Photoinduced, Copper-Catalyzed Enantioconvergent Alkylations of Anilines by Racemic Tertiary Electrophiles: Synthesis and Mechanism

Abstract

Transition-metal catalysis of substitution reactions of alkyl electrophiles by nitrogen nucleophiles is beginning to emerge as a powerful strategy for synthesizing higher-order amines, as well as controlling their stereochemistry. Herein, we report that a readily accessible chiral copper catalyst (commercially available components) can achieve the photoinduced, enantioconvergent coupling of a variety of racemic tertiary alkyl electrophiles with aniline nucleophiles to generate a new C–N bond with good ee at the fully substituted stereocenter of the product; whereas this photoinduced, copper-catalyzed coupling proceeds at −78 °C, in the a bsence of light and catalyst, virtually no C–N bond formation is observed even upon heating to 80 °C. The mechanism of this new catalytic enantioconvergent substitution process has been interrogated with the aid of a wide array of tools, including the independent synthesis of proposed intermediates and reactivity studies, spectroscopic investigations featuring photophysical and EPR data, and DFT calculations. These studies led to the identification of three copper-based intermediates in the proposed catalytic cycle, including a chiral three-coordinate formally copper(II)–anilido (DFT analysis points to its formulation as a copper(I)–anilidyl radical) complex that serves as a persistent radical that couples with a tertiary organic radical to generate the desired C–N bond with good enantioselectivity.

Graphical Abstract

INTRODUCTION

Amines play an important role in a wide array of disciplines, including biology, materials science, organic chemistry, and pharmaceutical chemistry.^1–4 Although the nucleophilic substitution of an alkyl halide by an amine is an attractive approach to the synthesis of higher-order amines, traditional S_N2 reactions have limited applicability in the case of less reactive (e.g., hindered and unactivated) electrophiles, which instead engage in undesired side reactions, such as the elimination of H–X.^5,6 Furthermore, conventional substitution pathways almost never enable the control of stereochemistry at the carbon of the new C–N bond, starting with a readily available racemic electrophile.

To address these shortcomings with respect to reactivity and enantioselectivity, a number of laboratories have pursued the use of transition metals to catalyze substitution reactions of secondary and tertiary alkyl electrophiles by nitrogen nucleophiles.^7–13 To date, catalytic enantioconvergent substitutions have only been described for a few families of (mostly secondary) alkyl electrophiles, e.g., allylic electrophiles,¹⁴ α-halocarbonyl compounds,^15–18 α-cyano-α-halocarbonyl compounds,^18,19 propargylic electrophiles,^18,20 benzylic electrophiles,¹⁸ and unactivated alkyl halides that bear a directing group.²¹

Whereas a variety of bioactive molecules include as a subunit an arylamine (aniline) wherein the nitrogen is attached to a stereocenter (Figure 1A),^22,23 to our knowledge there has been only one report of the synthesis of this motif via the enantioconvergent alkylation of an aniline by a racemic alkyl electrophile.²⁴ In the present study, we establish that, with the aid of a chiral copper catalyst and light, anilines can be coupled with a variety of racemic tertiary alkyl electrophiles to generate N-alkylanilines that bear a fully substituted stereocenter with good enantiomeric excess (Figure 1B). Mechanistic studies are consistent with (L1)CuCl acting as a photoreductant and with [(L1)Cu(NHAr)]Cl serving as a key intermediate in the catalytic cycle.

Figure 1.

Anilines wherein the nitrogen is attached to a stereocenter. A) Examples of bioactive compounds. B) This report: Photoinduced, copper-catalyzed enantioconvergent alkylations of anilines by racemic tertiary alkyl electrophiles.

RESULTS AND DISCUSSION

Reaction development.

