction of copper(II) to copper(I) during the lag phase of the reaction was supported by UV-visible spectroscopy which reveals the formation of a copper(I) band around 430 nm during these early reaction times (~5–15 min, Figure 2).

Figure 2.

Representative UV-visible spectra measured for the decarboxylation of complex 2 (50 mM) in DMSO at 140 °C for 2h. The initial increase in CuII due to dissolution is followed by its complete disappearance and the formation of CuI which reaches its maximum concentration at 14 min before decreasing.

The reduction of copper(II) to copper(I) in the absence of a reducing agent is a common phenomenon in copper catalysis²⁸ as well as materials synthesis.²⁹ In many of these cases a ligand, solvent, anion or base facilitates the reduction. With regards to decarboxylation reactions, Cohen and coworkers described the in-situ reduction of copper(II)-carboxylates to copper(I)-carboxylates prior to decarboxylation.²⁵ Under their conditions, the coordinating quinoline solvent served as the reducing agent. Under our reaction conditions, however, quinoline is not included and no phenderived oxidation products could be found, suggesting against the phen ligand as the reductant.

An alternative possibility is single-electron-transfer from the carboxylate ligand. Both aroyloxyl and aryl radicals have been proposed as intermediates in the photoredox catalytic decarboxylation protocols.³⁰ In these systems a light-induced carboxylate-to-copper(II) charge transfer (LMCT) results in the formation of an aroyloxyl radical and copper(I) (Scheme 4, Pathway A). The aroyloxyl radical rapidly decarboxylates to give an aryl radical which undergoes further reaction to produce the coupled product.

Scheme 4.

Possible Pathways for Copper(II)-Carboxylate Reduction and Radical Trapping Experiments.

Prior studies by the groups of Goossen¹⁴ and Cohen^25b,31 had ruled out radical intermediates in the thermal decarboxylation of copper(I)-benzoate complexes. Similarly, under our conditions no radical trapping products could be obtained when BHT or DHA were included as radical trapping agents (Scheme 4b), nor were there significant changes in the ¹H NMR reaction profiles (Figures S16–S17; BHT = butylated hydroxytoluene; DHA = 9,10-dihydroanthracene).³² In addition, no changes in the reaction time course were observed when the decarboxylation of 2 was conducted in the dark (Figure S20). Furthermore, the (phen)Cu^II(2-phenylbenzoate)₂ complex underwent smooth decarboxylation yielding biphenyl in 89% yield without the formation of any intramolecular trapping products (Scheme 4b, Figure S21).^{30a, 33} Although we cannot exclude the possibility of radical intermediates, these data suggest against the formation of trappable free radical intermediates.³⁴ Instead, we favor a two-electron reductive decarboxylation pathway that generates copper(I) and biaryl without the formation of trappable radical intermediates (Scheme 4, Pathway B). This pathway is analogous to that proposed for the Cu-catalyzed homocoupling of terminal alkynes.³⁵

The proposed reductive decarboxylation pathway was supported with a combined ReactIR / UV-visible / HPLC study. Under the standard reaction conditions the decarboxylation of 2 was monitored with ReactIR gas cell measurements. Simultaneously, aliquots of the same reaction mixture were withdrawn and subjected to UV-visible spectroscopy to quantify the copper(I) and copper(II) present, as well as HPLC analysis for quantification of the organic products (nitrobenzene and 2,2’-dinitrobiphenyl). Correlation of the CO₂ evolution with the copper speciation revealed that reduction of copper(II) to copper(I) occurs concurrently with the generation of CO₂ (Figure 3a), consistent with a reductive decarboxylation pathway. HPLC analysis of the organic products demonstrated formation of both 2,2’-dinitrobiphenyl and nitrobenzene with good mass balance (Figure S15). The formation of 2,2’-dinitrobiphenyl strongly parallels the initial release of CO₂ from decarboxylation of 2 (Figure 3b). In contrast, the nitrobenzene generation follows shortly after CO₂ formation and likely results from decarboxylation of the copper(I) that is formed in situ. Taken together, these data provide strong support that decarboxylation of 2 generates biaryl and copper(I) through a reductive coupling pathway.

Figure 3.

ReactIR, (a) UV-visible and (b) HPLC time course data for the decarboxylation of 2 (50 mM) in DMSO at 140 °C. In both plots, the CO₂ evolution measured by ReactIR headspace analysis is shown as open circles (○)

Autocatalytic Behavior of Copper(I).

