Synthesis of substituted (N,C) and (N,C,C) Au( III ) complexes: the in ﬂ uence of sterics and electronics on cyclometalation reactions †

Cyclometalated Au( III ) complexes are of interest due to their catalytic, medicinal, and photophysical properties. Herein, we describe the synthesis of derivatives of the type (N,C)Au(OAc F ) 2 (OAc F = tri ﬂ uoroace-tate) and (N,C,C)AuOAc F by a cyclometalation route, where (N,C) and (N,C,C) are chelating 2-arylpyridine ligands. The scope of the synthesis is explored by substituting the 2-arylpyridine core with electron donor or acceptor substituents at one or both rings. Notably, a variety of functionalized Au( III ) complexes can be obtained in one step from the corresponding ligand and Au(OAc) 3 , eliminating the need for organomer-cury intermediates, which is commonly reported for similar syntheses. The in ﬂ uence of substituents in the ligand backbone on the resulting complexes was assessed using DFT calculations, 15 N NMR spectroscopy and single-crystal X-ray di ﬀ raction analysis. A correlation between the electronic properties of the (N,C) ligands and their ability to undergo cyclometalation was found from experimental studies combined with natural charge analysis, suggesting the cyclometalation at Au( III ) to take place via an electrophilic aromatic substitution-type mechanism. The formation of Au( III ) pincer complexes from tridentate (N,C, C) ligands was investigated by synthesis and DFT calculations, in order to assess the feasibility of C(sp 3 ) – H bond activation as a synthetic pathway to (N,C,C) cyclometalated Au( III ) complexes. It was found that C(sp 3 ) – H bond activation is feasible for ligands containing di ﬀ erent alkyl groups (isopropyl and ethyl), although the C – H activation is less energetically favored compared to a ligand containing tert -butyl groups.

To understand and develop the chemistry of organometallic Au(III) complexes, robust synthesis protocols are of high importance.Varying the functionalization of the ligand backbone in coordination compounds is of interest as this can have a significant impact on their catalytic, [44][45][46][47] photophysical, 2,16-18,48-55 magnetic, 56,57 electrochemical [58][59][60][61][62] and biological [63][64][65] properties.Furthermore, it allows for the evaluation of the robustness of the metalation protocol.Therefore, we wanted to explore the possibility to synthesize functionalized derivatives of 2a-Au(OAc F ) 2 and 3b-AuOAc F (see general structures in Fig. 1).Variation of the ancillary 2-arylpyridine ligand is easily implemented, particularly through cross-coupling reactions, and by this a series of new ligands for Au(III) is readily available.In addition, cyclometalation as a strategy for the synthesis of (N,C) and (N,C,C) Au(III) complexes remains a somewhat underdeveloped field.6][67][68] Although efficient, this method suffers from the toxicity of mercury, creating a need to investigate and further develop alternative synthesis methods.We herein present the synthesis and characterization of a series of (N,C) and (N,C,C) Au(III) complexes.All complexes were conveniently prepared by microwave-heating using Au(OAc) 3 and the corresponding 2-arylpyridine ligand, with electron-donating or -withdrawing substituents at one or both rings.The formation and bonding properties of these complexes were assessed by DFT calculations, 15 N NMR spectroscopy and single-crystal X-ray diffraction analysis, in order to address any substituent effects.

