Catalytic Activity of trans-Bis(pyridine)gold Complexes

Gold catalysis has become one of the fastest growing fields in chemistry, providing new organic transformations and offering excellent chemoselectivities under mild reaction conditions. Methodological developments have been driven by wide applicability in the synthesis of complex structures, whereas the mechanistic understanding of Au(III)-mediated processes remains scanty and have become the Achilles’ heel of methodology development. Herein, the systematic investigation of the reactivity of bis(pyridine)-ligated Au(III) complexes is presented, based on NMR spectroscopic, X-ray crystallographic, and DFT data. The electron density of pyridines modulates the catalytic activity of Au(III) complexes in propargyl ester cyclopropanation of styrene. To avoid strain induced by a ligand with a nonoptimal nitrogen–nitrogen distance, bidentate bis(pyridine)–Au(III) complexes convert into dimers. For the first time, bis(pyridine)Au(I) complexes are shown to be catalytically active, with their reactivity being modulated by strain.


X-ray Crystallography
Single crystals of [3-Au(III)2], [4-Au(III)]2(AuCl4)2 and [4-Au(III)]2(BF4)2 were crystallized by slow diffusion of n-pentane into a dichloromethane solution of the complexes. A suitable crystal was selected and Bruker D8 APEX-II diffractometer equipped with a CCD camera using Mo Kα radiation (λ = 0.71073 Å). Data reduction was performed with SAINT, 5 Absorption corrections for the area detector were performed using SADABS. 5 The crystals were kept at 150(2) K during data collection. Using Olex2, 6 the structure was solved with the SHELXT 7 structure solution program using Intrinsic Phasing and refined with the SHELXL 8 refinement package using Least Squares minimization. [4-Au(III)]2(AuCl4)2 was refined a two-component twin.

Computational details
To gain structural parameters of the Au(III)/ Au(I)-complexes and insights into the process of acyloxy migration, we performed DFT calculations with the Gaussian 16 suite of programs (Revision A.03). 9 The calculations were carried out with dispersion-corrected B97X-D exchangecorrelation functional. [10][11] The SMD implicit solvation model was used to take into account the global solvation effects. 12 The solvent used in our calculations was dichloromethane following the reaction conditions. The ultrafine integration was employed to increase the accuracy of the numerical integration in all electronic structure calculations. Harmonic vibrational frequency calculations were utilized to identify the nature of the obtained structures. The vibrational analysis revealed that all located transition states have only one negative Hessian eigenvalue, while no imaginary frequencies were found for the reported minima.
The reported Gibbs free energies were obtained from B97X-D/Def2TZVPP electronic energies and all the additional terms computed at the B97X-D/Def2SVP level according to the following formula: In this formula, E0′ and E0 are electronic energies obtained using Def2TZVPP and Def2SVP basis sets, 13 respectively, G0 and Gsol are gas-phase and solution-phase Gibbs free energies obtained from ωB97X-D/Def2SVP calculations (T = 298.15 K, following the experimental conditions). The thermal and entropic contributions to the Gibbs free energies were computed by employing Grimme's quasi-RRHO approximation. This approach seems more appropriate than the standard ideal gas RRHO (rigid rotorharmonic oscillator) model, because most of the optimized structures possessed numerous low harmonic frequency modes. 14 The value of Gconc (0.003019 Hartree) corresponds to concentration correction to the Gibbs free energy when shifting from ideal gas standard state (p = 1 atm) to the standard concentration in solution phase (c = 1 mol/dm3).
The natural population analysis implemented in Gaussian 16 was performed at the B97X-D/Def2TZVPP level of theory.

Cis bis(pyridine) Au complexes
DFT calculations at the B97X-D/Def2SVP level predicts that the cis arrangement of the bis(pyridine) Au(III) complexes is higher in energy in all cases. The optimized structures are shown in Figure S5.

