The partial dehydrogenation of aluminium dihydrides

The reactions of a series of β-diketiminate stabilised aluminium dihydrides with ruthenium bis(phosphine), palladium bis(phosphine) and palladium cyclopentadienyl complexes is reported.


S-2
then allowed to warm to room temperature and H2 gas evolution was observed along with the formation of a colourless precipitate. After stirring overnight, the mixture was concentrated to dryness. Addition of diethyl ether (60 mL) resulted in a yellow solution with a colourless precipitate. After filtering the mixture via a cannula, the clear yellow filtrate was concentrated to half the volume and stored at -21 o C overnight. The product crystallised and filtration and drying in vacuo gave the product 1e as yellow crystals (0.55 g, 1.98 mmol, 25%).
Spectroscopic data for 1e matched that previously reported. 6
Sparing solubility of the isolated crystalline material of 5 in common NMR spectroscopy solvents (benzene, toluene) prohibited full analysis by solution phase NMR spectroscopy of purified 5.
Single crystal X-ray and neutron diffraction were used to characterise 7c in the solid-state.

Reaction of [Pd(PCy3)2] with 1e: Identified Decomposition pathway
The reaction of [Pd(PCy3)2] with 1e produced a mixture of species. The hypothetical complex 7e (equivalent to the isolated 7c) could not be experimentally isolated. However, a related decomposition product was isolated in very low yield as a few single crystals. It was not possible to obtain bulk characterisation data for this compound but the X-ray crystal structure of this product (S1) showed it to have a similar Pd2Al2 motif (the number and location of metal bound hydride ligands could not be unambiguously identified in this structure due to poor Xray data quality, figure S2.2). As in 7c, each Pd is ligated by one PCy3 but a reaction has occurred in one of the Al-bound β-diketaminate ligands making this structure different to 7c.
The structure shows C3 has become a tetrahedral sp 3 -centre through reduction of the βdiketaminate by hydride transfer (figure S2.1). This decomposition pathway was not observed for the other palladium complexes of aluminium hydrides 1a-c which we ascribe to the bulky ortho substituents on the aryl groups sterically blocking hydride transfer processes.

Figure S2.2:
The molecule of S1 (50% probability ellipsoids) modelled from single-crystal Xray diffraction. Hydrogen atoms and hexane solvent molecules omitted for clarity. Metal bound hydrides were not located in X-ray structure.

S-10
3. X-ray Crystallographic Data   Table S1 provides a summary of the crystallographic data for the structures of 3, 5, 6, 7c. Data were collected using Agilent Xcalibur 3 E (3, 6, 7c) and Xcalibur PX Ultra A (5) diffractometers, and the structures were refined using the SHELXTL and SHELX-2014 program systems. 7

S-12
3.2 X-ray crystal structures 3.2.1The X-ray crystal structure of 3 The two bridging Ru-H-Al and two terminal Ru-H hydrogen atoms in the structure of 3 were all located from ΔF maps and refined freely.

The X-ray crystal structure of 5
The molecule of 5 shows crystallographic Ci symmetry. The bridging Al-H-Pd hydrogen atoms in the structure of 5 were located from a ΔF map and refined freely.

The X-ray crystal structure of 6
The molecule of 6 shows crystallographic Ci symmetry. The bridging Al-H-Pd and hydrogen atoms in the structure of 6 were located from a ΔF map and refined freely. Figure S3.3: The molecule of 6 (50% probability ellipsoids). Selected hydrogen atoms omitted for clarity.

The X-ray crystal structure of 7c (form 1)
The molecule of 7c (form 1) was found to sit across a centre of symmetry at the middle of the Al2Pd2 ring. The unique bridging Al-H-Pd hydrogen atom was located from a ΔF map and refined freely. The included benzene solvent molecule was found to be disordered. Two orientations were identified of ca. 79 and 21% occupancy, their geometries were optimised, the displacement parameters of adjacent atoms were restrained to be similar, and only the non-hydrogen atoms of the major occupancy orientation were refined anisotropically (those of the minor occupancy orientation were refined isotropically).

The X-ray crystal structure of 7c (form 2)
The structure of 7c (form 2) contains 2 half molecules each sitting across a centre of symmetry at the middle of the Al2Pd2 ring. The unique bridging Al-H-Pd hydrogen atoms were located from a ΔF map and refined freely. One cyclohexyl ring was found to be disordered over 2 positions with the occupancies freely refined to ca. 70 and 30%. The C-C bonds in this ring were restrained to 1.54 Å using the DFIX command and the ellipsoids restrained by the ISOR command. Figure S3.5: One molecule of 7c (form 2) (50% probability ellipsoids). Selected hydrogen atoms and second molecule of asymmetric unit omitted for clarity.

Neutron Diffraction Study for 7c-d2
The Laue single-crystal neutron diffraction study reported here as each undertaken using the KOALA instrument standing at the end guide position of TG3, a thermal neutron beam produced from the OPAL reactor at the Australian Nuclear Science and Technology

