Cu(II) carboxylate arene C─H functionalization: Tuning for nonradical pathways.

We report carbon-hydrogen acetoxylation of nondirected arenes benzene and toluene, as well as related functionalization with pivalate and 2-ethylhexanoate ester groups, using simple copper(II) [Cu(II)] salts with over 80% yield. By changing the ratio of benzene and Cu(II) salts, 2.4% conversion of benzene can be reached. Combined experimental and computational studies results indicate that the arene carbon-hydrogen functionalization likely occurs by a nonradical Cu(II)-mediated organometallic pathway. The Cu(II) salts used in the reaction can be isolated, recycled, and reused with little change in reactivity. In addition, the Cu(II) salts can be regenerated in situ using oxygen and, after the removal of the generated water, the arene carbon-hydrogen acetoxylation and related esterification reactions can be continued, which leads to a process that enables recycling of Cu(II).

INTRODUCTION

As an important industrial chemical, phenol has been deployed in applications in various fields including household products, nylon, polymers, and the manufacturing of other chemical derivatives. Over 12 million metric tons of phenol are produced each year, and greater than 90% of phenol is produced through the Hock process (–). However, a major issue with the Hock process is the generation of a stoichiometric amount of acetone as an oversupplied side product (). Alternative methods for phenol production include a three-step route from benzene to phenol with cyclohexanone as a coproduct developed by ExxonMobil and the oxidation of benzene using N2O developed by Solutia (–). However, the source of N2O gas limits the location and amount of phenol production for the Solutia process, while neither of the other methods provide a desirable coproduct-free route for phenol production (Fig. 1). Other related benzene hydroxylation processes using copper rely on the formation of hydroxyl radical and/or require solid supports to form heterogeneous catalysts (, –), which suffer from low selectivity and/or low yield.

Fig. 1.

Current methods for phenol production and the proposed method via Cu(II)-mediated acetoxylation.

NHPI, N-hydroxyphthalimide.

Acetoxylation of benzene has been considered as a potential route for phenol production since hydrolysis of phenyl acetate generates phenol (). Palladium-catalyzed benzene acetoxylation has been studied (–), but only a limited number of examples of aromatic C─H acetoxylation/hydroxylation have been reported using low-cost transition metals, and these examples rely on functionalized aromatic substrates with “directing groups” to facilitate coordination to the catalyst (–). Yu and co-workers () demonstrated a series of C─H functionalizations using pyridine as the directing group with copper(II) acetate as the catalyst precursor, and the reaction was proposed to occur by the formation of a radical cation intermediate via a single electron transfer mechanism. Later, the Shi group () reported a study using a bidentate directing group derived from 2-(pyridine-2-yl)isopropylamine through a proposed Cu(III)-aryl intermediate. Subsequently, the Yu group () demonstrated a related C─H hydroxylation process using oxazolyamide as the directing group via a possible monomeric Cu(III)-aryl intermediate. Benzoic acid has also been reported as a directing group for the synthesis of hydroxylated arenes assisted by benzoyl peroxide (). Other related C─H functionalization processes by Cu(II) also require a directing group to facilitate the reaction (–).

To our knowledge, there are no reported examples of molecular Cu(II)-mediated arene acetoxylation without directing groups on the arene. Here, we report acetoxylation of benzene and toluene using simple Cu(II) salts, and we propose that the C─H functionalization occurs by a nonradical organometallic pathway that forms Cu-aryl intermediates under certain conditions. The Cu(II) salt offers recyclability using O2 or air, similar to the commercialized Pd-catalyzed ethylene oxidation Wacker process that has proven viable with Cu(II) recycling both in situ (with purified O2) and in a separate step/reactor from the ethylene oxidation process (using air) (–). This process can potentially lead to a coproduct-free synthetic route for phenol production (Fig. 1), since all of the coproducts are used for recycling of the Cu(II) salt, and, thus, the overall reaction is the conversion of benzene and dioxygen to phenol (Fig. 1).

RESULTS

Development of nondirected Cu(II)-mediated C─H acetoxylation of benzene

Phenyl acetate has been observed as a side product in our previously reported Rh-catalyzed arene alkenylation that incorporates Cu(OAc)2. Phenyl acetate formation is the result of a reaction of Cu(OAc)2 with benzene (). After initial optimization of conditions (sections S2 and S3), we found that heating (180°C) 10 ml of a benzene mixture with predried Cu(OAc)2 [0.48 mole percent (mol %) relative to benzene] under dinitrogen [75 psig (pound-force per square inch gauge)] produces phenyl acetate in 87(1)% yield based on the Cu(II)-limiting reagent as determined by gas chromatography with flame ionization detection (GC-FID) (Table 1, entry 2). The control experiment using CuCl2 showed no activity for C─H activation on benzene (Table 1, entry 12), which likely indicates that the carboxylate group is essential for arene C─H activation (see below). Similar to Cu(OAc)2, more soluble Cu(II) salts, such as Cu(II) pivalate [Cu(OPiv)2] and Cu(II) 2-ethylhexanoate [Cu(OHex)2], were shown to achieve similar reactions at even a lower temperature (Table 1, entries 6, 7, and 9). The Cu(II) species Cu(OPiv)2 was found to give a lower yield compared to the other Cu(II) salts, which we attribute to the lower thermostability of the pivalic group (section S2.5). The presence of air was found to inhibit the reaction (Table 1, entry 4), and the reaction was slowed down with the addition of an increasing amount of carboxylic acid (Table 1, entries 3, 5, 8, 10, and 11). Different ligands and solvents were tested for the reaction; however, none of them provided a beneficial effect (sections S5.1 and S6). These observations are distinct from the previously published directed Cu(II)-mediated acetoxylation processes (, –), in which the presence of acid, air, or ligands is commonly required by the reactions.

