Cyclometalated gold(iii) complexes: noticeable differences between (N,C) and (P,C) ligands in migratory insertion

Gold(iii) complexes are garnering increasing interest for opto-electronic, therapeutic and catalytic applications.


Introduction
The past few years have witnessed spectacular development in gold(III) chemistry. 1-4 Au(III) complexes display very interesting luminescence properties 5,6 and biological activities, 7,8 and increasing efforts are made to develop their opto-electronic and therapeutic applications. Au(III) complexes also show unique catalytic properties, for the electrophilic activation of p-CC bonds and carbonyl compounds, and as intermediates in Au(I)/ Au(III) redox cycles. 3, [9][10][11] Hard C-and N-based ligands occupy a forefront position in gold(III) chemistry. 1, 12 In particular, (N,C) cyclometalated complexes A, and (C,N,C) and (N,C,C) pincer complexes B and B 0 offer a unique balance between stability and properties (Scheme 1). 6,10c,d,13-15 Strikingly, so P donors have also proved recently to efficiently stabilize gold(III) species. 16 (P,C) cyclometalated gold(III) complexes C (which are readily available by chelate-assisted oxidative addition of C-X bonds) 16a have been shown to display rich reactivity. 16b-f Unprecedented elementary organometallic reactions such as migratory insertion of alkenes 16b,c,e and b-hydride elimination 16d have been evidenced, enlarging the portfolio of chemical transformations of gold complexes. 2 To progress and develop further the chemistry of gold(III) complexes, it is critical to better understand how their properties and reactivity are inuenced by ancillary ligands. It was precisely the focus of this work to bridge the gap between widespread (N,C) gold(III) complexes and recently introduced (P,C) gold(III) species, and to try to answer the following questions: (i) Is the unprecedented reactivity recently evidenced with (P,C) gold(III) complexes specic to this ligand set or is it general? In other words, are (N,C) gold(III) complexes also prone to migratory insertion and b-H elimination? a QBIS-CAT Group, Institut de Química Computacional i Catàlisi (IQCC), Departament de Química, Universitat de Girona, Campus Montilivi, Girona, E-17003, Catalonia, Spain. E-mail: xavi.ribas@udg.edu (ii) What is the precise inuence of the ancillary ligand on reactions with alkenes? Does it affect the kinetics and thermodynamics? Does it change the mechanism and/or fate of the reaction?
To this end, we studied a cationic tricoordinate (N,C) gold(III) complex, namely [(N,C)AuPh] + 1A (Scheme 2), and we report here a detailed investigation of its reactivity towards ethylene. This study draws some analogies with the related [(P,C)AuPh] + complex 1C, 16e but also reveals noticeable differences. The weaker donicity of N versus P makes the (N,C) ligand much more electronically dissymmetric than the (P,C) ligand. As shown by detailed DFT investigations of the reaction proles, these electronic properties noticeably inuence the balance between the different reaction paths and eventually change the outcome of the ethylene insertion at gold. To the best of our knowledge, this is the rst time that the impact of ancillary ligands on gold(III) reactivity is thoroughly explored.