S-18
The neutron diffraction data obtained for 7c-d2 has been modelled using the CRYSTALS 9 software package. Atoms of 7c-d2 are modelled with positional parameters included for all atoms. Anisotropic displacement parameters are refined for all non-hydrogen atoms including the deuteriums while the hydrogens are modelled with individual isotropic displacement parameters. Some limited evidence of disorder is apparent for one cyclohexyl group but this remains modelled by anisotropic displacement parameters as no successful alternative model was resolved. Data were supplemented by appropriate distance, angle and displacement parameter restraints to yield a final model refined against F with R = 0.090 Rw = 0.122 S = 1.23, minimum and maximum residual nuclear densities were -1.29 and 1.74 fermi Å 3 Full details of the experiment, data reduction and refinement are contained in the supplementary CIF file for this paper.   VT NMR data was fitted using line shape analysis with the DNMR programme integrated into Topspin v3.1. The 1 H and 31 P resonances of 2 were fitted over the 193 to 253 K range with an initial line broadening factor of 20 Hz for 31 P NMR and 2 Hz for 1 H. Fits for k were optimized to the experimental data with reasonable accuracy and the modelled data are presented below, a minor unassigned species observable was not included in the model. The activation parameters for the exchange process are as follows: 31 P DH ‡ = 10.8 ± 1.1 kcal mol -1 , DS ‡ = +6.6 ± 5.2 cal K -1 mol -1 , DG ‡ 298 K = 8.8 ± 2.7 kcal mol -1 .

S-24
The low temperature NMR spectroscopy data agreed with the assignment of the structure of 2 whereby the PCy3 ligands were in a cis-arrangement around the ruthenium, one in the axial and one in the equatorial position to give magnetically inequivalent phosphines. The positive entropy of activation (with the caveat of the error associated with the Eyring analysis) potentially suggests that the mechanism of exchange was through a dissociative process whereby the heterobimetallic complex could break apart to give just the aluminium and ruthenium fragments to allow for the rearrangement of the ligands around the ruthenium centre via C3 rotation. No coupling between the hydrides and phosphines were observed due to the proximity of the hydrides to the quadrupolar aluminium nuclei and the facile exchange process of the hydrides within the NMR spectroscopy acquisition time.  In the 31 P{ 1 H} VT data, in the low temperature regime (193 K) a broadened peak assigned to the proton coupled phosphine environment of 4 can be observed at δ = 37.0 ppm alongside

S-27
[Pd(PCy3)3] at δ = 23.5 ppm and free PCy3 at δ = 8.4 ppm. Comparison of the 31 P and 31 P{ 1 H} NMR spectra confirms the 2 JP-H coupling and supports the assignment ( Figure 5.13). The additional resonances have not been assigned. Upon warming 4 is consumed and the spectra is reminiscent of previously reported spectra for the room temperature reaction of [Pd (PCy3)2] with 1a which forms 7a as an insoluble red precipitate.   S-31

DFT calculations on the formation of 4
In order to gain further support to the formation of the σ-complex 4, the thermodynamics of these complexations were explored by DFT calculations (for methods see section 7), and their formation is energetically favourable. The proposed three-coordinate complex 4 with two phosphine ligands is more stable than its two-coordinate analogue (figure S5.16), although dissociation of one PCy3 is energetically feasible.

S-32
The NPA charges are very similar to complex 5 (vide infra), which supports our formulation of complexes 4 and 5 as σ-complexes of Pd(0) and Al(III).  In the WBI analysis it becomes apparent that the Al-H bond that coordinates to Pd is weakened and this is nicely illustrated in the Second Order Perturbation Theory Analysis. A significant donation from the bonding σ-(Al-H) orbital to an empty s orbital in Pd is present.

WBI Pd-Al
Also, a non-negligible back-donation interaction from a d orbital in Pd to the anti-bonding σ*-(Al-H) can be found. Based on the Dewar-Chatt-Duncanson model, this complex is thus a textbook example of a σ-complex.
QTAIM is also analogous to that of complex 5 (vide infra).

Computational details 7.1 Methods
The geometries of products were optimised with the M06L Minnesota DFT functional using the Gaussian09 and Gaussian16 program packages. 13 Stationary points were characterised depending on their imaginary frequencies (0 for minima). NBO analysis was performed using the NBO 6.0 version program. 14 QTAIM analysis was conducted with the AIMAll package. 15 Non-covalent interactions were analysed with the NCIPLOT 3.0 program. 16 Dispersion effects were included via single point energy corrections and were modelled using Grimme's D3 correction for M06L (EmpiricalDispersion=GD3). 17 The default numerical integration grid was also improved using a pruned grid with 99 radial shells and 590 angular points per shell (int=ultrafine). Solvent effects were not included since the reactions are carried out in rather nonpolar solvents. The level of theory employed in this study (M06L/BS1) was previously benchmarked by our group and it was shown to correlate accurately with experimental results. 18 The basis set employed (BS1) was built as follows. The SDD effective core potential was used for all metals (SDDAll). The split-valence 6-31G* basis set was used for C and H atoms. The

Bonding analysis
NBO analysis was performed on the optimised structures of a series of intermetallic complexes. For the octametallic complex 6, an optimised structure could not be obtained, due to unsurmountable convergence problems. This complex presented severe difficulties to converge the SCF equations. These problems could only be circumvented by using a quadratically convergent SCF procedure (scf=qc, or scf=xqc in G16). This approach increased dramatically the computational cost, making an optimisation inviable and hence only a single point calculation could be obtained. For this reason, the NBO analysis for 6 was performed using the geometry obtained from X-ray diffraction. In order to validate this approximation, the NBO analysis for the rest of complexes was performed on both the experimental and optimised geometries. The differences were small and qualitatively irrelevant.

S-44
As expected, the charges for the intermetallic complexes are inside the range comprised by the parent Al(III) and Al(I) complexes. This is consistent with the proposed picture of these complexes as snapshots of the dehydrogenation process from Al(III) to Al(I). The effect of reducing the sterics of the b-diketiminate ligands was found to be very small, although it appears that the charges are marginally reduced when the ligand becomes smaller.