Zn···Zn interactions at nickel and palladium centers

Zinc–zinc interactions on nickel and palladium centers are highly dependent on the co-ligands. These dependencies are also found for the formation of dihydrogen vs. dihydride complexes and underline the analogy [Zn2Cp*2] ↔ H2.


Introduction
Decamethyldizincocene, [Zn 2 Cp* 2 ], a metalla-analogue to the dihydrogen molecule H 2 , was reported in 2004 by Carmona et al. as the rst molecular compound with a covalent Zn I -Zn I s-bond, a bond that was rumored to be non-existent. 1 [Zn 2 Cp* 2 ] reveals a rich coordination chemistry. Reactions of [Zn 2 Cp* 2 ] with transition metal complexes containing labile ligands L are typically based on an initial homolytic cleavage of the Zn-Zn bond, followed by ligand exchange, resulting in the coordination of ZnCp* fragments. [2][3][4] As ZnCp* is isolobal to H, the most common reaction pattern of [Zn 2 Cp* 2 ] towards transition metal centers L a M can be well compared to the "oxidative addition" reaction of H 2 to electron-rich unsaturated transition metal fragments, forming dihydride complexes [L a M(H) 2 ]. Very recently we observed a different coordination mode of [Zn 2 Cp* 2 ] in the two triangular compounds [Zn 3 Cp* 3 ] + and [Zn 2 CuCp* 3 ]. Both species are described as metalla-analogues of the saromatic [H 3 ] + ion, again emphasizing the isolobal analogy between the fragment Cp*Znc and the hydrogen atom Hc. 5 In particular, we came across this conceptual analogy in 2008 with the discovery of [Mo(ZnMe) 9 (ZnCp*) 3 ]. 6 This unique icosahedral coordinated complex can be regarded as a stable and accessible analogue of the matrix isolated polyhydride complex [WH 12 ], a species which, differently from the [MoZn 12 ] analogue, could better be written as [WH 6 (H 2 ) 3 ]. 7 The unique formation reaction of [Mo(ZnMe) 9 (ZnCp*) 3 ] is based on the chemistry of homoleptic compounds [M(GaCp*) n/2 ] (n > 8; M ¼ Mo, Ni, Pd, Pt, Cp* ¼ pentamethylcyclopentadienyl) towards ZnMe 2 or ZnEt 2 , which leads to pseudo homoleptic all organo-zinc coordinated products with the general formula [M(ZnR) n ] (n $ 8; M ¼ Mo, Ru, Rh, Ni, Pd, Pt; R ¼ Cp*, Et, Me). 8 This concept is transferable to heteroleptic starting materials [L m M(GaCp*) n/2 ] (L ¼ Cp*, CO, X, PMe 3 .), where the ligands L have been able to control the reaction selectivity and product formation. [9][10][11][12] A reaction pattern dependent on the co-ligands of a metal complex and thus the electronic properties of a system is also reported for the coordination of H 2 to transition metal fragments. Not only are simple oxidative addition reactions, resulting in dihydride complexes known, the "side-on" coordination of H 2 under the preservation of the H-H bond has also been widely studied and reported, especially by Kubas et al. [13][14][15] The triangular clusters [Zn 3 Cp* 3 ] + and [Zn 2 CuCp* 3 ] can be regarded as the rst examples with an H 2 analogous "side-on" coordination mode of [Zn 2 Cp* 2 ] at the two isoelectronic, unsaturated fragments [ZnCp*] + and CuCp*. 5 With this background in mind the question arises, if a series of complexes featuring the Zn 2 M structural motive can be prepared exhibiting a (more or less) intact Zn-Zn interaction, i.e. dizinc complexes which are analogous to non-classical dihydrogen complexes of the Kubas type.

Results and discussion
In this contribution, phosphine and isonitrile ligated heteronuclear Ni/Zn and Pd/Zn complexes with different L/Zn ligand ratios are described (Scheme 1  4 (CN t Bu) 2 ] (4) and [Pd 3 Zn 6 -(PCy 3 ) 2 (Cp*) 4 ] (5). All new compounds 1-5 were characterized by NMR and IR spectroscopy, Liquid Injection Field Desorption Ionization (LIFDI) MS analysis, single crystal X-ray diffraction and elemental analysis. The electronic structures of 1 and 2 were investigated by quantum chemical calculations at the DFT level of theory.

