Evaluation of 3,3′-Triazolyl Biisoquinoline N,N′-Dioxide Catalysts for Asymmetric Hydrosilylation of Hydrazones with Trichlorosilane

A new class of axial-chiral biisoquinoline N,N′-dioxides was evaluated as catalysts for the enantioselective hydrosilylation of acyl hydrazones with trichlorosilane. While these catalysts provided poor to moderate reactivity and enantioselectivity, this study represents the first example of the organocatalytic asymmetric reduction of acyl hydrazones. In addition, the structures and energies of two possible diastereomeric catalyst–trichlorosilane complexes (2a–HSiCl3) were analyzed using density functional theory calculations.


Results and Discussion
We recently developed the modular method to synthesize axial-chiral 3,3′-triazolyl biisoquinoline N,N′-dioxides from readily available triazoles and optically pure 3,3′dibromo-biisoquinoline N,N′-dioxide [57] as part of our longstanding interests in developing new chiral Lewis-bases [58][59][60][61][62]. Since this new class of catalysts was found capable of activating trichlorosilane at relatively low temperatures, we envisioned that they might be able to catalyze the reduction of acyl hydrazones under conditions where no background reaction would take place. We set out on our investigation by employing benzoyl hydrazone 1a as a model substrate (Scheme 2). To our delight, the background reaction was found negligible at −40 °C, and catalyst 2a provided hydrazine (R)-3a in 48% yield with 53% ee (entries 1 and 2). Next, we looked at several solvents that are commonly used for trichlorosilane-mediated reactions. Chloroform provided the product with a lower yield but with a slightly higher enantioselectivity (34% yield, 66% ee). Acetonitrile gave 3a in a comparable yield but with a lower ee of 32%. The reaction in tetrahydrofuran afforded the opposite enantiomer (S)-3a with a much lower yield and selectivity (entry 5). Overall, dichloromethane was found optimum. We tested with twice as much solvent since benzoyl hydrazone 1a was not fully dissolved under the reaction conditions (entry 6). However, it did not improve the result. Previously, we found that 4 Å molecular sieve was an effective acid scavenger for adventitious HCl in trichlorosilane [63], but its use did not positively impact the outcome in the present case (entry 7). The use of 3.0 equivalent of trichlorosilane did not improve the yield, either (entry 8). As the protecting groups on hydrazones are known to influence their reactivities and enantioselectivities in many cases (e.g., see; [19]), we evaluated the Boc and Cbz protected hydrazones (1b and 1c in entries 9 and 10, respectively). While enantiomeric excesses of the corresponding products were slightly higher than that of the benzoyl counterpart, both Boc and Cbz protecting groups adversely affected the yields. Since the C=O unit of Boc or Cbz group is more Lewis basic than that of the benzoyl counterpart, we tested a less Lewis basic hydrazone (1d). However, the yield decreased to 14% albeit with a slightly higher enantioselectivity (entry 11). Overall, the present method was found to be quite sensitive to reaction solvents and the hydrazone protecting groups.
Next, we evaluated different triazolyl groups on the biisoquinoline that are expected to play important roles on the catalyst's reactivity and selectivity (Scheme 3, entries 1-4). Catalyst 2a was clearly superior to the other three catalysts (2b-d) in terms of their reactivity. Catalyst 2c was more enantioselective than others, albeit with a low yield. These results indicate that the reactivity and selectivity of this new class of catalysts can be tuned by changing the triazolyl groups. We also compared these triazolyl catalysts to conventional 3,3′-substituted biisoquinoline N,N′-dioxides (entries 5 and 6). To our surprise, neither 2e nor 2f promoted the reaction although 2f was as reactive as 2b-d for the hydrosilylation of an N-phenyl ketimine with trichlorosilane [57]. Nonetheless, these results clearly demonstrated that this new class of axial-chiral biisoquinolines is indeed complementary to the existing Lewis-base catalysts and bode well for the development of their applications.
