Synthesis, Structure and Bonding Analysis of the Zwitterionic PPP-Pincer Complex (6-Ph 2 P-Ace-5-) 2 P(O)AuCl 2

: The reaction of (6-Ph 2 P-Ace-5-) 2 P(O)H with (tht)AuCl 3 proceeds via elimination of tetrahydrothiophene (tht) and HCl, providing the zwitterionic PPP-pincer complex (6-Ph 2 P-Ace-5-) 2 P(O)AuCl 2 ( 1 ) as yellow crystals. The molecular structure of 1 was established and studied by X-ray crystallography. The electronic structure was computationally analyzed using a comprehensive set of real-space bonding indicators derived from electron and electron-pair densities, providing insight into the relative contributions of covalent and non-covalent forces to the polar-covalent Au–Cl, Au–P, and P–O − bonds; the latter being one of the textbook cases for strongly polarized covalent interactions. Partial spatial complementarity between both bonding aspects is suggested by the electronic properties of the distinctively di ﬀ erent Au–Cl bonds.


Introduction
Recently, we introduced bis(6-diphenylphosphinoacenaphth-5-yl)phosphine oxide (6-Ph 2 P-Ace-5-) 2 P(O)H (I) as a novel ligand in coordination chemistry [1]. Like many other secondary phosphine oxides, I is in equilibrium with the corresponding phosphinous acid (6-Ph 2 P-Ace-5-) 2 POH (I'), from which it proved to react with nickel and palladium compounds to give a number of well-defined PPP-pincer complexes [1]. In an effort to broaden the scope of the new ligand, we have now extended our study to (tht)AuCl 3 (tht = tetrahydrothiophene) which yielded the metastable zwitterionic complex (6-Ph 2 P-Ace-5-) 2 P(O)AuCl 2 (1) as yellow crystals. The spatial arrangement of the Au atom was studied by means of X-ray crystallography, whereas the electronic structure was studied by density functional theory (DFT) calculations, transferring the C-H-distance corrected experimental molecular structure into the gas phase. Topological, surface, and integrated electronic bonding parameters were determined according to the Atoms-In-Molecules (AIM), [2] non-covalent interactions (NCI) [3] index, and electron localizability indicator (ELI-D) [4] methods. Real-space bonding indicators (RSBI) complement the orbital picture and extend the Lewis-picture of chemical bonding (e.g., by uncovering the vital role of weak secondary intramolecular interactions on the overall stability of complex molecules), and afford information of tiny electronic rearrangements due to structural changes. The AIM bond topology includes all types and strengths of chemical interactions and provides atomic/fragmental charges and volumes, which has made it a more routinely used tool for bonding analysis in the last three decades [5]. However, atom-atom contacts within cage structures (e.g., boranes within cage structures (e.g., boranes or metal-cyclopentadienyl complexes) or those with sharp bonding angles (e.g., intramolecular O-H … O) may be hidden in the electron density (ED) not giving rise to the formation of a bond critical point (bcp) [6,7]. Analysis of the reduced density gradient, s(r) = [1/2(3π 2 ) 1/3 ]|ρ|/ρ 4/3 , according to the NCI method, is used to visualize (extended) areas of noncovalent bonding aspects and turned out to be useful to detect contacts which are not discernible by AIM [8]. Mapping the ED times the sign of the second eigenvalue of the Hessian (sign(λ2)ρ) on the iso-surfaces of s(r) facilitates the estimation of different non-covalent contact types to be steric/repulsive (λ2 > 0), van der Waals-like (λ2 ≈ 0), or attractive (λ2 < 0). NCI is greatly complemented by the ELI-D, which provides electron populations and volumes of bonding and lone-pair basins and is especially suitable for the analysis of covalent bonding aspects. Notably, the combination of NCI and ELI-D (or its predecessor, ELF [9]; electron localization function) suggests the partial spatial separation of both bonding aspects [10,11]. Bond polarities are affected by the interplay between covalent and non-covalent bonding aspects and can be estimated by the intersection of ELI-D bonding basins with AIM atomic basins according to the Raub-Jansen-Index, [12] which discloses how bonding electrons are distributed over adjacent atoms forming a bond. Accordingly, the combination of AIM, NCI, and ELI-D provides a wealth of information, which is not accessible by the sole inspection of the molecular geometry or orbital-based approaches. In this study, RSBI are applied to characterize the polar-covalent Au-Cl, Au-P, and P-O − bond types.