Synthesis and spectroscopic characterization of 1 and 2
The treatment of the heteroleptic Ni/(Al,Ga) starting compounds [Ni(ECp*) a (PMe 3 ) 4Àa ] (a ¼ 1, 2; E ¼ Ga, Al) with exact stoichiometric amounts ZnMe 2 (1.2 M solution) in toluene as the solvent leads to [Ni(ZnCp*)(ZnMe)(PMe 3 ) 3 ] (1) and [Ni(ZnCp*) 2 (ZnMe) 2 (PMe 3 ) 2 ] (2), independent of the element E used in the starting complexes. Both compounds were isolated in good yields of 67% (1) and 84% (2). E III containing by-products, Cp*EMe 2 species, were observed. They are well soluble in organic solvents like benzene, toluene, or n-hexane and are stable for several weeks in the pure crystalline form when stored under argon at À30 C. Yellow crystals of both compounds suitable for single crystal X-ray analysis were obtained from saturated hexane or toluene solutions, respectively. The elemental compositions were determined by combustion analysis (C, H) and atomic absorption spectroscopy (Zn), respectively. The empirical formulas were conrmed by mass spectrometric analyses using LIFDI, which show exclusively the presence of the molecular ion peaks at 568.09 m/z for 1 and 771.94 m/z for 2. The ne structure of the signals also match very well the calculated isotopic patterns. 1 H NMR studies are also in agreement with the determined molecular structures in the solid state (see the Experimental section for details on 31 P and 13 C NMR studies). In contrast, analogous reactions using the sterically more bulky phosphine ligands PPh 3 or PCy 3 lead to a mixture of homoleptic products [Ni(ZnCp*) 4 (ZnMe) 4 ] and [Ni(PR 3 ) 4 ] (R ¼ Ph, Cy) in non-stoichiometric reactions. The observed facial coordination of three PR 3 and two ZnR ligands in penta-coordinated 1 and two cis-PR 3 and four ZnR in hexa-coordinated 2 is obviously favored only for the less bulky ligands PR 3 . It should be noted that the missing member of the employed series of starting compounds, [Ni(GaCp*) 3 (PR 3 )], is only isolable for R ¼ Cy; not for R ¼ Me. 16 Thus, the monophosphine-substituted, heptacoordinated complex [Ni(ZnCp*) 3 (ZnMe) 3 (PMe 3 )] remains unknown so far, pointing to very delicate, small kinetic effects related to steric overcrowding. For the larger central metal atom Pd, the all-zinc ligated hepta-coordinated [Pd(ZnCp*) 4 (ZnMe) 2 {Zn(tmeda)}] 17 (tmeda ¼ N,N,N 0 ,N 0 -tetramethyl-ethane-1,2-diamine) is known, however. Here, six Zn atoms (acting as single electron ligands similar to ZnR in 1 and 2) are arranged in an ideal trigonal dodecahedron way, where the {Zn(tmeda)} moiety is located almost exactly between two "missing" vertices of the dodecahedron (S Q (P) ¼ 0.07, see ESI † for details of CShM). Conceptually, the {Zn(tmeda)} unit can be viewed as a two electron donor ligand with a steric bulk at least comparable to phosphine PR 3 or ECp*. Therefore, the existing [Pd(ZnCp*) 4

Molecular structures of 1 and 2 in the solid state
Nickel adopts penta-and hexa-coordination in 1 and 2, respectively, with different ratios of ZnR (R ¼ Cp*, Me) and phosphine ligands according to Fig. 1. In both cases phosphine and organo-zinc ligands are arranged fac (1) and cis (2) and the structures are best described as composed of a tetrahedron (NiP 4 ), in which one and two vertices are substituted by Zn 2 units, respectively. For 2 this results in a polyhedron which consists of half of a tetrahedron (NiP 2 ) and half of a dodecahedron (NiZn 4 ) (see ESI † for CShM values). All Ni-Zn and Ni-P distances are comparable to the distances found in related compounds, e.g. [Ni(ZnCp*) 4 (ZnMe) 4 ] (2.313(1)-2.371(1)Å for Ni-Zn) or [Ni(GaCp*) 2 (PMe 3 ) 2 ] (Ni-P: 2.139(1) A). 8 The P-Ni-P angles all lie between 106.12(2) and 121.10(3) , which is rather close to the ideal tetrahedral angle of 109.5 .
From a conceptual point of view, the substitution of PMe 3 by tert-butylisonitrile in 2 triggers elimination of Cp*ZnZnMe similar to a classic reductive elimination reaction, forming the (hypothetical) 16 valence electron (ve) fragment [Ni(CN t Bu) 2 -(ZnMe)(ZnCp*)], which subsequently dimerizes to yield 3. The heteroleptic organozinc(I) species Cp*ZnZnMe is thermally very unstable and thus spectroscopically not observable, and presumably disproportionates to yield the observed products, elemental Zn and Cp*ZnMe. Attempts to trap the monomeric 16ve species [Ni(CN t Bu) 2 (ZnMe)(ZnCp*)] by addition of excess PPh 3 failed. As expected, 1 can be used as a starting material for the formation of 3 too, and with PMe 3 as the only observable byproduct. The composition of 3 has been conrmed by elemental analysis and LIFDI-MS which shows the molecular ion peak at 1009.7 m/z exclusively (calcd 1010.2 m/z).

