Oxidative Cross Dehydrogenative Coupling of N-Heterocycles with Aldehydes through C(sp³)–H Functionalization

Abstract

Existing methodologies for metal-catalyzed cross-couplings typically rely on pre-installation of reactive functional groups on both reaction partners. In contrast, C–H functionalization approaches offer promise in simplification of the requisite substrates, however, challenges from low reactivity and similar reactivity of various C–H bonds introduce considerable complexity. Herein, the oxidative cross dehydrogenative coupling of α-amino C(sp³)–H bonds and aldehydes to produce ketone derivatives is described using an unusual reaction medium that incorporates the simultaneous use of di-tert-butyl peroxide as an oxidant and zinc metal as a reductant. The method proceeds with a broad substrate scope, representing an attractive approach for accessing α-amino ketones through the formal acylation of C–H bonds α to nitrogen in N-heterocycles. A combination of experimental investigation and computational modelling provides evidence for a mechanistic pathway involving cross-selective nickel-mediated cross-coupling of α-amino radicals and acyl radicals.

Graphical Abstract

Traditional approaches to cross-couplings utilize complementary functional groups on both reaction partners to enable the assembly of a desired C–C bond.¹ In contrast, C–H activation methods enable simplification of the approach by allowing one of the two substrates to lack a preinstalled functional group,² thus streamlining the construction of target molecules.³ Cross-couplings that accomplish C–C couplings through site-selective C–H activation on both substrates, however, are quite rare, while offering considerable advantages in the simplicity of the substrates used. Oxidative dehydrogenative cross-couplings of this type, however, present considerable challenges in reactivity, site-selectivity, and cross-selectivity.⁴

In the realm of cross-couplings that involve a single functionalized reaction partner, key contributions have come from many investigators, including initial findings from Doyle and MacMillan,⁵ across a range of reactions and catalyst types (Scheme 1A). The fields of photoredox catalysis and hydrogen atom transfer (HAT) have subsequently experienced rapid growth with broad scope being demonstrated in α-amino C(sp³)–H functionalization with different coupling partners, including halides,⁶ activated acids,⁷ acrylates⁸, azoliums,⁹ amides.¹⁰ Of these and related reports, work from Doyle,^5b Scheidt,⁹ Hong,¹⁰ Huo,⁷ Xu,¹¹ Ohmiya,¹² and Chi¹³ describe acylation processes using an array of electrophile classes. Among various HAT agents explored, peroxide oxidants have been demonstrated by Lei as effective in nickel-catalyzed C–H activation processes of heterocyclic C(sp³)–H bonds.¹⁴ Recent work from Stahl illustrating C–H methylation processes via cross-selective radical-radical couplings relying on methyl radical extrusion¹⁵ and from Gong demonstrating the co-utilization of a peroxide oxidant with a zinc reductant¹⁶ provide important examples of peroxide-mediated processes with nickel involving free radicals generated by HAT processes.

Scheme 1.

Strategies in C(sp³)–H functionalization

Methods that utilize cross dehydrogenative couplings (CDCs) through C–H functionalization on two different unfunctionalized reaction partners are highly desirable but quite rare.¹⁷ Extensive work from Li has examined cross dehydrogenative couplings using copper catalysis (Scheme 1B),¹⁸ while fundamental studies from Fagnou illustrated biaryl synthesis through palladium-catalyzed cross-couplings of two arenes with cross selectivity arising from electronic differences in the two arene substrates.¹⁹ C–H activations in CDCs typically operate via different modes, with one of the two substrates bearing acidic protons that enables the formation of a reactive species through acid-base chemistry, while the other substrate undergoes HAT-mediated C–H cleavage.²⁰

Based on the above advances, we envisioned that the versatile chemistry of nickel in mediating radical-based cross-coupling processes could potentially be paired with HAT chemistry to enable C–H activation methods on two different substrates, thus avoiding functional group pre-installation. A strategy of this type could potentially allow two different C–H activation processes to simultaneously operate even when the two substrates possess C–H bonds that are not easily distinguished by acidity or bond strength.²¹ Towards this goal, we set out to examine the coupling of aldehydes with N-heterocycles (Scheme 1C), with these studies being motivated by our longstanding interest in the development of nickel-catalyzed additions to aldehydes utilizing unconventional nucleophiles.²² Herein, we describe the formation of unsymmetrical ketones through the direct acylation of α-amino C(sp³)–H bonds with aldehydes in a nickel-catalyzed process utilizing peroxide oxidant as the source of the HAT agent.