The sigmoidal profile of the time course of decarboxylation of 2 suggested autocatalytic behavior.³⁶ Given the observed formation of copper(I) from the decarboxylation of 2, we wondered whether copper(I) might act as a catalyst in the decarboxylation of 2. Thus, the decarboxylation kinetics of 2 were measured in the presence of a catalytic amount of copper(I) (10 mol% CuCl and phen, Figure 4). Under these conditions, the reaction time course retained the sigmoidal profile and there was no change in the maximum rate (13.67 mM min⁻¹, Figure S40c), yet there was a dramatic reduction in the initial lag phase (from ~8 min to ~4 min). These data suggest an autocatalytic pathway for the decarboxylation of 2 in which copper(I), and possibly complex 1, is generated and subsequently catalyzes the decarboxylation (Scheme 5).

Figure 4.

CO₂ evolution measured by React IR headspace analysis for the decarboxylation of 2 in DMSO (50 mM) at 140 °C (black circles, ●) and with the inclusion of 10% copper power and phen (blue diamonds, ◆) and 10% CuCl and phen (red squares, ■)

Scheme 5.

Proposed pathway for the decarboxylation of complex 2.

Based on the proposed two-electron reductive decarboxylation of 2 (Scheme 4, Pathway B) and autocatalytic behavior of CuI, we expected biaryl formation to occur with generation of 1, and thus show a second-order dependence on [2]. Instead, analysis of the rates of formation of 2,2’-dinitrobiphenyl under conditions of varying [2] revealed a first order dependence on [2], suggesting intramolecular formation of the biaryl from a single copper(II)-bis(carboxylate) species (Scheme 5, Figures S26, S31, and S39). Such a reaction step would generate copper(0), not copper(I).

The formation of copper(0) is observed at later reaction times (~15–20 min), yet there is no visible formation of copper(0) at early reaction times.³⁷ Given these visual observations we supposed that at early reaction times copper(0) may be a short-lived intermediate that undergoes rapid comproportionation with copper(II) to generate copper(I) or that copper(0) may serve to accelerate the decarboxylation of 2 on its own. To explore these possibilities, the decarboxylation kinetics of 2 were measured in the presence of a catalytic amount of copper(0) (10 mol% Cu powder and phen). Under these conditions, the reaction time course retained the sigmoidal profile and there was no change in the maximum rate (12.91 mM min⁻¹, Figure S41d), however a dramatic reduction in the initial lag phase from ~8 min to ~4 min was observed (Figure 4). Notably, the induction periods were nearly identical in the presence of copper(I) and copper(0), suggesting that these two copper sources may access the same autocatalytic species. This result, combined with the first order dependence of biaryl formation on [2], is consistent with a pathway in which 2 undergoes a reductive decarboxylation step to generate biaryl and copper(0).

Overall, these data combined support a pathway in which a reductive decarboxylation of 2 generates 1 and biaryl, via initial formation of copper(0) followed by rapid comproportionation with 2. This step accounts for both the observed formation of 1 and biaryl as well as the first order dependence of biaryl formation on [2]. The newly formed complex 1 is an efficient catalyst for the decarboxylation of 2 leading to the sigmoidal time course kinetics observed. Once 2 is consumed, decarboxylation of 1 proceeds with generation of primarily arene (see Figure 3 above). Although other copper(I) species may also catalyze the decarboxylation of 2 (for example, complex 3, described below) and other pathways are plausible (see SI section VIII) the combined data for the decarboxylation of 2 in the presence and absence of additives are consistent with this model.

Formation and Isolation of [(phen)₂Cu^I][2-nitrobenzoate] (3).

Based on the observed reduction of copper(II) to copper(I) described above, we posited complex 1 as a reasonable intermediate that would form from 2 via decarboxylation. We sought to observe this species spectroscopically during the decarboxylation of complex 2. When complex 2 is heated at 140° C in DMSO-d₆ and monitored by ¹H NMR spectroscopy the resonances in the aromatic region sharpen, suggestive of the formation of copper(I). However, no signals corresponding to 1 are observed. Instead, the spectra indicate the formation of a single benzoate-containing species (3) with characteristic resonances at 7.31, 7.46, 7.51, and 7.55 ppm. The same signals for 3 are also observed when the decarboxylation of complex 1 is monitored by ¹H NMR spectroscopy (Figure 5b) suggesting 3 to be formed in both decarboxylation reactions.

