Article pubs.acs.org/Organometallics
A DFT Study on the Binuclear CuAAC Reaction: Mechanism in Light of New Experiments Yılmaz Ö zkılıç† and Nurcan Ş. Tüzün*,† †
Department of Chemistry, Faculty of Science and Letters, Istanbul Technical University, Maslak, Istanbul 34469, Turkey S Supporting Information *
ABSTRACT: In this DFT study, the mechanism of the copper(I)-catalyzed azide−alkyne cycloaddition (CuAAC) reaction is revisited in light of recent experimental findings that made significant contributions to unraveling this challenging and important reaction. The generally accepted binuclear mechanism was used as a framework to investigate two inquiries raised by new experiments. First, Fokin et al. have proposed ligand exchanges that can take place as possible alternative pathways to the generic path and in that way they have proved the binuclear nature of the CuAAC mechanism. In this study, the experimentally proposed ligand exchanges which deviate from the generic path were modeled with NHC as the ligand and the electronic nature of the mechanism was also investigated with the NBO analyses. The results in this study are compatible with the experimental proposals, since the ligand exchange and the generic pathways’ calculated energies are on the same order. Second, possible pathways for the formation of a recently isolated bis-copper triazolide intermediate were considered by DFT calculations to explain this mechanism thoroughly. It was shown that its formation is energetically highly unfavorable during the cycloaddition step, whereas it can be facile after the formation of the mononuclear triazolide. The calculations were performed at the M06-L/6-31+G(d,p) level with the LANL2TZ+ effective core potential for copper atoms.
■
INTRODUCTION Copper has been the best catalyst of choice in the exclusive 1,4-disubstituted 1,2,3-triazole formation since its use in the synthesis of azide−alkyne cycloaddition reactions.1 The CuAAC procedure has found widespread use in many applications. However, unraveling the precise nature of this catalytic cycle is still a challenging task. Understanding the path(s) of the reaction along with the various possible intermediates and transition states is crucial for both the development and the generalization of the currently employed procedures. To achieve this goal, both experimental2 and theoretical3 investigations have been conducted. Sharples et al. proposed a mononuclear mechanistic model which included a six-membered mononuclear copper(III) structure.1b Subsequent experiments showed that the nuclearity of this structure was highly debatable and more than one copper species may be actively involved in the reaction.2a,c,d,f−n Quantum mechanical investigations supported this idea by demonstrating that the reaction path with a binuclear structure has an energy barrier lower than that of a mononuclear structure whose energy barrier was found to be on the order of the noncatalyzed path.3a,b The source of the yellow color that formed upon addition of the copper catalyst and faded away toward the completion stages was associated with the formation of the binuclear copper species.2d The X-ray structures of homoleptic Cu(I) arylethynyl coordination polymers, [(PhCCCu)2]n, revealed4 that their Cu−Cu distances are between 2.49 and 2.83 Å. This result is in accordance with DFT investigations3a © XXXX American Chemical Society
carried out for binuclear Cu(I) complexes. This observation encouraged Heaney et al. to use phenylethynylcopper(I) polymer as a catalyst in the CuAAC reaction, where the anticipated triazole structure could be obtained in good yield.2d,f In the three latest experimental studies, the binuclear nature of the CuAAC mechanism was proven by isotopic labeling,2a isolation,2j and detection2k of bis(copper) intermeditates via electrospray ionization mass spectrometric techniques. Kinetic studies led to the conclusion that the CuAAC reaction is second order with respect to copper when copper is used in catalytic amounts. Otherwise, the rate tends to vary between first and second order, possibly due to the formation of copper aggregates.2b The rate law with respect to copper can also vary with the ligand and the solvent used.2c Indeed, the catalytic efficiency of the CuAAC reaction was observed to depend on the nature of the ligand and the solvent that can act as a ligand in the reaction medium. The inhibition effects of acetonitrile (CH3CN) as a competing ligand for copper is an example.2d,g Ligands are necessary to activate Cu(I), since their absence promotes aggregation and/or solvation-based stabilization. Additionally, if the ligand’s nature particularly promotes binuclear intermediates, ligand-based acceleration is expected.2g When the ligands are selected to be strongly coordinating in comparison to the environment, ligand-based inhibition is observed and the number of ligands that are present becomes important. The unique properties Received: April 7, 2016
A
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics of N-heterocyclic carbenes (NHC) have rendered them very efficient ligands for metal-based catalysts. In click reactions, NHC ligands generally stabilize copper; thus, their use as ligands allows the synthesis of triazole in good yields.2h,5 Their versatility for the CuAAC reaction was also shown in a DFT study.3h The generally accepted binuclear reaction path starts with the π coordination of alkyne to the first Cu center (Figure 1).
Figure 3. Experiment by Bertrand et al.2j (CAAC = cyclic alkyl amino carbene).
As a part of our ongoing interest in the CuAAC reaction, the mechanism of this important reaction is revisited in light of recent experimental findings that made significant contribution in unraveling this challenging and important reaction. Herein, we present our quantum mechanical calculations of the CuAAC reaction path which is based on previous binuclear computational studies and consider deviations from the generic path by the experimentally proposed ligand exchanges. The electronic nature of the mechanism was also investigated. Additionally, the possible pathways for the formation of the recently reported intermediate2j were considered to explain this mechanism thoroughly and to generalize it to a greater extent. This could lead to possible new pathways in click chemistry.
Figure 1. Generally accepted mechanism of CuAAC.
