Magnus T.
Johnson
a,
J.
Marthinus Janse van Rensburg
a,
Martin
Axelsson
a,
Mårten S. G.
Ahlquist
*b and
Ola F.
Wendt
*a
aCentre for Analysis and Synthesis, Department of Chemistry, Lund University, P.O. Box 124, SE-221 00, Lund, Sweden. E-mail: ola.wendt@organic.lu.se; Tel: +46 46 2228153
bDepartment of Theoretical Chemistry, School of Biotechnology, Royal Institute of Technology, S–106 91, Stockholm, Sweden. E-mail: mahlquist@theochem.kth.se
First published on 16th September 2011
The first example of the reaction of an isolated gold(I) complex with an aryl halide to form a C–C bond is reported. The reactivity of (NHC)Au(I)–R complexes towards a wide range of electrophiles was investigated. The Au–C σ-bond is shown to exhibit low nucleophilicity, but it is reactive towards MeI and MeOTf to form toluene, biphenyl and ethane, most likely through an oxidative mechanism. Carbon dioxide is completely unreactive. The experimental findings are supported by theoretical calculations.
Despite the increasing use of Au(I) in catalysis, there are few examples of the reaction of Au(I)–C σ-bonds with electrophiles, one such example being the alkylation of PPh3AuMe with alkyl iodides via observable Au(III) intermediates.9Au(I) methyl complexes can also be oxidised by iodine followed by subsequent C–I reductive elimination.10 Reactions of Au(I)–aryl complexes with aryl halides to produce catalytic cross-coupling chemistry have appeared futile and it was found that when the product was observed it was due to catalysis by trace amounts of palladium.7 Hashmi investigated the reactivity towards electrophilic halogenation reagents.11 Blum studied the relative kinetic basicities of organogold complexes towards protons and a higher reactivity was observed for the more electron rich aryl groups.12 Interestingly, unless strong acids (e.g.HCl) were used, the protonation of the aryl group proceeded relatively slowly and often 48 h was not enough for the reaction to go to completion. In this context, we were surprised to find that Nolan and coworkers reported that gold(I)–arene complexes with aryl groups of relatively low nucleophilicity, such as oxazolyl, reacted with CO2 at −78 °C.13 This was especially puzzling to us, since we had previously failed to achieve carboxylation reactivity in similar IMesAu–aryl complexes.14 In view of this and the general paucity of information on such reactions, we set out to study the reactivity of the IMesAu(I)–C σ-bonds and, in this communication, we report the synthesis of complexes 2–6 and the reactivity of 3, 4 and 6 towards various electrophiles (Figs 1 and 2).15
Fig. 1 The synthesis of 2–5. |
Fig. 2 The crystal structures of 2, 4 and 5. Selected bond distances (Å) and angles (deg); 2: Au1–C5 1.971(1), Au1–O1 1.990(8), C5–Au1–C5 178.0(4). 4: Au1–C1 2.042(10), Au1–C2 2.017(9), C1–Au1–C2 177.7(5). 5: Au1–C1 1.999(17), Au1–I2 2.5332(13), I2–Au1–C1 179.6(5). |
IMesAu–Ph 3 was found to react with strongly electrophilic methyl triflate, to produce toluene and biphenyl in a 1:1 ratio, together with some small amounts of ethane (Scheme 1 and Table 1, where all the results are presented). The reaction proceeded to completion within seven hours at room temperature. On the other hand, no reaction took place between 3 and 1 eq. of acetyl chloride, nor could we detect any reactivity toward carbon dioxide, not even at 80 °C for prolonged periods of time. In the latter case, we used both THF-d8 and C6D6 as solvents. It is hence clear that, although a reaction could be observed for MeOTf, the complex is only weakly nucleophilic and requires very strong electrophiles.
