Diverse reactivity of an iron–aluminium complex with substituted pyridines

The reaction of an Fe–Al complex with an array of substituted pyridines is reported. Depending on the substitution pattern of the substrate site-selective sp2 or sp3 C–H bond activation is observed. A series of reaction products are observed based on (i) C–Al bond formation, (ii) C–C bond formation by nucleophilic addition or (iii) deprotonation of the β-diketiminate ligand. A divergent set of mechanisms involving a common intermediate is proposed.


General Experimental
All manipulations were carried out using standard Schlenk-line and glovebox techniques under an inert atmosphere of argon or dinitrogen. A MBraun Labmaster glovebox was employed, operating at <0.1 ppm O2 and <0.1 ppm H2O. Solvents were dried over activated alumina from a SPS (solvent purification system) based upon the Grubbs design and degassed before use. Glassware was dried for 12 h at 120°C prior to use. C6D6 was dried over 3 Å molecular sieves and freeze-pump-thaw degassed thrice before use. Chemicals were purchased from Sigma Aldrich, Fluorochem, Alfa Aesar, and VWR. Pyridine substrates were dried over CaH2, distilled, and stored over activated 3 Å molecular sieves. 1 and 2 were prepared as reported previously by our group. 1   S-3

Synthesis of 3:
In a glovebox, 2 (15 mg, 0.020 mmol) was dissolved in C6D6 (0.6 mL) and transferred to a J-Young NMR tube.

S-4
Synthesis of 4 and 5: In a glovebox, a stock solution of 3,4-dimethylpyridine in C6D6 (0.1 M, 230 μL, 0.023 mmol) was added to a solution of 1 (15 mg, 0.023 mmol) in C6D6 (0.6 mL) and transferred to a J. Young NMR tube. Within 30 min, the colour of the reaction solution changed from red-orange to yellow. NMR analysis of the reaction solution revealed formation of 4 and 5 in a 4:1 ratio (>95 % NMR yield based on the relative integrals of in the 31 P NMR spectrum). The reaction solution was transferred back to a glovebox and the solvent was removed under reduced pressure. The remaining sticky solid was triturated in n-pentane and the solvent again removed under reduced pressure to afford an orange solid.

Synthesis of 8:
In a glovebox, 7 (20 mg, 0.031 mmol) was dissolved in C6D6 (0.6 mL) and transferred to a J-Young NMR tube.

X-Ray Data
Table S1 provides a summary of the crystallographic data for the structures of 6, 7, 8, 9, 10 and S1. Data were collected using an Agilent Xcalibur PX Ultra A diffractometer, and the structures were solved and refined using the OLEX2, 2 SHELXTL 3 and SHELX-2013 4 program systems. CCDC 2195829 to 2195834. [c] The complex has crystallographic CS symmetry.

X-ray crystallography
The three Al-H-Fe bridging hydrogen atoms in the structures of 6, 7, 9, 10 and S1, and the two unique Al-H-Fe bridging hydrogen atoms in the structure of 8 were all located from ΔF maps and refined freely.
The crystal of 8 that was studied was a noticeably weak scatter of X-rays, and so the data collection was designed with low target intensities for both the low and high angle images in order to give a reasonable data collection time. Similarly, the data collection was also designed to only collect unique data, but unfortunately, the close similarity of the a and c axis lengths [16.6412(4) and 16.6540(8) Å respectively, though they appeared closer at the time the experiment was being designed] led to the presumption that the correct S-13 crystal system was tetragonal, and so 4/m data was collected. Sometime after the crystal had already been removed from the diffractometer (and thus decomposed), but before the structure had been successfully solved, it was discovered that the correct crystal system was orthorhombic, making the data set incomplete (ca. 81% completeness to a resolution of 0.84 Å). Unfortunately, again, by this time the rest of the sample had also decomposed, making the incomplete data set all that was going to be collected. However, despite the incompleteness of the data set making the derived structure of inevitably lower quality than would otherwise be the case, there is still plenty of worthwhile information that can be gleaned from it. The structure of 8 was found to sit across a mirror plane that passes through C2, Al1 and Fe1, and bisects the N1···N1A vector. The C13-bound pyridyl ring and the methyl groups of the P20-based PMe3 moiety were both found to be disordered across this mirror plane, and in each case one complete 50% occupancy orientation was identified (with a second 50% occupancy orientation being generated by operation of the mirror plane).
The geometries of the two unique orientations were optimised, and all of the non-hydrogen atoms were refined anisotropically. The P24-based PMe3 group was also found to be disordered. Two orientations were identified of ca. 54 and 46% occupancy, their geometries were optimised, the thermal 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). Despite all the disorder and the incomplete data, the basic identity of the compound and its connectivity are clear, especially when considered with the other structures in this paper. The main effect of the incomplete data set is likely to be a degradation of the standard uncertainties, and this can be seen by a comparison of the Al···Fe separations across all six structures -the "well behaved" structures of 6, 7, 9 and 10 have su's of 0.0007, 0.0006, 0.0008 and 0.0008 Å respectively for this distance, whilst the structure of S1 (with the twinned data set) has an s.u. of 0.0013 Å and the structure of 8 (with the incomplete data set) has an s.u. of 0.0015 Å.
The backbone of the deprotonated β-diketiminate ligand in the structure of 10 was found to be disordered, with the location of the terminal C-Me and C=CH2 units being "swapped" between the C1/C4 and C3/C5 sites.
This was modelled by using two sets of idealised partial-occupancy hydrogen atoms in a ca. 51:49 ratiothe carbon atoms were unaffected, and the occupancies of the hydrogen atoms were allowed to vary with no restraints other than the members of each set having to have the same occupancy as each other, and the total occupancy being 100%. The Fe(PMe3)3 moiety was also found to be disordered, and two orientations with a common iron atom position were identified of ca. 54 and 46% occupancy. The geometries of the two orientations were optimised, the thermal parameters of adjacent atoms were restrained to be similar, and only the non-hydrogen atoms of the major occupancy orientation, and the phosphorus centres of the minor occupancy orientation, were refined anisotropically (the rest were refined isotropically).
The crystal of S1 that was studied was found to be a two component

