Radical-initiated alkene hydroauration as a route to gold(iii) alkyls: an experimental and computational study

The hydroauration of functionalised 1-alkenes by the gold(iii) hydride (C^NOMe^C)AuH is initiated by organic radicals and proceeds via (C^N^C)Au(ii) radical intermediates following a bimolecular outer-sphere mechanism. The outcome of these reactions is determined by the stability of the gold-substituted radicals, and chemoselectivity correlates with the degree of spin delocalisation in the alkylgold radical intermediates. The reaction is sensitive to steric as well as electronic factors; disubstituted alkenes and alkenes that form unstable radicals give product mixtures or are unreactive. As DFT calculations show, the reactions agree well with the calculated reaction enthalpies and the standard free energy change for the reaction of the gold(ii) radical with the respective alkene.


Introduction
We recently reported the synthesis of the rst example of a gold(III) hydride complex (C^N^C)AuH, based on the stabilisation provided by a C^N^C pincer ligand framework [(C^N^C) ¼ 2,6-(C 6 H 3 Bu t ) 2 pyridine]. 1 Cyclometallated C^N^C pincer complexes of gold(III) 2,3 have proved particularly useful for the stabilisation of otherwise non-isolable species, including gold(III) alkene, 4 alkyne, 5 CO 6 and peroxide complexes. 7 (C^N^C) AuH proved to be thermally stable and did not react with air, moisture or even acetic acid and was also unreactive to alkenes and alkynes. On the other hand, it did react with allenes to give gold vinyl complexes in high yield. 1 This lack of reactivity is not entirely unsurprising: gold(III) adheres strictly to a squareplanar coordination geometry and in (C^N^C)AuH all four coordination sites are occupied, so that these pincer compounds lack the ability to bind unsaturated substrates. However, we discovered that alternative reaction pathways become accessible in the presence of traces of organic radicals capable of abstracting the hydrogen ligand and thus generating (C^N^C)Au(II)c radical species. These Au(II) radicals can readily bind alkynes and lead to alkyne hydroauration in a bimolecular outer-sphere process (Scheme 1). Increasing the concentration of radicals greatly increases the rate of insertion reactions into Au-H bonds. This pathway allows the hydroauration of a range of substituted alkynes to give (Z)-vinylgold complexes (C^N^C) Au-C(R 1 )]CH(R 2 ) with almost quantitative stereo-and regioselectivity. These reactions are tolerant of a large variety of functional groups including hydroxide and carboxylic acid functions. 8 There is a growing realisation of the role that singleelectron transfer steps and gold(II) intermediates may play in gold-mediated reactions. 9 We report here on the reactivity of in situ generated gold(II) radicals with 1-alkenes, which leads to the formation of gold(III) alkyl complexes. Alkyl complexes of C^N^C gold pincer complexes are accessible in a variety of ways: by alkylation with Grignard reagents or aluminium alkyls, 10-12 by O-abstraction from (C^N^C)AuOMe with phosphines, 13 or by the reaction of (C^N^C)AuOH with allylic alcohols, 13 or by the reaction of (C^N^C)AuCl/base with C-H acidic alkanes CH 2 R 1 R 2 . This last method is very versatile and gives alkyls (C^N^C)AuCHR 1 R 2 which carry functional groups in the a-position. 12 Here we Scheme 1 Mechanism of radical-mediated alkyne hydroauration with (C^N^C)AuH pincer complexes. 8 describe the hydroauration of alkenes to give gold(III) alkyls with functional groups in b-position. The experimental and computational results provide insights into the factors inuencing radical-based hydroaurations of unsaturated substrates.