After deprotonation, the alkyne coordinates to another Cu center. This binuclear complex and azide form a six-membered ring, which subsequently contracts to a five-membered triazolide ring. Generally, slightly acidic conditions easily furnish the desired triazole from the triazolide ring. A seminal experiment by Fokin et al.2a introduced convincing evidence for the binuclear mechanism: the azide was reacted with σ-bound Cu(I) acetylide that had a natural ratio of copper isotopes in the presence of isotopically pure [63Cu(CH3CN)4]PF6 as catalyst (Figure 2). They observed 50% isotopic enrich-
■
METHODOLOGY
The optimizations and calculations involving structural properties were carried out with the M06-L functional,6 which is reported to be suitable for copper complexes.7 To achieve an optimum balance between the performance and the computational time required, the LANL2TZ+8 effective core potential was used for copper atoms along with the 6-31+G(d,p) basis set for the rest of the structure. DFT calculations were performed with Gaussian G09 software.9 Effective core potential data were retrieved from the EMSL Basis Set Exchange Web site. Calculations were performed in tetrahydrofuran (THF) as a continuum medium, which is based on the PCM method developed by Tomasi et al.10 Every reported transition state (TS) was confirmed to have only one imaginary frequency belonging to the reaction coordinate. Intrinsic reaction coordinate11 (IRC) calculations were performed to identify adjacent intermediates of the relevant transition states. The energies reported in this study are free energies at 298 K. The distances are in units of Å, and hydrogens are omitted for clarity in the figures. The experimentally utilized azide and alkyne in the work2a of Fokin et al. were computationally cumbersome; thus, model structures were used to decrease the computational time (Figure 4). Likewise, in our
Figure 2. Experiment conducted by Fokin et al.2a (NHC Nheterocyclic carbene).
ment of the product triazolide with the pure 63Cu. This result clearly demonstrates the demand for the active participation of both Cu centers in the catalytic process. The explanation for the isotopic enrichment was a possible ligand exchange between the two copper centers which would require the migration of NHC from Cu1 (63Cu:65Cu = 69%:31%) to isotopically pure Cu2 (63Cu). To pinpoint the ligand exchange on the catalytic cycle, Fokin et al. conducted additional test reactions. In the presence of only σ-bound Cu(I) acetylide and the catalyst, no enrichment was observed, which implies the necessity of an azide molecule. Additionally, the reaction of the product copper triazolide with Cu1 and the catalyst with Cu2 did not lead to enrichment. From their experiments, Fokin et al. concluded that the ligand exchange should take place during the cycloaddition steps. In 2015, Bertrand et al. isolated important intermediates from the CuAAC reaction. In their experiment, the CuAAC reaction was carried out with benzyl azide and the previously prepared binuclear acetylide−Cu−CAAC complex in dichloromethane (Figure 3).2j Intermediate EI (Figure 3), which had not been reported before as a possible intermediate or product in the literature, was isolated in their experiment. This intermediate was interpreted as one of the resting states of the catalytic cycle.
Figure 4. Corresponding models of the structures used in the experiment.2a previous study, the mechanism of CuAAC catalysis was considered for a model and an experimentally studied bulkier set and no significant difference was observed for the relative free energies of the transition states and the barriers.3d Natural bond orbital (NBO) data was used to analyze bonding trends throughout the mechanism. It is first applied to the generic mechanism to obtain a general picture which would later serve as a background for the ligand exchange states and let us track the deviations electronically by means of donor−acceptor interactions. Stabilization due to donor→acceptor delocalization energy E(2) is determined as
E(2) = ΔEij = qi B
F(i , j)2 εj − εi DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics
σ for the σ bond, n for the lone pair and ωN12−C6−N13 for three-center− four-electron π interactions involving the N−C−N array of the NHC. In NBO analysis an asterisk (*) signifies either poorly occupied or antibonding orbitals.
where qi represents the occupancy of the donor orbital, εi and εj represent orbital energies, and F(i,j) represents the off-diagonal NBO Fock matrix element. Coordinate bonding was traced from the donor→acceptor interactions, since they often do not appear explicitly in NBO data. E(2) values taken into consideration are therefore often large. Since the second-order perturbation theory becomes quantitatively unreliable for large values, the NBO data were analyzed qualitatively for these interactions whose E(2) values should be considered only internally consistent. The following shorthand notation was used throughout the discussion: Q A for the charge of the atom A, RAB for the distance between atoms A and B, AABC for the angle involving the A−B−C array, DABCD for the dihedral angle involving the A−B−C−D array, π for the π bond (especially, πAB is the first π bond and πAB is the second π bond),
■
RESULTS AND DISCUSSION In this study, binuclear copper acetylide, which is the initial structure that forms from deprotonation of alkynes in the regular click procedure (Figure 1), was considered initially (Figure 2). The generic reaction path is given in Scheme 1. To address the problem of solvent−ligand competition for the Cu centers raised by Finn et al.,2g the stabilities of their Cu(I) acetylide complexes were compared. The binuclear copper acetylide
Scheme 1. Energy Profile of Generic Pathway (Shown in Black) with Proposed Possible Ligand-Exchange Mechanism (Shown in Red)a
a
Energies are not to scale. Complex 9 (also seen in Scheme 3) is not observed in the experiment of Fokin et al.2a The possibility of its formation is discussed in the text. C
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics
nation sites: πC3C4→n*Cu2 and nCu2→π*C3C4 interactions (μ2-η1:η2 mode interactions in Table 1) cause slight distortion in the alkyne symmetry of C4, whose distortion can be understood by means of the Chatt−Dewar−Duncanson model12 and can be easily seen by inspecting AC3C4C5 (171°). ACu1C3C4 is narrowed to 173° by σCu1−C3→n*Cu2, nCu2→σ*Cu1−C3, and the cuprophilic interactions (nCu1→n*Cu2 and nCu2→n*Cu1). As a result, Cu2 settles closer to C3 in comparison to C4 (RCu2C3 = 1.95 Å, RCu2C4 = 2.22 Å). Back-donation (nCu1→ω*N12−C6−N13) is found to be small throughout the mechanism (Table 1), as reported in a detailed NBO study of various carbene complexes.13 The evidence of a van der Waals (vdW) complex similar to that in 2 (Figure 6) has been recently reported.2k As a consequence of the electrostatic characteristics of Cu2 and N9 atoms (QCu2 = 0.59, QN9 = −0.38, RCu2N9 = 2.16 Å), the nN9→ n*Cu2 interaction is rather small in comparison to σCu1−C3→ n*Cu2 (2 in Table 1). A recent topology analysis has revealed3g that the electron pair of σCu1−C3 should be regarded as the lone pair of C3 in 2. This can be seen by the polarization coefficients 0.35 and 0.95 for Cu1 and C3, respectively. Accordingly, the σCu1−C3→n*Cu2 interaction is quite large. The bond between Cu1 and C3 is weakened by the filling of its antibonding orbital through the nCu2→σ*Cu1−C3 interaction. Cuprophilic interactions (nCu1→n*Cu2 and nCu2→n*Cu1) become stronger and direct the two Cu centers to a closer proximity of 2.47 Å, which is the characteristic Cu−Cu bond distance.3a,4
Figure 5. 3-D model of cationic acetylide−Cu−CH3CN complex 1.