S-14
unique orientations were identified of ca. 30 and 20% occupancy (with two further orientations of the same occupancies being generated by operation of the inversion centre). The geometries of the two unique orientations were optimised, the thermal parameters of adjacent atoms were restrained to be similar, and all of the atoms of both unique orientations were refined isotropically. Figure S2. The crystal structure of 6 (50% probability ellipsoids). Figure S3. The crystal structure of 7 (50% probability ellipsoids). Figure S4. The crystal structure of the CS-symmetric complex 8 (30% probability ellipsoids). Figure S5. The crystal structure of 9 (50% probability ellipsoids). Figure S6. The crystal structure of 10 (50% probability ellipsoids). Figure S7. The crystal structure of S1 (50% probability ellipsoids).

Computational Methods
DFT calculations were run using Gaussian 09 (Revision D.01). 5 Geometry optimisation calculations were performed without symmetry constraints. Frequency analyses for all stationary points were performed using the enhanced criteria to confirm the nature of the structures as either minima (no imaginary frequency) or transition states (only one imaginary frequency). Intrinsic reaction coordinate (IRC) calculations followed by full geometry optimisations on final points were used to connect transition states and minima located on the potential energy surface allowing a full energy profile (calculated at 298.15 K, 1 atm) of the reaction to be constructed. 6,7 Solvent corrections were applied using the polarizable continuum model (PCM). 8 Dispersion corrections were applied using Grimme's D3 (GD3) correction. 9 All calculations were conducted using the B3PW91 10 functional including solvent and dispersion corrections directly in the optimisations. Al and Fe centres were described with Stuttgart SDDAll ECP and associated basis sets, and the 6-31G** basis sets were used for all other atoms. 11,12,13 The employed level of theory has been benchmarked against experimental data as reported previously by our group. 1 Natural Bond Orbital analysis was carried out using NBO 6.0. 14 Calculated thermodynamical data: Figure S8. Calculation on the thermochemistry of the reaction of 1 with different pyridine substrates showing the energy difference between the respective C-C and C-Al bound products. P' = PMe3, N' = N(2,4,6-MeC6H2). Gibbs free energies are given in kcal.mol -1 .

S-18
Calculated mechanism for the reaction of 9 to 10: Figure S9 shows the calculated free energy profile for the reaction of 9 to 10. The reaction sequence is initiated by a conformational change of the pyridyl unit in 9 (TS-1) followed by the of the C-C bond cleavage between the β-diketiminate ligand and the activated substrate (TS-2). In a final step, the β-diketiminate ligand is deprotonated by the activated substrate (TS-3) resulting in the formation of 10 and release of 2ethylpyridine. The overall activation energy required for this reaction is ΔG ≠ = 25.3 kcal mol -1 providing a reasonable barrier for a reaction that only occurs at elevated temperatures. Figure S9. Calculated free energy profile for the formation of 10. Gibbs free energies are given in kcal.mol -1 .
However, thermal corrections have been applied taking the entropic contribution to the Gibbs free energy into account. The corrected thermodynamical data for the actual reaction conditions (80°C) in fact agree with an exergonic process (ΔG°353K = -3.3 kcal mol −1 ) for the formation of 10 ( Figure S10).