Isolation of bis(copper) key intermediates in Cu-catalyzed azide-alkyne "click reaction".

The copper-catalyzed 1,3-dipolar cycloaddition of an azide to a terminal alkyne (CuAAC) is one of the most popular chemical transformations, with applications ranging from material to life sciences. However, despite many mechanistic studies, direct observation of key components of the catalytic cycle is still missing. Initially, mononuclear species were thought to be the active catalysts, but later on, dinuclear complexes came to the front. We report the isolation of both a previously postulated π,σ-bis(copper) acetylide and a hitherto never-mentioned bis(metallated) triazole complex. We also demonstrate that although mono- and bis-copper complexes promote the CuAAC reaction, the dinuclear species are involved in the kinetically favored pathway.

The copper-catalyzed azide-alkyne cycloaddition (CuAAC) (, ) is the flagship of the “click chemistry” arsenal (, ). Because of the straightforward procedures, the high functional group tolerance, and the usually high yields, CuAAC has affected all fields of chemistry from material (–) to life sciences (–); however, the mechanism of the CuAAC is still under debate (, ). The first postulated mechanism involved mono-copper complexes, although more recently, the participation of bis(copper) complexes has been suggested on the basis of kinetic experiments () and computational investigations (–). In 2013, Fokin and colleagues published a very elegant mechanistic study and stated that “monomeric copper-acetylide complexes 1 are not reactive towards organic azides unless an exogenous copper catalyst is added” (). The marked difference with the previously accepted mechanism is the introduction of the cationic π,σ-bis(copper) acetylide complexes of type 1 (Fig. 1) as the “catalytically active complex,” instead of 1. However, such a dinuclear species was qualified as a “non-isolable and highly reactive intermediate.” We have recently shown that the strong σ-donating and π-accepting properties of cyclic (alkyl)(amino)carbenes (CAACs) (–) allow for the isolation of electron-deficient (), electron-rich (), and even paramagnetic species (). This prompted us to attempt using CAAC ligands to stabilize elusive species involved in the CuAAC catalytic cycle. Here, we report the isolation of a π,σ-bis(copper) complex of type 1 and of an unprecedented catalytically active 3,5-bis(metallated) triazole 2. Thanks to the unusual stability of these complexes, we compare the mono- and dinuclear-based mechanisms and bring unambiguous evidence that the latter is indeed kinetically favored.

Fig. 1

Stepwise reproduction of the CuAAC reaction; isolation of mono- and bis(copper) intermediates involved in both catalytic pathways.

(A) Stoichiometric reactions reproducing the different steps of the postulated CuAAC catalytic cycles, which allow for the isolation of a previously postulated π,σ-bis(copper) complex of type 1 and of the never-mentioned bis(copper) triazole 2. (B) Kinetic profiles of the stoichiometric reactions of 1a and 1a with benzyl azide, which reveal the critical rate acceleration when the bimetallic complex 1a is used. (C) Molecular view of complexes 1a (left, table S1), 2a (center, table S2), and 1b (right, table S3) in the solid state (for clarity, H atoms and anions are omitted).

We first prepared the monomeric (, ) copper complex 1a (figs. S7 and S8) by reacting lithium phenylacetylide with (CAAC)CuOAc (figs. S1 and S2) in tetrahydrofuran (THF) (Fig. 1A). We found that 1a cleanly reacted with (CAAC)CuOTf (figs. S3 to S6) in methylene chloride, affording the desired cationic dinuclear complex 1a (figs. S9 and S10), which was isolated as a white solid in 95% yield. The 1H and 13C nuclear magnetic resonance (NMR) spectra show only one set of signals for both CAAC ligands, even at temperatures as low as −80°C, which is indicative of a very fast exchange between the two “Cu(CAAC)” units. However, in the solid state, a single crystal x-ray diffraction study shows two different types of coordination, π and σ (Fig. 1C, left). The Cu-Cu distance [2.9387(4) Å] is slightly longer than those computationally predicted by Fokin and colleagues for model compounds (2.85 to 2.88 Å) (). 1a proved to be air-stable and thermally robust [melting point (mp): 174°C].