As we determined the basic reaction parameters, we proceeded to evaluate the extent to which the present catalytic system could enantioselectively promote the hydrosilylation of various benzoyl hydrazones with trichlorosilane (Scheme 4). To our surprise, a paramethyl substitution (3e)-which is a minimal change from the model substrate (3a)-had a detrimental effect on the chemical yield while the corresponding meta-substitution (3f) did not. An ortho-methyl substitution (that is known to push the aromatic ring out of conjugation with a C=N bond) completely shut down the reaction (3g). Eventually, it was gleaned that the para-substitutions have adverse effects on the reactivity but not much on the enantioselectivity regardless of their electronic nature (3e, 3h-m) (these enantioselectivities are approximately the same). A heteroaromatic hydrazone was moderately less reactive and selective than the model substrate (3n). Although an ethyl group at the C=N bond is in general expected to lead to an increased steric demand in the TS, it did not affect the reactivity in the present case (3o). It is noteworthy that a cyclohexyl counterpart provided the opposite sense of enantioselection to the model substrate (3p). Differentiation of the two similar alkyl groups franking the C=N bond was difficult by the present catalytic system (3q). Unreacted hydrazones and corresponding ketones were the major components of the crude reaction mixtures besides the desired products, and no significant amounts of by-products were observed for 1a-1q. An α,β-unsaturated hydrazone was not a viable substrate for this method as the conjugate reduction took place (3r) [64].
We also tested a 1.0 mmol scale reaction with the model substrate. To our delight, it provided essentially the same result (Scheme 5), demonstrating a potential robustness of the method. Furthermore, catalyst 2a was quantitatively recovered after a flash column chromatography on silica gel (see Supplementary Materials for details). The recovered catalyst promoted the model reaction (1a on 0.25 mmol scale) with no loss in reactivity and enantioselectivity.
The structure of the active reducing species generated from a chiral catalyst and HSiCl 3 is considered to play a central role for the enantioselectivity of a reaction. Even with notable advances made in this area (selected references; [35][36][37][38][39][40][41][42][43][44][45]), it remains largely elusive and significantly challenging to control the relative populations and reactivities of diastereomeric reducing species that are reversibly produced from a chiral Lewis-base and trichlorosilane. C 2 -symmetric 2a and HSiCl 3 can give rise to two diastereomeric complexes that are expected to have different enantioselectivities (as long as 2a acts as a bidentate Lewis-base). Therefore, the binding geometry of 2a to HSiCl 3 was investigated computationally with the aim of shedding some light on the structure of the active reducing species. To our delight, 2a was found to bind to HSiCl 3 through its two oxygen atoms (i.e., a C 2 -symmetric bidentate ligand), generating two diastereomeric complexes ( Figure 1). Complex 1 was found to be 1.91 kcal/mol lower in energy than the complex 2. The analysis of their electrostatic potentials revealed an anion-π-type interaction between the hydrogen atom in complex 1 or one of the chlorine atoms in complex 2 and the phenyl ring. It should be mentioned that a pileup of electron density occurs at the peripheral atoms of a hypervalent silicon complex of this kind [59,65,66]. This anion-π-type interaction appears to effectively lock the conformation of the benzyl group at least at the ground state, leading to a welldefined chiral pocket around the hypervalent silicon atom. This computationally identified non-covalent attractive interaction could offer a possible basis to rationalize why 2a (benzyl) was as enantioselective as 2d (1-adamantyl), and why 2c (benzhydryl) was substantially more enantioselective than 2d (53% ee, 54% ee and 74% ee, respectively; Scheme 3).

Conclusions
Axial-chiral 3,3′-triazolyl biisoquinoline N,N′-dioxides offer potential for functioning as effective catalysts for the asymmetric hydrosilylation of acyl hydrazones with trichlorosilane. Since catalyst's triazolyl units indeed tuned the reactivity and enantioselectivity of the reaction and our modular synthesis allows ready access to a variety of 3,3′-triazolyl biisoquinoline N,N′-dioxides, potential for the identification of more effective catalysts than those presented herein clearly exits.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.

Funding:
Support is currently provided by the National Institute of Health (1R15 GM139087-01) and the Florida Institute of Technology.

Data Availability Statement:
Data is contained within the article and Supporting Information. Computed structures of the two lowest energy minima for the 2a-HSiCl 3 complex (i.e., two diastereomeric complexes) calculated with PBEh-3c//C-PCM (DCM). Both are shown with balls-and-sticks (left) and space filling (right) models. Molecular electrostatic potentials are also shown in the space filling models. Complex 1 (top) is 1.91 kcal/mol lower in energy than complex 2 (bottom).