Table 1.

Selected results for benzene acetoxylation and related reactions using Cu(II) salts.

Yield of the phenyl carboxylate is relative to the limiting reagent (CuX2) as determined by GC-FID. The theoretical maximum conversion of benzene will be 0.24%. Reaction conditions: benzene (10 ml, 112.3 mmol), CuX2 (0.48 mol %, 0.540 mmol), 75 psig of N2. CuX2 = Cu(OAc)2 anhydrous; Cu(OPiv)2 or Cu(OHex)2. N.D., not detected. SDs were calculated from at least three independent experiments.

Entry	Cu(II) salt	Temp. (°C)	HX (equiv.)	Air	Yield (%)	Time (hours)
1	Cu(OAc)2	180	0	None	78(0)	24
2	Cu(OAc)2	180	0	None	87(1)	40
3	Cu(OAc)2	180	1.0	None	80(1)	40
4*	Cu(OAc)2	180	0	1 atm	15(2)	24
5	Cu(OAc)2	180	4.0	None	65(6)	48
6	Cu(OPiv)2	180	0	None	49(1)	16
7	Cu(OPiv)2	170	0	None	52(0)	16
8	Cu(OPiv)2	170	1.0	None	36(1)	16
9	Cu(OHex)2	170	0	None	89(1)	24
10	Cu(OHex)2	170	1.0	None	71(1)	40
11	Cu(OHex)2	170	4.0	None	37(0)	72
12†	CuCl2	180	0	None	N.D.	24

*Using 1 atm of air, the total top pressure of air and N2 is 75 psig.

†Using half the amount of reagents in a custom made high-pressure steel reactor with Swagelok Vacuum Coupling O-ring (VCO) seal fitting reactor under 100 psig of Ar.

Mechanistic studies

To differentiate between radical and nonradical pathways, separate reactions of toluene or a mixture of benzene with cyclohexane were used as probes. These reactions offer intramolecular and intermolecular competition of weaker sp3 C─H bonds and stronger sp2 C─H bonds (Fig. 2, A and B). It is expected that a radical C─H abstraction reaction will select for the weaker sp3 C─H bonds. Under anhydrous conditions, all of the reactions were found to select for the stronger sp2 C─H bonds, which suggests that the reactions likely undergo a nonradical reaction pathway. Using 1 ml of benzene with 9 ml of cyclohexane resulted in ~10% of the rate compared to the reaction using neat benzene, consistent with the reaction being first order in benzene. A common radical trap, (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO), was tested as an additive in the reaction of toluene with Cu(OAc)2 (Fig. 2C). The reactions with TEMPO were slow and remained selective for tolyl acetates, but more benzyl acetate was formed compared to reactions without TEMPO. In a control experiment, we found that TEMPO can functionalize the benzylic position of toluene without Cu(II) salt (Fig. 2D). When using toluene as the substrate, the regioselectivity of the sp2 C─H bonds can be used to differentiate between a potential electrophilic aromatic substitution mechanism (ArSE, with no M─C bond formation) and a metal-mediated C─H activation mechanism via M─C bond formation (–). The Cu(II)-mediated reaction was found to be more selective for the meta-tolyl carboxylate than the ortho-tolyl carboxylate (Table 2, entries 1, 4, and 5), which we propose (see below) is a result of an organometallic reaction that forms Cu-tolyl intermediates. We propose that the Cu-mediated toluene C─H activation favors the meta-position over the ortho-position of toluene due to the steric effect of the methyl group.

Fig. 2.

Differentiate between radical and nonradical pathways.

(A) Intermolecular competition of sp2 versus sp3 C─H bonds using a mixture of benzene and cyclohexane. N.D., not detected. (B) Intermolecular competition of sp2 versus sp3 C─H bonds using toluene. (C) Meta-tolyl acetate and benzyl acetate yield versus time plot of reaction using different amounts of TEMPO. Reaction conditions: toluene (10 ml, 94.1 mmol), Cu(OAc)2 (98 mg, 0.540 mmol), TEMPO [0, 0.2, 0.5, and 1.0 equiv. relative to Cu(OAc)2], under 75 psig of N2 at 180°C. (D) Observed products of control experiment using TEMPO with no Cu(II) salt. Reaction conditions: toluene (10 ml, 94.1 mmol), TEMPO (71 mg, 0.452 mmol), under 75 psig of N2 at 170°C. SDs were calculated from at least three independent experiments.

Table 2.

Selected results for toluene acetoxylation and related reactions using simple Cu(II) salts.

