Role of σ,π-Digold(I) Alkyne Complexes in Reactions of Enynes.

Gold(I) acetylide and σ,π-digold(I) alkyne complexes derived from one prototypical 1,6-enyne and from 7-ethynyl-1,3,5-cycloheptatriene have been prepared and structurally characterized. Their possible role in gold(I)-catalyzed cycloisomerizations has been studied by experiment and by DFT calculations. Gold(I) acetylides are totally unproductive complexes in the absence of Brønsted acids. Similarly, no cyclizations were observed by heating σ,π-digold(I) alkyne digold(I) at least up to 130 °C. Theoretical studies provide a rationale for the much lower reactivity of digold species in reactions of enynes.

Introduction

Although digold(I) complexes have been known since the mid-1970s,[1] their relevance in homogeneous gold(I) catalysis has only been recognized recently.[2−4] They also display interesting luminescence properties and are important building elements for the design of supramolecular structures.[5] Alkenyl,[6−9] and aryl[10−12] digold complexes with Au2C three-center–two-electron bonds have been characterized.[13]

The vast majority of the gold(I)-catalyzed chemistry of alkynes under homogeneous conditions can be understood by the initial π coordination via ligand substitution to form π-alkyne gold(I) complexes 1, which then react with carbo- or heteronucleophiles (Scheme ).[14,15] Complexes of this type with nonterminal alkynes have been structurally characterized.[16−18] Alternatively, deprotonation of the terminal alkyne by the counterion X– can generate σ-alkynyl gold(I) complexes 2. Acetylides 2 react with 1 or with the initial catalyst [AuLL′]X to form finally σ,π-digold(I) alkyne digold(I) complexes 3,[19−21] which are stable species that have been structurally characterized.[22−29] The facile formation of complexes 3 is a result of the stronger binding of gold(I) to acetylides 2 than to free alkynes.[30]

Scheme 1

Formation of σ,π-Digold(I) Alkyne Complexes 3 from π-Gold(I) Alkyne Gold(I) Complexes 1

Digold complexes 3 are excellent catalysts in reactions of diynes in which one of the alkynes is a terminal one, by allowing the simultaneous formation of a nucleophilic σ-alkynyl gold(I) species and an electrophilic π-alkyne gold(I) species, which then react with each other in a dual-gold(I)-catalyzed process.[31−33]

The first example of the reaction of alkynes via σ,π-activation was proposed by Toste and Houk for the cycloisomerization of 1,5-allenynes catalyzed by [(Ph3PAu)3O]BF4 (Scheme ).[34] The experimental and computational data for the cyclization of allenynes such as 4 to form 5 suggested that σ,π-alkyne digold(I) intermediates 6 initiate a 5-endo-dig cyclization to form gem-diaurated species 7 and 8, which give rise to 5 by protodeauration. This and related mechanisms have been recently reexamined by Fensterbank, Gandon, and Gimbert in the context of a broader study on the ligand and anion effects in the cycloisomerizations of allenynes.[35]

Scheme 2

Proposed Role of gem-Diaurated Species in the Gold(I)-Catalyzed Cycloisomerization of 1,5-Allenynes

Although similar species have been observed in the cycloisomerization of 1,6-enynes by mass spectrometry under electrospray ionization, the real involvement of digold complexes as intermediates in this reaction has been questioned.[36] Furthermore, a theoretical study on the gold(I)-catalyzed hydroamination of alkynes suggested that the nucleophilic attack takes place on π-alkyne gold(I) complexes 1, rather than σ,π-digold(I) alkyne digold(I) complexes 3.[37] On the other hand, digold complexes of type 3 have been proposed to be superior catalysts in several gold(I)-catalyzed transformations.[22]

We have isolated σ,π-digold(I) alkyne digold(I) complexes 3 in the intermolecular [2 + 2 + 2] cycloaddition of alkynes with oxoalkenes,[24] as well as in the synthesis of cyclobutenes by intermolecular [2 + 2] cycloaddition of alkynes with alkenes.[38] However, these isolated complexes were shown to lay outside of the main catalytic cycle. Actually, by a change in the counterion of the cationic catalyst from SbF6– to the softer anion BAr4F–, the formation of digold species could be minimized,[24] which had a positive effect on the overall yields of the [2 + 2] cycloaddition and other related processes.[39]

The active role displayed by σ,π-digold(I) alkyne digold(I) complexes in the gold(I)-catalyzed cycloisomerization of 1,5-allenynes (Scheme )[34] is in apparent contradiction with the proposal that these species are actually “dead ends” in reactions of cycloisomerizations of enynes and in the cyclobutene synthesis.[38] In order to shed light on the actual role played by σ,π-digold(I) alkyne complexes (3) in catalytic transformations of 1,n-enynes, we reinvestigated several gold(I)-catalyzed cycloisomerization reactions by isolating the corresponding gold(I) acetylides and digold complexes.

