Why are the {Cu4N4} rings in copper(I) phosphinimide clusters [Cu{μ-N = PR3}]4 (R = NMe3 or Ph) planar?

The copper phosphinimide complexes [Cu{μ-N[double bond, length as m-dash]PR3}]4 (1, R = NMe2 and 2, R = Ph) were obtained in good yields from the reactions of Cu[Mes] (Mes = mesityl, C6H2Me3-2,4,6) with the corresponding iminophosphoranes, HNPR3. The molecular structures of 1 and 2 reveal the presence of planar eight-membered {Cu4N4} rings which contrasts with the saddle-shaped {M4N4} rings found in related metal phosphinimide complexes. According to computations, there is negligible aromaticity in the planar {Cu4N4} rings in 1 and 2 and the saddle shape observed in related {M4N4} rings is due to steric factors.


Introduction
The significance of iminophosphoranes is well established in both organic synthesis 1 and organometallic chemistry, 2 with metal phosphinimide complexes (especially those of titanium and some rare earth elements) having been exploited in the development of highly efficient of ' nonmetallocene' based catalysts, 3 of the general form (R′3PN)2MRx and (Cp)MRx(NPR′3) (R′ = alkyl or aryl, R = alkyl). In comparison, exploitation of metal phosphinimide complexes in organic synthesis is predominantly limited to the use of lithium phosphinimide systems, which find utility in a number of areas including, as an [NH2 -] synthon, in the preparation of non-ionic phosphazene bases, in dehydrocoupling of primary and secondary phosphines, in the synthesis of primary, secondary, cyclic or functional amines, as well as in the generation of heteroatomic linkages (P-N-P, P-N-As, P-N-S). 1a, 1b The chemistry of iminophosphoranes is intrinsically associated by an isolobal, isoelectronic and isoneutral relationship with phosphorus ylides and phosphine oxides. The P=E bonding (E = CH2, NH and O) in these systems being viewed as a resonance hybrid between a double bonded neutral 'ylene' form and a zwitterionic 'ylide' form ( Figure 1). 4

Figure 1.
Given the developing utility of lithium phosphinamide complexes, it has been suggested that the preparation and development of potassium, 5 magnesium, 6 nickel, palladium and copper derivatives may lead to promising applications in organic synthesis. 1a Indeed, the novel Co(I) and Ni(I) complexes [Co(2-NP t Bu3)]4 and [Ni(2-NP t Bu3)]4 have both been reported recently, along with their use as catalysts in the mild hydrogenation of alkenes and alkynes. 7 Until now, the isolation and unambiguous characterisation of a neutral homoleptic N-Cu(I)metallated iminophosphorane complexes has not, to our knowledge, been reported, although the

Scheme 1
The reaction of [Cu(Mes)] with HNP(NMe2)3 in toluene (Scheme 1) at low temperature (-78 o C) produced an immediate reaction with the solution turning from pale yellow to colourless.
Warming of the solution to ambient temperature followed by filtration, via cannula, and cooling gave a crop of pale yellow crystals (1)

X-ray crystallography
Single-crystal X-ray diffraction studies were carried out on crystals of 1 and 2 to determine their solid-state structures. Complex 1 crystallises in the space group P21/n with the molecule sitting on a centre of symmetry such that only half of complex 1 is present in the asymmetric unit. Complex 2 crystallises in the space group P21/c and one molecule of the complex is present in the asymmetric unit cell (along with half of a disordered toluene molecule residing on a centre of crystallographic symmetry such that one toluene molecule is present for two molecules of 2). The molecular structures of complexes 1 and 2 are shown in Figure 2 and selected structural parameters listed in Table 1. (1) (2)   (15) 170.14 N(1)-Cu(2)-N(2) 176.98 (8)    It has been suggested that the steric demands of the anionic ligand play a dominant role in the solid state conformation of the cluster rather than a saddle-like geometry being indicative of strong metallophilic interactions. 19a, 19c, 19d, 24 However, the planarity of the {Cu4} rings in related clusters (and analogous Ag and Au systems) has also been attributed to a contribution from transition metal based -aromatic stabilisation resulting from a degree of cyclic electron conjugation within the cluster bonding (vide infra). 28 In the cobalt and nickel phosphinimide complexes recently reported by Stryker et al, 7