Despite its thermodynamic stability, 1a reacts cleanly with benzyl azide in methylene chloride, affording 2a (figs. S11 and S12), which was isolated in 87% yield as an air- and moisture-stable, pale yellow solid. 1H and 13C NMR spectra show the presence of one benzyl and one phenyl group, as well as two magnetically nonequivalent CAAC moieties. X-ray diffraction studies of the corresponding tetrafluoroborate salt confirmed the 3,5-dimetallated-1,2,3-triazole structure 2a (Fig. 1C, center).

Because dinuclear complexes of type 2a had never been postulated in any CuAAC mechanism, the next question was whether it was involved in the catalytic cycle. Addition of a stoichiometric amount of phenyl acetylene to 2a led, after 5 hours, to the quantitative release of the 1-benzyl-4-phenyl-1,2,3-triazole 3 accompanied by the regeneration of the bis-copper complex 1a. As an additional confirmation, the same reaction was repeated using 2a and a different alkyne (Ph2CHC≡CH), which resulted in the isolation of 3 along with the corresponding π,σ-bis-copper complex 1b (Fig. 1C, right; figs. S13 and S14) in quantitative yields.

After having shown that the bis-copper complex 1a could be part of the catalytic cycle, we turned our attention to the corresponding mono-copper complex 1a. We found that a stoichiometric amount of benzyl azide slowly reacted with complex 1a, leading, after 16 hours, to the C-metallated heterocycle 2a (figs. S15 and S16), which was isolated as a pale yellow solid in 61% yield (). Addition of one equivalent of phenyl acetylene to 2a led, after 5 hours, to the quantitative isolation of triazole 3 and regeneration of the mono-copper complex 1a.

These results as a whole suggest that both the mono- and bis-copper complexes could lead to the triazole 3, at least under stoichiometric conditions. However, the kinetic profile of the reactions of 1a and 1a with benzyl azide revealed critical rate acceleration when the bimetallic complex is used [kobs (1a)/kobs (1a) > 94] (Fig. 1B and fig. S17) (). In contrast, the protodemetallation of 2a and 2a affording triazole 3 and regenerating 1a and 1a, respectively, proceeds at a similar rate (fig. S18). It should be noted that in both cases, the protodemetallation arises from the reaction with the terminal alkyne and, thus, does not necessitate an external Brønsted acid. These reactions lead back to the acetylide complexes 1a and 1a, which exclude the copper precatalyst (CAAC)CuOTf from the catalytic cycle.

The experiments discussed above reproduce parts of both postulated catalytic cycles under stoichiometric conditions. Each step proved to be extremely clean and high-yielding, and could be achieved either under strictly dry and anaerobic conditions or using nondried solvent under air. To confirm that the isolated complexes are active catalytic species, phenyl acetylene was reacted with benzyl azide at room temperature in methylene chloride using 10 mol % of 1a and 2a, and only 5 mol % of 1a and 2a. The kinetic profiles of the reactions reveal that the catalytic activity of the dinuclear complexes 1a and 2a is drastically higher (94 and 99% yields of 3 after 10 hours) than that of mononuclear complexes 1a and 2a (2 and 12% yields of 3 after 10 hours) (Fig. 2A and fig. S19). These results are in agreement with those obtained for the stoichiometric reactions. Note that although the reactions promoted by 1a and 2a proceed at the same rate, a short initiation period is observed with 1a, which is concomitant with the formation of 2a; after the initiation period, the concentration of 2a remains constant over the course of the catalytic reaction (Fig. 2B). Similar observations can be made from the reactions catalyzed by complexes 1a and 2a (Fig. 2C), demonstrating that metallated triazoles 2a and 2a are the resting states of their respective catalytic cycles.