Figure 5.

Complex 3 (blue circles, ●)observed in situ and isolated from the decarboxylation of 1 (125 mM) in DMSO-d₆ with H₂O (0.37 M) at 140 °C. Peaks arising from nitrobenzene are indicated with red squares (■). Phen resonances are unlabeled and are broadened due to exchange.

We considered several possible assignments of 3. For example, the decarboxylation of complex 1 is expected to form a copper(I)-aryl intermediate,^4,11,14 however the spectroscopic features of complex 3 are inconsistent with the previously prepared (phen)Cu(2-nitrophenyl).²⁴ We also considered [(phen)₂Cu^I][Cu^I(2-nitrobenzoate)₂] as a possible assignment of 3. Related phenanthroline-ligated copper(I) species have been demonstrated to exist as equilibrium mixtures of the neutral and ionic forms.³⁸ While conductivity measurements confirmed the presence of an equilibrium mixture of 1 and an ionic species (13.1 Ω⁻¹ mol⁻¹ cm² for 1, Table S35), the spectroscopic features of the independently prepared tetrabutylammonium copper(I) salt [NBu₄][Cu(2-nitrobenzoate)₂] were inconsistent with those of 3. Instead, complex 3 was identified to be [(phen)₂Cu][2-nitrobenzoate].

Complex 3 was isolated from the decarboxylation of complex 1. Under these conditions, nitrobenzene is formed as the predominant byproduct allowing for more facile isolation and purification. After heating complex 1 in DMSO-d₆ at 140° C for 10 min (Figure 5b), the volatiles were removed and the presence of 3 was confirmed spectroscopically (Figure 5c). Recrystallization from CH₂Cl₂/pentane provided single crystals of complex 3. The ¹H NMR spectrum of the resulting crystals contain the same characteristic benzoate resonances as those observed during decarboxylation of 1 and 2, confirming this species to be 3 (Figure 5d). In freshly crystallized samples the resonances of 3 are sharp and reveal a 2:1 ratio of phen:benzoate indicating a complex in which the two phen ligands are equivalent and each phen is symmetrical. Crystallographic characterization confirms these structural features in the solid state and reveals 3 to be [(phen)₂Cu][2-nitrobenzoate] (Figure 5).

A ¹H NMR pattern similar to that of 3 could also be generated from the addition of phen to copper(I)-(2-nitrobenzoate) in DMSO-d₆. Based on this result, complex 3 was prepared independently. Upon addition of a solution of copper(I)-(2-nitrobenzoate) to a solution of phen in CH₃CN, the color changed to dark reddish-brown, and a precipitate formed. Complex 3 was isolated from the reaction mixture in 70% yield as a dark purple solid (Scheme 6).

Scheme 6.

Synthesis of complex 3

Finally, to confirm a role for 3 in the decarboxylation reactions of 1 and 2, complex 3 was heated in DMSO-d₆ and the resulting solution monitored by ¹H NMR spectroscopy. Under these conditions, decarboxylation of 3 occurred readily to form nitrobenzene as the dominant product (62% yield) with only small amounts of biaryl observed (12% yield) after 150 min (Figure 6a, Figure S43). The time course for decarboxylation of 3 was also measured by ReactIR headspace analysis revealing the decarboxylation of 3 to proceed with a rate slower than that of 1 (1.29 mM min⁻¹ at [3]_o = 50 mM, Figures 6b and S46). This result is inconsistent with a pathway in which decarboxylation of 1 proceeds via generation of 3, and instead indicates 3 to be an off-pathway species and not a direct intermediate in the formation of nitrobenzene and biaryl.

Figure 6.

(a) The stoichiometric decarboxylation and (b) corresponding CO₂ evolution time course data of complexes 1 (blue circles, ● 50 mM) and 3 (red squares, ■ 50 mM) in DMSO (2 mL) at 140 °C under N₂. Each time course is an average of two runs and the error bars are given in the supporting information. The traces are not fits and are included to guide the eye.

Inhibition by 1,10-Phenanthroline.

