Article pubs.acs.org/joc
Cite This: J. Org. Chem. 2017, 82, 13572−13582
Au(I)-Catalyzed Dimerization of Two Alkyne UnitsInterplay between Butadienyl and Cyclopropenylmethyl Cation: Model Studies and Trapping Experiments Mathis Kreuzahler,† Sven Fabig,† Gebhard Haberhauer,*,† and Rolf Gleiter*,‡ †
Institut für Organische Chemie, Universität Duisburg-Essen, Universitätsstrasse 7, D-45117 Essen, Germany Organisch-Chemisches Institut, Universität Heidelberg, Im Neuenheimer Feld 270, D-69120 Heidelberg, Germany
‡
S Supporting Information *
ABSTRACT: In recent years, Au(I)-catalyzed reactions proved to be a valuable tool for the synthesis of substituted cycles by cycloaromatization and cycloisomerization starting from alkynes. Despite the myriad of Au(I)-catalyzed reactions of alkynes, the mono Au(I)-catalyzed pendant to the radical dimerization of nonconjugated alkyne units has not been investigated by quantum chemical calculations. Herein, by means of quantum chemical calculations, we describe the mono Au(I)-catalyzed dimerization of two alkyne units as well as the transannular ring closure reaction of a nonconjugated diyne. We found that depending on the system and the method used either the corresponding cyclopropenylmethyl cation or the butadienyl cation represents the stable intermediate. This circumstance could be explained by different stabilizing effects. Moreover, the calculation reveals a dramatic (>1012-fold) acceleration of the Au(I)-catalyzed reaction compared to that of the noncatalyzed radical variant. Trapping experiments with a substituted 1,6-cyclodecadiyne using benzene as a solvent at room temperature as well as studies with deuterated solvents confirm the calculations. In this context, we also demonstrate that trapping of the cationic intermediate with benzene does not proceed via a Friedel−Crafts-type reaction.
■
INTRODUCTION The Bergman cyclization of enediynes, discovered in the 1970s, represents a very interesting method for generating highly reactive diradical intermediates en route to new aromatic rings.1 The first investigations of this reaction included the thermal cyclization of (Z)-hex-3-en-1,5-diyne (1) at temperatures of >200 °C leading to 1,4-didehydrobenzene intermediate 2 (Scheme 1a). This temperature is necessary to overcome the repulsive interaction between the alkyne inplane π orbitals.2 1,4-Didehydrobenzene 2 represents a reactive 1,4-diradical intermediate and can irreversibly abstract hydrogen atoms from an appropriate donor to produce benzene.3 In the 1990s, Gleiter and co-workers4 showed that nonconjugated cyclic diynes undergo a transannular ring closure via a radical mechanism if the two parallel oriented alkyne units are close to each other. An example is the reaction of 1,6-cyclodecadiyne (3) to bicyclic 1,3-butadien-1,4diyl intermediate 4 shown in Scheme 1b.4−6 Abstraction of hydrogen atoms from donor molecules by intermediate 4 causes the formation of a bicyclic trans-butadiene system with two annulated six-membered rings. Recently, we were able to show that arylacetylenes substituted with an electron-withdrawing group at the triple bond dimerize via a radical mechanism around 120 °C (Scheme 1c).7 The activation barrier of this dimerization decreases with the increasing electronegativity of the substituent attached to the sp© 2017 American Chemical Society
Scheme 1. (a) Bergman Cyclization, (b) Transannular Ring Closure of 1,6-Cyclodecadiyne (3), and (c) Dimerization of Substituted Phenylacetylene 5a
a
The activation energies (Ea) are given in kilocalories per mole.