Fig. 2

Kinetic profiles of the CuAAC reaction of phenyl acetylene with benzyl azide using CAAC-supported mono- and bis(copper) catalysts.

(A) The kinetic profiles of the catalytic reaction of phenyl acetylene with benzyl azide demonstrate the superior catalytic activity of the dinuclear complexes 1a and 2a over their mononuclear counterparts 1a and 2a, and show that the (CAAC)CuOTf complex adopts the dinuclear pathway after an initiation period. (B) Evolution of the amount of dimetallated triazole 2a and free triazole 3 during the early period of the catalytic reaction using 1a as catalyst, showing that 2a is the resting state of the catalytic cycle. a.u., arbitrary units. (C) Same as (B), but of 2a and 3 using 1a as catalyst. (D and E) Et3N, as a proton scavenger, allows for the rapid formation of 1a from (CAAC)CuOTf and phenyl acetylene (D), and consequently shortens the induction period of the CuAAC reaction described in (A) using (CAAC)CuOTf (E).

The results described above demonstrate that the catalytic reaction operates efficiently once 1a is formed. When phenyl acetylene was reacted with benzyl azide at room temperature, in the presence of 10 mol % (CAAC)CuOTf, we observed the near quantitative formation of triazole 3, but only after 35 hours. The kinetic profile (Fig. 2A) shows a long initiation period, followed by drastic rate acceleration. Because the formation of the bis(nuclear) acetylide 1a from 1a is a fast reaction (vide supra), we conclude that the formation of 1a is the limiting step of CuAAC reaction when (CAAC)CuOTf is the precatalyst. Indeed, when two equivalents of (CAAC)CuOTf were added to phenyl acetylene, in the absence of benzyl azide, no apparent reaction occurred after 8 hours at ambient temperature (Fig. 2D). The trifluoromethanesulfonate anion is too weakly basic to efficiently promote the metallation of the alkyne (). Addition of one equivalent of triethylamine to the previous solution cleanly afforded 1a along with Et3NH+OTf− within a few minutes. The effect of the presence of an additional base is also apparent on the kinetic profile of the catalytic reaction using 10 mol % (CAAC)CuOTf and 5 mol % triethylamine (Fig. 2E). As expected, a significant shortening of the initiation period is observed.

In conclusion, the isolation of the mono- and bis-copper acetylide complexes 1a and 1a demonstrates that although both species are active in the catalytic cycle, the dinuclear complex is involved in the kinetically favored pathway (Fig. 3). The stability of 1a seems to be in contradiction with previous reports (). The peculiar electronic properties of the CAAC ligand possibly play a role (), but the use of a weakly coordinating trifluoromethanesulfonate ligand is crucial. Indeed, when (CAAC)CuX complexes bear the more popular chloride or acetate X ligand, no bis(copper) complexes could be observed, although kinetic data demonstrate their involvement. Moreover, when the trifluoromethanesulfonate is used, investigations of the dinuclear pathway permitted the isolation of a never-postulated bis(copper) triazole complex of type 2, which is the resting state of the catalytic cycle. The alkyne serves as the proton source for the demetallation of 2, which regenerates the π,σ-bis(copper) acetylide of type 1, leaving out complexes of type 1 from the catalytic cycle. These results led to the broader question of whether bis(metallic) complexes of type 1 are also key species in other copper-catalyzed organic reactions ().

Fig. 3

Mechanistic conclusions: Both the mono- and bis-copper pathways are active in the CuAAC reaction, but the latter is kinetically favored.

The protodemetallation is performed by the alkyne, which regenerates the metallated acetylide, thereby excluding the σ-copper acetylide from the preferred catalytic cycle.