Results and discussion
Reaction of the [(N,C)AuPh] + complex 1A with ethylene Initially, we attempted the selective mono-arylation of the [(N,C) AuI 2 ] complex 1, which was recently prepared via C AR -I oxidative addition to Au(I), 13f following the same synthetic procedure as that described for the [(P,C)AuI 2 ] complex. 16e However, treatment of 1 with PhMgBr led to an intractable mixture of products. Hopefully, exchange of iodines at gold for more labile triuoroacetate groups (complex 2) allowed for the synthesis of the desired [(N,C)AuBrPh] complex 3-Br(Scheme 3), applying the conditions reported by Tilset and co-workers. 17 Complex 3-Br was isolated as a white powder in 59% yield aer column chromatography and characterized by NMR spectroscopy and high-resolution mass spectrometry (HRMS). 18 The cationic [(N,C)AuPh] + complex 1A was then generated by abstracting the bromide at the Au(III) atom with AgSbF 6 in dichloromethane under ethylene pressure (2 bar). No changes were observed within 72 h at room temperature, whereas upon heating at 40 C for 18 h, complete conversion into a new species with aliphatic signals above d 1.8 ppm was achieved (Scheme 4). Moreover, crystals formed spontaneously in small amounts inside the NMR tube, and X-ray diffraction analysis revealed double insertion of ethylene plus coordination of one molecule of adventitious water (Scheme 4). At this point, aiming at generating a more stable neutral four-coordinate gold(III) complex to allow for easier handling and characterization, 16d,e we quenched the reaction crude with nBu 4 NCl. A white solid was isolated in 60% yield and characterized by multinuclear NMR spectroscopy and HRMS as 5A-Cl. The relative integration of the aromatic and aliphatic signals in the 1 H NMR spectrum (13 versus 8H), and the peak at m/z 508.1346 corresponding to the mass of [(N,C)AuPh] + augmented by two ethylene molecules, corroborated the double insertion into the Au-Ph bond. The solid-state structure of the complex 5A-Cl was also determined by X-ray crystallography. 18 The gold atom sits in a square-planar environment formed by the (N,C) chelate ligand, the Cl atom and the (CH 2 ) 4 Ph chain located cis to the aryl moiety.
The stability of the cationic gold complex 5A generated under these conditions contrasts with the instability of 1A. In the absence of ethylene, the latter fully decomposes within 6 h in solution at room temperature into several side-products and metallic gold. At this stage, and without knowing the precise structure of 5A due to its ill-dened 1 H NMR spectrum (the fourth coordination site at gold may be occupied by the SbF 6 counter anion, the pendant Ph ring, an H 2 O or ethylene molecule), it is difficult to rationalize this difference.
The double addition of ethylene to complex 1A is reminiscent of the insertion of two norbornene (NB) molecules into the Au-Me bond of the (P,C)-ligated gold(III) dimethyl complex. 16b,c In this process, the insertion of a second NB into the Au-C norbornyl bond gives access to a more thermodynamically stable complex, with the Au-C norbornyl bond at the trans position relative to phosphorus. In a likewise manner, the strong trans inuence of the benzoquinolinic carbon in 1A triggers the migration of the alkyl group by insertion of a second ethylene molecule at the cis position relative to the aryl moiety (see DFT calculations below). In spite of the presence of a hydrogen atom at a b position and the tendency of cationic low-coordinate gold(III) alkyl complexes to undergo b-H elimination, 16d complex 5A shows unusual thermal stability and no formation of styrene or higher olens was detected during the course of the reaction. This is in stark contrast to the behaviour of the [(P,C)AuPh] + complex 1C, where, aer ethylene insertion, a b-H elimination and re-insertion sequence takes place leading to the gold(III)-arene complex II, which aer 1 h at room temperature eventually evolves into the linear gold(I) complex III by reductive elimination, styrene release and ligand redistribution (Scheme 4, bottom). 16e

D-labeling experiment
To assess the possible occurrence of b-H elimination in competition with ethylene insertion, a D-labeling experiment was carried out using trans-ethylene-d 2 (Scheme 5). The reaction conditions have been adapted to enable the use of transethylene-d 2 which was supplied by Cluzeau Info Labo in a lecture bottle (1 atm, 95.2% D). Using an atmospheric pressure of ethylene, complete conversion was achieved aer two ethylene rellings and 2 Â 24 hours heating at 40 C. The reaction was then repeated with trans-ethylene-d 2 using the same experimental conditions. 18 The reaction mixture was then trapped with nBu 4 NCl and analysed by NMR spectroscopy. The 1 H NMR spectrum shows the expected integration and pattern for the Au(CHD) 4 Ph enchainment (Scheme 5): the two terminal CHD moieties resonate as doublets compared to triplets for the non-deuterated complex (5A-Cl). Consistently, the corresponding 13 C NMR signals all appear as 1 : 1 : 1 triplets in the 13 C{ 1 H} NMR spectrum. There is very little sign of H/D scrambling, if any, conrming that b-H elimination does not occur to a signicant extent under these conditions and that ethylene insertion prevails.