Yield of the tolyl or benzyl carboxylate is relative to the limiting reagent (CuX2). Reaction conditions: toluene (10 ml, 94.1 mmol), CuX2 [for Cu(OAc)2, 0.57 mol %, 0.540 mmol; for Cu(OPiv)2 or Cu(OHex)2, 0.48 mol %, 0.452 mmol), under 75 psig of N2. CuX2 = Cu(OAc)2 anhydrous, Cu(OPiv)2, or Cu(OHex)2. SDs were calculated from at least three independent experiments.

Entry	CuX2	Temp. (°C)	Additive (equiv.)	Yield (%)					Time (hours)
				Total	Benzyl	Para	Meta	Ortho
1	Cu(OAc)2	180	None	72(1)	4.0(5)	19(0)	31(1)	18(0)	40
2	Cu(OAc)2	170	None	71(2)	5.0(2)	20(1)	31(1)	15(1)	96
3*	Cu(OAc)2	180	Dried O2	95(6)	89(6)	1(0)	3(0)	2(0)	4
4	Cu(OAc)2	180	H2O (1.0)	37(2)	17(2)	5(0)	10(0)	5(0)	40
5	Cu(OAc)2·H2O	180	None	48(3)	23(0)	7(1)	13(2)	6(1)	96
6	Cu(OAc)2	180	HOAc (1.0)	50(3)	3.0(3)	12(1)	23(2)	11(1)	40
7	Cu(OAc)2	180	Ac2O (1.0)	3(1)	0.3(0)	1(0)	2(0)	1(0)	24
8	Cu(OPiv)2	170	None	33(0)	0.9(1)	10(0)	16(0)	7(0)	16
9	Cu(OHex)2	180	None	54(1)	2.4(2)	16(0)	24(0)	12(0)	8
10	Cu(OHex)2	170	None	72(0)	3.1(1)	18(1)	33(1)	18(0)	24
11	Cu(OHex)2	170	H2O (1.0)	41(3)	8.4(0)	9(1)	16(2)	7(1)	24
12	Cu(OHex)2	170	HOHex (1.0)	68(2)	2.9(1)	19(1)	33(1)	13(0)	40
13	Cu(OHex)2	170	HOHex (2.0)	56(1)	2.8(3)	15(0)	28(1)	10(3)	40
14	Cu(OHex)2	170	HOHex (4.0)	36(1)	2.3(2)	10(0)	19(1)	6(1)	48

*Using 1 atm of dried O2, the total top pressure of O2 and N2 is 75 psig.

In the presence of 1 atm of dried O2, the selectivity of toluene acetoxylation reaction using Cu(OAc)2 changed to form benzyl acetate (Table 2, entry 3), which is evidence for a radical process since the weaker benzyl C─H bond was functionalized. In addition, the inclusion of O2 suppresses reactivity with the stronger sp2 tolyl C─H bonds, similar to our observation when using benzene as the substrate that exhibited lower yield in the presence of O2 (Table 1, entry 4). The selectivity shifts from sp2 C─H bond functionalization to benzylic functionalization is expected since mixing Cu(II) with O2 is known to form Cu-peroxo and/or superoxo species (–), which likely leads to the formation of free radical species that favor weaker sp3 C─H bond functionalization over stronger sp2 C─H bonds. Since hexamethylbenzene (HMB), which has been used as the standard for GC analysis, contains benzylic C─H bonds and the Cu(II)-mediated functionalization of toluene activates benzyl C─H bonds under some conditions, control experiments were done using Cu(OAc)2 and toluene as the substrate at 180°C with in situ and ex situ addition of HMB for GC-FID analysis (fig. S6). As expected, under anaerobic conditions, the consumption of HMB is negligible, while under aerobic conditions, a competing radical-based reaction with HMB was observed, which is consistent with our proposal that reaction pathways are distinct under anaerobic versus aerobic conditions. Therefore, to avoid complications caused by the consumption of HMB, all the reported quantifications used ex situ addition of HMB for GC analysis. Water was also examined as an additive (Table 2, entries 4 and 11), and similar to O2, the presence of 1 equivalent of water (relative to Cu) switches the reaction selectivity for toluene activation toward benzylic C─H functionalization, although tolyl products are still observed (section S11). We suspect that water coordinates to Cu2(μ-OAc)4 to form copper(II) acetate hydrate, Cu2(μ-OAc)4·2H2O, which can convert to monomeric Cu(OAc)2·(H2O) () or other decomposition species such as Cu/Cu2O under reaction temperatures that could lead to a radical process(es) that competes with the organometallic Cu(II)-mediated pathway. The observation of a similar change in reaction selectivity using commercial copper(II) acetate monohydrate [Cu2(μ-OAc)4·2H2O] (Table 2, entry 5) is consistent with the observation of reactions with water added to anhydrous Cu2(μ-OAc)4. The addition of increasing amounts of carboxylic acid (Table 2, entries 6 and 12 to 14) decreases the reaction rate without changing the benzyl-to-tolyl selectivity. This suggests that carboxylic acid inhibits the organometallic reaction pathway but does not facilitate undesirable radical-based processes, which is different from the effect of water or O2.

