Computational exploration of copper catalyzed vinylogous aerobic oxidation of unsaturated compounds

Abstract

Selective oxidation is one of the most important and challenging transformations in both academic research and chemical industry. Recently, a highly selective and efficient way to synthesize biologically active γ-hydroxy-α,β-unsaturated molecules from Cu-catalyzed vinylogous aerobic oxidation of α,β- and β,γ-unsaturated compounds has been developed. However, the detailed reaction mechanism remains elusive. Herein, we report a density functional theory study on this Cu-catalyzed vinylogous aerobic oxidation of γ,γ-disubstituted α,β- and β,γ-unsaturated isomers. Our computational study unveils detailed mechanism for each elementary step, i.e. deprotonation, O₂ activation, and reduction. Besides, the origin of regioselectivity, divergent reactivities of substrates as well as reducing agents, and the byproduct generation have also been investigated. Notably, the copper catalyst retains the + 2 oxidation state through the whole catalytic cycle and plays essential roles in multiple steps. These findings would provide hints on mechanistic studies and future development of transition metal-catalyzed aerobic oxidation reactions.

Introduction

Selective oxidation has gained a preeminent position in both academic research and chemical industry^1–3. One particular class of selective oxidation reactions achieved by a combined use of air as an oxidant and copper as a catalyst is highly desirable due to the natural abundance of air and copper^4–6. Over the past few decades, great progress has been made in this field and many successful Cu-catalyzed aerobic reactions have been developed^7–23.

Catalytic vinylogous reactions are among the most important reactions in organic synthesis due to their extensive application in the synthesis of complex natural products and bioactive molecules^24–27. Despite significant advances in transition metal-catalyzed α- or β-functionalization of α,β- and β,γ-unsaturated compounds^28–32, the vinylogous version leading to synthetically valuable γ-substituted α,β-unsaturated compounds^33–36 has been rarely studied. For example, the vinylogous hydroxylation of α,β-unsaturated or β,γ-unsaturated compounds is a direct method to synthesize γ-hydroxy-α,β-unsaturated compounds^37–42 which are valuable biological active pharmaceuticals and important intermediates in organic synthesis. However, the catalytic aerobic vinylogous hydroxylation is highly challenging due to the control of the reaction selectivity^43,44, such as regioselectivity, chemoselectivity (hydroxylation vs. oxidative fragmentation, epoxidation, and other competitive oxidation reactions) and overoxidation problems.

In 2018, Yin and Newhouse’s group successfully realized an efficient and operationally simple copper-catalyzed vinylogous oxidation reaction by using air as an oxidant, which leads to a broad array of γ-hydroxy-α,β-(E)-unsaturated compounds⁴⁵. Reactions of both γ,γ-disubstituted α,β- and β,γ-unsaturated compounds produce γ-hydroxy-α,β-(E)-unsaturated compounds in high yield with perfect stereo- and regioselectivity (Scheme 1). The copper(II) triflate catalyst, the base (tetramethylguanidine, TMG), and the reducing agent (PPh₃) were found essential for the reaction. Their method was successfully applied to the vinylogous oxidation of unsaturated esters, aldehydes, ketones, amides, nitriles, and sulfones, demonstrating great potential in the synthesis of natural products and bioactive molecules.

Scheme 1.

Copper-catalyzed vinylogous aerobic oxidation of γ,γ-disubstituted α,β- and β,γ-unsaturated compounds.

The preliminary mechanistic study indicated that radicals might not be involved in the reaction because the reaction efficiency was not affected by the addition of radical scavengers^46–48. Based on the experimental observations, Yin et al. proposed a three-step pathway for the generation of the main γ-hydroxylated product (Scheme 2). This pathway consists of deprotonation (A + R + TMG → B), O₂ activation (B + O₂ → C) and PPh₃ participated reduction step (C + PPh₃ → P + P(O)Ph₃). The ketone side-product P′ was proposed to be formed through a four-membered endoperoxide intermediate D. Scheme 2 provides a general mechanism for this Cu-catalyzed aerobic vinylogous oxidation reaction, but the details of each elementary step are unknown. Besides, the origin of regioselectivity and the role of copper catalyst remain elusive. Furthermore, the inertness of γ,γ-dialkyl-substituted α,β-unsaturated compounds toward the aerobic oxidation reaction and the ineffectiveness of P(OEt)₃ as reducing agent have not been fully understood. The understanding of the reaction mechanism is essential for further reaction development. Herein, we perform density functional theory (DFT) calculations¹¹ on this Cu-catalyzed vinylogous aerobic oxidation reaction to elucidate the detailed reaction mechanism, to explore the role of copper catalyst and phosphine, and to understand the observed regioselectivity and side-product formation.