The slower rate of decarboxylation of complex 3 as compared with 1, suggests an inhibitory role of phen on the rate of decarboxylation. Inhibition by phen is surprising given the literature examples of copper-catalyzed decarboxylative coupling reactions in which a 2:1 phen:Cu ratio outperforms 1:1 phen:Cu conditions.^17c,17f,18d Thus, we sought to more directly probe the effect of phen on the rate of decarboxylation. The decarboxylation of the unligated CuI(2-nitrobenzoate) complex 4 was monitored in the presence of varying equivalents of phen (0, 1 and 2 equiv) and the reaction progress and decarboxylation rates compared (Figure 7). Decarboxylation is most rapid in the presence of 1 equivalent of the phen supporting ligand (rate = 6.67 mM min⁻¹, Figure S49) and an additional equivalent of phen leads to a reduction in decarboxylation rate (1.85 mM min⁻¹ in the presence of 2 equiv phen, Figure S50). Additionally, the rate of decarboxylation in the presence of 2 equiv of phen is similar to that measured for complex 3 (1.29 mM min⁻¹, Figure S46). In contrast, the decarboxylation in the presence of 1 equiv of phen (6.67 mM min⁻¹, Figure S49) is more rapid than that measured for complex 1 (4.4 mM min⁻¹, Figure S6) and may suggest slow complexation of copper(I) with phen. Overall, these data support an equilibrium binding of phen with 1 to generate 3, which undergoes more sluggish decarboxylation than 1.

Figure 7.

The stoichiometric decarboxylation of complex 4 (50 mM, gray diamonds, ◆) in the presence of 1 equiv (50 mM, blue circles ●) and 2 equiv of phen (100 mM, red squares ■) as well as the decarboxylation of 1 (50 mM, open blue circles ○) and 3 (50 mM, open red squares □). All reactions were run in DMSO (2 mL) at 140 °C under N₂. Each time course is an average of two runs and the error bars are given in the supporting information. The traces are not fits and are included to guide the eye.

Synthesis and Characterization of Substituted Copper(II) Benzoate Complexes.

Twelve additional phen-ligated copper(II)-bis(R-benzoate) complexes were synthesized, each with varying substitution on the benzoate ligands. The (phen)copper(2-nitrobenzoate)₂ complex (2) was previously synthesized by our group from the slow addition of sodium 2-nitrobenzoate to an ethanolic solution of copper(II) triflate and phen to yield a blue crystalline solid (Scheme 7, Route A).² This synthetic protocol, however, could not be extended to the preparation of other copper(II) benzoates due to the co-precipitation of residual copper(II) triflate. Instead, the complex bearing 2-fluorobenzoate ligands (5) was prepared from the treatment of copper(II) acetate with 2-fluorobenzoic acid and phen in aqueous tetrahydrofuran and isolated as blue crystals in 85% yield (Scheme 7, Route B). Complexes 6–16 were synthesized from freshly prepared copper(II) hydroxide, benzoic acid and phen in aqueous ethanol in a modification of a procedure reported by Xia (Scheme 7, Route C).³⁹ The complexes were isolated as green, blue or purple powdery solids in moderate to good yields (51–93%, Scheme 7).

Scheme 7.

Synthesis of Substituted (phen)Cu(R-benzoate)₂ Complexes.

The full series of phenanthroline-ligated copper(II) benzoate complexes (2, 5–16) were characterized using UV-visible and infrared spectroscopic techniques, elemental analyses and, where possible, X-ray crystallography. The vibrational spectra of the complexes feature characteristic asymmetric (ν_asym(CO₂⁻)) and symmetric (ν_sym(CO₂⁻)) stretching vibrations of the carboxylate group around 1520–1640 cm⁻¹ and 1300 cm⁻¹, respectively (Table S40). For each complex, the presence of two stretching bands that are red-shifted from that of the free benzoic acid (around 1680 cm⁻¹, Table S40) is indicative of coordinated benzoate ligands. Additionally, complexes 2 and 5-16 all feature Dn(CO₂⁻) values between 170 and 283 cm⁻¹, most consistent with κ¹-bound carboxylates in the solid state (Δn(CO₂⁻) = ν_asym(CO₂⁻) - ν_sym(CO₂⁻), Table S40).^40,41 The coordination of the phen ligand is supported by the aromatic C-H bending frequencies around 850 and 720 cm⁻¹.⁴² Upon coordination, the C=N and C=C stretches are red-shifted from those of free phen (1588–1621, 1501, and 1422 cm⁻¹)⁴² and are likely overlapping with the (ν_asymCOO⁻) bands as a result.