Results and discussion
For solubility reasons, from the library of differently substituted C^N^C gold(III) hydrides previously reported, 1,8 we chose to carry out the reactions reported here using the p-OMe substituted gold hydride, (C^N OMe^C )AuH (1). This compound is accessible following literature procedures from (C^N OMe^C ) AuCl and LiAlH 4 in 85% yield.
The reactivity of this complex towards different alkenes was investigated initially through scoping experiments carried out on a small scale, by mixing micromolar quantities of 1 and stoichiometric amounts of the alkene in toluene-d 8 in a J-Young NMR tube (Scheme 2). Two molar equivalents of azobisisobutyronitrile (AIBN) were added, the mixture was shaken and heated in the dark to 50 C to induce the decomposition of AIBN. The progress of the alkene hydroaurations was monitored by 1 H NMR spectroscopy. At the end of the reaction the volatile components were removed in vacuo, the residue was washed with n-hexane followed by MeOH to remove any unreacted alkene and excess AIBN, and the residue was dissolved in CD 2 Cl 2 . The product was characterized spectroscopically. This method led to the formation of the alkyls 2-8 in high yields.
For the alkenes CH 2 ]CHR [R ¼ CN, COOMe, COOH, Ph, 2-MeC 6 H 4 , 3-MeC 6 H 4 , C(O)Me] this resulted in the clean and facile formation of the corresponding gold-alkyl products 2-8. These reactions were also conducted on a preparative scale and allowed the isolation of the gold alkyls 2-8 as microcrystalline powders in moderate yields, with losses being mainly due to the washing steps during purication. Attempts to obtain crystals suitable for X-ray diffraction were unfortunately not successful.
Scheme 3 Proposed mechanism for the radical-initiated hydroauration of alkenes by 1.
There were therefore two classes of alkenes: those that gave clean insertions into the Au-H bond, and those that showed borderline or no reactivity. In order to rationalize the reactivity differences observed for the various alkene substrates, a computational investigation was undertaken using density functional theory (DFT). 14 It is proposed that the mechanism of alkene hydroauration follows the principles previously established for the correspondent alkyne reactions, 8 as shown in Scheme 3, and involves various radical intermediates a, b and c.
Our calculations focussed on the rst step of the mechanism described above, the formation of the intermediate radical species b* from the gold(II) radical a* and the alkene substrate. The model for the pincer ligand was simplied by omitting the Bu t and OMe substituents (denoted by *). The energetics of this reaction step were investigated by calculating the standard reaction enthalpies ðDH r Þ, Gibbs free energy of reaction ðDG r Þ and the total electronic energy differences (DE tot ) in the gas phase and under standard conditions. Very similar trends were calculated for the three parameters taken into consideration (see Table 1). In particular, the calculated values of DG r reect the experimental observations quite accurately: the formation of alkyl radicals b1*-b5* is energetically favourable, as observed by the clean, near-quantitative formation of gold alkyls in these cases, while reactions leading to b6*-b15* are close to DG r ¼ 0 or positive and are therefore not predicted to proceed.
To provide a further insight into the observed trends, the calculations were extended to the previously investigated alkyne 8 and allene 1 substrates, and in particular to the formation of some of the corresponding vinyl and allyl radicals d* and h*, respectively, from the reaction with a* (Scheme 4). For all of these systems, the calculated values of DH r , DE tot and DG r were more negative than in the case of the alkene substrates (Table 1 and Fig. 1). The reaction with allenes to give the allyl radical h1* proved energetically particularly favourable, in agreement with the experimentally observed facile hydroauration of allenes by (C^N^C)AuH. 1 The formation of the vinyl radical d1* is also strongly exergonic, while there is little energy difference between the other mono-and disubstituted alkynes in this series.
The chemoselectivity of the hydroauration was explored using the enynes 9 and 10 (Scheme 5), under analogous AIBNinitiated conditions. NMR spectroscopy showed that product mixtures are formed from attack on both the double and triple bonds, which in the case of 9 occurred with about equal probability, while 10 gave an approximately 80 : 20 mixture with predominant attack on C]C. In agreement with this, calculations of the reaction of species a* with 2-methylbuten-3-yne Table 1 Standard reaction enthalpies ðDH r Þ, Gibbs free energy of reaction ðDG r Þ, and total electronic energy differences (DE tot ) calculated in the gas phase at 298 K for the formation of gold(III) alkyl, vinyl and allenyl radicals from the corresponding unsaturated substrates and a* (Scheme 3) showed essentially identical DE tot values for the formation of bb* (DE tot ¼ À26.06 kcal mol À1 ) and dd* (DE tot ¼ À26.11 kcal mol À1 ) (Scheme 5). In order to rationalize observed alkene reactivity pattern, an investigation of the spin density in different b*, d* and h* radical intermediates was performed. As summarized in Fig. 2, in radicals b1*-b5*, d1*, and h1* a signicant degree of spin delocalization is observed, while such delocalization does not arise in b6*-b10*. This suggests that the reason for the energy difference in the formation of radicals on reaction with the gold(II) species reects these differences in spin delocalisation, the most stable radical intermediates being those stabilized by resonance. Accordingly, all of the alkyne and allene substrates previously explored, 1,8 whose radicals can in all cases be stabilized by resonance, were observed to undergo facile hydroauration, while for the alkenes the reactivity depends on the nature of the substituent.