Results and Discussion

Mechanistic studies of gold(I)-catalyzed cycloisomerizations of 1,n-enynes have been the benchmark for the understanding of the fundamental reactivity of alkynes and alkenes,[40−42] which have recently been extended to intermolecular systems.[39c] The gold(I)-catalyzed reaction of 1,6-enyne 9 at room temperature leads cleanly to diene 10a by a single cleavage rearrangement using complex A or B (Scheme ).[40] At longer reaction times, or using complex C as catalyst, mixtures of 10a and 10b, the products of double-bond isomerization, were obtained. We also examined the reaction of 7-ethynyl-1,3,5-cycloheptatriene (11), which reacts cleanly with A or B to give indene (12).[44]

Scheme 3

Cycloisomerizations of 1,6-Enyne 9 and 7-Ethynyl-1,3,5-cycloheptatriene 11

Gold(I) acetylide 13 was easily prepared by deprotonation of 1,6-enyne with n-BuLi at −50 °C followed by reaction with IPrAuCl (Figure ). Further reaction of 13 with B or C led to σ,π-digold(I) alkyne complexes 14a,b. Complexes 15–17 and 18a,b were similarly prepared from 7-ethynyl-1,3,5-cycloheptatriene (11). We also synthesized gold acetylide complex 19 from 3-methoxy-7-methylocta-5,6-dien-1-yne and IPrAuCl, although the corresponding digold complex could not be obtained in pure form by reaction with B or C.

Figure 1

Acetylide gold(I) and σ,π-digold(I) alkyne complexes 13–19 (CHT = cycloheptatriene).

The structures of 13, 14a,b, 15–17, and 18a,b were confirmed by X-ray diffraction. The structures of the two representative complexes 15 and 16 are shown in Figure . For the gold(I) acetylide complexes 13, 15, and 17, the C1–Au and C≡C bond distances are 1.98–2.01 and 1.18–1.21 Å, respectively. In the case of σ,π-digold(I) alkyne complexes (14a,b, 16, and 18a,b) the C1–Au (1.98–2.03 Å) and C≡C bond distances (1.21–1.23 Å) are similar. As expected, the second π-coordinated gold atom is symmetrically bonded to the alkyne: 2.25/2.20 Å for 14a, 2.21/2.23 Å for 14b, 2.22/2.25 Å for 16, 2.20/2.20 Å for 18a, and 2.21/2.20 Å for 18b.

Figure 2

ORTEP plots (50% thermal ellipsoids) for gold(I) acetylide 15 and σ,π-alkyne digold(I) Complex 16. The SbF6– anion of 16 is omitted for clarity.

Mononuclear species 13, 15, and 17 were heated under refluxing conditions in CDCl3. Gold(I) acetylide 13 remained unchanged, whereas 15 and 17 showed partial decomposition, although indene (12) could not be detected after 2 h. These results definitively confirm that gold(I) acetylides are not intermediates in these cycloisomerizations.

Digold complexes 14a,b, 16, and 18a,b were heated from 25 to 130 °C in CDCl2CDCl2,[45a] and any reaction progress was monitored by 1H, 31P, and 19F NMR. Complexes 14a and 18a with SbF6– as the counterion remained unchanged up to 130 °C. Complexes 14b, 16, and 18b bearing the weakly coordinating counterion BAr4F- were stable up to 100 °C. At this temperature, a slow conversion of the digold species into new gold complexes together with cyclization products were observed. In the case of 14b at 130 °C, a mixture of the starting complex and two new gold species were formed in a 1:1.2:1.5 ratio, together with cyclization product 10a, which was obtained in 58% yield (1H NMR yield, using mesitylene as internal standard). From the reaction mixture, we obtained a crystalline compound, whose structure was determined by X-ray diffraction analysis as the symmetrical gold(I) complex 20 (Figure ).[46]

Figure 3

Complexes 20 and 21.