NBO charges of pyridine ligands
Natural population analysis was carried out to determine the net atomic charges of the pyridine N atoms (n(N)) at the B97X-D/Def2TZVPP level of theory (Table S2). We then calculated the change of the net atomic charge (n(N)) with respect to the unsubstituted pyridine (1-H).  The quality of the correlations was examined by determining the regression coefficients R 2 for the correlation of the change of the measured N chemical shift ( 15 N) with the change of the natural atomic population of N atom (n(N)) ( Figure S6). The change in both cases was defined to the unsubstituted pyridine (1-H). The regression coefficients R 2 was found to be high (0.993). Consequently, the calculated NBO charge of the pyridine N atom adequately reflects the electronic effects of the substituents on the pyridine ring as the calculated changes upon substitution correlates excellently with the measured  15 N data.

Substrate exchange with ligands
In the course of the reaction, the first step is expected to be the coordination of the propargyl ester 5 to the Au(III) center. The formation of such a complex could take place via either the exchange of a Clanion or a pyridine ligand 1-H. To examine the possibility of the two pathways, we first calculated the relative stability of the Au(III)(1-H)2Cl-5 complex with respect to the reactant state ([(1-H)2-Au(III)] + + 5). The formation of the [(1-H)2-Au(III)Cl]-5 complex was computed to be highly exergonic (46.2 kcal/mol), rendering this pathway highly unlikely ( Figure S8). Therefore, only the pathway involving the exchange of a pyridine 1-H was considered in the further computational investigations (see below section 4.6.).

Figure S8
The

Conformation of the reactive intermediate
Given that several orientations of the aromatic group are feasible, the conformational space was explored by generating possible rotamers along the Au-C1-C3-O dihedral angles (see A in Figure S9) which were then subjected to optimization at the B97X-D/Def2SVP level of theory. The cis and the trans geometry of the chlorides were both considered in our calculations. Various conformers were identified as minima. The most stable and additional conformers of the cis and the trans complexes are presented in Figure S9 and S10, respectively. The most stable arrangement was found to be A ( Figure S9) wherein geometry of the chlorides is cis and the activated triple bond is in close proximity to the ester group. This complex is predicted to be 23.2 kcal/mol above the reactant state, indicating that the formation of such an intermediate upon pyridine dissociation is still thermodynamically unfavored, leading to a highly reactive species. The second most stable conformer A-1 lying 24.1 kcal/mol above the reactant state was derived through rotating the aromatic moiety by about 180°. The orientation of the aromatic moiety found in complex A-2 and A-3 significantly decreased the stability.  In the case of the trans geometry ( Figure S10), complex A-4 was computed to be the most stable, in which the oxygen atom of the ester group is facing towards the Au(III) center. However, complex A remained favored over A-4 by 2.8 kcal/mol, being the most stable conformer identified in our analysis. Consequently, we used complex A to study the process of the acyloxy migration which is a key step in the formation of the substituted-cyclopropane product.

Acyloxy migration
To model the migration of the acyloxy group, we performed a series of constrained geometry optimizations (energy scans) departing from the most stable structure A while the dihedral angle that rotates the acyloxy group towards the triple bond (C2-C3-O-C4) in the less hindered direction was gradually altered by 1° ( Figure S11). By carrying out the constrained optimizations in this fashion, the obtained potential energy curve revealed an energy barrier of 0.2 kcal/mol for this motion. Using the structure found as the energy maximum on the potential energy curves as initial geometry, we identified a transition state that was found to be 24.4 kcal/mol above the reactant state. The structure of the obtained transition state is shown in Figure S12. The normal mode associated with the imaginary frequency computed for this transition state corresponded to rotational motion. The IRC calculation towards the product side showed the formation of a cyclic product B having the acyloxy group enclosed in a ring ( Figure S13). This intermediate B was predicted to be more stable than the reactant state by 4.8 kcal/mol. These computational experiments indicated that the coordinated propargyl ester 5 is highly activated and undergoes practically spontaneous intramolecular cyclisation with the acyloxy group by a simple rotational displacement of that group. The same series of calculations were performed rotating the acyloxy group in the opposite direction ( Figure S14). In this case, the rotation gave rise to a cyclisation involving the acyloxy group through an energy barrier of 9.2 kcal/mol. The total barrier of such a cyclisation process, therefore, would be 32.1 kcal/mol, which is prohibitively high at ambient conditions. The resultant product of the energy scan was found to be the same product that we obtained in the previous case ( Figure S13).