Molecular structure of 3 in the solid state
The core moiety of [{Ni(CN t Bu) 2 (m 2 -ZnCp*)(m 2 -ZnMe)} 2 ] (3) can be described as a compressed Ni 2 Zn 4 octahedron, where the Ni atoms are located trans along the short axis (Fig. 1). Both Ni sites are additionally coordinated by two tert-butylisonitrile ligands with a coplanar arrangement of the two Ni centers, Zn2, Zn2 0 and all four CN moieties. Interestingly, despite a different ligand environment and a different cluster valence electron count (cve), the Ni 2 Zn 4 core structure is very similar to that of the related complex [Ni 2 Zn 4 Cp 6 ]. 18 3) , as well as C-N-C angles of 158.0(4) and 168.5(4) . The deviation from linearity can be explained by the electron-rich situation at the nickel centers leading to strong p-back-bonding. Also weak interactions of the CN groups with the adjacent ZnMe ligands (C-C distances of 2.554(3) and 2.578(3)Å) cannot be excluded.

Synthesis, spectroscopic and structural characterization of 4
Treatment of [{Pd(CN t Bu) 2 } 3 ] with [Zn 2 Cp* 2 ] in n-hexane at room temperature leads to the formation of [Pd(CN t Bu) 2 -(ZnCp*) 4 ] (4) as an orange microcrystalline solid. Recrystallization from toluene at À30 C leads to yellow, cubic crystals suitable for single crystal X-ray diffraction. The empirical formula of 4 was derived from elemental analysis (C, H) and atomic absorption spectroscopy (Zn), respectively, and is consistent with the spectroscopic and structural data. The 1 H NMR of compound 4 exhibits one signal for the tert-butylisonitrile methyl groups (d ¼ 1.18, s, 18H) as well as two signals for chemically inequivalent Cp* groups (d ¼ 2.10, s, 30H; d ¼ 2.27, s, 30H), which points to cis-coordination of the t BuNC groups in the hexa-coordinate complex, matching with X-ray structural data (Fig. 1, below). The 13 C NMR is also consistent with the suggested structure. The C-N and C^N vibration bands in the FTIR spectrum can be observed at wavenumbers of 1188 and 2092 cm À1 . Compound 4 crystallizes in the orthorhombic space group Pbca with Z ¼ 4 with two independent molecules in the asymmetric unit. As both molecules are virtually the same in terms of their bond length and angles only one of them is discussed here (Fig. 1, below). The six ligands are coordinated in a strongly distorted octahedral arrangement to the palladium center. The Pd-Zn distances are all between 2.428(1)-2.484(1)Å and well comparable with other Pd-ZnCp* units known in literature. 2,8 Similarly the Pd-CN t Bu bond lengths are all in a similar range to other palladium-isonitrile complexes. 9 2.6 Synthesis, spectroscopic and structural characterization of 5 Treatment of [Pd(PCy 3 ) 2 ] with two molar equivalents of [Zn 2 Cp* 2 ] in 5 mL toluene leads to a red solution which aer heating to 80 C for one hour and standard workup gives [Pd(PdPCy 3 ) 2 (Zn)(m-Zn 2 Cp*)(m-ZnCp*) 3 ] (5) as dark red/black cubic single crystals. The elemental analysis (C, H) and atomic absorption spectroscopy (Zn) data, respectively, are consistent with the composition derived from single crystal X-ray diffraction studies (Fig. 1). Also, the 1 H, 13 C and 31 P NMR spectra exhibit the expected signals, matching the molecular structure determined in the solid state (vide infra). The FTIR spectrum of 5 shows absorption bands for the C-H valence vibrations of the Cp* groups (n ¼ 2898 and 2827 cm À1 ) as well as a p-cylcohexyl vibration (n ¼ 1433 cm À1 ). Compound 5 crystallizes in the triclinic space group P 1. The core of complex 5 is a [Pd 3 Zn 6 ] unit, in which the three palladium atoms are adopting a bent structure with a Pd1-Pd3-Pd2 angle of 115.22 (2) . The Pd-Pd distances are 2.669 (1) (1)). The Cp* ring of the {Zn 2 Cp*} unit is disordered and thus binds asymmetrically to Zn4 and Zn5, the Zn-C bond distances suggesting h 1 -h 3 -binding modes for the two identical isomers. The Zn4-Zn5 distance (2.729(1)Å) in this unit is comparable to the Zn-Zn distance found in compound 2 which again may indicate weak Zn/Zn interactions. With the exception of Zn4-Zn5, all Zn-Zn distances are longer than 2.83Å and thus outside the range of bonding Zn-Zn interactions. The last Zn atom (Zn6) is not coordinated by an organic ligand and is found in a position with an almost equal distance to all three palladium atoms (Pd1-Zn6 2.512(1), Pd2-Zn6 2.504(1), Pd3-Zn6 2.494(1)). All Pd-Zn distances (2.471(1)-2.563(1)Å) are elongated as compared to the terminally  8 but well in the range of other ZnR ligands found in bridging positions. 10 The central atom (Pd3) is located in the center of a pseudo-hexagon with one vacant vertex, consisting of alternating zinc and palladium atoms. Thus, Zn1 (0.186Å), Zn3 (0.336Å) and Zn6 (0.093Å) as well as all three palladium atoms are almost perfectly coplanar with deviations from the plane of 0.186Å (Zn1), 0.336Å (Zn3) and 0.093Å (Zn6). The bond angles, however, strongly deviate from those of a perfect hexagon.

