Metal control of selectivity in acetate-assisted C – H bond activation : an experimental and computational study of heterocyclic , vinylic and phenylic C ( sp 2 ) – H bonds at Ir and Rh †

Kevin J. T. Carr, David L. Davies,* Stuart A. Macgregor,* Kuldip Singh and Barbara Villa-Marcos Acetate-assisted C(sp)–H bond activation at [MCl2Cp*]2 (M1⁄4 Ir, Rh) has been studied for a series ofN-alkyl imines, PrN]CHR, (R 1⁄4 N-methyl-2-pyrrolyl, H-L1; 2-furanyl, H-L2; 2-thiophenyl, H-L3a; C2H2Ph, H-L4; and Ph, H-L5) as well as phenylpyridine (H-L6) by both experimental and computational means. Competition experiments reveal significant variation in the relative reactivity of these substrates and highlight changes in selectivity between Ir (H-L4 z H-L2 < H-L3a z H-L5 < H-L1 z H-L6) and Rh (H-L2 z H-L1 < H-L3a z H-L4 < H-L5 < H-L6). Comparison of H-L3a with its N-xylyl analogue, H-L3b, gives a further case of metal-based selectivity, H-L3a being more reactive at Ir, while H-L3b is preferred at Rh. H/ D exchange experiments suggest that the selectivity of C–H activation at Ir is determined by kinetic factors while that at Rh is determined by the product thermodynamic stability. This is confirmed by computational studies which also successfully model the order of substrate reactivity seen experimentally at each metal. To achieve the good level of agreement between experiment and computation required the inclusion of dispersion effects, use of large basis sets and an appropriate solvent correction.


General.
Ligand H-L 6 was obtained commercially and used without further purification. [MCl 2 Cp*] 2 (M = Ir, Rh) was synthesised according to the literature procedure. 1 CH 2 Cl 2 was dried over CaH 2 and distilled prior to use. Dry MeOH was purchased for Acros Organics, extra dry over molecular sieves (MS), AcroSeal. 1 H and 13 C NMR spectra were recorded in ppm with reference to TMS as an internal standard at ambient temperature. FAB mass spectra were obtained on a Kratos concept mass spectrometer using NOBA as matrix. The electrospray (ES) mass spectra were recorded using a micromass Quattra LC mass spectrometer with MeCN as solvent.

General procedures
2a. General procedure for the synthesis of ligands L 1--L 5 .
1 equiv. of aldehyde was added to a round-bottomed flask equipped with a stir bar and 5 g of activated 4 Å MS. A volume of dry MeOH (10-20 mL) was then added to the flask, followed by slow addition of 3 equiv. of isopropylamine. The reaction was left stirring for 2-3 hours.
The solution was then filtered and dried on the rotary evaporator followed by high vacuum. No further purification was required.
2b. General procedure for the synthesis of ligand L 3b .
1.1 equiv. of thiophene-2-carbaldehyde was added to a round-bottomed flask equipped with a stir bar and 5 g of activated 4 Å MS. A volume of dry MeOH (10-20 mL) was then added to the flask, followed by slow addition of 1 equiv. of 3,5-dimethylaniline. The reaction was left stirring for 2-3 hours. The solution was then filtered and dried on the rotary evaporator. The crude product contain 5-7% of starting aldehyde and it was used for the cyclometallation reaction without further purification.
2c. General procedure for the cyclometallation reactions.
1 equiv. of [MCl 2 Cp*] 2 (M = Ir, Rh) and 4 equiv. of NaOAc (ratio Rh:NaOAc 1:2) was weighed and added to an oven-dried Schlenk tube equipped with a stirrer bar. The Schlenk tube was degassed three times and left under N 2 atmosphere. 7 mL of degassed solvent (CH 2 Cl 2 or MeOH) was then added with a syringe followed by a solution (3 mL) of the relevant ligand (2.2 equiv, ratio Rh:ligand 1:1.1) . The Schlenk tube was sealed and left stirring at rt for 3 h to 4 d. Once full conversion or equilibrium was achieved, the reaction solution was filtered through celite, and concentrated to afford the crude product. The iridium complexes achieved full conversion and were purified by recrystallization from CH 2 Cl 2 /hexane. The rhodium complexes were purified by column chromatography on silica gel eluting with CH 2 Cl 2 or CH 2 Cl 2 /MeOH (20/1), followed by recrystallization from CH 2 Cl 2 /hexane. For the synthesis of Rh-L 1 and Rh-L 2 a large excess of ligand was used (10 equiv., ratio Rh:ligand 1:5) to favour the forward reaction; however, isolation of pure samples of Rh-L 1 and Rh-L 2 was unsuccessful. Crystal data and structure refinement for Ir-L 1 (CCDC: 980293).
There are two molecule molecules in the unit cell, only one shown for clarity. H atoms omitted for clarity.