We began our experimental optimization with the coupling between benzaldehyde (1a) and N-Boc pyrrolidine (2a), where control experiments confirmed the necessity of catalyst, oxidant, and reductant (Table 1, entry 2). Ligand screening revealed that N-donor types together with Ni(II) pre-catalysts generally resulted in enhanced reactivities compared with NHC and phosphine ligands with Ni(COD)₂, though further modification of the bis(pyrazole) pyridine (bpp) scaffold failed to improve the yield (see supporting information S1). The thermal-activation was found to be unique with the combination of Zn and di-tert-butyl peroxide (DTBP), as using either metallaphotoredox or different reductants did not yield the desired product (entry 5). Different peroxides have also been tested under this dehydrogenative coupling manifold, and DTBP, a stable organic peroxide, performed the best (see supporting information S3). The use of silyl chlorides only yielded the silylated pinacol-coupled byproducts (entry 6).²³ whereas 2.0 equiv. of water resulted in product 3a in an improved 62% isolated yield (entry 7). Conducting the reaction in darkness resulted in a similar yield, excluded the possibility of photo-activation (entry 8). While initial screening showed little impact of the choice between chloride and bromide pre-catalysts, the use of NiBr₂•dme in an optimized protocol with water ultimately proved more robust and reproducible among the set of examples developed below (entry 9). Investigations also indicated that possible oxidative degradation of N-Boc pyrrolidine (2a) had minor influences on the reaction outcomes (see supporting information for details). Further examination of reaction temperature, concentration, and time led to optimal conditions for exploring the reaction scope.

Table 1.

Nickel-catalyzed N-heterocycle aldehyde coupling optimization^a

Entry	Deviation from standard condition	% yield
1	No	30-40
2	Absence of Ni or bpp or Zn or DTBP	<5
3	Ni(COD)2 instead of NiCl2•dme	0
4	Other N-donor, PR3, NHC ligands	0-40
5	[Ir], blue LED instead of Zn	0
6	With 1 eq TESCl	0
7	With 2 eq water	45-65
8	In dark	64b
9	20% NiBr2•dme and bpp	69c

^a1a (0.1 mmol, 1.0 eq), 2a (0.24 mmol, 2.4 eq), bpp (0.01 mmol, 0.1 eq), NiCl₂•dme (0.01 mmol, 0.1 eq), nano powder Zn (0.2 mmol, 2.0 eq), and DTBP (0.4 mmol, 4 eq) in MeCN (0.5 mL, 0.2 M) at 50 °C for 16h. Yield was determined by ¹H NMR using CH₂Br₂ as an internal standard.

^b2 eq H₂O was used as an additive.

^c≤ 10 μm Zn powder (0.6 mmol, 3.0 eq), H₂O (0.6 mmol, 3.0 eq) and DTBP (1.0 mmol, 5 eq) were used.

The transformation tolerated different electronic environments on benzaldehydes, while electron withdrawing groups generally led to lower yields (Table 2). Functional groups that were facile for activation with nickel catalysis under reductive conditions underwent chemoselective C–H functionalization, enabling orthogonal functionalization at a later stage (3c, 3d, 3h). Of note, by analyzing crude reaction profiles, remainders of mass balances were largely unreacted starting materials (1 and 2), and good mass recovery was demonstrated with the synthesis of 3d (93% brsm). The synthetic utility was illustrated with a 1.0 mmol scale synthesis of 3a in 54% isolated yield. Heteroaryl aldehydes were also compatible with this dehydrogenative coupling (3i-3l), and though electron-rich heterocycles gave lower yields, the absence of Minisici-type byproducts and clean reaction profile is worth highlighting. Further expanding the scope with aliphatic aldehydes, both cyclopropane carboxaldehyde and hydrocinnamaldehyde smoothly yielded the ketone products 3m and 3n, the former with the ring intact, and latter could be carried out with a 1:1 ratio of aldehyde and pyrrolidine. We envisioned that the potential of switching either coupling partner as the limiting reagent would facilitate late-stage applications in a complex setting. Along this line, an aldehyde derived from ricinoleic acid was well tolerated, leading to the formation of the desired coupling product 3o with a 41 yield.

Table 2.

Nickel-catalyzed N-heterocycle aldehyde coupling scope^a

^aNiBr₂•dme (0.04 mmol, 0.2 eq), bpp (0.04 mmol, 0.2 eq), ≤ 10 μm Zn powder (0.6 mmol, 3.0 eq), H₂O (0.6 mmol, 3.0 eq) and DTBP (1.0 mmol, 5 eq).

^bNiCl₂•dme (0.02 mmol, 0.1 eq), bpp (0.02 mmol, 0.1 eq), nano powder Zn (0.4 mmol, 2.0 eq), H₂O (0.4 mmol, 2.0 eq) and DTBP (0.8 mmol, 4 eq).

^cFormed as an 85:15 regioisomeric mixture.

^dTHF (6.0 equiv) was used.