Ligand synthesis
The ligands, substituted 2-arylpyridines (1c-1u), were readily available through the Suzuki-Miyaura cross-coupling of suitable 2-halogenated pyridine derivatives and arylboronic acids (Scheme 1A), using reaction conditions reported in the literature. 8,69,70Ligand 1r was prepared from 1q (Scheme 1B) according to a method developed by Fagnou and co-workers for the installation of pentafluorophenyl groups through Pdcatalysed C-H activation. 71

Synthesis of (N,C)-cyclometalated Au(III) di(trifluoroacetate) complexes
Having in hand a wide variety of potential ligands, we investigated their ability to cyclometalate at Au(III) under the same reaction conditions as utilized for the synthesis of 2a-Au (OAc F ) 2 (Scheme 2). 23he Au(III) complexes were obtained in yields ranging from 27 to 95%, and both electron-withdrawing (nitro, trifluoromethyl) and electron-donating (methyl, methoxy) substituents were tolerated.They were characterized by multinuclear NMR spectroscopy ( 1 H, 13 C, 19 F and 15 N NMR), MS, elemental analysis and single-crystal X-ray diffraction analysis.For Au(III) complexes carrying substituents on the pyridine ring, no clear trends in yields were observed.For complexes with substituents on the phenyl ring, certain trends in yields were found.For evaluation of the experimental observations, the carbon that undergoes cyclometalation in the protonated ligands (C2′, see Scheme 3) was investigated by natural charge analysis.In the following section, the natural charge of C2′ in the ligands are discussed relative to the charge of C2′ in the 2-phenylpyridinium cation, ppy-H + (ΔC2′ = 0) (for full details, see ESI †).The dimethyl-substituted complex 2h-Au(OAc F ) 2 was obtained in high yield (95%) like complex 2a-Au(OAc F ) 2 (94%). 19omplexes carrying either significantly electron-withdrawing (2e-Au(OAc F ) 2 , 2o-Au(OAc F ) 2 and 2r-Au(OAc F ) 2 ) or electrondonating (2k-Au(OAc F ) 2 ) groups were obtained in lower yields.The difluoro-substituted complex 2c-Au(OAc F ) 2 was obtained in 85% yield, supporting an electrophilic aromatic substitution-type of cyclometalation mechanism (Scheme 3).3][74][75][76] Despite being inductively electron-withdrawing, fluorine groups are ortho/para-directing and activating substituents in aromatic electrophilic substitution reactions, causing the C(sp 2 )-H acti- vation to proceed more easily at ligand 1c compared to e.g.1e, 1o and 1r.
Neither di(trifluoromethyl)-nor dimethoxy-substituted ligands 1f and 1j provided the desired cyclometalated products.The reaction of 1j and Au(OAc) 3 furnished a multitude of species as seen in the 1 H NMR spectrum of the crude product (Fig. S89, ESI †), and no single, clean compound could be isolated.It was however possible to obtain crystals suitable for single-crystal XRD, showing an unusual dinuclear M 2 L 3type complex (2j 3 Au 2 (OAc F ) 2 , see Fig. S146 and Fig. S147, ESI †).Having said this, we do not believe this to be the main reaction product, as no other characterization data can substantiate this.We surmise that the ligand is too electron rich and reactive to yield clean formation of a Au(III) complex under the investigated reaction conditions.This assumption is further supported by the calculated natural charge for C2′ in 1j-H + (ΔC2′ = −0.120).The natural charge for C2′ in 1j-H + is significantly larger than the ones found for ligands 1c-H + (ΔC2′ = −0.084)and 1h-H + (ΔC2′ = −0.008),which both have activating substituents in the 3′-and 5′-positions.For di(trifluoromethyl)-substituted 1f, the N-coordinated adduct 1f-Au (OAc F ) 3 (which is a likely precursor for the cyclometalation step) was isolated (Scheme 4).
The failure of obtaining cyclometalated 2f-Au(OAc F ) 2 can be related with the poor electrophilicity of 1f (ΔC2′ = +0.061for 1f-H + ).It was previously reported by our group that pincer complex 3b-AuOAc F (derived from the sterically encumbered and electron rich ligand 1b) proceeds via the corresponding (N,C)-cyclometalated complex 2b-Au(OAc F ) 2 (see below). 8As the trifluoromethyl group is smaller than the tert-butyl group, 77,78 the formation of 2f-Au(OAc F ) 2 should be feasible from a steric point-of-view.If the Au-C bond formation takes place by electrophilic C(sp 2 )-H bond activation (Scheme 3), (a) strongly electron-withdrawing substituent(s) in the aryl ring of the ligand might impede the reaction.Reaction of the monotrifluoromethyl-substituted ligand 1g with Au(OAc) 3 gave a ca. 2 : 1 mixture of the two regioisomers 2g-p-Au(OAc F ) 2 and 2g-o-Au(OAc F ) 2 (Scheme 5).Thus, cyclometalation of Au(III) is compatible with the steric demands of an ortho-trifluoromethyl group.This result therefore suggests that the electronic influence is the main reason for the failure to produce 2f-Au (OAc F ) 2 , as the di(trifluoromethyl)-substituted ligand 1f is more electron deficient than the corresponding mono-trifluoromethyl-substituted ligand 1g (ΔC2′ = +0.025for 1g-p-H + and ΔC2′ = +0.038for 1g-o-H + ).