Seeking to expand the rather limited scope of enantioconvergent substitutions of alkyl electrophiles by nitrogen nucleophiles, we chose to explore reactions of α-halonitriles with anilines to generate α-aminonitriles. In particular, we decided to examine the use of tertiary α-halonitriles as electrophiles so as to produce α-disubstituted α-aminonitriles. Whereas catalytic enantioselective Strecker reactions of aldimines provide a versatile approach to the synthesis of enantioenriched α-monosubstituted α-aminonitriles,^25,26 corresponding reactions of ketimines to afford α-disubstituted α-aminonitriles are less well-developed.^27–31

Upon irradiation (blue LED) in the presence of CuCl and DTBM-SEGPHOS (L1), a racemic tertiary α-chloronitrile undergoes substitution by p-toluidine (1.2 equiv) to furnish the desired α-disubstituted α-aminonitrile in good yield and ee (Table 1, entry 1: 77% yield, 92% ee; CuCl, L1, and BTPP are all commercially available). Control experiments establish that CuCl, L1, and light are critical for coupling under these conditions (entries 2–4; >95% recovery of the alkyl halide). The yield of the reaction is not particularly sensitive to small amounts of water or of air, which have no impact on enantioselectivity (entries 5 and 6).

Table 1.

Effect of reaction parameters. All data are the average of two runs.

When a lower catalyst loading is employed (2.0 mol% copper), a turnover number greater than 25 is observed (entry 7). A discrete, isolable copper–L1 complex, (L1)CuCl (Cu_A)^32,33 may be used in place of CuCl/L1 (entry 8). Whereas the photoinduced, copper-catalyzed coupling occurs in good yield at –78 °C, the corresponding S_N2 reaction does not proceed to a significant extent even at 80 °C (entry 9).

An array of racemic tertiary α-halonitriles serve as suitable electrophiles in these enantioconvergent N-alkylations. For example, consistently good enantioselectivity is observed as the R¹ group varies from Me to i-Bu (Figure 2A, 2a–2c; on a gram scale (1.28 g of product), the coupling to produce 2b proceeds in 64% yield, 92% ee), although the yield is sensitive to the size of R¹. Functional groups such as an ester, olefin, unactivated alkyl chloride, aryl fluoride, aryl chloride, aryl bromide, amide, and phosphonate are compatible with the method (2e–2l; also, an acetal, alcohol, alkylboronate ester, alkyl bromide, alkyl iodide, aryl iodide, aryl triflate, benzofuran, epoxide, ketone, sulfide, and tertiary amine: see the Supporting Information). Not only α-aryl-α-halonitriles (2a–2j), but also α-acyl- and α-phosphonyl-substituted (2k and 2l; Br as the leaving group) α-halonitriles, serve as suitable coupling partners.

Figure 2.

Tertiary α-halonitriles as electrophiles. A) Scope of electrophiles. B) Scope of nucleophiles. Reactions were conducted on a 0.8-mmol scale, and yields are for purified compounds. All data are the average of two runs. X=Cl, unless otherwise noted. ^a X=Br. Bpin, pinacolboryl.

The scope of this photoinduced, copper-catalyzed asymmetric N-alkylation is also reasonably broad with respect to the arylamine nucleophile. In the case of aniline itself, a substantial amount of addition of the electrophile to the para position is observed (C–C coupling: 24% yield, 27% ee; C–N coupling: 18% yield, 86% ee). However, if the para position bears a substituent, moderate-to-good yields and good enantioselectivities are obtained with either an electron-poor or an electron-rich aniline as the nucleophile (Figure 2B, 2m–2r). Furthermore, meta substitution of the aniline can sufficiently impede addition to the para position such that the desired N-alkylation proceeds in fair yield as well as good enantioselectivity (2s–2u).

It is noteworthy that enantioconvergent N-alkylation can be achieved with electrophiles that bear a leaving group other than chloride or bromide. In the case of a carbonate or a fluoride, although substitution does not occur in good yield and ee under the standard conditions, the addition of TBACl/B(OMe)₃ enables the desired C–N bond formation to proceed smoothly with good enantioselectivity (eq 1). To our knowledge, this is the first example of an alkyl fluoride serving as a suitable electrophile in a metal-catalyzed enantioconvergent substitution by a nitrogen nucleophile.