complex with a CH3CN ligand (1 in Figure 5) is 9.2 kcal/mol more stable than the analogous complex with THF. Additionally, intermediate 1 is 11.2 kcal/mol more stable than the acetylide with a free [Cu(CH3CN)4]+ cation. Since the catalyst Cu2 was not used in a catalytic amount in the experiment,2a all of the acetylides are expected to ligate to Cu2 in the early stages of the reaction. Accordingly, a copper acetylide complex with a CH3CN ligand (1) is the variant of the experimentally observed2j starting intermediate for the reaction path. In 1 (Figure 5), acetylide coordinates asymmetrically to Cu1 and Cu2 via σ and π coordinations (μ2-η1:η2), respectively. NBO data displayed in Table 1 demonstrates these coordi-
Table 1. E(2) Values (kcal/mol) of the Categorized Interactions for the Structures Mentioned in the Results and Discussion state interaction
1
2
3
4
5
6
7
8
4L
XCA
4′
5′
6′
7′
8′
2L
2L‑L
XCB
2′
3′
μ2-η1:η2 Mode πC3C4→n*Cu2 nCu2→π*C3C4
46.64 13.19 7.62
10.22 26.87 45.60 10.19
6.94
3.48
7.04
7.25
2.42
μ2-η1:η1 Mode σCu1−C3→n*Cu2
148.53 415.76
462.10 340.75
183.49
14.69 13.04
10.58
582.69 395.05
246.18 142.47
179.79 314.01 318.88
n′C3→n*Cu2 nCu2→σ*Cu1−C3
44.19 7.39
6.94
19.98
6.36
13.55 14.77
13.41
n′C3→n*Cu1
9.71 33.96
36.10
σCu2−C3→n*Cu1
223.08
142.47
nCu1→σ*Cu2−C3
22.98
14.77
495.67
183.49
345.95
19.44
10.58
21.59
Cuprophilic nCu1→n*Cu2 nCu2→n*Cu1
27.41 71.27
0.00 63.65 57.71 15.83 55.86
6.94 19.82 77.37 21.16 11.41
146.56 33.13
7.04 14.92
6.84 38.81
7.04 14.92
10.88 10.18 85.00 18.36 15.83 55.86
37.59 71.83 101.19 58.14 34.81 10.65
0.27
9.73 38.74 168.71
Azide−Acetylide nN9→n*Cu2
37.54
4.59
σC3−N9→n*Cu2 n*C4→π*N7N8
5.17
8.12
0.00
96.32 62.52
1.41
5.89 2.48
7.44
0.18
79.33 111.51 58.96 53.58
3.41
5.14
3.23
2.91
2.47 49.20 35.50
4.78
0.00 1.34
592.06
620.56 πC3C4 Breaking
nC3→n*C4
750.38
745.64 Ligand Exchange
σCu1−C6→n*Cu2
1.95
1.59
1.26
2.66
1.56
6.71
2.15
41.15
nCu2→σ*Cu1−C6
0.17
0.45 21.70
0.64
0.08
7.30
0.63
7.94 106.80
8.67
8.39
6.71
2.15
nCu1→ω*N12−C6−N13
5.55
6.62
5.57
5.20
5.44
6.07
6.68 7.39
3.27
4.56
0.00
0.00
0.00
0.00
nCu2→ω*N12−C6−N13
0.00
0.00
0.18
0.00
0.00
0.00
0.00
0.78
3.16
4.53
4.77
6.07
σCu2−C6→n*Cu1
D
45.94 22.41 160.26 16.74
5.45
6.46
5.47
3.86
2.72
0.10
0.00
6.68 7.39 1.14
0.71
3.35
4.87
6.22
20.30 15.08
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics
Figure 7. 3-D model of cycloaddition TS 3. Figure 6. 3-D model of vdW complex 2.