Scheme 2.

Proposed reaction mechanism for the copper-catalyzed vinylogous aerobic oxidation.

Results and discussion

The resting state of copper catalyst

The Cu^II(OTf)₂ catalyzed reaction of methyl-(E)-4-phenylpent-2-enoate E-1a with 1-equivalent TMG and 1-equivalent PPh₃ was selected as a representative case for the DFT calculations (Scheme 3a). We first examined the resting state of the copper catalyst. The Cu^II(OTf)₂ used in the experiment could be coordinated to the TMG base, the reducing agent PPh₃ or the solvent THF. As shown in Scheme 3b,c, the coordination of one TMG molecule to the Cu^II(OTf)₂ complex is exergonic by 33.9 kcal/mol. The subsequent coordination of a second TMG molecule to form the Cu^II(OTf)₂(TMG)₂ complex would release 22.1 kcal/mol of energy. The computational results suggest TMG is a stronger ligand than PPh₃ or THF. The binding with one molecule of PPh₃ or THF is less favorable than that of TMG by 3.1 or 22.4 kcal/mol, respectively. In addition, the generation of cationic copper species from the dissociation of an OTf^ɵ anion of the corresponding neutral copper catalysts are all endergonic (Scheme S1). Thus, the four-coordinated Cu^II(OTf)₂(TMG)₂ (Scheme 3c) is considered to be the resting state of the copper catalyst in this reaction.

Scheme 3.

(a) The Representative Reaction for the DFT Calculation. (b) Binding of TMG, PPh₃, and THF with Cu^II(OTf)₂. Binding Free Energies Are in kcal/mol. (c) The Structures of Cu^II(OTf)₂ and Cu^II(OTf)₂(TMG)₂. Bond Distances Are Given in Angstroms (Å). Note CYLview, 1.0b, https://www.cylview.org/download.html.

Reaction mechanism of copper-catalyzed γ-hydroxylation

Having determined the resting state of the copper catalyst, we then explored the mechanism of each elementary reaction step (i.e. deprotonation, O₂ activation, and reduction) of the copper-catalyzed γ-hydroxylation of E-1a. For each step (except for O₂ activation), both reaction pathways with and without the involvement of copper catalyst were computed to examine the role of copper catalyst.

As of the deprotonation step, the direct deprotonation of substrate E-1a by the base (TMG) via TS1′ in the absence of copper catalyst requires an energy barrier of 22.3 kcal/mol and this step is energetically uphill by 17.0 kcal/mol, which is thermodynamically unfavorable (Scheme S2). In the copper involved pathway as shown in Fig. 1, the Cu^II(OTf)₂(TMG)₂ catalyst first dissociates an OTf^ɵ ligand and binds with E-1a, generating a cationic copper-substrate species INT1. Subsequently, TMG abstracts a proton from INT1 via transition state TS1 with a free energy barrier of 19.8 kcal/mol and leads to a stable Cu^II σ-complex, INT2 (−11.1 kcal/mol). The computational results thus suggest that the coordination of copper with the carbonyl group facilitates the deprotonation of E-1a and the formation of stable σ-complex. Further Hirshfeld population analysis demonstrates the C_γ hydrogen atoms carries more positive charges when copper catalyst is bound to E-1a (Scheme S2). This indicates that the polarity-induced effect by the copper catalyst makes the hydrogen more acidic and thus easier to be activated^49–51. Hence, copper acts as a acid Lewis to mediate the first deprotonation step.

Figure 1.

Computed energy profile of copper-catalyzed vinylogous aerobic oxidation of α,β-unsaturated ester. Relative free energies (electronic energies) are in kcal/mol. The Hirshfeld charges of O atoms (highlighted in red) on INT4 and INT8 are listed.