(1)

The standard reaction conditions can be applied to the enantioconvergent N-alkylation of anilines by racemic tertiary electrophiles that lack a cyano group. Thus, α-haloamides are also suitable substrates, leading to coupling with good ee for both N-alkyl and N-aryl secondary amides (Figure 3, 2v–2x). Furthermore, the chiral catalyst can provide promising enantioselectivity when it is necessary to distinguish between two alkyl substituents on the electrophilic carbon, such as secondary vs. methyl (2y) or branched primary vs. methyl (2z).

Figure 3.

Tertiary α-haloamides as electrophiles. Reactions were conducted on a 0.8-mmol scale, and yields are for purified compounds. All data are the average of two runs. X=Cl, unless otherwise noted. ^a X=Br.

Mechanistic studies.

Our current working hypothesis is that these photoinduced, copper-catalyzed enantioconvergent C–N couplings may be proceeding through the pathway outlined in Figure 4. Thus, copper(I) complex Cu_A, which is a competent catalyst for the coupling (Table 1, entry 8; 10 mol% Cu_A, rather than 10 mol% CuCl/12 mol% L1, was utilized in all catalyzed reactions for our mechanistic studies) undergoes excitation upon blue-LED irradiation to generate Cu_A*, which reacts with electrophile R–Cl to afford organic radical R• and copper(II) complex Cu_B. Complex Cu_B then undergoes base-induced substitution by the aniline to furnish copper complex Cu_C, which combines with R• to provide the coupling product and regenerate copper(I) complex Cu_A.

Figure 4.

Outline of a possible mechanism for the photoinduced, copper-catalyzed enantioconvergent alkylation of anilines by racemic tertiary electrophiles. P⏜P = L1

Cu_A.

A solution of Cu_A in toluene at –78 °C shows a characteristic absorption at 376 nm (ε = 3890 M⁻¹ cm⁻¹), as does the reaction mixture (the coupling depicted in Table 1) prior to irradiation (Figure 5A); because the concentrations of copper for the two UV–vis spectra are the same, virtually all of the copper in the reaction mixture appears to be present as Cu_A prior to irradiation (see the Supporting Information). As illustrated in Figure 5B, room-temperature ³¹P NMR spectroscopic analysis is consistent with this conclusion (>95%).³⁴ During a catalytic C–N coupling, the concentration of Cu_A decreases somewhat as the reaction progresses, to 54% of total copper at 64% conversion, according to ¹H NMR spectroscopy at –78 °C (see the Supporting Information).

Figure 5.

Mechanistic studies: Cu_A and Cu_B (in toluene at r.t., unless otherwise noted). A) Absorption and emission spectra of Cu_A (0.5 mM), reaction mixture ([Cu_A] = 0.5 mM, [2-chloro-2-phenylbutanenitrile] = 5 mM, [p-toluidine] = 6 mM, and [BTPP] = 6 mM), blue-LED lamp. B) ³¹P NMR spectroscopy (162 MHz): comparison of a reaction mixture prior to irradiation versus Cu_A. C) Luminescence lifetime of Cu_A* (λ_probe = 480 nm). D) Stern-Volmer quenching of Cu_A* by 2-chloro-2-phenylbutanenitrile (λ_pump = 355 nm, λ_probe = 580 nm). E) Redox potentials (0.1 M TBAPF₆ in THF; scan rate = 20–50 mV/s). F) Steps 1 and 2 of the catalytic cycle: X-band CW-EPR spectra of Cu_B from a photoinduced reduction of 2-chloro-2-phenylbutanenitrile by Cu_A ([Cu_A] = 5 mM and [2-chloro-2-phenylbutanenitrile] = 50 mM; black), independent synthesis of Cu_B (red; X-ray crystal structure: thermal ellipsoids at 50% probability (solvents and hydrogen atoms are omitted for clarity)), and a catalyzed reaction after 30 min at –78 °C (blue); acquisition parameters: MW frequency = 9.37 GHz, MW power = 140 μW, modulation amplitude = 0.4 mT, conversion time = 5.02 ms, and temperature = 77 K.