donor−acceptor manner, which allows large electron transfer interactions toward N7 (nC3→n*C4 and n*C4→π*N7N8). As a result of these interactions, σC4−N7 is formed in the subsequent intermediate 4. Although NBO analysis agrees in many respects with the recent topology investigation by Quirante et al.,3g the direction of the electronic transition in the formation of the σC4−N7 bond in NBO analysis herein is reversed. Table 2 sum-
These interactions indicate C3’s comparable coordination to Cu2 and Cu1 in 2. The π system of the alkyne coordinates to Cu2 with weaker πC3C4→n*Cu2 and nCu2→π*C3C4 interactions in comparison to those in 1. Weakening of these interactions re-forms AC3C4C5 to an ideal alkyne angle of 179°. These interactions lead to a more symmetric σ:σ coordination mode (μ2-η1:η1): e.g., bond distances RCu1C3 = 1.93 Å and RCu2C3 = 1.92 Å and bond angles ACu1C3C4 = 130° and ACu2C3C4 = 150°. In the mechanism, there are cases when C3 coordinates to Cu1 more strongly than to Cu2 (1, 2, 4, 5, 7, 4L, XCA, 5′, 6′,2L, 2L‑L, XCB, and 2′). In these cases, σCu2−C3 does not appear as a bond in NBO data; instead, it appears as donor→acceptor interactions (σCu1−C3→n*Cu2 and nCu2→σ*Cu1−C3) with large E(2) values. Similarly, there are cases when C3 coordinates to Cu2 more strongly than to Cu1 (3, 6, 4′, 7′, 3′). In these cases donor→acceptor interactions (σCu2−C3→n*Cu1, nCu1→σ*Cu2−C3) with large E(2) values appear instead of σCu1−C3 in the NBO data. It should be reemphasized that the absence or presence of these bonds is due to reinterpretation of molecular orbitals in terms of NBOs with the corresponding Lewis structures, which cannot accommodate the C3 lone pair’s simultaneous interaction with both Cu centers. As far as the μ2-η1:η1 coordination mode is concerned, the C3 lone pair coordinates to both Cu centers. In subsequent sections, these NBO interactions will not be mentioned explicitly. Instead, they will be referred to as “μ2-η1:η1 mode interactions”, for they tend to promote the magnitude of C3 coordination to Cu centers in a symmetric way. These interactions are present even when the complex is not in the μ2-η1:η1 mode (e.g., in intermediate 1, the complex is in μ2-η1:η2 mode but, as discussed, μ2-η1:η1 mode interactions cause a considerable difference between RCu2C3 and RCu2C4). However, when the complex is in μ2-η1:η1 mode, its signature E(2) values become significantly higher (Table 1) and μ2-η1:η2 mode interactions vanish. In TS 3, Cu2 interacts more strongly with both C3 (RCu2C3 = 1.88 Å) and N9 (RCu2N9 = 2.00 Å) in comparison to that in 2. These interactions emerge as σCu2−C3 and σCu2−N9 bonds in azide addition TS 3 (Figure 7), which has a relative energy of 16.4 kcal/mol. The μ2-η1:η1 mode interactions magnify the equivalent coordination of C3 to Cu centers. This effect is reflected in bond distances RC3Cu1 = 1.91 Å and RCu2C3 = 1.88 Å and bond angles ACu1C3C4 = 142° and ACu2C3C4 = 136°. The partially broken πC3C4 bond emerges as nC3 and n*C4 lone pair orbitals, which are populated as 1.03 and 0.88, respectively. The lone pair orbitals continue to interact with a strong
Table 2. Occupancy of Selected NBOs for 2−4 state NBO
2
3
πC3C4 πN7N8 n7 n8 σC4−N7
1.91 1.98 1.96
1.93 1. 90
4
1.83 1.75 1.91
marizes the important NBOs which may be responsible for the formation of the σC4−N7 bond. According to Table 2, the existence of n7 remains invariant. Therefore, this NBO is not involved in the formation of σC4−N7. The annihilation of πC3C4 and πN7N8 and the creation of previously nonexistent σC4−N7 and n8 in 4 show that the migration of πN7N8 toward N8 causes the formation of n8. The total NBO charge of the azide fragment changes by −0.41 in 4 in comparison to 2. These results prove the self-consistency of the NBO analysis. Calculations of electrostatic potential (ESP) derived charges for the azide fragment confirm the NBO analysis (−0.50 < ΔQ azide < −0.32, Table S1 in the Supporting Information). It should be pointed out that this is the only part where previous QTAIM results contradict the NBO analysis. However, both methods have their own drawbacks; thus, the results should be considered cautiously. In intermediate 4 (Figure 8), the donor−acceptor interaction between C3 and N9 appears as a σC3−N9 bond with polarization coefficients 0.41 and 0.91 for C3 and N9, respectively, despite the long range (RC3N9 = 2.63 Å). This pseudo bond is mostly comprised of a N9 lone pair, which is responsible for the vdW interaction between N9 and Cu2 in 2. Mainly, its shift toward C3 causes the migration of πC3C4 electrons toward N7 in TS 3. In intermediate 4, the strength of the interaction between Cu2 and N9 is increased in comparison to that in intermediate 2, thanks to the σC3−N9→n*Cu2 interaction. As a result, RCu2N9 decreases to 1.94 Å. Meanwhile, C3 coordinations to Cu1 and Cu2 are preserved by μ2-η1:η1 mode signature interactions. E
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics
Figure 11. 3-D model of vdW complex 7.
Scheme 2. Energy Diagram of the Reaction When THF Is Used as the Ligand Instead of CH3CNa Figure 8. 3-D model of intermediate 4.
Figure 9. 3-D model of ring contraction TS 5.