Synthesis and characterization of the [(N,C)Au(nBu)] + complex 6A
To gain a better understanding of the reluctance of complex 5A to undergo b-H elimination, the analogous gold(III) n-butyl complex 6-Cl was synthesized and the behaviour of its cationic version was investigated (Scheme 6). The selective monoalkylation of [(N,C)Au(OAc F ) 2 ] complex 2 was successfully accomplished using 2.1 equiv. of the Grignard reagent nBuMgCl under identical conditions to those applied for the preparation of the [(N,C)AuBrPh] complex 3-Br. Aer column chromatography, the target [(N,C)AuCl(nBu)] 6-Cl complex was isolated as a bench-top stable white powder in 73% yield and characterized by multinuclear NMR spectroscopy and high-resolution mass spectrometry. 18 We then proceeded with the generation of the corresponding cationic gold(III) complex 6A by chloride abstraction with AgSbF 6 in CD 2 Cl 2 . As indicated by 1 H NMR, it is instantaneously formed at À60 C and it remains unchanged upon warming up to room temperature. No signs of degradation or b-H elimination (formation of butenes and butane) were observed by 1 H and 13 C NMR, even aer heating at 40 C for 24 h. In consequence, the cationic complex was fully characterized by NMR spectroscopy at room temperature without particular precautions. The 13 C NMR data reveal a signicant shi to high eld of the aromatic carbon directly bound to gold upon cationization (from d 145.37 ppm in the neutral complex 6-Cl to d 133.76 ppm in the cationic complex 6A). On the other hand, the alkylic carbon of the n-butyl chain attached to gold shis to low eld with Dd ¼ 4.6 ppm. Overall, the salient thermal stability of the nbutyl cationic complex is in accordance with the lack of b-H elimination in the ethylene insertion reaction of [(N,C)AuPh] + complex 5A.
In order to explain the double insertion of ethylene and selective formation of 5A, and to understand the different outcomes observed with the (N,C) and (P,C) complexes, a thorough DFT study was performed.
Computational investigation of the energy prole for the reaction of 1A with ethylene DFT calculations were carried out at the B3PW91/SDD + f(Au),6-31G**(other atoms) level of theory. Solvent effects (DCM) were taken into account by means of a polarizable continuum model (PCM). 18 Two energy minima corresponding to the cis and trans isomers of 1A were found on the potential energy surface (PES) (Fig. 1). By convention, cis and trans refer to the relative position of the phenyl (or alkyl group) at gold with respect to the carbon atom of the benzoquinolinic ligand. The cis isomer 1A is much more thermodynamically stable (by 36.2 kcal mol À1 ), in line with the detrimental trans arrangement of the phenyl and benzoquinolinic carbon atoms in t-1A.
1. First ethylene insertion. Compared to 1A in its groundstate cis geometry, the p-complex of ethylene 2A is slightly downhill in energy (DG À3.4 kcal mol À1 ) (Fig. 1). The alkene is coordinated in a symmetric fashion (two short Au/C contacts at 2.40Å). It is oriented perpendicular to the metal coordination plane 16c,19,20 and the CH 2 moiety atoms remain quasi-planar, in line with strong p(CC) / Au donation and weak Au / p*(CC) backdonation, as apparent from the NBO second order perturbation analysis (stabilization energies for the donor-acceptor interactions of 45.7 and 14.1 kcal mol À1 , respectively). Insertion of ethylene into the Au-Ph bond then proceeds via a planar 4center transition state. The resulting Au complex 3A is stabilized by p-arene coordination trans to the weak N donor (Au/C ortho and Au/C ipso contacts of 2.21 and 2.71Å, respectively and NBO second order stabilization energies for the p(C ipso C ortho ) / s*(AuN) donor-acceptor interaction of 31.0 kcal mol À1 ). 21 The activation barrier for the migratory insertion is 20.7 kcal mol À1 and the reaction is endergonic (DG +5.6 kcal mol À1 ). The involvement of the trans isomer t-1A was also considered, but as mentioned above, it lies very high in energy (36.2 kcal mol À1 above the cis isomer) and is thus very unlikely to participate. Note however that the very unfavourable trans arrangement of the two organic fragments makes the coordination-insertion of ethylene into t-1A very easy and strongly exergonic. Although the resulting insertion product c-3A is more stable than 3A by 22.0 kcal mol À1 , its involvement is unlikely because the barrier for trans / cis isomerization is also very high (28.2 kcal mol À1 ) due to the necessity to break the strong p-arene interaction.