Substitution effect
Next, we examined the relative stabilities of structures A, TS and the cyclic intermediate B when the pyridine ligand was substituted with CF3-, CH3-, OCH3-groups. The obtained minima and transition states along with the energy data are shown in Figure S15, S16, S17 respectively.

Structures of the dimeric complex
In our study, a ligand having two pyridines linked by a phenanthrene ring was also examined. Following the synthetic protocol applied for the preparation of the [(1-H)2-Au(III)] + complex, a mixture of two Au(III)-complexes was formed based on the 15 N-NMR data (for the analysis of the 15 N-NMR data see section 1). From the mixture, we obtained a crystal structure which was assigned as the [3-Au(III)2] complex by single crystal X-ray crystallography (for crystallographic data see section 2). Due to the lack of a crystal structure of the other complex detected in the mixture, we performed DFT-calculations to identify its plausible structure. Its NMR spectrum suggests that the Au(III)-complex has a high degree of symmetry. Keeping that in mind, the following two symmetrical complexes could be obtained as minima at DFT level ( Figure S18). Complex [3-Au(III) + ]2was computed to be the most stable form of a dimeric structure. This is likely to be a result of the stabilizing - stacking between the phenanthrene rings in [3-Au(III) + ] 2 complex. The second most probable structure [3-Au(III) + ] 2-a, in which - stacking is absent, was predicted to be 5.9 kcal/mol higher in free energy.

S19
The dimeric complex shown above is not the only symmetrical structure that could be envisioned. First, we did not exclude the possibility of a monomeric complex in which the AuCl2 + fragment was situated between the two pyridines resulting in a highly symmetrical structure. DFT calculations were carried out to assess the relative stability of such a complex. The optimized structure of the monomeric complex is shown in Figure S19. It is apparent from the relative free energy that the presence of a monomeric complex is highly unlikely compared to the dimeric counterpart. Inspection of complex [(3-Au(III)] + reveals that the phenanthrene ring is highly distorted, which may explain its high instability.

Helix structures
DFT computations predicted the preferential formation of dimeric and helical structure for Au(I) (see Scheme 4), instead of a monomeric helix wherein both of the pyridines belong to two different ligands. The predicted preference was corroborated by the obtained single crystal structure of [4-Au(I)] + (see crystallographic data in section 2). We suggest that the difference found in the relative stabilities of the weakly interacting monomeric helix and the dimeric helix structures of [4-Au(I)] + and [4-I(I)] + systems could be associated with the difference in the ionic radii of Au + and I + ions. Indeed these radii are notably different (r(Au + ) = 1.99 Å and r(I + ) = 2.21 Å). [10] The shorter Au•••Au distance in [4-Au(I)] + allows closer van der Waals contacts (-stacking interactions) between the aromatic rings of the ligand as compared to those in [4-I(I)] + . To demonstrate the more favorable stacking interaction in [4-Au(I)] + , we selected two characteristic distances between the aromatic groups, highlighted in red in Figure S20. The distance between Cα and N as well as C1 and C2 atoms (indicated in red) is 0.21 and 0.12 Å shorter, respectively, in the preferred structure of Au(I) complex. This significant difference indicates the presence of a stronger - stacking in complex [4-Au(I)] + .

Cartesian coordinates of the reported structures
Cartesian coordinates of the optimized geometries are given below in standard XYZ format (units are in Å). The first line indicates the total number of atoms and the second line is the molecule name (as defined above in Table S3).        . Figure S38.