hybridized carbon atom,7 which can be explained by Bent’s rule.8 The dimerization reaction of methoxyarylacetylene 5 can even be used to synthesize bicyclic compounds on a preparative scale.7 Received: November 9, 2017 Published: November 17, 2017 13572
DOI: 10.1021/acs.joc.7b02843 J. Org. Chem. 2017, 82, 13572−13582
Article
The Journal of Organic Chemistry
rearrangement of 1,5-dienes.13,14 These investigations were inspired by the recent explosion of Au(I)-catalyzed cycloaromatization and cycloisomerization reactions of alkynes and alkenes.15,16 For example, the Bergman cyclization can be accelerated by dual Au(I) catalysis,17 whereby one gold atom is bound at the triple bond as an acetylide and the second one forms a π complex with an alkyne unit (Scheme 2b).18 However, despite the myriad of Au(I)-catalyzed reactions of alkynes, the mono Au(I)-catalyzed pendant to the radical dimerization of nonconjugated alkyne units4−7 has not been investigated by quantum chemical calculations to the best of our knowledge. In most experimentally investigated Au(I)catalyzed reactions of two nonconjugated alkyne units, the proposed mechanism does not represent a classical dimerization, as the alkyne units are activated by propargylic substituents16b,15e leading to cycloisomerization cascade reactions or an initial addition of a nucleophile19 takes place. In the case of the dual Au(I) catalysis of nonconjugated and nonactivated alkynes, experiments and computational studies reveal gold−vinylidene20 and gold−butadiene21 intermediates. The assumed mechanisms for the mono Au(I)-catalyzed classical dimerization (i.e., neither initial addition of a nucleophile nor involvement of a propargylic group) of two nonconjugated alkyne units are contradictory,22 and a butadienyl cation is often described as an intermediate:22 For the mono Au(I)-catalyzed dimerization of the alkyne units of 12, two possible intermediates (13a and 13b) are proposed (Scheme 2c).22 This inspired us to investigate the mono Au(I)-catalyzed dimerization of nonconjugated alkyne units by model calculations. This paper focuses on the two possible cationic intermediates (butadienyl vs cyclopropenylmethyl cation) and their relative energies. Moreover, we studied the transannular ring closure of 1,6-cyclodecadiynes by means of quantum chemical calculations and trapping experiments. This transannular cyclization with subsequent trapping by an appropriate nucleophile could establish an attractive route to highly substituted bicyclic systems.
Recently, computational studies regarding the formation of aryl and vinyl cations starting from alkynes with one gold center were performed.9,10 As an example, dos Passos Gomes and Alabugin10 have impressively shown by means of quantum chemical calculations that the use of Au(I) as a catalyst tremendously decreases the activation barrier of the Bergman cyclization. For the mono Au(I) catalysis, they obtained an activation barrier of 14.6 kcal/mol at the CCSD(T) level of theory (Scheme 2a). This is a decrease Scheme 2. (a) Mono Au(I)-Catalyzed Bergman Cyclization (Ea in kilocalories per mole), (b) Au(I)-Catalyzed Ring Closure of Enediyne 9 Followed by an Intermolecular Reaction with Benzene Yielding Arenes 10 and 11, and (c) Two Possible Proposed Intermediates (13a and 13b) for the Mono Au(I)-Catalyzed Dimerization of the Alkyne Units of 12
■
of >14 kcal/mol compared to that of the noncatalyzed radical cyclization11,12 (Scheme 1a). The catalytic power of Au(I) for the Bergman cyclization was explained by a combination of two sources: stereoelectronic assistance of C−C bond formation (LUMO umpolung) and crossover from a diradical to a zwitterionic mechanism.10 An acceleration of the reaction rate by Au(I) as a catalyst was also proven for the Cope
RESULTS AND DISCUSSION Quantum Chemical Model Studies of Au(I)-Catalyzed Dimerization of Alkyne Units and Transannular Cyclization. The first model system represents two acetylene molecules. Although it is known that acetylene and terminal
Scheme 3. (a) Model Studies of the Mono Au(I)-Catalyzed Dimerization of Two Alkyne Units to Cations 17 and 18, (b) Rearrangement of Butadienyl Cation 17 to the More Stable Cation 19, and (c) Model Studies of the Au(I)-Catalyzed Transannular Ring Closure of 20a to Butadienyl Cation 22a and Cyclopropenylmethyl Cation 23a
13573
DOI: 10.1021/acs.joc.7b02843 J. Org. Chem. 2017, 82, 13572−13582
Article