Synthesis of (N,C,C)-cyclometalated Au(III) trifluoroacetate pincer complexes
Following the successful microwave-mediated synthesis of pincer complex 3b-AuOAc F via C(sp 2 )-H and C(sp 3 )-H bond activation of ligand 1b (Fig. 3), 8 we sought to investigate related tridentate ligands (1s-1u) in order to get more insight into the scope and limitations of the pincer formation.DFT calculations on the formation of 3b-AuOAc F suggested the directing effect of the bulky tert-butyl group in ligand 1b to be a key element in the successful synthesis of the complex. 8herefore, ligands 1s and 1t with less sterically demanding substituents, but otherwise an identical substitution pattern to that of 1b were investigated as tridentate ligands for Au(III).Additionally, the mono-substituted analogue of 1t, ligand 1u, was investigated in order to probe the selectivity of pincer formation relative to (N,C) cyclometalation.
Attempts of synthesizing 3s-AuOAc F in an analogous manner (microwave heating at 120 °C for 30 min) to 3b-AuOAc F failed, and the N-coordinated adduct of ligand 1s, 1s-Au(OAc F ) 3 , was the main species observed in the 1 H NMR spectrum of the crude product.The combination of lower reaction temperature (80 °C) and longer reaction time (3.5 h) did however furnish the pincer complex 3s-AuOAc F in moderate yields (38%) (Scheme 6).It is noteworthy that the C(sp 3 )-H bond activation of the isopropyl group introduces a chiral Scheme 4 Synthesis of 1f-Au(OAc F ) 3 .The corresponding (N,C) complex 2f-Au(OAc F ) 2 was not observed.centre in close proximity to gold in 3s-AuOAc F , which has previously been observed for a structurally related Pt(IV) complex. 79We did not attempt to resolve the enantiomers, but the accessibility to chiral centres through cyclometalation is a topic that deserves further investigation, especially if an enantiopure complex can be obtained. 80y employing the reaction conditions in Scheme 6, the ethyl analogue 3t-AuOAc F was obtained in a good yield (65%) from ligand 1t.In addition to the synthesis of 3s-AuOAc F and 3t-AuOAc F , mono-ethyl-substituted 1u was explored as a tridentate ligand for Au(III).The reaction of 1u with Au(OAc) 3 yielded the desired pincer complex 3u-AuOAc F as a mixture with the corresponding (N,C)-cyclometalated complex 2u-p-Au(OAc F ) 2 (Scheme 7).The other possible (N,C)-cyclometalated complex 2u-o-Au(OAc F ) 2 could not be observed, supporting the involvement of 2u-o-Au(OAc F ) 2 as an intermediate for the formation of 3u-AuOAc F (see ESI † for details).To summarize, the successful syntheses of 3s-AuOAc F and 3t-AuOAc F show that the sterically induced pre-orientation of the C-H bond that is activated (as seen in ligand 1b) is not a strict requirement for C(sp 3 )-H activation at Au(III), although the experimental observations suggest that the process is more feasible for 1b compared to 1s and 1t.Importantly, the results show that cyclometalation of Au(III) through C(sp 3 )-H bond activation takes place at a β position regardless of the nature of the alkyl group in the ligand.
tallized with distorted square planar geometry around Au(III) (τ′ 4 = 0.09 for 3s-AuOAc F and τ′ 4 = 0.11 for 3t-AuOAc F ).The bond lengths and angles of the complexes are similar to those reported for 3b-AuOAc F (Table 2).Interestingly, only one enantiomer of the racemic mixture of pincer complex 3s-AuOAc F could be modelled as the major component during refinement of the single-crystal X-ray structure.
