Hydrodenitrogenation of pyridines and quinolines at a multinuclear titanium hydride framework

Investigation of the hydrodenitrogenation (HDN) of aromatic N-heterocycles such as pyridines and quinolines at the molecular level is of fundamental interest and practical importance, as this transformation is essential in the industrial petroleum refining on solid catalysts. Here, we report the HDN of pyridines and quinolines by a molecular trinuclear titanium polyhydride complex. Experimental and computational studies reveal that the denitrogenation of a pyridine or quinoline ring is easier than the ring-opening reaction at the trinuclear titanium hydride framework, which is in sharp contrast with what has been reported previously. Hydrolysis of the pyridine-derived nitrogen-free hydrocarbon skeleton at the titanium framework with H2O leads to recyclization to afford cyclopentadiene with the generation of ammonia, while treatment with HCl gives the corresponding linear hydrocarbon products and ammonium chloride. This work has provides insights into the mechanistic aspects of the hydrodenitrogenation of an aromatic N-heterocycle at the molecular level.

(1) (100 mg, 0.14 mmol). The mixture was stirred at room temperature for 10 min, and an immediate color change from dark green to dark purple was observed. The solution was concentrated and cooled at -33 °C to give 2 as dark-purple crystals (105 mg, 0.13 mmol, 93% yield). Single crystals suitable for X-ray diffraction studies were obtained by recrystallization from a hexane solution at -33 °C. The preparation of the 15  . 13 C NMR (C 6     vacuum and recrystallized in hexane/THF (5:1) at -33 °C to give 4 as dark-purple crystals (60 mg, 0.073 mmol, 89% yield), which were suitable for X-ray diffraction studies. 1 H NMR (THF-d 8

Reaction of 3 and 5 with Hydrochloric Acid
Hydrochloric acid (3.0 M HCl aqueous solution, 82 µL, 0.25 mmol) was slowly added to a C 6 D 6 solution (0.8 mL) of 3 (20 mg, 0.025 mmol) via a syringe. An orange-red solution was formed immediately together with solid precipitates. After 10 min, the reaction mixture was separated by a trap-to-trap technique. The volatile part was dried with Na 2 SO 4 . The organic products were confirmed by GC-MS and NMR analyses through comparison with authentic samples. The 1 H NMR integration against the 1,2,4,5-tetramethylbenzene internal standard showed 1-pentene and 2-pentene in 48% and 14% yields, respectively. To the residue Et 2 O was added to give a suspension, which was then centrifuged and filtered. The white solid was confirmed to be NH 4 Cl (76% yield) by 1 H NMR analysis and phenol-hypochlorite titration. The solution was evaporated under vacuum. The resulting orange-red solid was washed with hexane and dried under vacuum to give Cp′TiCl 3 (23 mg, 0.066mmol, 88% yield).

Kinetic Studies of Conversion of 2 to 3
In a typical experiment, a 5-mm NMR tube with a J. Young valve was charged with 10 mg of 2 (0.0112 mmol) and toluene-d 8 (0.70 ml) in a glovebox, and was then subjected to the 1 H NMR measurements. The relative concentration was calculated based on the integration of the resonances of the SiMe 3 groups in the C 5 Me 4 SiMe 3 ligands. The activation parameters for the conversion of 2 to 3 were determined by the rate constants obtained at various temperatures ( Supplementary Fig. 43). An Eyring plot ( Supplementary Fig. 44) was constructed from these data (Supplementary Table 1). The activation enthalpy and entropy were extracted as follows:

X-ray Crystallographic Studies
Crystals for X-ray diffraction studies were obtained as described in the preparations. The crystals were manipulated in a glovebox under a microscope, and were sealed in thin-walled glass capillaries. X-ray diffraction data collections were performed on a Bruker D8 QUEST   6 (1). All the stationary points were optimized by TPSSTPSS functional 7 , which has been successfully used for the calculations of trinuclear titanium complexes 1,8 . In the geometrical optimizations, the 6-31G(d) basis set was considered for C, H, and N atoms and the Stuttgart/Dresden effective core potentials (ECP) as well as the associated valence basis sets 9,10 MWB10 and MDF10 were used for the Si and Ti atoms, respectively. The basis set of the Si atom was also augmented by one d polarization function (exponent of 0.284) 11 . Considering that antiferromagnetic coupling singlet ground state may exist for some structures, the combination of the spin unrestricted strategy and scf=nosymm keyword was used in the optimizations. Each optimized structure was subsequently analyzed by harmonic vibration frequencies for characterization of a minimum (N imag = 0) or a transition state (N imag = 1). To obtain more reliable relative energies, the single-point calculations of optimized structures were carried out at the level of TPSSTPSS/BSII. In the BSII, the 6-311G(d,p) basis set was used for C, H, N, and Si atoms and MDF10 basis set together with ECP was considered for Ti atoms. In these single point calculations, the solvation effect of toluene was considered with the CPCM solvation model 12,13 and dispersion correction was included through GD3BJ approach 14 . Actually, it was previously reported that TPSSTPSS also often shows the best performance for the structure and energy calculations for transition-metal-containing systems [15][16][17] . The stabilities of wave functions were tested by the keyword of stable=opt in this study. All of the structures were optimized without any geometrical symmetry restriction. The calculated activation free energy is well in line with the experimental value (vide infra), suggesting that the theoretical strategy adopted here is reliable for such reaction. The relative Gibbs free energies in solution were used for discussion of the mechanism in this study. The relative Gibbs free energies in gas-phase of all stationary points are also provided in Supplementary Table 11 for reference. The results suggest that the solvation effect has little influence on the overall reaction.
Triplet state pathway was also checked for some important intermediates and the ratedetermining transition state. The results shows that triplet states of the structures such as 1m, B, 2m, TS CD , D, E, and 3m are higher in free energy than their corresponding single states by 6.5, 14.2, 7.7, 12.2, 9.9, 4.0, and 12.4 kcal/mol, respectively. Therefore, the triplet state pathway was not discussed in the text 8 . All calculations were performed with Gaussian 09 program 18 .

Results and Discussion
As shown in Supplementary Fig. 57, the Kohn-Sham orbital analysis revealed that the red parts (same sign) of LUMO of (Cp′Ti) 3 H 7 and HOMO of pyridine concentrate on the Ti1 atom and the N atom, respectively, suggesting that the N atom of pyridine could coordinate to the Ti1 atom of 1m to provide good orbital overlap. This led us to model the coordination of pyridine to the Ti1 atom via site 1 or site 2 ( Supplementary Fig. 57). The site 1 is at the same side of the Ti1−Ti2−Ti3 plane as the µ 3 -H is, while the site 2 is at the opposite site. During this reduction process, the N−C1 bond was elongated from 1.36 Å in a2 to 1.45 Å in B.

Pyridine-HOMO
In complex B, there is an unambiguous agostic interaction between C1−H8 bond and Ti2 atom.

Subsequently, the C1−H bond in [C 5 H 4 N] moiety was further activated via transition state TS B2m
to give 2m, which is equivalent to 2 obtained experimentally, accompanied by elimination of another molecule of H 2 (H6−H8). The whole process of the formation of 2m from 1m and pyridine overcomes a low energy barrier of 13.8 kcal/mol and is exergonic by 21.2 kcal/mol. Such a low energy barrier is in agreement with experimental observation that the formation of 2 by the reaction of 1 and pyridine could occur rapidly at room temperature.   Supplementary Fig. 59b). Therefore, the latter path is more likely to work here.
The whole process of transformation of 2m to 3m is significantly exergonic by 35.9 kcal/mol and the rate-determining step is the cleavage of C1−N bond with an energy barrier of 26.5 kcal/mol, which is well in agreement with experimental result (ΔG ‡ (298 K) = 25.5 kcal/mol) obtained from kinetic studies for the transformation of 2 to 3.