The Journal of Organic Chemistry alkynes undergo a fast formation to σ gold intermediates, we exclude this aspect in these calculations as we are interested only in the mono Au(I) catalysis of two alkyne units via C−C bond formation. As a second model system, we chose 1,6cyclodecadiyne. The latter is an ideal candidate for an experimental investigation of the mono Au(I)-catalyzed reaction. For the mono Au(I)-catalyzed dimerization of two acetylene units, the reaction of acetylene with 15, which corresponds to a π complex of acetylene with the Au(I) catalyst, was investigated. This model was chosen to allow a better comparison to the transannular cyclization of 20 (Scheme 3c) and the Au(I)-catalyzed Bergman cyclization (Scheme 2a).10 Please note that because of additional geometric constraints, transannular cyclizations are quite different from their intermolecular counterpart, and a comparison should be made with caution. As feasible products, butadienyl cations 17 and 22a as well as cyclopropenylmethyl cations 18 and 23a were presumed. The stationary points of this Au(I)-catalyzed dimerization were optimized using the PBE023 method. Basis set 6-311+ +G(d,p)24 has been used for C, H, N, Cl, and P. For Au, the def2-TZVP+ECP25 basis set has been employed. This level of theory proved to give the most reliable results for similar Au(I)-catalyzed cyclizations.10,26 Furthermore, single-point calculations by means of B2PLYP27 and CCSD(T)28 were performed on the PBE0-optimized structures. Here again, basis set 6-311++G(d,p) has been used for C, H, N, Cl, and P. For Au, the def2-TZVP+ECP basis set has been employed. In the case of the transannular cyclization of 20a, the geometric parameters were additionally calculated using density functionals PBE0-D3,29 B3LYP,30 B3LYP-D3,29 and M06-2X.31 To determine the effect of the solvent on reactivity, the transannular cyclization of 20a was optimized using benzene as a solvent. Gold catalyst Au+-PMe3 was used unless stated otherwise. In addition, the transannular cyclization was calculated with AuCl and a gold carbene complex as catalysts (see Tables 1 and 2 and Table S1).
Table 2. Energies (ΔE in kilocalories per mole) of 21a− 23a Relative to the Starting Material 20a Calculated by Different Methodse
a
B2PLYP//PBE0. bB2PLYP//B3LYP. cCCSD(T)//PBE0. dCCSD(T)//B3LYP. eBasis set 6-311++G(d,p)/def2-TZVP+ECP has been used.
Table 1. Energies (ΔE in kilocalories per mole) of 16−19 Relative to the Starting Material (acetylene and 15) Calculated by PBE0 and CCSD(T)a 16 17 18 19
R(C1−C3)
E (PBE0)b
E [CCSD(T)]c
2.277 1.521 1.592 1.347
6.69 9.25 −14.76 −34.83
9.79 19.30 −4.56 −23.65
Figure 1. Energy profiles for the Au(I)-catalyzed dimerization of two alkyne units via transition state 16 to cyclopropenylmethyl cation 18 and butadienyl cation 17 calculated using PBE0/6-311++G(d,p)/ def2-TZVP+ECP (black bars) and CCSD(T)/6-311++G(d,p)/def2TZVP+ECP (red bars). The geometries were optimized using PBE0/ 6-311++G(d,p)/def2-TZVP+ECP. [Au]+ = Au+-PMe3.
a
The distances R (angstroms) of the C1−C3 bond are also given. [Au] + = Au + -PMe 3 . b PBE0/6-311++G(d,p)/def2-TZVP+ECP. c CCSD(T)/6-311++G(d,p)/def2-TZVP+ECP//PBE0/6-311++G(d,p)/def2-TZVP+ECP.
whereas cation 17 was identified as a transition state. The calculated imaginary frequency of 17 describes the rearrangement of 17 to cation 19 (Scheme 3b), which is ∼23.7 kcal/ mol more stable than the starting material (acetylene and 15). While the energy difference for transition state 16 between the PBE0 and CCSD(T) values is rather small, the data for cations 18 and 17 differ significantly (∼10 kcal/mol). Obviously, the PBE0 approximation overestimates the stabilization of the cations. In the case of the transannular ring closure of 20a with Au+-PMe3 as a catalyst, the obtained structures strongly depend on the method used (Table 2 and Table S1). In the case of PBE0 and M06-2X, only cyclopropenylmethyl cation 23a was found as an intermediate. However, the usage of
Frequency calculations were performed at each of the structures to verify the nature of the stationary point. It turned out that all transition states as well as 17 have exactly one imaginary frequency, whereas all other structures have none. The calculated data are summarized in Tables 1 and 2 and Table S1 as well as in Figures 1 and 2. In the case of the dimerization of the alkyne units, both cationic species 17 and 18 were found. However, there is a huge energetic difference amounting to 23.9 kcal/mol on the CCSD(T) level (Figure 1). The frequency analysis shows that 18 is a global minimum 13574
DOI: 10.1021/acs.joc.7b02843 J. Org. Chem. 2017, 82, 13572−13582
Article
The Journal of Organic Chemistry
Figure 2. Energy profiles for the Au(I)-catalyzed transannular ring closure of 20a via transition state 21a to cyclopropenylmethyl cation 23a (left) calculated using PBE0/6-311++G(d,p)/def2-TZVP+ECP (black bars) and CCSD(T)/6-311++G(d,p)/def2-TZVP+ECP (red bars). The geometries were optimized using PBE0/6-311++G(d,p)/def2-TZVP+ECP. Energy profiles for the Au(I)-catalyzed transannular ring closure of 20a via transition state 21a to butadienyl cation 22a (right) calculated using B3LYP/6-311++G(d,p)/def2-TZVP+ECP (blue bars) and CCSD(T)/6-311++G(d,p)/def2-TZVP+ECP (red bars). The geometries were optimized using B3LYP/6-311++G(d,p)/def2-TZVP+ECP. [Au]+ = Au+-PMe3.