Conclusion
The hydroauration of 1-alkenes with the gold(III) hydride pincer complex (C^N OMe^C )AuH is initiated by radicals and appears to follow the same bimolecular outer-sphere mechanism that has previously been established for the regio-and stereoselective hydroauration of alkynes. The process involves the generation of a (C^N^C)Au(II) radical which reacts with alkenes to give a gold-substituted alkyl radical. According to DFT calculations, the determining factor for the reaction appears to be the energy change associated with the attack by a gold(II) radical species on the alkene. Alkenes leading to alkyl radicals with restricted spin Fig. 1 Trend of DG r , (T ¼ 298 K) calculated for the reactions depicted in Scheme 3 (kcal mol À1 ). The substituents for the radical species b1*-b15*, d1*-d5* and h1* are as listed in Table 1.

Scheme 5 Reactions with enynes.
delocalisation either reacted slowly to a mixture of products, or did not react at all. While this limits the scope of the method to some extent, the hydroauration of activated alkenes is a facile method for the metal alkyl-free generation of gold(III) alkyl complexes bearing a variety of functional groups in b-position, including cyano, keto, ester and carboxylic acid functions. Moreover, the present study suggests that the hydroauration by gold(III) hydrides can be extended to different classes of unsaturated species, and that the reactivity trend of different substrates can be rationalized and/or predicted based on the spin delocalization of the radical intermediates involved. The scope of such alkyls for C(sp 2 )-C(sp 3 ) coupling reactions by reductive elimination using (aryl)(alkyl)gold(III) complexes is currently under investigation.

Experimental
When required, manipulations were performed by using standard Schlenk techniques under dry nitrogen or a MBraun glove box. Nitrogen was puried by passing through columns of supported P 2 O 5 with moisture indicator, and of activated 4 A molecular sieves. Anhydrous solvents were freshly distilled from appropriate drying agents. Elemental analyses were carried out at London Metropolitan University. AIBN (BDH Chemicals) was degassed by evacuation and stored under N 2 in the glovebox before use. The alkenes (Sigma Aldrich) were degassed by freeze-pump-thaw cycles and stored over activated 4 A molecular sieves before use. Solvents, toluene-d 8 and CD 2 Cl 2 (Apollo Scientic) were degassed by three freeze-pump-thaw cycles and stored over activated 4 A molecular sieves prior to use. 1 H and 13 C { 1 H} NMR spectra were recorded using a Bruker Avance DPX-300 spectrometer equipped with a 1 H, BB smartprobe. 1 H NMR spectra (300.13 MHz) were referenced to the residual protons of the deuterated solvent used. 13 C{ 1 H} NMR spectra (75.47 MHz) were referenced to the D-coupled 13 C resonances of the NMR solvent.
Preparation of (C^N OMe^C )AuH (1) Complex 1 was prepared by a modication of a literature procedure. 8 Under a N 2 atmosphere, 0.40 g (0.66 mmol) of the chloro complex (C^N OMe^C )AuCl was charged in a dry Schlenk ask with 40 mL of dry toluene. The mixture was cooled to À78 C and a solution of LiAlH 4 in dry THF (0.05 M, 13 mL, 0.66 mmol) was added dropwise. The mixture was stirred at À78 C in the dark for 15 min, yielding a dark suspension which was ltered under N 2 . The ltrate was evaporated to dryness to afford a brown powder. This was taken up in dichloromethane, and the resulting suspension was ltered over cotton in the dark, to give a pale-yellow ltrate. The solvent was removed under reduced pressure, and a white solid was obtained, yield 0.38 g (0.56 mmol, 85%).
Synthesis and characterization of insertion products 2-8 NMR scale reactions. As a general procedure, a solution of 1 (0.009 mmol) in toluene-d 8 (0.7 mL) was prepared inside a glovebox in a J-Young NMR tube. The desired olen (1 molar equivalent) was then added using a microlitre syringe, followed by 2 equivalents of AIBN. The tube was shaken and heated in the dark to 50 C. The reactions were monitored by 1 H NMR spectroscopy. The volatile components were removed in vacuo at the end of the reaction, and the residue was washed with n-hexane and with MeOH, and eventually redissolved in CD 2 Cl 2 for the NMR characterization. Yields were calculated from the NMR integration.
Reactions on a preparative scale. As a general procedure, to a solution of 1 (0.07 mmol) in dry and degassed toluene (4 mL) under nitrogen was added 1 molar equivalent of the desired olen using a microlitre syringe, followed by 1 equivalent of AIBN. The mixture was heated in the dark to 50 C. The reactions were monitored by 1 H NMR spectroscopy. The volatile components were removed in vacuo at the end of the reaction, and the residue was washed with n-hexane and with MeOH.