In the case of 16, a 1:8 ratio between starting digold complex and a new gold species was observed at 130 °C, together with indene (12), which was formed in 40% NMR yield. The major product was determined by X-ray diffraction to be complex 21, arising by transmetalation from the tetrakis[3,5-bis(trifluoromethyl)phenyl]borate (BAr4F-) counterion to the gold cation. In fact, gold(I) is known to be able to heterolytically cleave the C–B bond of BAr4F, leading to the aryl transfer from boron to gold.[47]

The presence of cyclization products 10a and 12 in the thermolysis of 14b and 16 suggests two possible pathways. One is the aryl transfer from the BAr4F– to σ-coordinated gold to form LAuAr complexes together with alkynyl borates, which undergo cyclization promoted by the second, π-coordinated gold center. Hydrolysis of the C–B bonds could take place at the enynes or the final products by reaction with water present in the solvent. The other involves first a protodeauration of the σ-coordinated gold with the water present in the solvent, leading to the corresponding alkyne species and subsequent aryl transfer from the BAr4F– to this gold to form LAuAr complexes. The alkynes could then undergo cyclization promoted by the second π-coordinated gold center. Similar results were obtained when digold complexes 14b and 16 were heated in CDCl3[45b] under refluxing conditions for 1 h.

We also studied the catalytic activity of the isolated gold(I) acetylide and digold complexes in the cyclizations of substrates 9 and 11 (Tables and 2). Gold(I) acetylide 13 was not catalytically active (Table , entry 1), although addition of HSbF6·6H2O, which cleaves the Au–C bond, generates a catalytically active cationic gold(I) species (Table , entry 2).[3]b,[24] On the other hand, digold complexes 14a, were moderately active in the cycloisomerization of 9. However, even under the best conditions (Table , entry 5), the reaction required 24 h to furnish 10a in 79% yield, while catalyst A or B provided 10a in higher yields (93–98%) after just 1 h.

Table 1

Cycloisomerization of 1,6-Enyne 9 with Complexes 13a,b and 14a,b

entry	[Au] (amt (mol %))	additive (amt (mol %))	time (h)	10a:10b yield (%)a
1	13 (5)		24
2	13 (5)	HSbF6·6H2O (5)	8	21:34
3	14a (2.5)		1	11:0b
4	14a (2.5)		2	15:0b
5	14a (2.5)		24	79:0
6	14a (5)	HSbF6·6H2O (5)	8	23:33
7	14b (3)		24	51:0
8	14b (5)	HSbF6·6H2O (5)	8	21:34

Yields determined by 1H NMR using mesitylene as internal standard.

An 84–87% of enyne 1 was recovered.

Table 2

Cycloisomerization of Cycloheptatriene 11 with Complexes 17 and 18a,b

entry	[Au] (amt (mol %))	Aadditive (amt (mol %))	time (h)	12 yield (%)a
1	17 (5)		24
2	17 (5)	HSbF6·6H2O (5)	8	61
3	18a (2)		24	40
4	18a (3)		24	46
5	18a (5)	HSbF6·6H2O (5)	8	50
6	18b (2)		24	37
7	18b (5)	HSbF6·6H2O (5)	8	48

Yields determined by 1H NMR using mesitylene as internal standard.

Very similar results were obtained in the reaction of 11 to give indene (12) (Table ). Again, gold(I) acetylide 17 was catalytically inactive (Table , entry 1), whereas in the presence of a strong Brønsted acid (Table , entry 2) or using digold complexes 18a,b (Table , entries 3–7), indene (12) was obtained. Nevertheless, all of these reactions are slower than those catalyzed by A or B (1.5 h, 25 °C, Scheme ).

We also performed experiments by first mixing 1,6-enyne 9 with 3 mol % of gold acetylide 13 in CD2Cl2, followed by addition of 3 mol % of catalyst B (Table ). After 15 min, only traces of product 10a together with digold complex 18a were formed (Table , entry 1), suggesting a higher affinity of the gold(I) for the gold acetylide species, in accordance with the precedents.[30,36] After 20 h (Table , entry 2) product 10a was formed in 45% yield. At this stage, addition of an extra 3 mol % of catalyst B led to a very fast conversion, affording 10a in 95% yield (Table , entry 3). In the case of ethynylcycloheptatriene (11), after addition of 3 mol % of gold acetylide 17 and 3 mol % of catalyst B, 12 was obtained in 78% yield, along with digold complex 18a, which suggests that the ligand exchange between the catalyst B and 11 is faster than that between catalyst B and gold acetylide 17.

Table 3

Competition Studies between 1,6-Enyne 9 and Gold Acetylide 13

entry	amt of B (mol %)	time	10a yield (%)a
1	3	15 min	trace
2	3	20 h	45b
3	6	15 min	95

Yields determined by 1H NMR using mesitylene as internal standard.

A 51% yield of enyne 9 was recovered.

Computational Results

DFT calculations (M06, 6-31G(d) (C, H, P, N) and SDD (Au), CH2Cl2) were performed in order to compare the energy barriers for the previous gold(I)-catalyzed cycloisomerizations.[48] In all cases, the calculations were performed with three different ligands: L1 = trimethylphosphine, L2 = triphenylphosphine, and L3 = 1,3-bis(2,6-diisopropylphenyl)imidazol-2-ylidene.