2. Second ethylene insertion versus b-H elimination. In line with experimental observations, the insertion of a second ethylene molecule is highly favoured. Displacement of the parene coordination at 3A by ethylene is about thermoneutral and migratory insertion then proceeds with a very low energy barrier (6.6 kcal mol À1 for 3A) to give the (CH 2 ) 4 Ph complex 5A (Fig. 2). The extended linear form 5A is slightly more stable (by 2.5 kcal mol À1 ) than the g-CH agostic stabilized one 5A-g [a 7.1 kcal mol À1 s(CH) / s*(AuC) donor-acceptor interaction is found at the second-order perturbation level in NBO]. 18,22 b-H elimination from 3A was also considered. In line with that observed for [(P,C)Au(n-alkyl)] + complexes, the activation barrier for b-H elimination is low (only 2.8 kcal mol À1 ) and because of the unfavourable trans arrangement of 3A, the formation of the gold hydride styrene complex 7A is thermodynamically favoured (by 14.4 kcal mol À1 , DG). Rotation of styrene and re-insertion into the Au-H bond may occur to give the branched complex 8A as in the case of the (P,C) complex, but the corresponding energy barrier is relatively high (20.9 kcal mol À1 ). From a kinetic viewpoint, the second ethylene insertion and b-H elimination from 3A are likely to compete, but the absence of noticeable H/D scrambling upon reaction with ethylene-d 2 (Scheme 5) indicates a minimal occurrence, if any, of b-H elimination. The easiness of the re-insertion of styrene into 7A (DG s 17.2 kcal mol À1 ) combined with the high exergonic character of the second ethylene insertion into 3A drives the reaction to the formation of 5A. For the sake of completeness, b-H elimination, g-H transfer and third ethylene insertion into 5A, as well as second ethylene insertion into 8A have been considered but all these transformations were found to be kinetically or thermodynamically disfavoured ( Fig. S7 and S9-S11 †). 18 The same holds true for the second ethylene insertion and b-H elimination from c-3A, the more stable isomer of the rst ethylene insertion. It is in particular worth noting that if the activation barrier for b-H elimination from 5A is not prohibitively high (19.0 kcal mol À1 ), the reaction is strongly endergonic (DG 18.5 kcal mol À1 ) due to the unfavourable trans arrangement of the resulting gold hydride olen complex. This probably explains the stability of the [(N,C)Au(nBu)] + complex 6A noticed experimentally. The activation barrier for the third ethylene insertion is ca. 17 kcal mol À1 higher than that of the rst insertion, in line with the higher reactivity of the Au-C sp 2 vs. Au-C sp 3 bond, as previously observed with the [(P,C)Ph] + and [(P,C)AuMe] + complexes. 16d,e Comparison of the (N,C) and (P,C) complexes The above discussion has pointed out noticeable differences between the (N,C) and (P,C)-ligated gold complexes which are mainly associated with the weaker donor character of N versus P. As a result, the (N,C) ligand induces signicant electronic dissymmetry on gold compared to the (P,C) ligand, and the two reaction sites trans to N or C (cf. square-planar geometry of the Au(III) complexes) behave quite differently, as previously noticed by Tilset, Nova and co-workers for the tolylpyridine (tpy) ligand. 13c, d To conrm and highlight the difference in the electronic properties of the ligand, calculations were performed on the naked [(N,C)Au] 2+ and [(P,C)Au] 2+ fragments (Fig. 3). 18 In both cases, the LUMO corresponds to the vacant orbital at Au at the trans position to the N or P [main contribution of the s*(Au-N)/ (Au-P) orbitals]. It is signicantly lower in energy (by 1 eV) with the (N,C) ligand. In contrast, the LUMO + 1 orbitals of the two fragments, which correspond to the vacant orbitals at Au at the trans position to the aryl moiety, lie close in energy. Thus, the  (N,C) ligand induces a signicantly larger LUMO-LUMO + 1 gap than the (P,C) ligand and hence a stronger electronic dissymmetry of the coordination sphere at gold. Note also that the weaker donicity of the (N,C) ligand results in a signicantly more electrophilic gold center than with the (P,C) ligand: the LUMO is noticeably lower in energy and the Bader atomic charge at gold is much higher (1.05 versus 0.58).