The absorption spectra of the complexes exhibit broad absorption bands around 700 nm consistent with the weak d-d transitions of Cu^II complexes existing in square planar or tetragonally distorted coordination environments (Table S41).⁴³

X-ray quality single crystals of complexes 6, 7, 9, 11, 12 and 16 were obtained from the slow evaporation of the reaction filtrates over several weeks. The crystallographic data confirmed the incorporation of one phen and two benzoate ligands for each copper(II) center in all complexes. The X-ray analyses highlight the structural diversity of these complexes in the solid state, with some adopting monomeric forms (6, 9, 11 and 16) and others crystallizing as dimers with bridging benzoate ligands (7 and 12). For all complexes, the Cu-O bond lengths are between 1.91–2.18 Å and the Cu-N bond lengths are between 1.98 and 2.03 Å. Similarly, the O-Cu-O bond angles are all between 91.69 and 97.92°.³⁷ Additionally, some of the crystal structures feature the incorporation of a coordinated water (6, 9, and16), consistent with the presence of an O-H stretching band centered around 3400 cm⁻¹ in the IR spectra.^43a The structures of the complexes as well as selected bond lengths and angles are included in Figure 8 and Table S42.

Figure 8.

Solid state structures of selected (phen)CuII(R-benzoate) complexes: (a) 6•H₂O, (b) 7₂, (c) 9•H₂O, (d) 11, (e) 12, and (f) 16•H₂O with thermal ellipsoids drawn at 50% probability. Hydrogen atoms have been omitted for clarity, with the exception of those on water molecules (in 6•H₂O, 9•H₂O, and 16•H₂O). For complex 7, the structure of [(phen)₂Cu₂(μ,κ¹–2-methoxy-benzoate)₂(μ,κ²–2-methoxybenzoate)][2-methoxybenzoate] (7₂) is shown with one half of the dimeric unit, [(phen)Cu(2-methoxybenzoate)₂] (7), grayed out for clarity and emphasis. For complex 12, the unit cell, which is one half of the dimeric unit, is shown. For complex 16, one position of the nitro group is shown. Atom colors are shown as Cu = gray, C = black, H = white, O = red, N = blue, F = green, S = yellow. Full structures and metric data for complexes 6•H₂O, 7₂, 9•H₂O, 11, 12 and 16•H₂O are provided in the supporting information.

Decarboxylation of Copper(II) Benzoate Complexes.

The decarboxylation reactions of complexes 5-16 were monitored by ¹H NMR spectroscopy in DMSO-d₆ to probe the possible formation of a reaction intermediate analogous to 3. In these reactions, new species with spectroscopic features similar to those of 3 were formed and consumed during the reaction, consistent with the formation of the corresponding [(phen)₂Cu^I](R-benzoate) intermediates.

The kinetics of decarboxylation of complexes 5-16 were also monitored with ReactIR gas cell measurements. All complexes displayed the same sigmoidal reaction profile as observed with complex 2, above (Figure 9a). The maximum rate was obtained from the first derivative of [CO₂]_t and the induction period was defined as the time at which the initial rate and the maximum rate intersect (Figure 9b).^36a Each of these quantities was found to vary with the benzoate substitution (Figure 9, Tables S57 and S58). In general, electron-withdrawing substituents on the benzoate led to reduced induction periods and enhanced reaction rates, consistent with the reduction of copper(II) during both the initial reaction times as well as during the autocatalytic decarboxylation. The three-parameter Fujita-Nishioka correlation^27a revealed steric and resonance contributions to be minor (Figures S64a and S78a). Instead, the major influence on both the induction period and the maximum rates arises from the field effect (F) of the ortho-substituent (Figure 9c and 9d). We have previously identified the F value, a measurement of proximity polar effects,²⁷ to govern the rate of decarboxylation of well-defined silver-carboxylate complexes.^26,44 These data indicate that this substituent feature also governs reactivity in copper(II)-mediated decarboxylations. This finding supports the importance of the field effect in decarboxylation reactions beyond those that are silver-mediated and may provide opportunities to further understand and surpass the existing substrate limitations.

Figure 9.

(a) Kinetic profiles for the decarboxylation of substituted copper(II)-benzoates complexes (2 and 5–16, 50 mM) in DMSO at 140 °C under argon. (b) Graphical depiction of the maximum rate and induction period determination. Correlation of (c) the ln(k) and (d) the ln(induction period) with the field effect parameter (F).