The presence of bi- or tridentate ligand [e.g., bipyridine, terpyridine, tetramethyl ethylenediamine, and 1,2-bis(dimethylphosphino)ethane] shuts down the benzene functionalization reaction, and many ligands result in a switch of the reaction selectivity for toluene activation toward benzylic C─H functionalization (section S5.1). However, using (MeO)3P═O, which we presume is a weakly coordinating ligand, only slows down the arene functionalization process without changing the sp2-to-sp3 selectivity (Fig. 3A). By refluxing Cu(OAc)2 and (MeO)3P═O in toluene, the resulting product was found to be the bis-Cu complex [(MeO)3P═O]2Cu2(μ-OAc)4. The structure of [(MeO)3P═O]2Cu2(μ-OAc)4 was confirmed by a single-crystal x-ray diffraction study (Fig. 3B). We speculate that (MeO)3P═O does not disrupt the dimeric Cu(II) structure and hence only slows down the functionalization of arenes by competing with the arene substrate for Cu coordination. In contrast, multidentate ligands likely convert bis-Cu(II) to monomeric Cu(II) complexes of the type LCuIIX2 that are not active for the functionalization of arenes (Fig. 4A). By monitoring the initial rate of the reaction using different amounts of soluble Cu2(μ-OHex)4 and 2-ethylhexanoic acid (HOHex), the reaction was found to be first order in Cu2(μ-OHex)4 concentration (fig. S26) and inverse first order in HOHex concentration (fig. S29). On the basis of these observations, we propose a plausible reaction pathway and rate equation (Fig. 4B). Reversible arene C─H activation by Cu2(μ-X)4 (X = OAc, OHex, or OPiv) and dissociation of carboxylic acid are followed by an overall rate-limiting reductive coupling/elimination of aryl ester. This proposed reaction pathway is consistent with the results of density functional theory (DFT) calculations (see below).

Fig. 3.

Effects of weakly coordinating (MeO)3P═O ligands.

(A) Sum of tolyl/benzyl OHex yields and benzyl OHex yield versus time plot with the presence of different amounts of (MeO)3P═O ligands. Reaction conditions: toluene (10 ml, 94.1 mmol), Cu(OHex)2 (0.48 mol %, 0.452 mmol), (MeO)3P═O (0, 0.5, 1.0, and 2.0 equiv. relative to Cu(OHex)2), under 75 psig of N2 at 170°C. SDs were calculated from at least three independent experiments. (B) Oak Ridge thermal ellipsoid plot of {(MeO)3P═O}2Cu2(μ-OAc)4. Ellipsoids are drawn at 50% probability level.

Fig. 4.

Ligand effects and reaction pathway.

(A) Proposed competing pathway in the arene acetoxylation reaction in the presence of weakly coordinating or bidentate ligand. (B) Proposed reaction pathway and associated rate law.

On the basis of the proposed rate equation (Fig. 4B), the rate constant for the C─H activation step (k1) can be calculated via a curve fitting of rate versus [HOHex] (Fig. 5C) or the intercept from a 1/rate versus [HOHex] plot (fig. S31), where rate is the slope of the concentration of tolyl OHex versus time plot (fig. S28). Both plots gave good fits, which is consistent with the proposed reaction pathway. The rate constant for the C─H activation step (k1) and the related activation free energy (ΔG1‡443K) can be estimated from the intercept as 2.7(4) 10−6 s−1 M−1 and 38(1) kcal mol−1 (section S9). A slight induction period was observed for all of the Cu(II)-mediated acetoxylation reactions without addition of carboxylic acid. The reaction rates of the induction period and the initial linear region were monitored at different reaction temperatures using soluble Cu(OHex)2 with toluene as the substrate (fig. S23 and table S11). It was found that a decrease in reaction temperature led to an increase in the induction period, and at 180°C, there is no observable induction period. The induction period of the reaction was also studied in the presence of variable concentrations of HOHex (fig. S24), and with the addition of more acid, the induction period becomes less obvious. We propose that the induction period is due to the process of dissolving Cu(II) salt under acid-free conditions.

Fig. 5.

Mechanistic studies.

(A) Measurements of KIE using parallel reactions and intermolecular competition experiments. (B) H/D exchange experiment using acetic acid-d1 (DOAc) as the deuterium source. (C) Curve fitting of observed reaction rates (rate) versus HOHex concentration plot for 2-ethylhexanoation of toluene using Cu(OHex)2 with the addition of different amounts of HOHex at 170°C. (D) Eyring plot for toluene functionalization using Cu(OHex)2 calculated with kobs (in black) and k1 (in red). Reaction rates were measured for production of tolyl OHex over the temperature range 150° to 180°C under the conditions with addition of variant amounts of HOHex. Rate equation: rate = k″obs[toluene][Cu dimer]. The rate constant k1 was calculated using curve fitting of observed reaction rates versus HOHex concentration.

To gain insight into the Cu(II)-mediated arene C─H activation step, we studied the kinetic isotope effect (KIE) using C6H6 and C6D6 as the substrates with soluble Cu(OHex)2 at 170°C. The KIE value was determined to be 3.0(1) for parallel reactions using benzene and benzene-d6 and 2.8(1) for the intermolecular competition reactions using a mixture of benzene and benzene-d6 with a 1-to-1 molar ratio (Fig. 5A). This type of KIE has typically been interpreted as C─H cleavage that occurs during or before the rate-determining step (–), which is consistent with the proposed reaction pathway (Fig. 4B). In the presence of eight equivalents of DOAc [relevant to Cu(OAc)2], a small amount of deuterium incorporation (3.0(3)%) was found in the product (Fig. 5B), which likely indicates that the C─H activation step is partially reversible and is also consistent with the proposed reaction pathway (Fig. 4B).