Results and Discussion
The reaction of the secondary phosphine oxide (6-Ph2P-Ace-5-)2P(O)H (I) [1] with (tht)AuCl3 in CH2Cl2 at -40°C occurred via the equilibrium of the phosphinous acid (6-Ph2P-Ace-5-)2POH (I'), and proceeded with the elimination of tetrahydrothiophene (tht) and HCl, and afforded the zwitterionic PPP-pincer complex (6-Ph2P-Ace-5-)2P(O)AuCl2 (1) as yellow crystals, see Scheme 1. In solution, the stability of 1 is very limited; it decomposed within minutes at rt and within a day at −40 °C, which effectively precluded any characterization by NMR spectroscopy. The molecular structure of 1 was established by X-ray crystallography and is shown in Figure 1. Selected bond lengths are collected in the caption. The spatial arrangement of the Au(III) atom of 1 is square planar when taking into account the P3Cl donor set of the first coordination sphere.  Considering also the second chlorine atom, the spatial arrangement can alternatively be described as distorted square pyramidal (4 + 1 coordination), as shown by the related geometry index, 5 ( - / 60° with  = 175.2(1)° and  = 164.1(1)°) = 0.185 (5 is 0 for an ideal square pyramidal structure and 5 is 1 for an ideal trigonal bipyramidal structure) [13]. The secondary Au2···Cl2 contact (2.830(2) Å ) is 0.438(2) Å longer than the primary Au-Cl bond (2.392(2) Å ). The coordination of the phosphorus atoms is also asymmetric. The Au1-P1 and Au1-P2 bond lengths (2.345 (2)  The electronic structure of 1 was studied by DFT calculations. Figure 2a displays the AIM bond topology of 1, showing the two primary Au-Cl and three Au-P bonds, as well as three secondary Cl···H and one Cl···Cπ contact, along with all of the C-C, C-H, P-C, and P-O-bonds. The iso-surface representation of the NCI, however, proves the existence of additional C-H···C contact patches between the organic fragments, which do not give rise to the formation of a bond critical point (bcp), shown in Figure 2b. This points towards an additional London-dispersion stabilization of the complex. The quantitative properties at the P-O -, Au-Cl, and Au-P bcps are given in Table 1, and those of the secondary contacts in Table 2. The AIM atomic and fragmental charges are listed in Table  3. In the strong and short P-Obond, covalent and non-covalent bonding aspects are both pronounced [14]. Covalent contributions are indicated by a high value for the electron density (ED, ρ(r)bcp) at the P-O bcp of 1.64 eÅ −3 and a highly negative total energy density over ED ratio (H/ρ(r)bcp: -0.93 a.u.). Moreover, an ELI-D bonding basin is formed, which contains 1.64e (NELI) in a volume of 2.0 Å -3 (VELI), as seen in Figure 2c, whereas no NCI basin is formed, as shown in Figure 2d. Noncovalent contributions are indicated by a strongly positive Laplacian of the ED ( 2 ρ(r)bcp: 29.3 eÅ -5 ) and a kinetic energy density over ED ratio of G/ρ(r)bcp = 2.18 a.u. With 77.8%, the Raub-Jansen-Index (RJI) is half-way between nonpolar-covalent (50-60%, e.g., C-C) and ionic or dative (> 90%, e.g., Li + For N→B). Considering also the second chlorine atom, the spatial arrangement can alternatively be described as distorted square pyramidal (4 + 1 coordination), as shown by the related geometry index, τ 5 (β -α/60 • with β = 175.2(1) • and α = 164.1(1) • ) = 0.185 (τ 5 is 0 for an ideal square pyramidal structure and τ 5 is 1 for an ideal trigonal bipyramidal structure) [13]. The secondary Au2···Cl2 contact (2.830(2) Å) is 0.438(2) Å longer than the primary Au-Cl bond (2.392(2) Å). The coordination of the phosphorus atoms is also asymmetric. The Au1-P1 and Au1-P2 bond lengths (2.345 (2) and (2.365(1) Å) are slightly longer than the Au1-P3 bond length (2.298(2) Å) that is affected by the trans-effect of the Cl1 atom. The P3-O1 bond (1.483(4) Å) is very short; even shorter than those of the other two known zwitterionic complexes [(5-Ph 2 P-Ace-6-) 2 P(O)NiCl] (1.497(5) Å) and [(5-Ph 2 P-Ace-6-) 2 P(O)PdCl] (1.509(2) Å) [1].