The six-membered metallacycle closes to a five-membered triazolide ring via TS 5 (Figure 9). TS 5 has a very low barrier (2.5 kcal/mol) due to the emergent σC3−N9 bond that is already present in 4. The rotation of the σC3−Cu2 bond (DN7C4C3Cu2 = 41° in 5) furnishes the exergonic ring contraction product 6, in which the emergent σC3−N9 becomes a proper bond and σC3−Cu2 aligns perpendicularly to the triazolide ring (Figure 10). a
Energies are calculated relative to 1.
For completeness, the cycloaddition mechanism with THF as ligand rather than CH3CN is considered (Scheme 2). Although the reaction barrier with THF is comparable to that of the generic path, formation of 1(THF) is unlikely to occur, since the exchange with CH3CN will be endergonic by 9.2 kcal/mol. Furthermore, all of the structures on this path are relatively higher in energy than the generic path. According to these results THF cannot be an efficient competing ligand. Next, we considered how to incorporate an EI type of a product (Figure 3) as observed in the experiment of Bertrand et al. in the model considered in this study. The presence of EI might be evaluated via two different ways, depending on which stage the ligation of the triazolide nitrogen to [(CAAC)Cu]+ cation takes place: it can be formed during or after the cycloaddition. To test the possibility of the formation of an EI type of product during the cycloaddition step (path C in Scheme 3), the model system used in this study was considered. Our results showed that path C requires TS C3 with a very high barrier energy (38.8 kcal/mol). Therefore, path C was refuted in this model system. The negative charges on N9 and C3 (QN9 = −0.25 and QC3 = −0.21) prevent an effective overlap for bond formation and the azide is approximately 15° out of the alkyne plane, rendering this addition energetically unlikely.
Figure 10. 3-D model of intermediate 6.
Ligation of another CH3CN to Cu2, which is less hindered and electronically less occupied in comparison to Cu1 (QCu2 = 0.64, QCu1 = 0.45 in 6), weakens Cu2’s interaction with the remaining complex. Intermediate 7, which is a vdW complex with distances of RCu2C3 = 2.15 Å and RCu1Cu2 = 2.51 Å, is obtained via this ligation (Figure 11). Coordination of two more CH3CN ligands to practically detached Cu2 restores the original catalyst complex while the product triazolide 8 is liberated. In conventional CuAAC reactions, triazole is generated via protonation by either an alkyne or another proton source. Therefore, the absence of a proton donor prevents the formation of the triazole. F
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics Scheme 3. Path Ca
Figure 13. 3-D model of unobserved product 9.
complex corresponding to model structure 9. The identification of a triazolide corresponding to model structure 8 was achieved via an LC-MS analysis (using a TOF MSD) of the reaction mixture, in which relatively weak [(CH3CN)Cu]+ substitution could be removed due to fragmentation. Therefore, the presence of 9 might have not been detected in the experiment of Fokin et al. The ligand exchange requires a triggering factor, which can cause transient deviations from the usual pathway. This may stem from transient expansion of the coordination number of the complex by ligation of another CH3CN molecule. The energetics obtained by extra ligation of a ligand to the structures of the generic path are shown in Scheme 4. Generally, ligation of an extra ligand to Cu1 (states with subscript ligands) or to Cu2 a
Scheme 4. Effect of Additional Ligand (THF or L = CH3CN) on the Energies of the Structures of the Generic Patha
Energies are not to scale.
Although a possible reaction via C3 (Scheme 3) requires a very high barrier, the model system of Bertrand et al. was also modeled in order to check if the two NHC ligands as in their system facilitated a C3-like transition state. C3-likeTS and 3-likeTS leading to an EI type product (Figure 3), in which NHC was used instead of CH3CN as ligand, were compared: C3-likeTS was found to require 25.1 kcal/mol more energy than 3-likeTS (Figure 12). These results reinforce the thesis
Figure 12. Free energies of 3-like TS and C3-like TS. Calculations are performed in the CH2Cl2 phase, and the free energies are relative to 3-like TS.
that C3-likeTS is not electronically favored. This leads us to deduce that the ligation of the complex to the [(CAAC)Cu]+ cation should take place after the formation of mononuclear triazolide in the experiment of Bertrand et al. In the model system of Fokin et al., this would correspond to the ligation of 8 to [(CH3CN)Cu]+, giving the product 9 (Figure 13 and Scheme 1). Although an NBO investigation of the mononuclear triazolide 8 revealed that its N7 position can be a likely donor (QN7 = −0.34) to [(CH3CN)Cu]+, Fokin et al. did not report a
a
Subscript ligands are ligated to Cu1, and superscript ligands are ligated to Cu2.