Following the deprotonation step, the molecular oxygen approaches the Cu^II center and attacks the C_γ of E-1a substrate in a concerted manner via a six-membered chair-like transition state (TS2, Fig. 2) which leads to a peroxide bridge between Cu and substrate^52–55. The distances of O–Cu and O–C_γ in TS2 are 2.23 and 2.17 Å, respectively. Moreover, the O–O bond increases from 1.21 Å in O₂ to 1.26 Å in TS2, which indicates that the O₂ has been activated. This oxygen activation step needs to overcome a free energy barrier of 16.3 kcal/mol and results in a γ-peroxy copper intermediate INT3. To understand the regioselectivity, the formation of α-hydroxylated product by the oxygen addition at the α-carbon via transition state α-TS2 was also calculated (Figure S1). The free energy of α-TS2 is 1.3 kcal/mol higher than that of TS2 which leads to the γ-hydroxylated product, in line with the experimental observation that γ-hydroxylated product is more favorable.

Figure 2.

The deprotonation, O₂ activation and reduction transition states. Relative free energies (electronic energies) are in kcal/mol. The bond distances are given in Angstroms (Å). Note CYLview, 1.0b, https://www.cylview.org/download.html.

To explore the possibility of oxygen activation by binuclear copper species^{52,53,56–58}, we computed the reactions of O₂ with Cu^II(OTf)₂(TMG)₂ and Cu^I(OTf)(TMG)₃, respectively, which afford peroxide bimetallic copper compounds (Scheme S3). The former reaction is endergonic with 88.5 kcal/mol and thus can be ruled out. Although the latter one has a relatively low reaction free energy (14.9 kcal/mol), the formation of Cu^I(OTf)(TMG)₃ by the disproportionation of Cu^II(OTf)₂(TMG)₂ needs 73.5 kcal/mol of energy which excludes the involvement of Cu^I. The computational results thus suggest that the oxygen molecule is unlikely to be activated by binuclear copper complexes.

After the O₂ activation step, the generated γ-peroxy copper species INT3 is reduced to the γ-hydroxylated product. The reduction process with and without copper (black and blue path in Fig. 1, respectively) was examined. In the black path, INT3 first undergoes rapid ligand exchange with PPh₃ to yield a complex INT4 which is further reduced by PPh₃ through transition state TS3. In this transition state, PPh₃ attacks the distal peroxide oxygen while the copper transfers to the proximal oxygen in a concerted way to yield INT5. The subsequent proton transfer of INT5 via TS4 gives intermediate INT6 which proceeds ligand exchange to release the γ-hydroxylated product and regenerate Cu^II(OTf)₂(TMG)₂ catalyst to complete the catalytic cycle. Alternatively, in the reduction process without the participation of copper (blue path), the INT3 first abstracts a proton from the protonated TMG and dissociates from copper to yield the hydroperoxyl compound INT8. Then INT8 undergoes reduction with PPh₃ via a concerted proton shift transition state⁵⁹ (TS5) leading to the γ-hydroxylation product. The calculated activation barriers for the reduction process with and without the involvement of copper (TS3 vs. TS5) are 18.0 and 23.2 kcal/mol, respectively, which indicates that copper facilitates the reduction step. This could be attributed to two main reasons: (1) the incorporation of copper serves to withdraw electron density from the peroxy which helps to polarize the O–O bond as suggested by a larger difference in charges of O atoms on INT4 than that for INT8 (Fig. 1); (2) additional interaction between Cu and the phosphorous atom stabilizes TS3.

Overall, the black path of Fig. 1 that involves substrate association, deprotonation, O₂ activation, reduction, proton transfer and product dissociation was calculated to be the most favorable pathway for the γ-hydroxylation of E-1a. The copper-assisted deprotonation step is the rate-determining step, where TMG base plays a pivotal role in hydrogen abstraction. This result coincides with the experimental observation that in the absence of TMG there is no product generated⁴⁵. Furthermore, our computational results suggest the activation of O₂ molecule proceeds via a six-membered chair-like transition state, which is different from the common end-on or side-on O₂ activation model⁷ and accounts for the regioselective γ-carbon activation. In addition, a copper-mediated reduction process in which copper helps to polarize the O–O bond was unveiled by our computation. The reduction of peroxy complexes by phosphine has been reported in many aerobic reactions^60–65, while the detailed mechanism is rarely studied. The transition-metal assisted reduction model established in this work provides a possible mechanism for similar reactions. Notably, copper remains as Cu^II oxidation state through the whole catalytic cycle and plays vital roles in multiple steps to facilitate the deprotonation, O₂ activation, as well as reduction.