Scheme 1 C–C bond formation directly from Au(I)–Ph and R–X. |
Entry | IMesAu–R | E+ | Products | t (h)/T (°C) |
---|---|---|---|---|
a Product distribution determined by 1H-NMR. b Product distribution determined by GC-MS. c A maximum of 25% conversion was reached. | ||||
1a | Ph | MeOTf | Ph–Ph (54%) PhMe (46%) Me–Me | 7/25 |
2a | Ph | MeI | Ph–Ph (24%) PhMe (76%) Me–Me | 27/110 |
3 | Ph | PhI | Ph–Ph | 50/110 |
4b | Ph | p-tol-I | 4,4′-bitoluene (3%) 4-methyl-biphenyl (85%) biphenyl (12%) | 24/110 |
5c | Ph | PhOTf | biphenyl | 100/110 |
6 | Ph | 4-Br–PhOMe | no reaction | 48/100 |
7 | Me | MeI | Me–Me | 6/110 |
8 | Me | MeOTf | Me–Me | 3.5/25 |
9 | Me | PhI | no reaction | 48/110 |
10 | Me | PhOTf | no reaction | 24/110 |
11 | Me | CO2 (5bar) | no reaction | 24/50 |
12b | p-tol | MeOTf | 4,4′-bitoluene (60%) p-xylene (40%) Me–Me | 4.5/25 |
13 | Ph | PhBr | no reaction | 48/100 |
14 | Ph | AcCl | no reaction | 4/25 |
15 | Ph | CO2 | no reaction | 48/110 |
PPh3AuMe has previously been shown to form Au(III) intermediates when reacted with methyl iodide.9 When we reacted 3 with methyl iodide, however, no stable gold(III) intermediates were observed by 1H-NMR spectroscopy.
Instead, toluene and biphenyl formed in an approximate ratio of 1:1 together with ethane, as shown both by 1H-, and 13C-NMR spectroscopy,16 using 13C-labelled methyl iodide. No phenyl iodide as a product of reductive elimination was observed in the product mixture.17 Simultaneously, a decrease of phenyl complex 3 took place concomitantly with a formation of the complex IMesAu–I 5 (Scheme 1). The identity of 5 was confirmed by its independent synthesis from 1, see Fig. 1 and the ESI†. During the progress of the reaction at 110 °C, 5 slowly decomposed, as observed by 1H-NMR and by the formation of Au(0) on the walls of the J. Young NMR tube used to follow the reaction. In the reaction with methyl triflate, no IMesAuOTf was observed due to immediate decomposition. Instead, the decrease of starting material 3 and increase of toluene and biphenyl was used to follow the reaction. To further investigate the possibility of the oxidative addition of an aryl halide to molecular gold(I), we tested the reaction between 3 and iodobenzene. Much to our surprise, after 50 h at 110 °C the reaction resulted in the complete formation of biphenyl and 5. To test the possibility of a Lewis acid catalysis mechanism of a Friedel–Crafts type, we synthesized the corresponding p-tolyl complex IMesAu-p-C6H4Me 6. This was reacted with MeOTf and the products obtained were p-xylene and 4,4′-dimethylbiphenyl, which are consistent with a mechanism in which the reactivity is on the carbon covalently bonded to gold (entry 12).
Methyl complex 4 also reacts with MeOTf, resulting in the formation of a new organic compound, which we assign as ethane-based on the 1H-NMR spectrum. The reaction with 4 was faster compared to that of 3 and completed within five hours at room temperature. Similar to the reaction of 3, ethane formation is accompanied by the disappearance of the starting material without the formation of a new complex detectable by 1H-NMR. Possibly the highly unstable complex IMesAuOTf is formed prior to immediate decomposition. The reaction of 4 with MeI went to completion within six hours at 110 °C, forming ethane and 5 as the only products. A small amount of ethane was detectable by 13C-NMR when 13C-labelled methyl iodide was used. Given the reactivity of 3, we were somewhat surprised to see the lack of reactivity between 4 and both phenyl iodide and phenyl triflate at 110 °C up to 48 h.