DFT Studies
In order to provide further insight as to whether the planarity of the {Cu4N4} ring present in the X-ray geometries of 1 and 2 is due to steric and/or electronic factors, density functional theory geometry optimisation calculations at the B3LYP/6-311G(d,p) level were carried out on 1 and 2.
Using the molecular geometries obtained from single crystal X-ray diffraction experiments as starting geometries, a cis, trans, cis, trans-(c,t,c,t-)orientation (conformer A, Fig 4) and planarity was retained for complex 1, but for complex 2 molecular rearrangment to a trans, trans, trans, trans-(t,t,t,t-) configuration (conformer B , Fig 4) was observed upon optimisation with an average saddle angle of 159.4º. Selected parameters, for comparison between the experimental and computed geometries, are listed in Table 1 and reveal that bond lengths are consistently longer by 0.1 Å in the computed values giving some confidence in the accuracy of B3LYP/6-311G(d,p) for copper phosphinimides. Table 2  Geometry optimisation of complex 1 starting with a t,t,t,t-conformer (B, Fig. 4) As B3LYP/6-311G (d,p) optimisations on the much more complex molecule 1 gave geometries in good agreement with experimental data (Table 1), B3LYP/6-311G (d,p) was used on simpler models with tetrahedral ring nitrogens to predict whether planar or saddled forms are in accord with experimental data. The results of Cu4(NR2)4 are summarised in Table 2 where R is H, Me and Et and the optimised molecular geometries ae shown in Figure 6.
With B3LYP/6-311G (d,p), the parent molecule Cu4(NH2)4 is planar and attempts to locate the saddled form by starting with saddled geometries all resulted in the planar form. While th is parent molecule has not been structurally determined experimentally, the methyl and ethyl analogues have been determined by X-ray crystallography. As already noted, the ethyl analogue Cu4 (NEt2)4 is saddled while the methyl analogue Cu4 (NEt2) A-C were looked at (see Figure 6). Conformer B was found to be the most stable conformer and saddled whereas the other two are planar. This suggests that the sterics of the ethyl groups are not a determining factor in this case.
Since our experimental results of 1 and 2 concern {Cu4N4} systems with three-coordinate ring nitrogens several complexes (table 1)  (av) and 150.5º (av) respectively. Their planar forms could not be located from various starting planar geometries. It seems that even the less bulky PMe3 groups are responsible for steric interactions leading to saddled {Cu4N4} rings (Fig. 7). The planar forms observed experimentally for 1 and 2 seem to occur due to favourable packing of the PR3 groups leading to planar {Cu4N4} geometries.
As noted above, there have been theoretical studies on {Cu4} ring systems that suggest aromatic stabilisation resulting from cyclic electron conjugation within the planar ring. 28 Here, the nucleus-independent chemical shift (NICS) 31 calculations were carried out as a measure of (anti)aromaticity in 1, 2 and the related {Cu4N4} systems listed in Table 2. At the B3LYP/6-311G (d,p)

Experimental Section
General Remarks: All manipulations were carried out under an atmosphere of dry dinitrogen or argon using standard Schlenk and glove-box techniques. Toluene and hexane were dried using an

Single Crystal X-ray Crystallography
Experimental details relating to the single-crystal X-ray crystallographic studies are summarised in Table 2. For all structures, data were collected on a Nonius Kappa CCD diffractometer at 150(2) K using Mo-K radiation ( = 0.71073 Å). Structure solution and refinements were performed using SHELX86 35 and SHELX97 36 software, respectively.
Corrections for absorption were made in all cases. Data were processed using the Nonius Software. 37 Structure solution, 38 followed by full-matrix least squares refinement 36b was performed using the WINGX-1.80 suite of programs throughout. 39 For all complexes, hydrogen atoms were included at calculated positions. Crystals of the complex 2 were both small and weakly diffracting, with intensity loss at higher 2-theta angle. Hence a data completeness of > 93.5 % (max 2 = 25.0 °) could not be met. CCDC reference numbers 955629-955630.

Computational Studies
DFT-Calculational studies were carried out using the Gaussian09 package. 40 All starting geometries of 1, 2 and related systems were optimised without symmetry constraints at B3LYP/6-311G(d,p) level of theory. 41 No imaginary frequencies were found from frequency calculations on these optimised geometries and indicate that the geometries are true minima. Symmetry constraints were however applied to conformers C (C4v) and D (C4h) of Cu4(NPH3)4. NICS values were obtained from dummy atoms placed in the centre of the {Cu4} rings using the GIAO 42 -NMR method at B3LYP/6-311G (d,p). Calculated 31 P GIAO-NMR chemical shifts were obtained using the δ( 31 P) = 310.0 -σ( 31 P) scale while the 13 C shifts were calculated using the δ( 13 C) = 182.5σ( 31 C) scale.