Having identified Cu_A as a major component of the reaction mixture, we carried out an investigation of its photophysical properties. The steady-state emission spectrum of Cu_A* shows a single emission band (λ_max = 480 nm) at temperatures ranging from 77 K to room temperature (see the Supporting Information). The luminescence decay of Cu_A* observed at 480 nm at room temperature shows two sets of decay curves, with lifetimes of 3.8 ns and 0.36 μs (Figure 5C).

To avoid complications arising from the two competing luminescence-decay pathways, we measured the lifetime of an excited state of Cu_A by transient absorption spectroscopy (λ_pump = 355 nm, λ_probe = 580 nm) as a function of electrophile concentration at room temperature (Figure 5D). Addition of 2-chloro-2-phenylbutane-nitrile results in a decrease of the excited-state lifetime of Cu_A* that allows for the determination of the second-order rate constant, k_q = 3.7 × 10⁸ M^–1 s^–1, for the quenching process. The excited-state reduction potential of Cu_A* is estimated to be E_1/2 (Cu^II/I*) ~ –2.6 V (vs Fc/Fc⁺), based on E⁰⁰ (3.0 eV) and E_1/2 (Cu^II/I; ~ 0.4 V vs Fc/Fc⁺; irreversible), which are derived from emission spectroscopy and electrochemical characterization by cyclic voltammetry (CV); a CV of the model electrophile shows an irreversible feature at E_P ~ –2.2 V (Figure 5E).

Steps 1 and 2 of the catalytic cycle (Figure 4).

No reaction between Cu_A and 2-chloro-2-phenylbutanenitrile occurs in toluene at –78 °C after 30 min in the dark, as determined by ¹H NMR spectroscopy. However, when this mixture is irradiated at –78 °C for 30 min, the initially colorless solution turns dark purple, and analysis by ¹H NMR spectroscopy shows that a C–C coupled dimer derived from the electrophile is formed (Figure 5F: 43% yield based on Cu_A; ~ 1:1 mixture of diastereomers). Formation of the dimer is readily accommodated by the reaction mechanism illustrated in Figure 4; specifically, excitation of Cu_A (step 1), followed by chlorine atom transfer, furnishes R• (step 2), which engages in radical–radical coupling in the absence of aniline (this R–R dimer is observed as a minor side product in the photoinduced, copper-catalyzed C–N coupling of this electrophile under the standard conditions; see the Supporting Information).

Cu_B.

X-band continuous-wave (CW) EPR spectroscopy of the dark-purple solution obtained from this photoinduced reduction of 2-chloro-2-phenylbutanenitrile by Cu_A shows a signal characterized by large ³¹P hyperfine couplings from two phosphorous atoms (Figure 5F, black spectrum). We hypothesized that this paramagnet is (L1)CuCl₂ (Cu_B), which we then independently synthesized from L1 and CuCl₂ (toluene, r.t. to –78 °C) and crystallographically characterized (Figure 5F; see the Supporting Information); the CW-EPR spectrum of this Cu_B at 77 K matches that of the photoinduced reduction of 2-chloro-2-phenylbutanen-itrile by Cu_A (Figure 5F, black and red spectra). Upon monitoring the standard catalyzed C–N coupling reaction (Table 1), we determined that Cu_B is the predominant, but not the sole, paramagnetic species after 30 min of irradiation (Figure 5F, blue spectrum; 82% conversion of the electrophile, 66% yield of the coupling product).

Cu_C and step 3 of the catalytic cycle (Figure 4).