The gold(I)-catalyzed cycloisomerization of 1,6-enynes can proceed by two pathways, leading to five- or six-membered rings.[14,40,41] The 5-exo-dig cyclization of Ia–c leads to intermediates IIa–c, whereas IIIa–c are formed in a 6-endo-dig process that is less favorable kinetically (Scheme ). In most cases slightly higher energy barriers were observed for the cyclization with the more electron donating carbene ligand L3, which reduces the electrophilicity of AuL+.

Scheme 4

Cycloisomerizations of Monogold (Ia–c) and Digold 1,6-Enyne (IVa–c) Complexes,

Free energies in kcal/mol.

The energy of TS was calculated by freezing the following distance: d(C8–C55).[49]

The cyclization in digold complexes IVa–c has a considerably higher energy barrier (5-exo-dig, L1 22.3 kcal/mol, L2 20.1 kcal/mol, L3 22.1 kcal/mol; 6-endo-dig, L1 23.3 kcal/mol, L2 21.5 kcal/mol, L3 33.4 kcal/mol), leading to gem-diaurated species Va–c and VIa–c in thermodynamically unfavorable reactions. These results are consistent with the experimental data and clearly show that digold species are incompetent in the cycloisomerization process. Gold acetylides VIIa–c can only evolve by an ene reaction, via 1,5-hydrogen transfer trough TS, leading to VIIIa–c (Scheme ). Nevertheless, this transformation shows prohibitively high energy barriers (ca. 38 kcal/mol).

Calculations also show higher energy barriers for the reaction of digold complexes XIIa–c in comparison with the reaction of monogold species IXa–c (Scheme ). These reactions actually proceed by cycloisomerization of the norcaradiene complexes Xa–c, which are in tautomeric equilibrium with the cycloheptatrienes.

Scheme 5

Cycloisomerizations of Monogold (IXa–c) and Digold Cycloheptatriene (XIIa–c) Complexes

Free energies in kcal/mol.

The same reactivity pattern is reproduced in the reaction of 1,5-allenynes (Scheme ).[34,35] Thus, much higher barriers were found in the first step of the reaction through digold complexes XVIIIa–c in comparison to that initiated by the cyclization of XVa–c to XVIa–c. However, in this case, the second step actually makes the digold pathway more favorable, in agreement with a previous theoretical study employing PH3 as ligand.[34] Thus, the second step for the monogold pathway (XVIa–c to XVIIa–c) has much higher barriers (28–29 kcal/mol) in comparison to that of the digold pathway (XIXa–c to XXa–c), which only requires from around 5 to 10 kcal/mol. In this case aurophilic interactions are observed that could lower the energy of the transition states. Therefore, since both XVa–c and XVIIIa–c would be in equilibrium, the system will evolve almost exclusively through digold species XIXa–c and XXa–c. Products XXa–c possess strong aurophilic interactions and thus displace the reaction toward the products.

Scheme 6

Cycloisomerizations of Monogold (XVa–c) and Digold 1,5-Allenyne (XVIIIa–c) Complexes,

Free energies in kcal/mol.

The energy of TS was calculated by freezing the following distances: d(C1–C79) and d(C1–C80).[50]

Conclusions

We have prepared gold(I) acetylides and σ,π-digold(I) alkyne digold(I) complexes from substrates that undergo isomerization reactions under mild conditions in the presence of gold(I) catalysts. Gold(I) acetylides are unproductive complexes in cycloisomerization reactions, unless a strong Brønsted acid is added to cleave the Au–C bond leading to a cationic gold(I) complex. Well-characterized digold(I) complexes are very robust and fail to cyclize on heating in solution up to 100–130 °C. These results confirm that these digold(I) species are not catalytic intermediates; rather, these complexes are “dead ends” in catalytic reactions of enynes, in full agreement with previous studies.[36,38] Only in the case of cationic digold(I) complexes with BAr4F– as the counterion were cyclization products observed by heating at 100 °C, although the observed reactivity is due to the decomposition of the complexes by transmetalation with the tetraarylborate that results in the transfer of one of the aryls from boron to gold.

Theoretical studies fully support the experimentally observed inertness of σ,π-digold(I) alkyne complexes in intramolecular reactions with alkenes. σ-Coordination by a second gold(I) raises the energy of the C–C bond formation by ca. 10–20 kcal/mol in the nucleophilic attack of the alkene to the π-activated alkyne. Finally, σ,π-digold(I) alkyne complexes were shown to be only moderate catalysts in intramolecular reactions of enynes, being less efficient than cationic complexes bearing nitriles as weakly coordinating ligands.