The effect of different reducing agents

The reducing agent plays an important role in this copper-catalyzed vinylogous γ-hydroxylation. A high-yield (82%) of the γ-hydroxylated product was obtained with PPh₃ as a reducing agent, however, the reaction did not occur when PPh₃ was changed to P(OEt)₃ (Fig. 3a). In line with the experiment, our computation in Fig. 3b shows that the reduction with P(OEt)₃ (TS3′) is disfavored by 4.4 kcal/mol compared to the corresponding reduction process with PPh₃ (TS3), which supports the PPh₃ is a better reducing agent than P(OEt)₃ for this reaction. The calculated electrostatic potential surfaces (ESPs)^66–69 in Fig. 3c clearly indicate the P atom of PPh₃ is more electron-rich. Therefore, the PPh₃ could better stabilize the electron-deficient Cu center than P(OEt)₃, accounting for the lower barrier of TS3.

Figure 3.

(a) The reaction yields of γ-hydroxylated productusing different reducing agents. (b) Theoretically calculated reduction process using different reducing agents. Relative free energies (electronic energies) are in kcal/mol. (c) The electrostatic potential surfaces (ESP) of PPh₃ and P(O)Et₃ mapped between − 6.096e−2 to + 6.096e−2. Red as negative extreme and blue as positive extreme. Note GaussView6, https://gaussian.com/gv6main/.

The reactivity of different substrates

The original experimental work reported that γ-aryl-γ-alky-disubstituted α,β-unsaturated compound E-1a and γ-aryl-γ-alky-disubstituted β,γ-unsaturated compound E-1b are both reactive and generates the same γ-hydroxylated product (Table 1, entry 1 and 2). The γ,γ-dialkyl-substituted β,γ-unsaturated substrate 2b also exhibits good reactivity (entry 3). However, the γ,γ-dialkyl-substituted α,β-unsaturated compound E-2a was completely inert (entry 4). Further computation was performed to understand the observed reactivities of different α,β- or β,γ-unsaturated substrates. As indicated by our DFT studies, the deprotonation process is the rate-determining step. Thus, we examined the deprotonation transition states for different substrates to evaluate their reactivities. The calculated energy barriers of deprotonation follow the order: ∆∆G^≠(TS1_E-1b) < ∆∆G^≠(TS1_E-1a) < ∆∆G^≠(TS1_2b) < ∆∆G^≠(TS1_E-2a), which is consistent with the trend of yield (Table 1). To obtain deep insights, the bond dissociation energy (BDE)⁷⁰ of the corresponding C–H bonds (highlighted in red) were calculated to evaluate the intrinsic acidity. As shown in Figs. 4, a good linear association between the deprotonation free energy barriers (∆∆G^≠) and the BDE values were observed. This indicates that the inherent strength of the C–H bond is a key factor affecting the deprotonation process, which is related to the substrate reactivity. In addition, the (E)-α,β-unsaturated esters E-1a and (E)-β,γ-unsaturated esters E-1b generated the same γ-hydroxylated product in experiments. Based on the understanding of reaction mechanism and computational results, we concluded that the deprotonation of both (E)-α,β-unsaturated ester and (E)-β,γ-unsaturated ester would generate the same Cu^II σ-complex and thus leads to the same γ-hydroxylated isomer. However, the reactivities of these two substrates (E-1a and E-1b) are different. E-1a has a higher deprotonation energy barrier than E-1b by 2.2 kcal/mol corresponding with a lower yield of E-1a and further supports the deprotonation is the rate-determining step.

Table 1.

The reaction yields and the calculated energy barriers of deprotonation transition states for different substrates.

Figure 4.

The relationship between deprotonation the energy barriers and the corresponding C–H bond dissociation energy (BDE) for different substrates.

The mechanism of byproduct generation

One of the main problems for the hydroxylation of α,β-unsaturated compounds is the formation of oxidative fragments. In the original experiment, a side product acetophenone was detected and proposed to be generated by oxidative fragmentation of a four-membered endoperoxide intermediate D as depicted in Scheme 2. We also explored the formation of acetophenone byproduct. As shown in Scheme 2, the main product path and the byproduct path differentiate from the γ-peroxy copper intermediate.