MATERIALS AND METHODS

General

1H and 13C NMR spectra were recorded on Bruker Avance 300, Jeol ECA 500, and Varian Inova 500 spectrometers. NMR multiplicities are abbreviated as follows: s = singlet, d = doublet, t = triplet, q = quartet, sept = septet, m = multiplet, br = broad signal. Chemical shifts are given in parts per million (ppm) and are referenced to SiMe4 (1H, 13C) and CFCl3 (19F). All spectra were obtained at 25°C in the solvent indicated. Coupling constants J are given in hertz (Hz). Mass spectra were performed at the University of California San Diego Mass Spectrometry Laboratory. Melting points were measured with an electrothermal MEL-TEMP apparatus. Single crystal x-ray diffraction data were collected on a Bruker Apex II CCD (charge-coupled device) detector using Mo-Kα radiation (λ = 0.71073 Å). Crystals were selected under oil, mounted on nylon loops, and then immediately placed in a cold stream of nitrogen. Structures were solved and refined using Olex2 and SHELXTL. The CAAC and 3,3-diphenyl-1-propyne were prepared following literature procedures, whereas all other starting materials were purchased from commercial sources.

Synthesis of (CAAC)CuCCPh 1Cua

n-BuLi (2.4 mmol, 2.5 M in hexane) was slowly added to a solution of phenyl acetylene (234 mg, 2.3 mmol) in THF (15 ml) at −78°C. After 30 min, a solution of (CAAC)CuOAc (1.0 g, 2.3 mmol) in THF (10 ml) was added, and the mixture was stirred at ambient temperature for 1 hour. Volatiles were removed under vacuum, and the remaining solid was extracted with benzene (3 × 10 ml). After removal of benzene under vacuum, the solid was washed with hexane (20 ml). 1a was obtained as a light yellow solid. Yield 88% (970 mg). mp 152°C [decomposition (dec.)]. 1H NMR (CD2Cl2, 500 MHz): δ = 7.48 (t, J = 7.8 Hz, 1 H, p-H), 7.34 (d, J = 7.8 Hz, 2 H, m-H), 7.23 (br d, J = 7.1 Hz, 2 H, o-HC6H5), 7.14 (br t, J = 7.1 Hz, 2 H, m-HC6H5), 7.08 (br t, J = 7.1 Hz, 1 H, p-HC6H5), 2.88 (sept, J = 6.9 Hz, 2 H, CHMe2), 1.98 (s, 2 H, CH2), 1.96 to 1.87 (m, 2 H, CH2), 1.86 to 1.74 (m, 2 H, CH2), 1.39 to 1.30 (m, 18 H, CH3, CHCH3), 1.11 (t, J = 7.5 Hz, 6 H, CH3); 13C NMR (CD2Cl2, 125 MHz): δ = 253.5 (Ccarbene), 145.6 (Cq), 135.1 (Cq), 131.8 (CHC6H5), 129.9 (CHAr), 128.1 (CHC6H5), 127.7 (PhCCCu), 125.7 (CHAr), 125.0 (CHC6H5), 121.9 (Cq-C6H5), 106.3 (PhCCCu), 81.3 (Cq), 63.3 (Cq), 42.8 (CH2), 31.4 (CH2), 29.5, 29.4, 27.3, 22.4, 9.9; high-resolution mass spectrometry (HRMS) [electrospray ionization–time-of-flight mass spectrometry (ESI-TOFMS)]: mass/charge ratio (m/z) calculated for C30H40CuNNa+ 500.2349, found 500.2355.