A broad scope of amine coupling partners has also been developed. A proline derivative was successfully converted to the corresponding acylated product 3p in 77% yield. Different protecting groups were well tolerated, including CBz, Ac, and benzoyl derivatives (3q-t). Variations in the nitrogen heterocycle included piperidine, morpholine, piperazine, and γ-lactam derivatives (3u-w, 3ag) in addition to representative four-, seven-, and eight-membered derivatives (3ac-3af). Morpholine derivative 3v was isolated as an 85:15 mixture of regioisomers, and piperazine derivative 3w was isolated as a single regioisomer taking advantage of the deactivating influence of the tosyl protecting group.^8,24 The desired acylation was also observed in tetrahydrofuran (3ab) and in acyclic systems with carbamate, urea, benzamide, and benzylic functionality (3x, 3y, 3z, 3aa).

To provide mechanistic insight, experimental and computational studies were performed. A series of radical trapping experiments were carried out to probe the proposed nickel catalyzed radical-radical cross-coupling pathway (vide infra). Benzyl acrylate was selected as an electron deficient alkene for the Giese reaction²⁵ wherein the α-amino radical addition product was isolated in 30% yield (Scheme 2A). Swapping the radical scavenger to diphenyl diselenide, which has been used as a trapping agent for acyl radicals,²⁶ afforded the acyl-selenide product in 15% yield based on GCMS analysis. These experiments indicated that both carbon-centered radicals were involved in the transformation, and the use of TEMPO as an additive completely inhibited the reaction.

Scheme 2.

Experimental mechanistic investigations.

To further understand roles that Zn and DTBP played in this transformation, several control experiments were conducted. In the absence of either nickel catalyst or Zn, only trace decomposition of DTBP was observed at 50 °C over 16h. Moreover, we were able to correlate the stoichiometry of Zn and product yield based on GC analysis, indicating that Zn played a crucial role in the catalytic cycle rather than merely participating in a reductive event during reaction initiation (Scheme 2B).²⁷ Notably, the rate of consumption of DTBP in the absence of substrates is increased in the presence of water.

With these experimental observations in hand, a plausible Ni^I/Ni^III redox cycle is proposed as an operative pathway for catalytic conversion (Scheme 3). Quantum chemical simulations (see supporting information) revealed a feasible pathway to convert the precatalyst to the Ni^I(O^tBu)(bpp) species (A). A tert-butoxy radical from the decomposition of DTBP allows A to transform to a thermodynamically stable Ni^II(O^tBu)₂(bpp) intermediate B. Potentially, Zn-based reduction of intermediate B back to species A could effectively serve as a redox-buffering process, leading to further consumption of DTBP. Detailed recent studies from Sevov describe electrocatalytic studies of a nickel bpp catalyst system including related one-electron redox conversions.²⁸ The reversible B to A interconversion would increase the concentration of tert-butoxy radicals to enable productive catalysis. The increased rate of consumption of DTBP in the absence of substrates when water is included suggests that water may facilitate the B to A conversion and thus increase the rate of production of tert-butoxy radicals. For the productive pathway, B can undergo radical addition with either 1R or 2R generated from off-cycle HAT processes governed by tert-butoxy radicals or other contributing HAT agents. Trace observed levels of acetone and products from methylation suggest that β-scission of t-butoxy radical occurs at only a low level. The addition with acyl radical 1R was found to be favorable by 13.8 kcal/mol compared to the alternative α-amino radical (2R) addition, largely due to steric interactions with radical 2R in complex B (see supporting information Figure S1). Such energetic differences give rise to the high cross selectivity, as the trapping of B with 1R lowers the concentration of acyl radical in solution,²⁹ kinetically prohibiting the addition of second equivalent of 1R. A similar phenomenon has also been observed by MacMillan and co-workers in their radical sorting pathway.^7b,30 Following the formation of intermediate C, another Zn-based reduction could transform the Ni^III species to a Ni^II intermediate D, which could undergo subsequent α-amino radical (2R) addition. The resulting Ni^III hetero-adduct E undergoes reductive elimination to produce the desired cross-coupled product, while regenerating the nickel complex A. While alkoxide complexes were computed as most energetically feasible, the transient coordination of halogens is also possible. A complete energy profile for this catalytic cycle is presented in the supporting information.

Scheme 3.

Proposed catalytic mechanism involving a Ni^I/Ni^III redox cycle, based on experimental and computational investigations.

In summary, we report a method for nickel-catalyzed N-heterocycle C(sp³)–H functionalization using aldehydes as coupling reagents. We have demonstrated a highly cross-selective dehydrogenative coupling with a broad scope of aldehydes and N-heterocycles. An unusual combination of peroxide and zinc metal has been shown to be crucial for reaction success, with mechanistic investigations supporting the formation of radicals from both substrates generated from t-butoxy radical as the hydrogen atom transfer (HAT) agent. This method offers a strategy for exploring chemical space via a late-stage C(sp³)–H acylation approach. Further mechanistic study and investigation of the factors giving rise to the high cross-selectivity are currently underway.