Starting with this intermediate (INT3), three possible pathways leading to the acetophenone byproduct were examined (Scheme S4). A reaction pathway involving a peroxide radical^71–78 which is formed via the homolysis of Cu–O bond of INT3 needs to overcome a reaction energy barrier of 29.2 kcal/mol (path I in Scheme S4). Alternatively, INT3 could undergo an intramolecular alkene insertion into the Cu–O bond via TSB1 to form four-membered endoperoxide INTB1 which is similar to the endoperoxide intermediate D proposed in the original experimental work (Fig. 5). But the further homolysis of the C–C and O–O bond of INTB1 via TSB2 to cleavage the four-membered ring which affords the byproduct is highly unfavorable with an energy barrier of 32.8 kcal/mol. Instead, our computational results demonstrated that copper can facilitate the cycloelimination process by transferring to the β-oxygen atom of endoperoxide (TSB3) to maintain the conjugated α,β-unsaturated structure, lowering the activation barrier by 7.5 kcal/mol. The direct cycloelimination of four-membered peroxide complexes was proposed in many works^79–83 and our results suggested a possible role of transition metal to facilitate the oxidative fragmentation process and to provide the theoretical basis for further reaction improvement.

Figure 5.

The possible formation routes of acetophenone byproduct. Relative free energies (electronic energies) are in kcal/mol.

Conclusion

In summary, we have conducted DFT studies on the Cu-catalyzed vinylogous aerobic oxidation of γ,γ-disubstituted α,β- and β,γ-unsaturated compounds. As summarized in Fig. 6, computational results unveiled a detailed reaction mechanism of γ-hydroxylation reaction that includes six steps: substrate association, deprotonation, O₂ activation, reduction, proton transfer and product dissociation (black path), and the deprotonation is the rate-determining step. The regioselectivity is controlled by the O₂ activation step which prefers to proceed via a six-membered chair-like transition state, leading to a γ-oxidation intermediate. Besides, the inefficiency of P(OEt)₃ and inertness of γ,γ-dialkyl substituted β,γ-unsaturated ester were also understood by computations. A pathway consisting of intramolecular alkene insertion, cycloelimination, and product dissociation (brownish-green path) was revealed to account for the acetophenone byproduct generation. The understanding of the reaction mechanism laid a theoretical foundation for further reaction development.

Figure 6.

The proposed catalytic cycle for the copper-catalyzed vinylogous aerobic oxidation of α,β- and β,γ-unsaturated compounds.

The copper retains the + 2 oxidation state (Cu^II) and participates in the whole catalytic cycle of both main product and byproduct formation. Notably, it plays vital roles in multiple steps: (1) facilitates the substrate deprotonation by increasing the acidity of C–H; (2) activates O₂ via a six-membered chair-like model which is different from the common end-on or side-on O₂ activation model; (3) assists the reduction of peroxyl intermediate through a 1,2-migration transition state; and (4) promotes the cycloelimination of endoperoxide by transferring to oxygen to maintain the conjugated α,β-unsaturated structure. The understanding of mechanism for O₂ activation and O–O bond cleavage are essential for the development of transition metal-catalyzed aerobic reactions. The six-membered chair-like transition state for O₂ activation and copper-mediated O–O bond cleavage models (TS3 and TSB3) have not been documented in literature to the best of our knowledge and may provide hints for the mechanistic studies and future development of transition metal-catalyzed aerobic oxidation reactions.

Computational details

All the calculations were performed with Gaussian 09 package⁸⁴. Geometries were optimized in gas phase by using unrestricted B3LYP-D3^85–88 and a mixed basis set of SDD^89,90 for Cu and P, and 6-31G(d)^91,92 basis set for all other atoms. Optimized geometries were verified by frequency computations as minima (zero imaginary frequencies) or transition state (a single imaginary frequency) at the same level of theory. The transition states (TSs) were also confirmed by viewing normal mode vibrational vector. Solvent effect was included by single-point energy calculation using SMD model with tetrahydrofuran (THF) as the solvent and B3LYP-D3 method with def2-TZVP basis set for Cu and P, and 6-311+G(d, p) basis set for other atoms^93–97. All relative Gibbs free energies and electronic energies (at 298.15 K and 1 atm) were reported in kcal/mol. The Hirshfeld charges⁹⁸ were obtained from the B3LYP-D3 single-point calculation. The 3D structures were generated by CYLview⁹⁹.

Author contributions

T.W. performed calculations and data analysis. Y.Z. did the calculation of dimer copper. Y.X. checked for grammatical errors and revised the paper. G.-J.C. proposed this subject and wrote the paper. All authors reviewed the manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

is available for this paper at 10.1038/s41598-020-80188-2.