X-band CW-EPR spectroscopic analysis of a catalytic reaction after 10 min of irradiation shows that a different copper(II) species is dominant at the beginning of the coupling process (Figure 6A, red spectrum; 32% conversion of the electrophile, 22% yield of the coupling product; see the Supporting Information). We have determined that a matching EPR spectrum is observed upon treating Cu_B with p-toluidine in the presence of BTPP at –90 °C (Figure 6A, black spectrum; if either p-toluidine or BTPP is absent, the black EPR spectrum is not observed). Under the same conditions, use of a different copper halide complex (Br instead of Cl) or a different base (2-tert-butyl-1,1,3,3-tetramethylguanidine (BTMG) instead of BTPP) leads to no noticeable change in the black EPR spectrum (including superhyperfine structures resolved in their second derivatives; see the Supporting Information), consistent with the paramagnet being a copper(II) complex in which neither the halide nor the Brønsted base is bound (see the Supporting Information). We therefore postulated that this complex might be three-coordinate [(L1)Cu(NHAr)]X (Cu_C), (L1)Cu=NAr, or four-coordinate (L1)Cu(NHAr)₂.

Figure 6.

Mechanistic studies: Cu_C. A) X-band CW-EPR spectra of a catalyzed reaction after 5 min at –78 °C (red) and of Cu_C prepared independently (black); acquisition parameters: MW frequency = 9.36 GHz, MW power = 140 μW, modulation amplitude = 0.4 mT, conversion time = 5.02 ms, and temperature = 77 K. B) X-band CW-EPR spectra (top panel) and 2^nd derivative (bottom panel) of Cu_C in toluene generated from isotopologues of p-toluidine (black traces); acquisition parameters: MW frequency = 9.372–9.374 GHz, MW power = 140 μW, modulation amplitude = 0.1 mT, conversion time = 5.3 ms, and temperature = 77 K. C) Q-band HYSCORE of Cu_C in toluene generated from isotopologues of p-toluidine (left) measured at 1206 mT (g = 2.015) with overlay of ^14/15N or ²H simulation contours (right, red) with experimental contours (right, gray); experimental conditions: MW frequency = 34.005 GHz, τ = 128 ns, t₁ = t₂ = 100 ns, Δt₁ = Δt₂ = 12 ns, shot repetition time (srt) = 1.5 ms, and temperature = 30 K. D) Calculated spin-density plot of Cu_C viewed perpendicular to the NHAr plane and at a 45° angle (bp86 def2-TZVP; contour value = 0.005).

This paramagnetic copper complex (Cu_C) is unstable at –78 °C, decomposing over 30 min (as observed by CW-EPR spectroscopy) and forming (E)-1,2-di-p-tolyldiazene (this dimer is observed as a minor side product in photoinduced, copper-catalyzed N-alkylations of p-toluidine under the standard coupling conditions). Because the thermal instability of Cu_C frustrated our attempts at crystallographic characterization, we investigated its structure through EPR spectroscopy (Figure 6B). The X-band CW-EPR signal of Cu_C is dominated by hyperfine coupling to ^63/65Cu (I = 3/2) and two inequivalent ³¹P (I = 1/2) nuclei, without any additional hyperfine splitting being clearly resolved (see the Supporting Information).