15 N NMR spectroscopic studies of cyclometalated Au(III) complexes Some of the complexes and ligands discussed herein were investigated by 15 N NMR spectroscopy, and coordination shifts Δδ 15 N (δ 15 N complex -δ 15 N ligand ) were obtained in order to gain insight about the Au-N interactions.Furthermore, a selection of previously reported tpy-ligated Au(III) complexes 8,14,19,22,25,95 with varying substituents cis and trans to pyridine-N was studied by 15 N NMR spectroscopy to shed light on which factors influence the 15 N NMR chemical shifts of N-ligated square planar d 8 metal complexes (Fig. 6).Δδ 15 N were found in the range of −88.2 ppm to −104.5 ppm for the (N,C) di(trifluoroacetate) Au(III) complexes (except for 2m-Au(OAc F ) 2 ; see discussion below).7][98] The 15 N NMR data can be interpreted in a similar manner as the data from the single-crystal X-ray diffraction analysis of the complexes, where the Au-N bond lengths were found to be little dependent of the substituents in the ligand backbone.This reflects that other factors, such as the ligand trans to pyridine-N, 99 affects Δδ 15 N stronger than the substituents on the pyridine ring.Similar observations have also been reported by Pazderski for square planar Pd(II) and Pt (II) complexes with substituted bipyridine and phenanthroline ligands. 96he coordination shift of the 6-methoxy-substituted complex 2m-Au(OAc F ) 2 (Δδ 15 N = −66.1 ppm) is significantly smaller than those of the other substituted di(trifluoroacetate) complexes.It seems likely that the relatively small Δδ 15 N found for 2m-Au(OAc F ) 2 is a result of the weak interaction between the methoxy-oxygen and gold.This potential interaction was also observed in the single-crystal structure of the Au(III) complex.The coordination shift for the 6-methyl-substituted complex 2i-Au(OAc F ) 2 (−88.2ppm) was in the same range as those obtained for the other di(trifluoroacetate) complexes studied herein, although the Au-N bond length in the crystal structure of 2i-Au(OAc F ) 2 (2.034(3) Å) was similar to the one observed for 2m-Au(OAc F ) 2 (2.0426(17) Å).The very similar Au-N bond lengths in the two complexes strengthens the argument that the relatively small coordination shift obtained for 2m-Au(OAc F ) 2 is caused by an interaction between oxygen and gold, rather than steric repulsion between gold and the 6-substituent. 84If this was the case, it would be expected to yield a comparable coordination shift for 2i-Au(OAc F ) 2 .
For complex 1f-Au(OAc F ) 3 , Δδ 15 N was found to be −114.1 ppm, larger than what was observed for the cyclometalated complexes and also larger than what has been reported for pyridine-ligated AuCl 3 complexes in the literature (ca.][105] To further evaluate the effect of the identity of the ligand trans to nitrogen on the coordination shift of pyridine-N, Δδ 15 N for 2a-Au(OAc) 2 and 2a-Au(CH 3 ) 2 were obtained, and compared to the one found for 2a-Au(OAc F ) 2 .For the three complexes, Δδ 15 N was found to decrease in the order 2a-Au (OAc F ) 2 ∼ 2a-Au(OAc) 2 > 2a-Au(CH 3 ) 2 , agreeing with the established trans influence of the corresponding ligands; 19,103,104 Coordination shifts were also obtained   The corresponding data for 3b-AuOAc F are included for reference purposes. 8For 3s-AuOAc F , metric data for one of the two molecules in the asymmetric unit are listed.See Fig. S160 (ESI) † for metric data for both molecules in the asymmetric unit.For 3t-AuOAc F , metric data for one of the four molecules in the asymmetric unit are listed.See Fig. S162 (ESI) † for metric data for all molecules in the asymmetric unit.