An Eyring analysis was performed using reaction rates measured at seven temperatures between 150° and 180°C (Fig. 5D, in black), which provided an activation enthalpy (ΔH‡) of 38(1) kcal mol−1. The activation entropy (ΔS‡) was found to be −1(3) cal mol−1 K−1 assuming an overall second-order rate law (rate = k″obs[toluene][Cu2(μ-OHex)4]), which indicates that the entropy has a negligible effect on the overall reaction. The small entropy of activation is potentially due to the use of excess toluene as the solvent and/or that the C─H activation step may only be partially rate limiting (see DFT calculations below). Using an Arrhenius plot (fig. S47), the activation energy (Ea) was determined to be 39(1) kcal mol−1. As described above, the rate constant of the C─H activation step (k1) can be determined; therefore, a set of reactions has been done over the temperature range of 150° to 180°C under the conditions with addition of variant amounts of HOHex to calculate k1 at different temperatures (table S25). The Eyring analysis performed using k1 at different temperatures as it shown in Fig. 5D (in red), and these data indicate a ΔH‡ of 29(2) kcal mol−1 and a ΔS‡ of −19(4) cal mol−1 K−1. These activation parameters are consistent with the proposed mechanism for arene C─H activation and the results of computational modeling (see below).

Computational modeling

To further understand the reaction mechanism, we used DFT and time-dependent (TD)–DFT calculations to examine closed-shell (C─H activation) and open-shell (e.g., electron transfer, hydrogen atom transfer, and proton-coupled electron transfer; section S16) reaction pathways. Both dinuclear Cu2(μ-OAc)4 and mononuclear Cu(OAc)2 models were examined for each pathway. Figure 6 presents unrestricted MN15L () functional results. This method was chosen on the basis of its general accuracy for thermodynamics, barriers, and spin states for first-row transition metals. While there are no overall different mechanistic conclusions using alternative functionals [e.g., M06L ()] and wave function methods [DLPNO-CCSD(T) (–)], the relative energies of structures do change (see the Supplementary Materials, Section S28 for comparison table S36).

Fig. 6.

Computational modeling of possible reaction pathways.

(A) UMN15L/Def2-TZVPP//UMN15L/Def2-SVP calculated relative enthalpies and Gibbs energies (in kilocalories per mole) at 170°C of the calculated reaction pathways with Cu2(μ-OAc)4 and Cu(OAc)2 monomer. (B) Relative energies of ortho, para, and meta C─H activation transition states.

Using the results of DFT calculations in combination with our experimental results (see above and the Supplementary Materials), we propose the reaction mechanism outlined by the calculated energy landscape for the para position of toluene shown in Fig. 6A (in black). In an initial step, Cu2(μ-OAc)4 (1) coordinates and activates a C─H bond of toluene (we modeled the para C─H bond) in a one-step metalation process to form (p-tolyl)CuII(HOAc)(μ-OAc)2CuII(κ2-OAc) (2) with a calculated Gibbs activation energy of 41.4 kcal mol−1 (at 170°C). This C─H activation step involves concerted formation of the Cu–p-tolyl bond with simultaneous proton transfer to the acetate ligand in a pathway (TS1). The structure TS1 shown in Fig. 6A is similar to many other transition states reported for arene C─H activation by metal acetate complexes and is often called concerted metalation deprotonation (). The energies for TS1 at the ortho- and meta-positions of toluene are very similar (within 1.6 kcal mol−1 or closer) to the transition state energy for para C─H activation, which is consistent with the very small experimental selectivity (see Fig. 6B).

Dissociation of HOAc precedes C─O reductive coupling between tolyl and acetate ligands (TS2), which occurs with a calculated Gibbs activation energy of 39.2 kcal mol−1 and forms (tolyl acetate)Cu2(μ-OAc)2 (5). TS2 has a typical three-centered reductive coupling/elimination geometry where the Cu─OAc and Cu─Ar bonds are broken with simultaneous formation of the new C─O bond. Dissociation of the tolyl acetate from 5 gives the final organic product. Inspection of structure and bonding after HOAc dissociation and during the reductive elimination step suggests that this process occurs through a Cu(III)/Cu(I) intermediate rather than a Cu(II)/Cu(II) intermediate. This leads us to conclude that the second Cu atom is important in facilitating the C─O reductive coupling step by an intramolecular oxidation, a proposal that is consistent with calculated energetics using the monomeric Cu(OAc)2 (7) species (see below and Fig. 6A in blue).