CONCLUSION

In summary, this work describes an experimental mechanistic study of the decarboxylation of copper(II) benzoates. The decarboxylation of copper(II)-bis(benzoate) species was found to generate the homocoupled biaryl product and a copper(I) benzoate. The sigmoidal reaction profile paired with the reduction of the lag phase in the presence of copper(I) and copper(0) additives support an autocatalytic pathway in which the initially generated copper(I) species is capable of catalyzing the decarboxylation of the copper(II) benzoates. This same copper(I) benzoate also undergoes a redox-neutral decarboxylation to generate the arene as a protodecarboxylation product. Further kinetic studies reveal an inhibitory effect of phen on the decarboxylation at copper(I), in contrast to literature studies which report enhanced reactivity in the presence of phen.

In these studies, we have also evaluated the role of benzoate substitution on the rate of decarboxylation from copper(II)-benzoates and demonstrated the field-effect parameter to be the dominant factor influencing decarboxylation in these systems. Overall, these findings provide valuable insight into the mechanistic complexity of the decarboxylation of copper(II) benzoates and the structural features that enhance and hinder decarboxylation. We anticipate that this work will further enable the development of efficient copper-catalyzed oxidative decarboxylative coupling reactions.

EXPERIMENTAL SECTION

General Procedure for Decarboxylation Kinetics.

An oven-dried 2-necked 15-mL pressure tube was charged with either complex 1 or complex 2 (100 μmol) and a magnetic stir bar. Reaction vessels containing complex 1 were prepared in a N₂-filled glovebox, while those containing complex 2 were prepared on the benchtop. The container was sealed with a greased rubber septum, which was secured with electrical tape. The pressure tube was connected to a greased Teflon adapter which was connected to a T-junction that holds a Vernier Pressure Sensor 400, an inlet for argon gas, and outlet leading to the gas cell. After checking the setup for leaks by monitoring a passive vacuum in the pressure tube, the tube was evacuated and backfilled with argon three to five times. Anhydrous DMSO (2.0 mL, stored in a N₂-filled glovebox) was added to the reaction flask under a positive pressure of Ar with a syringe that was kept under air-free conditions. The headspace was equilibrated with the evacuated gas cell (under static vacuum) and refilled with Ar three to five times. The ReactIR spectrometer was set up as described below with data scans collected every 2 min. The reaction vessel was stirred at room temperature and after about 15 seconds of the flask headspace equilibration with the evacuated gas cell, the ReactIR data collection was initiated. Two additional room-temperature headspace samples were taken, after which the flask was immersed into a silicone oil bath at 140 °C and the headspace sampling commenced.

General Procedure for the Measurement of CO₂ Evolution by ReactIR.

Decarboxylation kinetics were measured by IR analysis of the headspace to quantify CO₂ evolution over time. CO₂ absorptions were recorded at room temperature on a ReactIR^™ ic15 equipped with a CRV Gas Cell using iC IR 7.1 software. Spectra were obtained under the following conditions: resolution = 16 cm⁻¹, gain = 1X (normal), 512 scans and spectral range = 650–3000 cm⁻¹. The optical path outside the gas cell was continuously purged with a flow of UHP-grade argon. A diagram of the experimental set up is provided in Figure S3.

Prior to each experiment, the gas cell and lines were evacuated for about 1 h, and the instrument was equilibrated for about 1 hour. To obtain a headspace sample for either a calibration or reaction monitoring data point, the gas-cell valve leading to the vacuum pump was closed and the connections to the reaction vessel were opened for 5 seconds. Data collection began after an additional 10 seconds of wait time during which the gas sample resided in the gas cell. The reaction flask was re-filled with UHP-grade argon after each headspace sampling step. After the completion of the scans, the gas cell and lines were put under vacuum until the next sampling step. The pressure in the line was continuously measured using a Vernier Pressure Sensor 400 and Logger Lite software.

Instrument control and raw data processing was achieved using the iC IR 7.1 software. Data processing included a baseline correction (offset at ~2000 cm⁻¹). The areas of the peak corresponding to CO₂ were obtained by using a two-point baseline (baseline = 2410 to 2220 cm⁻¹, peak location = 2410 to 2220 cm⁻¹), before being exported into Excel for data analysis. Additional details on data collection and analysis can be found in the Supporting Information.

Supplementary Material

The supporting information is available free of charge at https://pubs.acs.org

Experimental procedures and characterization data for all new compounds (PDF)

X-ray crystallographic data for 6, 7, 9, 11, 12, and 16