(states with superscript ligands) resulted with the weakly bound complexes which are less stable than that of the generic path’s original state. In 1, ligations to Cu2 are more likely than those to Cu1. However, in every case, the ligation is weak. Beginning from 2, the ligation of an extra ligand was not possible to Cu2, G
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics
interactions, RCu2C6 and RCu1Cu2 shorten to 3.20 and 2.40 Å, respectively. 4L maintains the μ2-η1:η1 coordination mode by its signature interactions (Table 1). In addition, the highly polarized pseudo σC3−N9 bond in 4 is preserved. A ring contraction TS starting from 4L was not possible. This might be due to the additional rigidity acquired from the emerging interactions between Cu2 and C6, which prevent the rotation of the σC3−Cu2 bond as described in the discussion of TS 5. The ligand exchange transition state (XCA in Figure 15) following 4L has a barrier of 11.9 kcal/mol, which is almost
since it is already a vdW complex formed with the azide. Ligations to Cu1 were weak and energetically demanding with one exception; ligation of CH3CN to Cu1 in 4 resulted in the more stable state 4L. Naturally, the first choice of structure as a host for the incoming ligand should be 4 (Scheme 1). The presence of an additional ligand changes the complex location on the energy profile by decreasing its relative free energy to 6.1 kcal/mol (4L in Scheme 1 and Figure 14). This stabilization by an additional CH3CN ligand (L′) can be traced back to the experiments conducted by Finn et al.,2g Heaney et al.2d and even the original paper by Kubas,14 who first described the synthesis of the tightly bound [Cu(CH3CN)4][PF6] complex. As noted in the Introduction, Finn et al. suggests that, with the increasing concentration of ligands, the inhibition probability of CuAAC increases due to the metal−ligand interaction. Coordination of L′ to 4 (originating from [Cu(CH3CN)4][PF6] which can introduce 3 equiv of extra CH3CN ligands) may prevent TS 5 by initiating the formation of the more stable intermediate 4L and slightly inhibiting the CuAAC reaction. L′ causes the NHC ring to align perpendicularly to the otherwise planar complex (DCu2Cu1C6N12 = 81° in Figure 14).
Figure 15. 3-D model of ligand exchange TS XCA.
the same as that of the generic path’s maximum barrier (11.0 kcal/mol). The NHC moiety migrates toward Cu2 in the attraction of both Cu centers and coordinates to Cu2 more strongly (RCu2C6 = 2.13 Å). Although σCu1−C6 is broken and instead emergent σCu2−C6 is formed, NHC and Cu1 are still interacting via σCu2−C6→n*Cu1, which can be read as nC6→ n*Cu1. The back-donation nCu2→ω*N12−C6−N13 is stronger in comparison to its value in 4L. In XCA, reduced charge on Cu2 (QCu2 = 0.45; the lowest yet) and competing ligand NHC diminish the interaction of Cu2 and the original CH3CN ligand (L). While L′ binds to Cu1 and RCu1N11 decreases to 1.88 Å, RCu2N10 increases to 2.05 Å. At the end of this transformation L is emitted from the complex (Figure 16).
Figure 14. 3-D model of ligand exchange initiating intermediate 4L.
This perpendicular alignment directs the NHC ring’s poorly occupied ω*N12−C6−N13 orbital (occupancy of ω*N12−C6−N13 is 0.40 in 4L) toward nCu2, whose NBOs consist of d orbitals. This direct alignment allows back-donation through the nCu2→ ω*N12−C6−N13 interaction. Its magnitude is also small, as in the aforementioned back-donation interaction in the generic intermediates (nCu1→ω*N12−C6−N13, Table 1). It can be used to evaluate the relative extent of interaction between C6 and Cu2. Beginning from 4L, nCu2→ω*N12−C6−N13 starts to grow evidently in XCA, 4′, 6′, 7′, and 8′, in comparison to the null E(2) values in the former generic intermediates (1, 2, 4, 6, 7) (Table 1). This interaction and perpendicular alignment of NHC ring are synergic events which cause the original backdonation interaction (nCu1→ω*N12−C6−N13) to recede and σCu1−C6 to weaken. The weakening of σCu1−C6 is also due to the antibonding orbital σ*Cu1−C6, which is being populated via the nCu2→σ*Cu1−C6 interaction (Table 1). Quantitatively speaking, among the ligand exchange initiating interactions, the most significant one is σCu1−C6→n*Cu2, which can be read as nC6→n*Cu2, since σCu1−C6 is mainly populated by a carbene electron pair (polarization coefficients for Cu1 and C6 are 0.39 and 0.92, respectively). This last interaction shows nC6’s increasing interest to Cu2. In agreement with these emerging
Figure 16. 3-D model of analogous intermediate 4′.
4′ (Figure 16) and 5′ are analogous ligand-exchanged structures of 4 and 5, respectively, and they have similar roles in CuAAC catalysis. Comparison of 4′ and 5′ with 4 and 5 shows how NHC reduces the charge of the atom to which it is ligated: charges of Cu atoms in 4′ (QCu1 = 0.71 > QCu2 = 0.47) and in H
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics 5′ (QCu1 = 0.70 > QCu2 = 0.49) are ranked in reverse order in comparison to charges in 4 (QCu1 = 0.56 < QCu2 = 0.65) and in 5 (QCu1 = 0.55 < QCu2 = 0.65). There are no significant changes in the energies of 4′ and 5′ (10.0 and 12.5 kcal/mol, respectively) in comparison to those of 4 and 5 (10.1 and 12.6 kcal/mol, respectively). Especially, equivalency of 4 and 4′ in the energy profile (Scheme 1) is very compatible with the experimentally observed 50% enrichment.2a Subsequent intermediates 6′ and 7′ are followed by 8′ which is the isotopically enriched variant of 8 (Scheme 1). Although 2L (Figure 17) was found to be significantly unfavorable (11.1 kcal/mol in Scheme 4) in comparison to the generic intermediate 2, the ligand exchange possibility starting from 2L was explored (Scheme 5) as an alternative to that in Scheme 1. If the azide coordinates more strongly to Cu2 of intermediate 2L, L is emitted from the system to form 2L−L (Figure 18), which is only 0.9 kcal/mol above the intermediate
Figure 18. 3-D model of ligand exchange initiating intermediate 2L−L.
Scheme 5. Alternative Ligand-Exchange Patha
Figure 19. 3-D model of ligand exchange TS XCB.