Figure 3. (a) Bond lengths, natural charges, and LUMO of gold-capped cations 17 and 18. (b) Bond lengths and natural charges of cationic intermediates 22a and 23a for the transannular ring closure of 20a. The data stem from PBE0/6-311++G(d,p)/def2-TZVP+ECP calculations.
functional M06-2X is an energy value of 16.7 kcal/mol obtained. The inclusion of empirical dispersion correction D3 leads to no substantial change in energy (Table 2). The usage of other gold complexes also enhances the activation barrier whereby Au(I)Cl represents the most ineffective catalyst having an activation energy of 22.2 kcal/mol. A similar trend has already been observed for the Au(I)-catalyzed Bergman cyclization.10 With CCSD(T) as the golden standard, the activation barrier for the dimerization with Au+-PMe3 amounts to 12.6 kcal/mol, which is lower by 2 kcal/mol than the activation barrier of the Au(I)-catalyzed Bergman cyclization calculated at the same level of theory.10 Compared to the barrier of the radical transannular ring closure of cyclodecadiyne (3) (Scheme 1b), this represents a decrease of 17
B3LYP as a functional results exclusively in butadienyl cation 22a as an intermediate (Figure 2). Please note that with no method were both cations (22a and 23a) simultaneously found as minima on the energy surface. The usage of benzene as a solvent results in a small increase in the energy of cations 22a and 23a [∼2 kcal/mol (Table 2)]. However, the obtained structures for the functional used remain the same: 22a for B3LYP and 23a for PBE0 and M06-2X (Table 2). Despite the fact that the products differ with regard to the method, the geometry of transition state 21a is almost independent of the employed level of theory. The energy values of transition state 21a with Au+-PMe3 as a catalyst and without a solvent model range between 11 and 13.6 kcal/mol (Table 2) for almost all methods. Only in the case of 13575
DOI: 10.1021/acs.joc.7b02843 J. Org. Chem. 2017, 82, 13572−13582
Article
The Journal of Organic Chemistry kcal/mol.5 Thus, according to our calculations, the catalytic effect of the Au(I) complex corresponds to a 1012-fold acceleration of the transannular cyclization. The relative energies of products 22a and 23a are in the range of 5−16 kcal/mol and strongly depend on the methods used (Table 2). With CCSD(T) as the golden standard, the difference between the single-point energies of 22a (optimized using B3LYP) and 23a (optimized using PBE0) amounts to 2 kcal/mol in favor of 23a (Figure 2). This result indicates the formation of cyclopropenylmethyl cation 23a as an intermediate during the Au(I)-catalyzed transannular cyclization of 20a; however, the existence of butadienyl cation 22a as an intermediate cannot be completely ruled out because of the small energy difference. This is extraordinary considering the fact that the energetic difference between parent systems 17 and 18 amounts to 23.9 kcal/mol. For a closer look at the discrepancy in the energetic difference between cations 17 and 18 on one side and cations 22a and 23a on the other, the structural parameters of 17, 18, 22a, and 23a are shown in Figure 3. The corresponding natural charges are also listed. Cations 17 and 22a exhibit a butadiene system having the cationic center at the C4 atoms. The empty p orbital of the cationic centers is perpendicularly oriented to the butadiene system. The natural charge at the C4 atoms amounts to 0.290 and 0.399, respectively. All other carbon atoms come up with a natural charge of