for the pincer complexes 3b-AuOAc F , 3s-AuOAc F and 3t-AuOAc F which all have an alkyl group trans to pyridine-N.As expected from the differences in relative trans influence strength of an alkyl ligand and a [OAc F ] − ligand, the coordination shifts for the three pincers were significantly smaller (Δδ 15 N from −37.1 to −39.5 ppm) than those obtained for the di(trifluoroacetate) complexes.Surprisingly, they were also found to be smaller than the coordination shifts of 2a-Au(CH 3 ) 2 , 2a-Au (CH 2 CHvCH 2 )Br, 2a-Au(CH 3 )Br and [2a-Au(C,N)] + [OAc F ] − , although it could be anticipated that the relative trans influence of the ligands trans to pyridine-N would be similar for all these complexes.The observations may be explained from differences in the relative cis influence of an alkyl ligand, a halide ligand and a carboxylate ligand.A smaller Δδ 15 N was found for 2a-Au(CH 3 )Br (Δδ 15 N = −46.7 ppm) compared to 2a-Au(CH 3 ) 2 (Δδ 15 N = −56.1 ppm) being in accordance with the reported higher cis influence of halide ligands compared to alkyl ligands. 106,107In summary, the findings from the 15 N NMR spectroscopic studies show that functionalization of the ligand backbone has little effect on the interaction between pyridine-N and gold, whereas the nature of the ligands cis and trans to pyridine-N has a significantly larger effect.
DFT calculations on the formation of (N,C,C)-cyclometalated Au(III) complexes 3s-AuOAc F and 3t-AuOAc F In order to gain understanding of pincer formation for tridentate (N,C,C) ligands 1s and 1t, DFT calculations were performed.The formation of 3s-AuOAc F and 3t-AuOAc F starting from complexes 2s-Au(OAc F ) 2 and 2t-Au(OAc F ) 2 via the same mechanism proposed for the formation of 3b-AuOAc F 8 was explored (Fig. 7).As the formation of 2s-Au(OAc F ) 2 was found to be very similar to that of 2b-Au(OAc F ) 2 , the energies for this first cyclometalation step are not included in the figure (see Table S17, ESI † for details), and was not calculated for 2t-Au (OAc F ) 2 .
Looking at the C-H activation step, a clear difference in energy for TS1 was observed for the three complexes, illustrating the effect of the substituent on the formation of the desired pincer complex.The endergonic dissociation step forming the agostic intermediates [2s-Au-OAc F ] + [OAc F ] − and [2t-Au-OAc F ] + [OAc F ] − is more than 10 kcal mol −1 higher in Fig. 6 Overview of (N,C) and (N,C,C) Au(III) complexes studied by 15 N NMR spectroscopy herein.All data were collected in CD 2 Cl 2 at either 600 or 800 MHz.See also Table S1 Fig. 7 Free energy profile in kcal mol −1 for the formation of 3b-AuOAc F , 3s-AuOAc F and 3t-AuOAc F from the corresponding di(trifluoroacetate) complexes.In the figure, 2s-Au(OAc F ) 2 , [2s-AuOAc F ] + [OAc F ] − and 3s-AuOAc F are displayed as structural examples.The energies of all minima and transition states in brackets are computed in CH 2 Cl 2 (SMD) for 3s-AuOAc F and 3t-AuOAc F .The energies and transition states for 3b-AuOAc F were computed in HOAc F . 8 energy for 3s-AuOAc F and 3t-AuOAc F (24.0 kcal mol −1 and 25.0 kcal mol −1 , respectively), compared to 3b-AuOAc F (13.6 kcal mol −1 ).The following proton abstractions by [OAc F ] − that furnish 3s-AuOAc F and 3t-AuOAc F are barrier-free, which is in accordance with what was found for 3b-AuOAc F .
The energy barrier associated with the formation of the agostic intermediates is depending on the bulkiness of the alkyl substituent that undergoes C-H activation.The lower energy barrier for [2b-Au-OAc F ] + [OAc F ] − is due to the higher energy of 2b-Au(OAc F ) 2 relative to TS1 due to steric interaction with the tert-butyl group in the (N,C)-cyclometalated complex.In order to highlight the difference in stability of the (N,C)cyclometalated complexes 2b-Au(OAc F ) 2 and e.g.2s-Au(OAc F ) 2 , which is responsible for the significantly lower TS1 found for the tert-butyl system, an isodesmic reaction of a formal chelate ligand exchange on 2b-Au(OAc F ) 2 with 1s was investigated (Scheme 8).The formation of 2s-Au(OAc F ) 2 and 1b is favoured by 9.9 kcal mol −1 which illustrates the negative effect the large, bulky substituents have on the stability of di(trifluoroacetate) complexes.On the other hand, this characteristic of the tertbutyl substituent ultimately facilitates the C(sp 3 )-H bond activation step and subsequent pincer formation.