Figure 6A also shows a comparison of the dinuclear Cu pathway to arene activation by monomeric Cu(OAc)2 (7), which is potentially viable because the Gibbs energy for formation of the monomer only requires 1.1 kcal mol−1 relative to Cu2(μ-OAc)4 (1). While the toluene C─H activation by monomeric Cu(OAc)2 (7) has a Gibbs activation barrier close to the experimental value (39.1 kcal mol−1), the calculated barrier (67.4 kcal mol−1) for the C─O reductive coupling step is too high to be feasible. Therefore, a mononuclear Cu pathway would require conversion of the mononuclear Cu–p-tolyl (8) to a dinuclear structure for reductive elimination or aryl acetate product formation. However, the low concentration of the mononuclear Cu–p-tolyl structure (8) is because it is highly endothermic, which makes this set of reaction steps unlikely. Another possible reaction pathway is C─H activation by Cu(OAc)2 (7), followed by an intermolecular disproportionation to form a monomeric Cu(III) species 6 along with a Cu(I) compound, which avoids the high-energy aryl ester reductive elimination from the Cu(II) intermediate 8 (see above). The disproportionation pathway is similar to the previously proposed mechanism for directed Cu(II)-mediated C─H acetoxylation/hydroxylation (, ). However, the calculated activation barrier for C─O reductive coupling from the monomeric Cu(III)-tolyl intermediate via TS3 (51.0 kcal mol−1) indicates that this is an uncompetitive pathway. In addition, if the reaction undergoes the intermolecular disproportionation route, then the addition of 0.5 equivalents of ligand (e.g., bipyridine, tetramethyl ethylenediamine, etc.) might be expected to accelerate the reaction since the donor ligand could facilitate the oxidation of LCu(II)-tolyl to form the monomeric LCu(III)-tolyl intermediate; however, our data show that all ligands studied inhibit the Cu2(μ-OAc)4-mediated organometallic arene functionalization.

The calculated activation enthalpy and Gibbs energies for the dinuclear Cu C─H activation step (ΔH‡ = 26.2 kcal mol−1 and ΔG‡443K = 41.4 kcal mol−1) are consistent with the experimental values from the Eyring analysis using k1 (ΔH‡ = 29(2) kcal mol−1 and ΔG‡443K = 38(3) kcal mol−1). The calculated KIE for the C─H(D) activation transition state using deuterium at all five aromatic positions and corresponding zero-point energies is 4.5, which is slightly larger than but still consistent with the experimental KIE value of ~3.0, especially considering that zero-point energy is only one of several factors that control KIE values. In addition, the KIE study was measured on the basis of the overall rate of the reaction, instead of the individual rate constant k1. The reductive coupling/elimination step has energies of ΔH‡ = 40.8 kcal mol−1 and ΔG‡443K = 39.2 kcal mol−1. The nearly equal energies for C─H activation and reductive elimination suggest that both transition states may control the reaction rate and one transition state may dominate under specific conditions. This viewpoint provides an explanation for the observed primary KIE value, partially reversible C─H activation step, and possible overall rate-limiting reductive elimination with a small activation entropy. We propose that the C─H activation step is reversible; therefore, at low acid concentrations, the equilibrium of the C─H activation step may contribute to the isotope effect. There are several possible reasons for a very small activation entropy, and one possible origin is provided by the energy landscape. Figure 6A shows that the Gibbs energy for the reductive elimination transition state (ΔH‡ = 40.8 kcal mol−1 and ΔG‡443K = 39.2 kcal mol−1) is very close to the barrier for C─H activation. This could be explained by a scenario where C─H activation, HOAc dissociation, and reductive elimination steps all contribute to control the reaction rate. This proposal is also consistent with the experimentally observed difference in ΔS‡ when performing the Eyring analysis using kobs and k1, as well as the inverse first-order dependence of the reaction rate on acid concentration.

In addition to the mononuclear and dinuclear C─H activation reaction pathways outlined in Fig. 6, we also examined several other possible mechanisms for toluene C─H functionalization that are much less likely based on calculated energies. While details of these pathways are presented in the Supplementary Materials, it is useful to mention a few key alternative pathways. For example, outersphere single-electron transfer from toluene to Cu(II) requires >140 kcal mol−1 since Cu(II) is only a moderate oxidant, and the resulting charged species are not significantly stabilized in toluene solvent. In addition, we estimated innersphere one-electron transfer from the toluene π system to the Cu metal center using TD-DFT excitation energies calculated from the Cu-toluene π-coordination complex (Fig. 7). This excitation energy is 36.9 kcal mol−1, and the orbitals responsible for this excitation are shown in Fig. 7 (π-orbital 70β to the Cu-ligand antibonding orbital 71β). While this excitation may be competitive with TS1, subsequent C─H activation from this excited state still require about 39.6 kcal mol−1 and therefore an overall barrier of >60 kcal mol−1 (for alternative pathways, see the Supplementary Materials for details), which suggests that electron transfer chemistry is not occurring. We also considered reactions that proceed through the initial formation of radical species, such as ·OH or ·OAc. However, these pathways are unlikely based on calculated Gibbs energies, which require 52.8 and 60.5 kcal mol−1, respectively, to form these radical species from the initial Cu complex.

Fig. 7.

Computational modeling of single-electron transfer pathways.

UMN15L/Def2-TZVPP//MN15L/Def2-SVP Calculated energies for TD-DFT estimation for excitation between occupied orbital 70β (toluene-centered) to unoccupied orbital 71β (Cu-centered).