The electronic structure of 1 was studied by DFT calculations. Figure 2a displays the AIM bond topology of 1, showing the two primary Au-Cl and three Au-P bonds, as well as three secondary Cl···H and one Cl···C π contact, along with all of the C-C, C-H, P-C, and P-O-bonds. The iso-surface representation of the NCI, however, proves the existence of additional C-H···C contact patches between the organic fragments, which do not give rise to the formation of a bond critical point (bcp), shown in Figure 2b. This points towards an additional London-dispersion stabilization of the complex. The quantitative properties at the P-O − , Au-Cl, and Au-P bcps are given in Table 1, and those of the secondary contacts in Table 2. The AIM atomic and fragmental charges are listed in Table 3. In the strong and short P-O − bond, covalent and non-covalent bonding aspects are both pronounced [14]. Covalent contributions are indicated by a high value for the electron density (ED, ρ(r) bcp ) at the P-O bcp of 1.64 eÅ −3 and a highly negative total energy density over ED ratio (H/ρ(r) bcp : -0.93 a.u.). Moreover, an ELI-D bonding basin is formed, which contains 1.64e (N ELI ) in a volume of 2.0 Å −3 (V ELI ), as seen in Figure 2c, whereas no NCI basin is formed, as shown in Figure 2d. Non-covalent contributions are indicated by a strongly positive Laplacian of the ED (∇ 2 ρ(r) bcp : 29.3 eÅ −5 ) and a kinetic energy density over ED ratio of G/ρ(r) bcp = 2.18 a.u. With 77.8%, the Raub-Jansen-Index (RJI) is half-way between nonpolar-covalent (50-60%, e.g., C-C) and ionic or dative (>90%, e.g., Li + F − or N→B). The three Au-P bonds are about 0.8 Å longer, but still show the signatures of both bonding aspects, although much weaker: ρ(r)bcp is significantly below 1 eÅ -3 and H/ρ(r)bcp is on average -0.45 a.u., which is balanced by  2 ρ(r)bcp being close to zero and G/ρ(r)bcp being about 0.5 a.u. Just like the P-Obond, the Au-P bonds form an ELI-D but no NCI basin (disregarding the tiny ring-shaped and red-colored surfaces perpendicular to the Au-P axes). There is a trend that covalent bonding aspects are most strongly pronounced in the shortest Au-P bond (2.297 Å), for which the Laplacian already turns negative and │H/ρ(r)bcp│ > │G/ρ(r)bcp│, and that ionic bonding aspects are most strongly pronounced in the longest Au-P bond (2.365 Å), for which │G/ρ(r)bcp│ > │H/ρ(r)bcp│. In the even longer Au-Cl bonds (2.392 and 2.830 Å), ionic bond contributions tend to dominate: ρ(r)bcp is as small as 0.58 or 0.25 eÅ -3 and │G/ρ(r)bcp│ >> │H/ρ(r)bcp│.
For the shorter Au-Cl bond, a tiny ELI-D basin of 0.3 e in 0.4 Å 3 is still formed (see Figure 2c), which is more polarized than the Au-P bonding basins (RJI = 82.1%), but no NCI basin is formed yet. In contrast, the longer Au-Cl bond leads to the formation of a disc-shaped and blue-colored NCI surface, which is typical for ionic bonds, as seen in Figure 2b, but no ELI-D basin is formed, supporting the spatial separation of both bonding aspects. The three Au-P bonds are about 0.8 Å longer, but still show the signatures of both bonding aspects, although much weaker: ρ(r) bcp is significantly below 1 eÅ −3 and H/ρ(r) bcp is on average −0.45 a.u., which is balanced by ∇ 2 ρ(r) bcp being close to zero and G/ρ(r) bcp being about 0.5 a.u. Just like the P-O − bond, the Au-P bonds form an ELI-D but no NCI basin (disregarding the tiny ring-shaped and red-colored surfaces perpendicular to the Au-P axes). There is a trend that covalent bonding aspects are most strongly pronounced in the shortest Au-P bond (2.297 Å), for which the Laplacian already turns negative and |H/ρ(r) bcp | > |G/ρ(r) bcp |, and that ionic bonding aspects are most strongly pronounced in the longest Au-P bond (2.365 Å), for which |G/ρ(r) bcp | > |H/ρ(r) bcp |. In the even longer Au-Cl bonds (2.392 and 2.830 Å), ionic bond contributions tend to dominate: ρ(r) bcp is as small as 0.58 or 0.25 eÅ −3 and |G/ρ(r) bcp | >> |H/ρ(r) bcp |.