Conclusions
In this work, a detailed experimental and computational study of a series of 2-arylpyridine-based (N,C)-and (N,C,C)-cyclometalated Au(III) complexes has been presented.For the (N,C) systems, it was found that the scope of microwave-mediated synthesis of cyclometalated Au(III) complexes is broad and that a large variety of different functional groups is tolerated.This makes it an attractive method for the synthesis of Au(III) complexes of substituted arylpyridine ligands without having to resort to any organomercury intermediates.The efficiency of the reaction is strongly dependent on the electronic features of the (N,C) ligand, being consistent with an electrophilic aromatic substitution-type mechanism for cyclometalation at Au (III).Natural charge analysis performed on the protonated (N, C) ligands was found to correlate with the experimental obser-vations of their reactivity towards Au(OAc) 3 .Single-crystal X-ray diffraction and 15 N NMR spectroscopy data suggest that the (N,C) Au(III) complexes are structurally similar species, meaning that the ligand scaffold is flexible to changes without drastically affecting the coordination sphere around Au. Detailed 15 N NMR spectroscopic studies of different cyclometalated Au(III) complexes with varying ligands trans to the (N,C) backbone show that these ligands have a much stronger influence on the Au-N interaction than any substituents in the backbone, with the exception of the 6-methoxy-substituted complex 2m-Au(OAc F ) 2 .For this complex, the relatively small coordination shift may be explained by a possible weak interaction between the methoxy-oxygen and gold.In addition to the studies of (N,C)-cyclometalated Au(III) complexes, C(sp 3 )-H bond activation as a synthetic feasible method to yield (N,C,C)cyclometalated Au(III) complexes was expanded.Earlier we have shown that the ligand 2-(3,5-di-tert-butylphenyl)pyridine (1b) functions as a tridentate ligand for Au(III), yielding pincer complex 3b-AuOAc F .We have broadened the scope of Au(III) pincer formation, via C(sp 3 )-H bond activation to include less sterically encumbered ligands.These ligands contain either isopropyl (1s) or ethyl (1t and 1u) groups, where the C-H bond that is activated is able to rotate away from gold.In a combined experimental and computational effort we have shown that these ligands indeed undergo C(sp 3 )-H bond activation, but that the process is less facile than for the tert-butyl-substituted system.Further work will focus on broadening the scope of (N,C,C) pincer formation from alkyl-substituted 2-arylpyridines, as well as investigating their reactivity and optical properties.

Computational details
Calculations were carried out at the DFT level as implemented in the Gaussian16 software package. 116The hybrid PBE0+GD3 functional 117,118 including Grimme's model for dispersion forces was used to optimize all geometries.This methodology was selected based on previous studies which have proven its solid performance in the modelling of Au(III) complexes. 7,24,25,119,120C, H, F, N and O were described with the all-electron triple-ζ 6-311+G** basis set, 121,122 whereas Au was described with the Stuttgart-Köln basis set including a small-core quasi-relativistic pseudopotential. 123,124NBO7 calculations were performed in order to analyse the natural charges. 125Geometries were fully optimized without any constraint.Vibrational frequencies were computed at the same level of theory to classify all stationary points as either saddle points (transition states, with a single imaginary frequency) or energy minima (reactants, intermediates and products, with only real frequencies).The Gibbs free energy used in the discussion includes both the thermochemistry and the refined energy.All optimizations were carried out in solvent (CH 2 Cl 2 or HOAc F ) using the SMD solvation model. 126HOAc F was defined as eps = 8.55, epsinf = 2.26 and rsolv = 13.7.In the bimolecular steps, the energies were corrected for the 1 M standard state.

Fig. 8
Fig. 8 Numbering scheme used for reporting the NMR data.