To further compare the two systems and pinpoint the differences, calculations on the [(P,C)AuPh] + complex 1C were re-performed 16e at the same level of theory as for the [(N,C) AuPh] + complex 1A and completed to include the second ethylene insertion (Fig. S12 †). 18 The difference between the (N,C) and (P,C) ligands is spectacular in the energy gap between the cis and trans isomers of 1A and 1C (as well as of related species). As mentioned above, the cis form of 1A is more stable than the trans isomer by 36.2 kcal mol À1 (DG). The energy gap falls down to 7.1 kcal mol À1 for 1C. The difference is also striking in the rst ethylene insertion as well as in the competition between second ethylene insertion and b-H elimination. With the (N,C) ligand, the rst ethylene insertion is slightly uphill in energy (DG +2.2 kcal mol À1 ) while the second insertion is strongly exergonic (DG À27.9 kcal mol À1 ) and favoured over b-H elimination. In contrast, ethylene insertion into the (P and C) complex 1C is downhill in energy (DG À10.0 kcal mol À1 ) but the second ethylene insertion involves a high activation barrier (21.6 kcal mol À1 versus 15.1 kcal mol À1 for the rst insertion) and does not compete with b-H elimination/re-insertion leading to the branched product II. In addition, reductive elimination of the gold hydride 7A (C sp 2 -H coupling) was found to be signicantly more demanding energetically than that of the corresponding (P,C) complex 7C (the activation barrier is about 10 kcal mol À1 higher in energy, see

Conclusions
To sum up, the novel cyclometalated Au(III) complex [(N,C) AuPh] + 1A was shown to react with ethylene to selectively give a double insertion product (5A), without b-H elimination nor rearrangement of the linear (CH 2 ) 4 Ph chain. Trapping with chloride affords complex 5A-Cl which has been fully characterized, including by X-ray diffraction. The energy prole for the reactions of 1A with ethylene has been thoroughly investigated computationally, considering competitive paths, and the inuence of the ancillary ligand has been delineated by comparing the (N,C) and (P,C) complexes 1A and 1C.
Because of the weak donor character of nitrogen (compared with carbon and even phosphorus), the benzoquinoline ligand is very electronically dissymmetric. As a result, the two reactive sites at gold are quite different. This situation contrasts with the more symmetric nature of the (P,C) chelate and explains the peculiar reactivity of 1A. The electronic dissymmetry of the (N,C) ligand explains why b-H elimination is not observed upon reaction of 1A with ethylene, neither upon cationisation of 6-Cl. The activation barrier for b-H elimination is not prohibitively high, but the reaction is uphill in energy due to the unfavourable position of the hydride (trans to the carbon atom of the ligand) in the resulting complex.
The detailed mechanistic study of migratory insertion and b-H elimination has revealed that the properties of the ancillary ligand, and in particular the electronic dissymmetry that the (N,C) ligand induces at gold play a prominent role in reactivity. This situation is somewhat reminiscent of that arising with phosphine-sulfonate bidentate ligands which, thanks to their strong electronic dissymmetry, bestow unique properties to Pd(II) complexes towards the coordination-insertion polymerization of ethylene and polar olens. 23 This study reveals that (N,C) Au(III) complexes possess rich reactivity. It generalizes migratory insertion of alkenes at gold, an elementary process unknown with gold until very recently, but involved in many synthetically useful transformations (such as the Mizoroki-Heck reaction, the oligomerization/ polymerization of olens, etc.). The ancillary ligand markedly inuences the outcome of the reaction and thus, modulation of its structure paves the way for tuning and optimizing the properties of gold(III) complexes.
Accordingly, future work will seek to take advantage of well-dened (N,C) and (P,C) gold(III) complexes in catalysis and to expand further the variety of chelating ligands.

Conflicts of interest
There are no conicts to declare.