Structural comparisons
Our discussion whether two adjacent Zn atoms in compounds 1-5 show a signicant Zn-Zn interaction or not is based in the rst place on the evaluation of Zn-Zn distances as extracted from X-ray single crystal structure datacomputational studies will complement the picture (vide infra). The distance of two Zn atoms in the (distorted) hexagonal closest packed structure of metallic zinc (Table 2) . In both clusters, the three metal atoms are held together by two delocalized electrons. Theoretical investigation of the bonding situations in these molecules hinted at s-aromaticity. However, in an alternative point of view the triangular structures may be also regarded as coordination compounds of [Zn 2 Cp* 2 ] to the fragments [ZnCp*] + or [CuCp*], respectively. In a way, this situation is similar to the coordination of H 2 to transition metal fragments: unsaturated electron-rich metal fragments usually interact with the H 2 molecule giving a dihydride species by oxidative cleavage of the hydrogen-hydrogen bond ("classical dihydride complexes"), while electron-poor unsaturated metal fragments lead to the formation of H 2 -adduct-complexes, without cleavage of the hydrogen-hydrogen bond ("non-classical" or "Kubas-type dihydrogen complexes"). 14 However, while the oxidative addition of H 2 includes oxidation of the transition metal center, the addition of [Zn 2 Cp* 2 ] to metal fragments proceeds without formal change of oxidation state. The distinction between pure Zn 2 R 2 coordination without Zn-Zn-bond cleavage and products with distinct ZnR ligands exhibiting no Zn-Zn interaction is not solely based the Zn-Zn distances as discussed above, but also other structural parameters can be used: The Zn-Zn-R as well as M-Zn-R bond angles for instance are important indicators for the presence of Zn-Zn interactions. While a strictly side-on coordinated Zn 2 R 2 should ideally exhibit a linear R-Zn-Zn-R geometry (not regarding steric effects in the rst place), the Zn-Zn-R bond angles gradually decrease towards 120 for perfectly symmetric MZn 2 R 3 complexes with a weaker Zn-Zn interaction. At the same time, the M-Zn-R bond angles increase towards 180 (Scheme 2).
With this background in mind, it is worth to have a closer look at the Zn-Zn distances measured for compounds 1-4. Most signicant is the short distance between the Zn atoms in 1 (2.525(1)Å), which is 13% shorter as compared to the higher coordinated [Ni(ZnCp*) 4 (ZnMe) 4 ] (2.746(1) to 2.912(1)Å). For the latter complex very weak Zn/Zn interactions were proposed, however, with no direct Zn-Zn bond paths. The covalent Zn-Zn single bond length in [Zn 2 Cp* 2 ] of 2.305(3)Å, which may serve as a reference for a strong Zn-Zn s-bond, is only 9% shorter than the one observed in 1. Obviously, the Zn-Zn distance of compound 1 lies almost exactly in between a weak tangential Zn-Zn interaction in the cluster-like compound [Ni(ZnCp*) 4 (ZnMe) 4 ] and the classical, unsupported s-bond in [Zn 2 Cp* 2 ] (see Table 2). Similar to 1, the shortest Zn-Zn distances in 4 (2.595(2)Å for Zn1-Zn2 and 2.609(2)Å for Zn3-Zn4) are only 12-13% longer than the covalent Zn-Zn interaction in [Zn 2 Cp* 2 ]. 1 All other Zn-Zn distances in 4 are much longer (3.093-3.753Å) and outside the range of weak Zn-Zn interactions. The Zn/Zn contacts in 1 and 4 are shorter than in metallic zinc (2.664Å for the closest Zn-Zn distances). 22 In contrast to 1 and 4 the shortest Zn-Zn distances in complexes 2 (2.718(1)Å) and 3 (2.817(1)Å) are outside the expected range for bonding Zn-Zn interactions and similar to those found in [Ni(ZnCp*) 4 (ZnMe) 4 ]. Thus, based on the above comparisons, the Zn/Zn interactions follow the trend 1 z 4 < 2 < 3 z [Ni(ZnCp*) 4 (ZnMe) 4 ]. The assumption of two signicant Zn/ Zn interactions in 1 and 4 is further supported by the respective bond angles: The M-Zn-Cp* centroid angle for 1 (175.08 ) deviates only slightly from linearity, while those of 4 (144. 22-150.17 ) are indeed closer to 150 , as expected for a perfectly symmetric M 3 triangle. 5 While in the case of 4 steric repulsion of the Cp* rings, which might inuence the M-Zn-Cp*centroid angle, cannot be excluded, no steric Cp*/Cp* repulsion is present in 1 (see Fig. S11 and S12 † for the depiction of space lling models of 1 and 4). As consequence of the bent structure and the involved steric repulsion, the Cp* rings of 4 are not h 5 coordinated to the Zn centers but are closer to h 2 (Zn2), h 3 (Zn1, Zn3) and h 4 (Zn4) bonding modes. Furthermore, the Zn1-Pd-Zn2 (63.76 (3) ) and Zn3-Pd-Zn4 (64.24 (3) ) angles of 4 are also clearly closer to a triangular geometry (60 ) than an octahedral geometry (90 ) (Fig. 2).