For the shorter Au-Cl bond, a tiny ELI-D basin of 0.3 e in 0.4 Å 3 is still formed (see Figure 2c), which is more polarized than the Au-P bonding basins (RJI = 82.1%), but no NCI basin is formed yet. In contrast, the longer Au-Cl bond leads to the formation of a disc-shaped and blue-colored NCI surface, which is typical for ionic bonds, as seen in Figure 2b, but no ELI-D basin is formed, supporting the spatial separation of both bonding aspects. * For all bonds, d is the geometric contact distance, ρ(r) bcp is the electron density at the bond critical point (bcp), 2 ρ(r) bcp is the corresponding Laplacian, ε is the bond ellipticity, G/ρ(r) bcp and H/ρ(r) bcp are the kinetic and total energy density over ρ(r) bcp ratios, N ELI and V ELI are electron populations and volumes of related ELI-D basins, γ ELI is the ELI-D value at the attractor position, and RJI is the Raub-Jansen Index. Sorted by increasing bond distances. * For all contacts, d is the geometric contact distance, d 1 and d 2 are the distances between atom 1 or 2 and the bcp, dt is the topological bond distance (d t = d 1 + d 2 ), ρ(r) bcp is the electron density at the bcp, 2 ρ(r) bcp is the corresponding Laplacian, ε is the bond ellipticity, and G/ρ(r) bcp and H/ρ(r) bcp are the kinetic and total energy density over ρ(r) bcp ratios. Sorted by increasing contact distances. Bond ellipticities are below 0.05 for all P-O − , Au-Cl, and Au-P bonds, which is expected for all primary atom-atom contacts apart from contacts in ring or cage structures. The secondary Cl···H and Cl···C π contacts form bcps in the range from 2.48 to 3.27 Å, and follow the trend established by the primary bonds, in that they are weak and dominated by ionic bonding aspects ( Table 2). The ED at the bcp is 0.1 eÅ −3 or smaller and H/ρ(r) bcp is already positive. These non-directed interactions are characterized by curved bond paths, resulting in a topological atom-atom distance (d t ), which is 0.002-0.03 Å longer than the geometric distance as well as bond ellipticities larger than 3 in the Cl···C π case. Due to the fact that 1 is zwitterionic, formally, a positive charge is situated at the gold atom and a negative charge is located at the oxygen atom. AIM finds a charge of  (Table 3). Equally strong charge separations are observed within the organic diphenyl-acenaphthyl fragments, but the overall charges of these parts are almost zero. P atom charges typically are strongly positive and can vary largely (by about 1 e), which is less common for most other atoms, suggesting their important role in the charge-balancing processes via ligand exchange or structural modifications.
In summary, (6-Ph 2 P-Ace-5-) 2 P(O)AuCl 2 (1) is a rare example of a zwitterionic pincer compound. The series of atom-atom contacts analyzed facilitated the deconvolution between bond strength, which decreases with increasing bond distance, and bond character, which becomes more and more ionic with increasing bond distance. In this connection, the two distinctively different electronic bond properties of the Au-Cl bonds support the concept of at least partial spatial separation between covalent and non-covalent bonding aspects.

X-ray Crystallography
Intensity data of 1·3 CH 2 Cl 2 were collected on a Bruker Venture D8 diffractometer with graphite-monochromated Mo-Kα (0.7107 Å) radiation. The structure was solved by direct methods and difference Fourier synthesis with subsequent full-matrix least-squares refinements on F 2 , using all data and OLEX2 [16]. All non-hydrogen atoms were refined using anisotropic displacement parameters. Hydrogen atoms attached to carbon atoms were included in geometrically-calculated positions using a riding model. Crystal and refinement data are collected in Table 4. Figures were created using DIAMOND [17]. Crystallographic data for the structural analysis have been deposited with the Cambridge Crystallographic Data Centre, CCDC number 2005193 [18].

Computational Methodology
Density functional theory (DFT) calculations were performed for 1 in the gas phase at the B3PW91/6-311+G(2df,p) [19,20] level of theory using Gaussian09 [21]. For the Au atom, an effective core potential (ECP60MDF) and the corresponding cc-pVTZ basis set were utilized [22,23]. Only the atomic coordinates of the H atoms were optimized, whereas all other positions were kept at the XRD positions. The wavefunction files were used for a topological analysis of the electron density according to the Atoms-In-Molecules space-partitioning scheme [2] using AIM2000, [24] whereas DGRID [25] was used to generate and analyze the electron localizability indicator (ELI-D)-related [4] real-space bonding descriptors applying a grid step size of 0.05 a.u. (0.12 a.u. for visualization). The NCI [3] grids were computed with NCIplot (0.1 a.u. grids) [26]. Bond path visualizations were generated with AIM2000 and ELI-D and NCI figures were produced with MolIso [27].