Synthesis of bis(copper) acetylide complexes 1Cu2a

(CAAC)CuOTf (530 mg, 1 mmol) was added to a solution of 1a (480 mg, 1 mmol) in methylene chloride (20 ml). The reaction was stirred for 5 min at ambient temperature. After removing the solvent under vacuum, 1a was obtained as a white solid. Yield 95% (952 mg). mp 174°C (dec.). 1H NMR (CD2Cl2, 500 MHz): δ = 7.44 (t, J = 7.8 Hz, 2 H, p-H), 7.26 (d, J = 7.8 Hz, 4 H, m-H), 7.22 (t, J = 7.6 Hz, 1 H, p-H), 7.04 (t, J = 7.6 Hz, 2 H, m-H), 6.41 (d, J = 7.6 Hz, 2 H, o-H), 2.82 (sept, J = 6.9 Hz, 4 H, CHMe2), 2.05 (s, 4 H, CH2), 1.96 to 1.83 (m, 4 H, CH2), 1.80 to 1.68 (m, 4 H, CH2), 1.38 (s, 12H, CH3), 1.29 (d, J = 6.9 Hz, 12 H, CHCH3), 1.18 to 1.02 (m, 24 H); 13C NMR (CD2Cl2, 125 MHz): δ = 249.4 (Ccarbene), 145.3 (Cq-Aro), 134.7 (Cq-Aro), 132.0 (CH-Aro), 130.4 (CH-Aro), 129.4 (CH-Aro), 128.7 (CH-Aro), 125.4 (CH-Aro), 122.2 (PhCCCu2), 118.9 (Cq-Aro), 110.8 (PhCCCu2), 82.5 (Cq), 63.1 (Cq), 42.1 (CH2), 31.5 (CH2), 29.4, 27.2, 22.4, 9.9; HRMS (ESI-TOFMS): m/z calculated for C52H75Cu2N2+ 853.4517, found 853.4507.

Synthesis of bis(copper) triazole complexes 2Cu2a

Benzyl azide (40.0 mg, 0.30 mmol) was added to a solution of 1a (250 mg, 0.25 mmol) in methylene chloride (0.5 ml). After 1 hour, diethyl ether (15 ml) was added to induce the precipitation of the product. After filtration, 2a was isolated as a pale yellow solid. Yield 87% (246 mg). mp 175°C (dec.). 1H NMR (CD2Cl2, 500 MHz): δ = 7.48 to 7.38 (m, 2 H, p-H), 7.34 to 7.14 (m, 12 H), 6.82 (br, 2 H), 5.11 (s, 2 H, CH2Ph), 2.77 (sept, J = 6.9 Hz, 4 H, CHMe2), 2.01 (s, 2 H, CH2), 1.96 (s, 2 H, CH2), 1.85 to 1.73 (m, 2 H, CH2), 1.73 to 1.61 (m, 4 H, CH2), 1.61 to 1.51 (m, 2 H, CH2), 1.34 (br, 12 H, CH3), 1.29 to 1.20 (m, 12 H, CHCH3), 1.04 to 0.86 (m, 24 H); 13C NMR (CD2Cl2, 125 MHz): δ = 252.7 (Ccarbene), 249.3 (Ccarbene), 156.7 (Ctriazolide), 155.1 (Ctriazolide), 145.4 (Cq), 145.3 (Cq), 136.7 (CqAr), 135.2 (Cq), 134.8 (Cq), 133.3 (CqAr), 130.3 (CHAr), 130.2 (CHAr), 129.1 (CHAr), 128.8 (CHAr), 128.2 (CHAr), 128.1 (Cq-Ar), 128.0 (CHAr), 127.1 (CHAr), 125.2 (CHAr), 125.1 (CHAr), 82.3 (Cq), 81.9 (Cq), 63.3 (Cq), 63.0 (Cq), 57.1 (CH2Ph), 42.4 (CH2), 31.3 (CH2), 29.4, 27.0, 22.3, 9.7. HRMS (ESI-TOFMS): m/z calculated for C59H82Cu2N5+ 986.5157, found 986.5161.

Protodemetallation of 2Cu2a leading to the regeneration of 1Cu2a

Phenyl acetylene (15 mg, 0.15 mmol) was added to the solution of 2a (136 mg, 0.12 mmol) in methylene chloride (5 ml). After stirring for 16 hours at room temperature, the solvent was removed under vacuum. The remaining solid was washed with diethyl ether (3 × 10 ml) to remove triazole 3, and after drying, 1a was obtained as a yellow solid. Yield 91% (108 mg). NMR data are identical to those of a sample prepared as described above.