2L on the energy profile (Scheme 5). In TS XCB (Figure 19), The NHC moiety aligns perpendicularly (DCu2Cu1C6N12 = 95°) and this alignment allows stronger ligand exchange initiating interactions (Table 1) which facilitate the breaking of the loose σCu1−C6 bond (RCu1C6 = 2.04 Å). TS takes place via a scissoring angle distortion (AC6Cu1Cu2 = 66°), which brings Cu2 and C6 closer (RCu2C6 = 2.44 Å), while the coordination of L′ to Cu1 (RCu1N11 = 1.93 Å) is enhanced. The following intermediate 2′ (Figure 20) and TS 3′ are analogous structures of vdW complex
a
The energies are relative to 1 in Scheme 1.
Figure 20. 3-D model of analogous vdW complex 2′.
2 and TS 3, respectively, where only the ligands are exchanged. The relative energy of TS 3′ is similar (15.7 kcal/mol) to that of its analogous TS 3 (Scheme 1; 16.4 kcal/mol). IRC calculation of TS 3′ gives the intermediate 4′ (Figure 16 and Scheme 1), and therefore, the mechanism evolves identically with that in Scheme 1 from there on. Although the energy of XCB (14.6 kcal/mol) is lower in comparison to that of XCA (18.0 kcal/mol) and even to that of 3 (16.4 kcal/mol), ligand exchange via Scheme 5 can hardly be an alternative path to Scheme 1 since 2L’s low stability renders it an unlikely resting state.
Figure 17. 3-D model of ligand exchange initiating intermediate 2L. I
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics This is also in accordance with the experiment of Fokin et al., where they did not observe isotopically enriched copper acetylide in the samples withdrawn from the reaction mixture while the reaction was still in progress.
■
■
CONCLUSION In this DFT study, the binuclear CuAAC mechanism was thoroughly modeled. This investigation was carried out with NBO analysis that allowed association of various transformations involved in the CuAAC catalysis with the categorized NBO interactions. The effect of stabilization due to delocalization on the intermediates and the transition states modeled in the reaction path was discussed, and eventually their roles in the mechanism and the energetics were enlightened. The experimental isotopic enrichment was explained by considering the possible pathways leading to the ligand exchanges. The investigation of the ligand exchange revealed that this transformation could spring from the generic mechanism at two different stages: either from intermediate 4 or from vdW complex 2 with facile transition state and barrier energies. In both cases, the initiation of the exchange comes from the ligation of another CH3CN molecule to Cu1. This ligation induces NBO interactions that facilitate the migration of the NHC ring toward Cu2. In two alternative ligand exchange mechanisms (Scheme 1 and Scheme 5), while ligand exchange via TS XCB is more speculative than conclusive, the mechanism via XCA is a facile path. These results are compatible with the structures identified in the experiments, and therefore the observed isotopic enrichment can be successfully explained. However, substitution on azide or alkyne may still have an effect on the possible reaction paths. From an experimental point of view, this investigation reveals an important point concerning the effect of the strength of the surrounding solvent or ligand molecules on the reactivity of the copper acetylide complexes: the formation of a stable intermediate such as 4L has the potential to hamper the catalysis if it becomes more stable with the incorporation of a ligand that can stabilize it even further. Although the CuAAC reaction has become a very successful catalytic scheme, this fact exhibits that the reaction system is still nontrivial and requires care in choosing the reaction medium and the form of the catalyst. Additionally, if a structure such as 4L were more stable, it could possibly be isolated. In that case, the binuclear mechanism and its stepwise nature would be further ascertained by demonstrating the formation of the elusive six-membered metallacycle 4, which has not been observed or isolated in any experiment. In this study, formation of the recently isolated intermediate was accounted for by the DFT calculations. It was shown that the formation of this new bis-copper triazolide intermediate is energetically highly unfavorable during the cycloaddition steps; however, it can be facile after the formation of the mononuclear triazolide 8. With this work, we hope to complement the seminal computational and experimental work in the literature of the CuAAC mechanism so that this can encourage creative experiments with both mechanistic and practical scopes in click chemistry and beyond.
■
Short discussion about the bonding situation between C4 and N7 and calculated energies for all optimized stationary points (PDF) All computed molecule Cartesian coordinates (XYZ)
AUTHOR INFORMATION
Corresponding Author
*E-mail for N.S.T.:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS ̇ AK 109T251 This research was made possible by the TÜ BIT project, National Center for High Performance Computing ̇ AK ULAKBIM, High Per(Grant number 10982010), and TÜ BIT formance and Grid Computing Center (TR-Grid e-Infrastructure).