Quantum chemical investigations for compounds 1 and 2
In order to answer the question if there are signicant differences between the metal-ligand interactions of the ZnR ligands in 1 and 2, we rst optimized the geometries of the complexes at the BP86/TZVPP level. Aer that, QTAIM calculations, natural bond orbitals (NBO) and energy decomposition analyses with the "Natural Orbitals for Chemical Valence" extension (EDA-NOCV) were performed to get insight into the bonding situation of the adducts. The optimized structures of 1 and 2 ( Fig. S13 and S14 †) are in good agreement with the experimental data given by single crystal X-ray diffraction measurements. The calculated bond lengths are generally about 0.040Å too large compared to experiment, with the exception being the Ni-ZnCp* bonds where the deviation is about 0.080Å. The differences can be explained with packing effects in the crystal which are absent in the gas phase. The topological QTAIM analysis of 1 shows a bond path between the zinc atoms (Fig. 3, top) while there are no Zn-Zn bond paths in 2 (Fig. 3, bottom). This suggests a Zn-Zn bond in 1 which is absent in 2. Although it is possible that there are interactions between atoms without bond paths according to the QTAIM, 30   a 1.2 M solution in toluene) was added to the yellow suspension and stirred for 10 min. Aer 5 min the suspension turned into an orange solution during heating to RT and in the meantime the solvent was reduced in vacuo. The residue was dissolved in a small amount of toluene and single crystals were obtained at À30 C within a few days. Yield: 67% (80 mg, 0.14 mmol).
[Ni(PMe 3 ) 3 (AlCp*)] can also be used as the starting material giving 1 in a poorer yield of 62%. 1  Method 1: 2 (200 mg, 0.26 mmol) was dissolved in toluene (5 mL) and an excess of CNt-Bu (10 eq. 0.28 mL, 2.60 mmol) was added. The reaction solution was heated to 100 C overnight whereupon metal precipitation occurred. All volatile materials were evaporated in vacuo and the residue was extracted from toluene (2 Â 3 mL) giving a clear orange solution, which gave aer evaporation of all volatile materials an orange powder in a yield of 90% (116 mg, 0.12 mmol). Single crystals can be obtained from a saturated solution in toluene overnight when stored at À30 C. Method 2: 1 (100 mg, 0.18 mmol) can also be used as the starting material, giving under the same reaction conditions no metal precipitation 82 mg (0.08 mmol) of an orange power in a yield of 89%. 1