To investigate the possibility of the catalysis being performed by trace amounts of metals, an ICP-MS analysis was performed to analyse the Cu and Pd contents of a typical experiment involving complex 3. The results indicated 303 ppm Cu and 616 ppm Pd, see ESI†. These are not negligible amounts and, to further investigate the possibility of colloidal palladium catalysis, we repeated the reaction between 3 and phenyl iodide and methyl triflate in the presence of Hg(0) and noticed no difference in the rate of coupling product formation. Clearly, the order of the rates in which reactivity occurs, MeOTf > MeI > PhI, is the opposite to that expected with respect to oxidative addition in a Pd(0)/Pd(II) catalytic cycle.18
Initial results using TEMPO as a radical scavenger in the reaction between 3 and phenyl iodide indicates that the reaction also does not proceed with radical intermediates, as the reaction times and products remained indifferent. We also hypothesized the possibility of the formation of colloidal gold. In a recent study by Lambert and coworkers,19PPh3AuCl was used as the precatalyst in a Sonogashira reaction at 145 °C and after an initiation period of 80 h the reaction started, which clearly speaks in favor of nanoparticle catalysis. However, using the more stable IMesAuPh in a stoichiometric reaction with phenyl iodide, as shown in Scheme 1, the reaction proceeds without an observable induction period at 110 °C, supporting a molecular non-catalytic reaction. Furthermore, the absence of reactivity between 4 and phenyl iodide in similar conditions seem to indicate a less straightforward path than the decomposition of the gold complex to nanoparticles. This is under the assumption that the catalytic activity of the possible nanoparticles stemming from both 3 and 4 exhibit identical activity. The mercury experiment also speaks against any participation of gold(0).20 Furthermore, the selectivity of the reaction in entry 4, forming 85% of the cross-coupling product also suggests a molecular mechanism. On the other hand, the seemingly complete conversion of the gold complex into a distinct product, 5, cannot be taken as unambiguous proof for a molecular mechanism, as was shown for palladium.21
To gain a further insight into these reactions, we undertook a computational study. The oxidative addition of MeOTf to 3′22 could, in principle, occur both at the gold centre and at the C1 carbon of the phenyl group. Both reactions were found to occur with inversion at the methyl carbon in SN2-like reactions. The reaction at the gold centre was calculated to have a free energy of activation of 21.7 kcal mol−1, which is considerably lower than the 25.3 kcal mol−1 barrier calculated for reaction at the carbon. The reaction to form the gold(III) intermediate 8 is calculated to be close to thermoneutral with ΔG = 0.4 kcal mol−1. The formation of toluene from 8 has a low barrier of 12.9 kcal mol−1, explaining why Au(III) intermediates could not be observed. It is also possible that the C–C bond forming step is preceded by the dissociation of the triflate ligand, but the error in the absolute solvation energy of the two ionic species formed in that step is likely too large for any quantitative comparison. Regardless, the barrier is so low that the reaction should proceed very rapidly. Based on previous studies, which showed that transmetallation between gold complexes are rapid,9a we believe that the biphenyl and ethane could be formed viatransmetallation between 8 and 3′, followed by reductive elimination. Due to the size of the system, we were not able to get any reliable barrier for the full ligand, but a calculation using a smaller NHC ligand indicated that the reaction could be facile.23 The reactions are outlined in Schemes 2 and 3. It is also possible that NHCAu(III)Ph2Me and NHCAu(I)OTf are formed in the transmetallation; however, the barriers for reductive elimination of biphenyl or toluene were high (25.1 kcal mol−1 for PhPh and 34.1 kcal mol−1 for PhMe). We also considered the possibility that PhOTf is reductively eliminated from 8 to give NHCAu(I)Me. PhOTf could then react with 1′ to give biphenyl. However, the barrier from 8 is calculated to be 45.5 kcal mol−1 and we thus regard this mechanism as highly unlikely.
Scheme 2 The free energy of activation in kcal mol−1 of reactions between 3′ and electrophiles. |
Scheme 3 Reductive elimination of toluene and transmetallation followed by reductive elimination of biphenyl. |
So far, both experiments and calculations point towards a mechanism involving a molecular gold species. We therefore investigated the oxidative addition of PhI to 3′, as well as the direct C–C coupling with DFT. However, in this case, both reactions were found to have relatively large barriers. The oxidative addition was calculated to have a free energy of activation ΔG‡ = 40.4 kcal mol−1, which is too high even at elevated temperatures. If we exclude the entropy and calculate just the enthalpy of activation, we get ΔH‡ = 27.0 kcal mol−1, which also indicates that the reaction is not likely to proceed rapidly. We tried the B3LYP and M06 functionals for both optimizations and energy calculations, as well as the larger ERMLER2**++ basis set on iodine, however, it only lead to minor changes in the energies. The direct C–C coupling was even less favorable, with a free energy barrier of 51.6 kcal mol−1.