■
REFERENCES
(1) (a) Tornoe, C. W.; Christensen, C.; Meldal, M. J. Org. Chem. 2002, 67, 3057−3064. (b) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B. Angew. Chem., Int. Ed. 2002, 41, 2596−2599. (2) (a) Worrell, B. T.; Malik, J. A.; Fokin, V. V. Science 2013, 340, 457−460. (b) Lewis, W. G.; Magallon, F. G.; Fokin, V. V.; Finn, M. G. J. Am. Chem. Soc. 2004, 126, 9152−9153. (c) Rodionov, V. O.; Fokin, V. V.; Finn, M. G. Angew. Chem., Int. Ed. 2005, 44, 2210−2215. (d) Buckley, B. R.; Dann, S. E.; Heaney, H. Chem. - Eur. J. 2010, 16, 6278−6284. (e) Smith, N. W.; Polenz, B. P.; Johnson, S. B.; Dzyuba, S. V. Tetrahedron Lett. 2010, 51, 550−553. (f) Buckley, B. R.; Dann, S. E.; Harris, D. P.; Heaney, H.; Stubbs, E. C. Chem. Commun. 2010, 46, 2274−2276. (g) Presolski, S. I.; Hong, V.; Cho, S. H.; Finn, M. G. J. Am. Chem. Soc. 2010, 132, 14570−14576. (h) Nolte, C.; Mayer, P.; Straub, B. F. Angew. Chem., Int. Ed. 2007, 46, 2101−2103. (i) Makarem, A.; Berg, R.; Rominger, F.; Straub, B. F. Angew. Chem., Int. Ed. 2015, 54, 7431−7435. (j) Jin, L.; Tolentino, D. R.; Melaimi, M.; Bertrand, G. Science advances 2015, 1, e1500304. (k) Iacobucci, C.; Reale, S.; Gal, J.F.; De Angelis, F. Angew. Chem., Int. Ed. 2015, 54, 3065−3068. (l) Rodionov, V. O.; Presolski, S. I.; Gardinier, S.; Lim, Y.-H.; Finn, M. G. J. Am. Chem. Soc. 2007, 129, 12696−12704. (m) Rodionov, V. O.; Presolski, S. I.; Díaz Díaz, D.; Fokin, V. V.; Finn, M. G. J. Am. Chem. Soc. 2007, 129, 12705−12712. (n) Berg, R.; Straub, J.; Schreiner, E.; Mader, S.; Rominger, F.; Straub, B. F. Adv. Synth. Catal. 2012, 354, 3445−3450. (3) (a) Ahlquist, M.; Fokin, V. V. Organometallics 2007, 26, 4389− 4391. (b) Straub, B. F. Chem. Commun. 2007, 7, 3868−3870. (c) Himo, F.; Lovell, T.; Hilgraf, R.; Rostovtsev, V. V.; Noodleman, L.; Sharpless, K. B.; Fokin, V. V. J. Am. Chem. Soc. 2005, 127, 210−216. (d) Ozen, C.; Tuzun, N. S. J. Mol. Graphics Modell. 2012, 34, 101− 107. (e) Cantillo, D.; Avalos, M.; Babiano, R.; Cintas, P.; Jimenez, J. L.; Palacios, J. C. Org. Biomol. Chem. 2011, 9, 2952−2958. (f) CalvoLosada, S.; Pino, M. S.; Quirante, J. J. J. Mol. Model. 2014, 20, 1−7. (g) Calvo-Losada, S.; Pino-Gonzalez, M. S.; Quirante, J. J. J. Phys. Chem. B 2015, 119, 1243−12458. (h) Straub, B. F.; Bessel, M.; Berg, R. Dicopper Catalysts for the Azide Alkyne Cycloaddition: A Mechanistic DFT Study. In Modeling of Molecular Properties; Comba, P., Ed.; Wiley-VCH: Weinheim, Germany, 2011; pp 207−214. (i) Wang, C.; Ikhlef, D.; Kahlal, S.; Saillard, J. Y.; Astruc, D. Coord. Chem. Rev. 2016, 316, 1−20. (4) Chui, S. S. Y.; Ng, M. F. Y.; Che, C. M. Chem. - Eur. J. 2005, 11, 1739−1749. (5) (a) Diez-Gonzalez, S.; Correa, A.; Cavallo, L.; Nolan, S. P. Chem. - Eur. J. 2006, 12, 7558−7564. (b) Goj, L. A.; Blue, E. D.; MunroLeighton, C.; Gunnoe, T. B.; Petersen, J. L. Inorg. Chem. 2005, 44, 8647−8649. (c) Díez-González, S.; Nolan, S. P. Angew. Chem., Int. Ed. 2008, 47, 8881−8884. (d) Lal, S.; Rzepa, H. S.; Díez-González, S. ACS Catal. 2014, 4, 2274−2287. (e) Diez-Gonzalez, S.; Escudero-Adan, E.
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.organomet.6b00279. J
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
Article
Organometallics C.; Benet-Buchholz, J.; Stevens, E. D.; Slawin, A. M. Z.; Nolan, S. P. Dalton Transactions 2010, 39, 7595−7606. (6) Zhao, Y.; Truhlar, D. G. J. Chem. Phys. 2006, 125, 194101− 194118. (7) Tsipis, A. C. RSC Adv. 2014, 4, 32504−32529. (8) Roy, L. E.; Hay, P. J.; Martin, R. L. J. Chem. Theory Comput. 2008, 4, 1029−1031. (9) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M. J.; Heyd, J.; Brothers, E. N.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A. P.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, N. J.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, Ö .; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09; Gaussian, Inc., Wallingford, CT, USA, 2009. (10) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105, 2999−3093. (11) (a) Gonzalez, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154− 2161. (b) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523− 5527. (12) (a) Dewar, M. Bull. Soc. Chim. Fr. 1951, 18, C71−C79. (b) Chatt, J.; Duncanson, L. A. J. Chem. Soc. 1953, 2939−2947. (c) Chatt, J.; Duncanson, L. A.; Venanzi, L. M. J. Chem. Soc. 1955, 4456−4460. (13) Comas-Vives, A.; Harvey, J. N. Eur. J. Inorg. Chem. 2011, 2011, 5025−5035. (14) Kubas, G. J.; Monzyk, B.; Crumbliss, A. L. Tetrakis(Acetonitrile)Copper(I) Hexafluorophosphate. In Inorganic Syntheses; Wiley: Hoboken, NJ, 2007; pp 90−92.
K
DOI: 10.1021/acs.organomet.6b00279 Organometallics XXXX, XXX, XXX−XXX