Isolation of products and recycling of Cu(II)

The reaction using Cu(OHex)2 was scaled to 1.1 g where 78% isolated yield of PhOHex and 83% recovery of the Cu(OHex)2 were achieved (Fig. 8A). We demonstrated recycling of the recovered Cu(OHex)2 using toluene as the substrate, and no difference in tolyl-to-benzyl selectivity was observed upon use of the recycled Cu(OHex)2 (section S17.1). PhOAc can also be isolated with 69% yield along with 93% recovery of the used Cu(OAc)2. With knowledge of the effect of molecular sieves, water, and O2 (sections S10 to S12), we pursued a demonstration that the Cu(II) salt could be recovered and recycled after a completed reaction, similar in some ways to one version of the commercialized Wacker process (–). To do so, after each reaction, 0.5 equivalents of HOHex was added to the reaction solution and stirred under air until the solution turned dark blue, indicating that Cu(I) is oxidized to Cu(II). Then, 3-Å molecular sieves were added to remove the generated water that was formed from the regeneration of Cu(OHex)2. After filtration to remove the sieves and trapped water, Cu(OHex)2 was isolated and reused for the reaction under anaerobic conditions and demonstrated a similar rate as before either using benzene or toluene as the substrate (Fig. 8B), and 2.5 turnovers (TOs) (based on bis-Cu not monomeric Cu) of PhOHex can be achieved. The hydrolysis step of the phenyl carboxylate products was performed (section S21) using a previously reported procedure (). In addition, phenol can be formed in >95% yield by reacting MeOH with phenyl carboxylate product, so that the formed methyl ester and remaining MeOH can be readily isolated from the reaction mixture by distillation. Overall, in the recycle experiments, 240% total yield from benzene to phenol was achieved [based on Cu(II)] with 0.6% total conversion of the initial benzene. By changing the molar ratio of benzene and Cu(II) salts along with using solvent, 2.4% conversion of benzene can be reached; however, there is likely a thermodynamic inhibition to achieving higher conversion (section S22).

Fig. 8.

Isolation of products and extension to Cu(II) recycling.

(A) Isolation of the phenyl OHex/acetate product and recovery of the Cu2(μ-X)4 used in the reaction. Reaction conditions: benzene (20 ml, 224.6 mmol), Cu2(μ-X)4 [0.24 mol %, 0.540 mmol; Cu2(μ-OAc)4, 196 mg; Cu2(μ-OHex)4, 378 mg], under 75 psig of N2 at 180°C for Cu2(μ-OAc)4 and at 170°C for Cu2(μ-OHex)4. (B) Cu(II)-mediated arene 2-ethylhexanoation with Cu(II) recycling. Reaction conditions: benzene (10 ml, 112.3 mmol), Cu2(μ-OHex)4 (189 mg, 0.270 mmol), under 75 psig of N2 at 170°C; toluene (10 ml, 94.1 mmol), Cu2(μ-OHex)4 (158 mg, 0.226 mmol), under 75 psig of N2 at 170°C. Yield of the phenyl carboxylate is relative to the limiting Cu reagent. SDs were calculated from at least three independent experiments.

DISCUSSION

We report a process using copper(II) carboxylate salts to oxidize benzene and toluene without installation of a directing group through a proposed organometallic Cu(II)-mediated C─H activation pathway. While Cu-mediated C─H activation to form intermediates with Cu─C bonds have been proposed previously (, ), to our knowledge, evidence for nonradical Cu-mediated functionalization of hydrocarbons with strong C─H bonds is rare. Further, we have demonstrated that the copper(II) carboxylate salts used for the reaction can be isolated, recycled, and reused for the reaction with little change in reactivity, which is similar to one version of the commercial Wacker process for ethylene oxidation (–). Regeneration of the copper(II) carboxylate salt is achieved in solution using O2 after each reaction, and with the removal of the generated water, the reaction maintains a similar rate using recycled Cu(II) (section S20). By adding a hydrolysis step of the phenyl carboxylate product, which is a known reaction (), this process could provide a coproduct-free synthetic route for phenol production using cheap air-recyclable Cu(II) salts.

Our experimental and computational mechanistic studies demonstrate that the dimeric Cu(II) structure of Cu2(μ-OAc)4 lowers the activation barrier for the product forming C─O reductive coupling step from the Cu(III)-Ar intermediate. This proposal shares some similarities to our previously published catalysis for styrene production using benzene and ethylene for which we propose the incorporation of Cu(II) in a Rh/Cu/Rh structure reduces the activation barrier for an O─H bond forming reductive coupling step (). Thus, the incorporation of a Lewis acidic and oxidizing Cu(II) center (and possibly related metals) into a multimetallic species might represent a general strategy to facilitate catalytic hydrocarbon oxidation.

MATERIALS AND METHODS

General considerations

All reactions were performed under a dinitrogen atmosphere using Schlenk line techniques or inside a dinitrogen filled glovebox unless specified otherwise. GC-FID was performed using a Shimadzu GC-2014 system with a 30 m–by–0.25 mm DB-5ms capillary column with 0.25-μm film thickness. GC–mass spectrometry (MS) was performed using a Shimadzu GCMS-QP2010 Plus or a Shimadzu GCMS-QP2020 NX with a 30 m–by–0.25 mm Rxi-5ms capillary column with 0.25-μm film thickness using electron impact ionization method. Toluene was dried using sodium-benzophenone/ketyl stills under dinitrogen atmosphere and stored inside the glovebox. Methanol was dried using a calcium hydride still under dinitrogen atmosphere and stored inside a glovebox. Benzene, acetonitrile, and methylene chloride were dried using a solvent purification system with activated alumina. Toluene-d8, acetonitrile-d3, benzene-d6, and methylene chloride-d2 were dried and stored over activated 3-Å molecular sieves inside a glovebox. Except for the Cu(II) salts (see below), all other chemicals were purchased from commercial sources and used as received. Elemental analyses were performed by the University of Virginia Chemistry Department Elemental Analysis Facility.