Since our results on the carbon dioxide reactivity were contradictory to the ones reported by Boogaerts and Nolan13 for a less nucleophilic IPrAu–oxazolyl complex we set out to study this computationally. When calculating the insertion of CO2 in the Au–C bond of 3′, the free energy barrier was 26.7 kcal mol−1, which is relatively high and in agreement with the experiments. Moreover, the reaction is calculated to be slightly endergonic, with ΔG = 0.5 kcal mol−1. Since the oxazolyl group is most likely less nucleophilic than a phenyl group, we expect the Au(I)-oxazolyl complex 10 to be less reactive toward CO2 than 3′. Indeed, the barrier for CO2 insertion was calculated to 38.9 kcal mol−1, indicating that the reaction should not be possible to observe unless a different mechanism is operating. However, we also calculate the reaction to be very endergonic, with ΔG = 9.3 kcal mol−1, which indicates that the reaction could never be observed, regardless of the mechanism. To investigate the possibility that the IPr ligand would drastically promote carboxylation relative to the IMes ligand, we synthesised the corresponding IPrAuPh complex, but no reaction could be observed at 3 bar pressure CO2 and up to 100 °C for 5 h in C6D6. We set out to reproduce the carboxylation of the IPrAu–oxazolyl complex, but unfortunately we were unsuccessful in reproducing the synthesis through the published route.13 Also, attempts on the preparation of (NHC)Au–oxazolyl complexes via the reaction of the corresponding Li or Mg reagents with 1 were unsuccessful in our hands. Furthermore, Nolan's experimental setup of applying CO2 pressure to the reaction mixture before addition of the substrate in the catalytic runs allows for the complete conversion of the KOH and IPrAuOH into the corresponding carbonates, which undoubtedly contradicts the proposed catalytic cycle.24 Clearly, carboxylation of electron-deficient heterocycles can also be achieved using carbonates as a stoichiometric base under transition metal-free conditions25 and our results contradict the results of Nolan, where the Au(I) oxazolyl complex was claimed to react irreversibly at low temperature with CO2.
Recently, Nolan and co-workers further reported26 that decarboxylation of aromatic carboxylic acids occurred in the presence of the (IPr)Au–OH complex. This reaction is simply the reverse reaction of CO2 insertion and, since they provided calorimetric data, as well as different reactivities for a range of substrates, we decided to use this reaction as a test of our computational model. The measured reaction enthalpy for (IPr)Au–OH and 2,6-dimethoxybenzoic acid to give (IPr)Au-(2,6-dimethoxyphenyl), carbon dioxide and water was −9.4 kcal mol−1. We calculated the corresponding reaction enthalpy (with the xylyl–NHC model ligand) to be −10.3 kcal mol−1. The free energy barrier was calculated at 18.3 kcal mol−1, which indicates that the reaction should be relatively rapid at room temperature. The corresponding reaction with benzoic acid as the substrate was calculated to have a barrier of 26.2 kcal mol−1 and should proceed much more slowly and, in the experiment, they do not observe any reaction of benzoic acid and very slow reaction of para-methoxy benzoic acid.
To conclude, we have investigated the catalytic competence of Au(I)–C σ-bonds and report the first example where an isolated Au(I)-nucleophile reacts with an aryl halide to give the cross-coupling product. With stronger electrophiles an oxidative mechanism is clearly operating, with transmetallation from Au(I) to Au(III) being responsible for homocoupling products. For aryl halides, we cannot exclude nanoparticle catalysis. It is also clear that the Au hydrocarbyl complexes investigated here are unreactive towards carbon dioxide and any involvement of such complexes in catalytic carboxylation reactions seems unlikely.
Financial support from the Swedish Research Council, the Knut and Alice Wallenberg Foundation and the Royal Physiographic Society in Lund is gratefully acknowledged. We also thank Mr. Lennart Möhlmann for experimental assistance.
Footnote |
† Electronic supplementary information (ESI) available, including experimental and computational details, X-ray data and NMR spectra for all new compounds. CCDC reference numbers 832959–832962. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c1sc00428j |
This journal is © The Royal Society of Chemistry 2011 |