We investigated the incorporation of p-toluidine in the spin system using isotopically labeled p-toluidine (p-toluidine-¹⁵N and p-toluidine-ND₂). The X-band CW-EPR spectra of these isotopologues show only very subtle differences in the second-derivative plots (Figure 6B, bottom panel). For quantitative determination of the natural abundance ¹⁴N and introduced ²H and ¹⁵N couplings, we turned to pulse EPR spectroscopy. Field-dependent Q-band hyperfine sublevel correlation (HYSCORE) spectra of ¹⁵N labeled Cu_C show a single set of elongated correlation ridges in the (–,+) quadrant, indicating that the coupling falls into the strongly coupled regime where A > 2ν_I (Figure 6C, middle panel).³⁵ These features are well simulated by a single class of highly anisotropic ¹⁵N hyperfine tensor with A(¹⁵N) = ±[6, 90, 6] MHz, a_iso(¹⁵N) = ±24.3 MHz. The ¹⁴N signals present in the analogous HYSCORE spectra of natural-abundance Cu_C also fall into the strong-coupling regime but are additionally complicated by the influence of ¹⁴N nuclear quadrupole interaction (Figure 6C, top panel). These spectra are well simulated by scaling the ¹⁵N hyperfine tensor determined from ¹⁵N HYSCORE by the proportion of ¹⁵N/¹⁴N gyromagnetic ratios (|γ¹⁵N|/|γ¹⁴N| = 1.403), with further variation of the ¹⁴N nuclear quadrupole parameters. The simulated ¹⁴N quadrupole parameters are identical to those determined for the amide nitrogen of structurally relevant Ni(III)-amide and -anilido species using similar HYSCORE spectroscopic methods,³⁶ and they are consistent with the presence of a trisubstituted nitrogen with a lone pair oriented roughly orthogonal to the trigonal plane, as expected for a Cu–NHAr moiety. Q-band HYSCORE spectroscopy of ²H-labeled Cu_C shows a single set of intense cross-peaks in the (+,+) quadrant that thus fall into the weakly coupled regime where A < 2ν_I (Figure 6C, bottom panel), which are well-simulated by a single class of relatively anisotropic ²H hyperfine tensor with A(²H) = ±[8.9, 4.3, 1.5] MHz, a_iso(²H) = ±4.9 MHz. The presence of multiple equivalent NHAr ligands could be ruled out by simulations that showed that the equivalent ^14/15N nuclei would generate combination peaks^37,38 that are absent in the observed spectra. Additionally, multiple equivalent ¹H couplings of the class detected directly from the ²H HYSCORE could not be accommodated by simulations of the X-band CW-EPR spectra (see the Supporting Information). Collectively, the spectroscopic data are consistent with the presence of a single trisubstituted nitrogen with a single N–H, i.e., Cu_C as [(L1)Cu(NHAr)]Cl.

Further support for the proposed structure of Cu_C has been obtained by comparing the experimentally derived CW-EPR parameters to values predicted by DFT calculations for the various possible structures. Only three-coordinate [(L1)Cu(NHAr)]Cl is predicted to have EPR parameters similar to those observed (see the Supporting Information).

DFT calculations of Cu_C indicate that there is considerable spin density on the aniline ligand (0.33 e⁻ on NH, 0.32 e⁻ on the aromatic ring) and less spin density on copper (0.15 e⁻) (Figure 6D), indicating that Cu_C is more accurately viewed as a copper(I)–(anilidyl radical) complex, rather than as its formal assignment as a copper(II)–anilido complex.³⁹ The substantial spin density at the para carbon of the aniline (0.17 e⁻) is consistent with our observation of significant C–C bond formation at that position when it is not blocked.^39–41

CONCLUSION

We have developed a photoinduced, copper-catalyzed method for the enantioconvergent N-alkylation of anilines by an array of racemic tertiary alkyl electrophiles that generates fully-substituted stereocenters with good ee. The catalyst, composed of commercially available components, effects asymmetric C–N bond formation at –78 °C, whereas the corresponding uncatalyzed coupling does not proceed at a significant rate even at 80 °C. Although the use of alkyl chlorides as electrophiles is the primary focus of this study, examples have been presented of the use of uncommon related electrophiles, specifically, an alkyl fluoride and an alkyl carbonate, under similar conditions. Mechanistic studies have provided support for key intermediates and elementary steps in the proposed catalytic cycle. Future investigations will explore the development of other enantioselective bond-forming processes that employ catalysts based on earth-abundant copper.