General procedures for pretreatment of Cu(II) salts

The commercial Cu(II) salts was ground and heated in a Schlenk flask under ~10-mTorr vacuum in a 110°C oil bath for at least 1 day or 90°C for 3 days. During the drying process, the condensation of a colorless liquid was observed on the side of the round bottom flask, and a heat gun was used to completely transfer the liquid into the liquid dinitrogen trap on the Schlenk line. After drying, the needle valve was sealed, and the flask containing Cu(II) salts was directly transferred into the glovebox without opening to air.

Alternate method to purify Cu(OHex)2

Commercial Cu(OHex)2 (10 g) was dissolved in 60 ml of ethanol (EtOH) and stirred until a homogeneous solution was formed. Then, deionized water was added until no more precipitation. The mixture was filtered to collect the fine greenish-blue solid. After washing with water a few times, the solid was placed in a vacuum oven at 100°C until no mass change to yield the final product (9.3 g). This method particularly useful for removing high boiling point 2-ehtylhexanoic acid in the commercial Cu(OHex)2. For more details, please see section S2.3.

Synthesis of Cu(OHex)2

To a solution of NaOH (10 g, 250 mmol) in 300 ml of EtOH, OHex acid (42 ml, 263 mmol) was added and stirred for 2 hours at room temperature. Then, Cu(NO3)2·3H2O (30.2 g, 125 mmol) was added to the solution. The color of the solution was blue and then changed to a greenish-blue color with solid precipitation. The mixture was then stirred at room temperature for at least 2 hours, and then deionized water was added to precipitate out the solid product as a greenish-blue fine solid. The solid was washed with deionized water to remove the water-soluble salts. Then, the solid product was placed in a vacuum oven-heated to 100° to 110°C under vacuum until no mass changes to yield the final Cu(OHex)2 product (39.4 g, 90% isolated yield). The solid not only can be further dried using the same procedure described above but can also be used directly without further drying process.

Synthesis of (MeO)3P═O Cu(OAc)2

To a solution of Cu(OAc)2 (100 mg, 0.551 mmol) in 5 ml of toluene, trimethyl phosphate (1 ml, 8.639 mmol) and 1 ml of HOAc were added. After heating at reflux for approximately 1 hour, the suspension was filtered hot to remove the undissolved Cu(OAc)2 [most of the Cu(OAc)2 was not dissolved], and crystals suitable for x-ray analysis were obtained by slow evaporation from light-blue filtrate (20 mg, 11% isolated yield). Combustion elemental analysis calculated (Anal. Calcd) for C14H30O16P2Cu2: C, 26.13; H, 4.70. Found: trial#1, C, 25.84; H, 4.48; trial#2, C, 25.92; H, 4.49.

Synthesis of (MeO)3P═O Cu(OPiv)2

To a solution of Cu(OPiv)2 (100 mg, 0.389 mmol) in 5 ml of toluene, trimethyl phosphate (45 μl, 0.389 mmol) was added. After heating at reflux for approximately 1 hour, the solution was filtered hot, and crystals suitable for x-ray analysis were obtained from slow evaporation from the blue filtrate (82 mg, 53% isolated yield). Anal. Calcd for C26H54P2O16Cu2: C, 38.47; H, 6.71. Found: C, 38.53; H, 6.96.

General procedure for Cu(II)-mediated arene C─H acetoxylation

Under an atmosphere of dry dinitrogen inside a glovebox that the purity was maintained as O2 < 5 parts per million (ppm) and H2O < 1 ppm, the copper(II) salt [CuX2, X = OAc, 0.540 mmol (0.48 mol % to relative to benzene and 0.57 mol % relative to toluene); for soluble CuX2, X = OPiv or OHex, 0.48 mol % relative to arene (0.540 mmol for reaction using benzene and 0.452 mmol for reaction using toluene)] was added into a dried Andrews Glass Lab-Crest Fisher-Porter tube with a stir bar, and then 10 ml of arene (benzene or toluene) was measured by a syringe and added to the reaction tube. The Fisher-Porter tube was then sealed and pressurized with 75 psig of dinitrogen. Then, the mixture was stirred in an oil bath at reaction temperature (the oil bath temperature was confirmed by an external mercury thermometer), and the stir plate stirring speed was set to 550 rpm. The reaction time was recorded starting at the time the oil bath temperature reached the setting point, which typically required 5 to 10 min after placing the Fisher-Porter tube into the 1 liter of oil bath with 500 to 600 ml of silicone oil (usable range from −40° to 200°C). The reaction was monitored periodically by taking a 100-μl aliquot of the reaction solution, mixing with 100 μl of an HMB stock solution in arene (benzene or toluene) with known HMB concentration, and then quickly washing the mixture with saturated sodium hydroxide solution (~0.2 ml). The resulting organic layer was subjected to GC-FID and/or GC-MS for quantitative analysis, using relative peak areas versus HMB as the standard.