Computational details
DFT calculations were run with Gaussian 03 (Revision D.01) 7 and Gaussian 09 (Revision A.02). 8 Geometry optimisations were performed with Gaussian 03 and the BP86 functional 9, 10 and employed a smaller basis set, BS1, in which Rh, Ir and S centers were described with the Stuttgart RECPs and associated basis sets 11 with additional d-orbital polarisation on S ( = 0.503) 12 and 6-31G** basis sets were used for all other atoms. 13,14 The 'grid=ultrafine' option was used throughout as it was found to be necessary to achieve consistent results, in particular for the geometries of the Cp* rings. Different orientations of the N-i Pr and N-xylyl substituents and the vinylimine substrate H-L 4 were tested via rotation profiles around the appropriate bonds in both the free substrate and the N-bound metal complexes. Geometries computed with BS1/BP86 compared well with the experimental data that are available for several of the cyclometalated products (see Table S4). All stationary points were fully characterized via analytical frequency calculations as either minima (all positive eigenvalues) or transition states (one negative eigenvalue). IRC calculations and subsequent geometry optimizations were used to confirm the minima linked by each transition state. Corrections for the effects of methanol (ε = 32.6) and dichloromethane solvent ( = 8.93) were run with Gaussian 09 and used the polarisable continuum model.
In order to obtain improved energetics (in particular for Cl -/OAcexchange processes) all energies were recomputed with a larger basis set, BS2, featuring cc-pVTZ-PP on Rh and Ir 6-311++g** on all other atoms. Further details on the basis set dependency are given in Table S5. The effects of functional choice were also tested (see Table   S6). For this the BP86-optimised geometries were reoptimised with the selected functional in Gaussian 09 using BS1. Frequency calculations were again used to confirm the nature of all stationary points. Single point energy dispersion corrections were also assessed using Grimme's D3 parameter set 15

Basis Set Effects
Calculations were run to assess the basis set dependency of the computed results and how these relate to the experimental observations. The energetics of Cl -/OAcexchange were found to be particularly sensitive to basis set choice and two such processes are considered in detail here. Experimentally, two observations suggest that the free energy of Cl -/OAcexchange should be close to zero: (i) opening of the [RhCl 2 Cp*] 2 dimer by NaOAc in the absence of substrate led to the observation of both the chloro-acetate species [RhCl(OAc)Cp*], Rh II , and the bis-acetate species [Rh(OAc) 2 Cp*], Rh III , (see Eqn. S1 16 ); (ii) in the competition experiment between substrates H-L 1a and H-L 3a the cyclometalated product derived from H-L 3a was observed as an equilibrium mixture of its chloro (VII Rh -L 3a ) and acetate adducts (VII Rh -L 3a -OAc, Eqn. S2.).
The results of the basis set testing are given in Table S5 and show that the use of diffuse functions on the ligand heavy atoms is crucial to obtain a reasonable value for the Cl -/OAcexchange energy. Without these the relative stabilities of the OAc adducts III Rh and VII Rh -L 3a -OAc are considerably over-estimated (Entries 1b, 1 and 2). Entry 4 shows that including diffuse functions on Cl, C, N and O improves the situation, while additional diffuse functions on H has little further effect (Entry 5). Use of larger basis sets on either the ligands (Entries 6-8) and rhodium (Entries 5, 9-12) did not significantly change the outcome. On the basis of these data the basis set associated with Entry 9 was employed for the final single-point calculations. Table S5. Computed SCF energy change, (kcal/mol) for Cl -/OAcexchange at II Rh (Eqn. S1) and VII Rh -L 3a (Eqn. S2) as a function of the basis set employed. Negative value indicates a greater stability of the OAc species involved in the exchange.  2 Cp*], Rh III , were assessed, as well as the energy of the final cyclometallated product VII Rh formed with substrates H-L 1-6 . Relative free energies are given in Table S6 and include corrections for solvent (MeOH) and basis set effects. An additional correction for dispersion effects is included for all functionals, with the exception of M06, M06L, B97D and B97XD for which a treatment of dispersion is included in the original functional.