Bis(imido)vanadium(V)-Catalyzed [2+2+1] Coupling of Alkynes and Azobenzenes Giving Multisubstituted Pyrroles

Abstract

The combination of VCl₃(THF)₃ and N,N-bis(trimethylsilyl)aniline (1a) is an efficient catalyst for the [2+2+1] coupling reaction of alkynes and azobenzenes, giving multisubstituted pyrroles. A plausible reaction mechanism involves the generation of a mono-(imido)vanadium(III) species as an initiation step, where 1a served as an imido source with concomitant release of 2 equiv of ClSiMe₃, followed by a reaction with azobenzene to form a catalytically active bis(imido)-vanadium(V) species via N═N bond cleavage.

Multisubstituted pyrroles are synthetically valuable heteroaromatic compounds in terms of their versatility as building blocks of pharmaceuticals,¹ natural products,² functional materials,³ and dyes.⁴ Although cyclocondensation reactions such as the Paal–Knorr⁵ and Hantzsch⁶ reactions are well-established routes for synthesizing multisubstituted pyrroles, transition-metal-catalyzed multicomponent coupling reactions have been recently investigated as potent alternatives.⁷ Among them, the [2+2+1] cycloaddition reaction of alkynes and primary amines is attractive because many alkynes and primary amines are commercially available or easily synthesized.⁸ As shown in Scheme 1, AgBF₄ catalyzed the oxidative coupling of activated alkynes and primary amines in the presence of stoichiometric amounts of iodobenzene diacetate (Scheme 1a),^8g Cu(OAc)₂ combined with Cs₂CO₃ under O₂ catalyzed the coupling of 2-butynedioate and primary amines (Scheme 1b),^8c and PdCl₂ combined with super-stoichiometric amounts of CuCl and Na₂CO₃ catalyzed the coupling of diarylalkynes and primary amines (Scheme 1c).^8d In these [2+2+1] cycloaddition reactions, activated alkynes bearing ester groups and stoichiometric amounts of oxidant and base were inevitably required.

Scheme 1.

Transition-Metal-Catalyzed [2+2+1] Pyrrole Formation Reactions

Recently, we reported a formal [2+2+1] coupling of alkynes and azobenzenes that was catalyzed by (imido)titanium complexes, either preformed or in situ generated from TiCl₄(THF)₂ and reductant (Scheme 1d).⁹ In an effort to expand the classes of catalysts and utility of this type of oxidative coupling, we herein report that vanadium complexes facilitate the [2+2+1] coupling of alkynes and azobenzenes, where an in situ generated catalyst system of VCl₃(THF)₃ combined with N,N-bis(trimethylsilyl)aniline (1a) showed the best catalytic activity (Scheme 1e). Furthermore, we confirmed that an in situ generated mono(imido)vanadium(III) intermediate cleaves the N═N double bond of azobenzene to give a bis(imido)vanadium(V) active species, which is responsible for the catalytic reaction, in contrast to the previously reported mono(imido)titanium(IV) active species for the Ti(II)/Ti(IV) catalytic cycle. These results highlight that diverse catalyst metals, architectures, and oxidation states can potentially play a role in complex, oxidative heterocycle-forming reactions.¹⁰

We have proposed that the previous Ti catalyst system for the [2+2+1] pyrrole formation reaction involved an (imido) titanium as the catalytically active species.⁹ Accordingly, we sought to find another effective catalyst system by varying transition metal chlorides upon activation by N,N-bis-(trimethylsilyl)aniline (1a), since some imido complexes of early transition metals have been synthesized by treating metal chlorides with 1a with concomitant loss of 2 equiv of ClSiMe₃.^9b,11. Thus, we searched for the best catalyst system for [2+2+1] pyrrole formation by reacting 5-decyne (2a) and azobenzene (3a) in the presence of various metal chlorides (10 mol %) and 1a (10 mol %) in o-dichlorobenzene at 170 °C for 16 h.

The results of our initial catalyst screening are summarized in Table 1. The [2+2+1] coupling product, N-phenyl-2,3,4,5-tetra(n-butyl)pyrrole (4aa), was obtained in 23% yield with TiCl₃(THF)₃ and 1a, whereas the yield of 4aa was 4% for TiCl₄(THF)₂ and 1a (entries 1 and 2);¹² both reactions produced a small amount of alkyne cyclotrimerization product, hexa(n-butyl)benzene. In contrast, VCl₃(THF)₃ quantitatively gave 4aa without any contamination of hexa(n-butyl)benzene (entry 3). CrCl₃(THF)₃ and MnCl₂(THF)_1.5 resulted in no reaction (entries 4 and 5). In the absence of 1a, the reactivity of VCl₃(THF)₃ was significantly suppressed, giving 4aa in 9% yield (entry 6). Among other early transition metal chlorides we screened in Table S1, WCl₆ showed moderate catalytic activity to give 4aa in 65% yield.

Table 1.

Catalyst Screening ^a

entry	cat.	additive	yield [%]b
1	TiCl3(THF)3	PhN(SiMe3)2 (1a)	23
2	TiCl4(THF)2	PhN(SiMe3)2 (1a)	4
3	VCl3(THF)3	PhN(SiMe3)2 (1a)	>99 (71)c
4	CrCl3(THF)3	PhN(SiMe3)2 (1a)	trace
5	MnCl2(THF)1.5	PhN(SiMe3)2 (1a)	n.d.d
6	VCl3(THF)3		9
7	V(═NPh)Cl3	PhN(SiMe3)2 (1a)	>99
8	V(═NPh)Cl3		6
9	V(═NtBu)Cl3	PhN(SiMe3)2 (1a)	98
10	V(═N(2,6-iPr2C6H3))Cl3	PhN(SiMe3)2 (1a)	33

^aReaction conditions: 2a, 3.00 mmol; 3a, 0.500 mmol; o-C₆H₄Cl₂, 2.5 mL.

^bDetermined by GC using pentadecane (internal standard).

^cIsolated yield.

^dNot detected.

Noteworthy was that the yield of 4aa by a preformed (imido)vanadium(V) complex, V(═NPh)Cl₃, was sensitively affected by the addition of 1a: in the presence of 1a, V(═NPh)Cl₃ produced 4aa in quantitative yield, whereas in the absence of 1a, V(═NPh)Cl₃ resulted in poor yield of 4aa (entries 7 and 8). This result indicates that the formation of bis(imido)vanadium(V) species may be important for catalysis (vide infra). Bulkiness of the substituent on the imido group of the precatalyst was another factor to control the catalytic yield: V(═N^tBu)Cl₃/1a showed high yield, while bulkier V(═N(2,6-ⁱPr₂C₆H₃))Cl₃/1a resulted in significantly lower yield (entries 9 and 10). We thus selected VCl₃(THF)₃/1a as the best catalytic system for further substrate scope.¹³

With the best catalytic system in hand, we explored the scope and limitations of alkynes for the vanadium-catalyzed pyrrole synthesis with 3a in C₆D₅Br, and the results of the NMR yields are shown in Table 2. Coupling reactions of symmetrical aliphatic alkynes such as 3-hexyne (2b) and 4-octyne (2c) afforded the corresponding pyrroles 4ba and 4ca in 74% and 92% yield, respectively (entries 1 and 2), whereas the yield of 4da from diphenylacetylene (2d) and 3a was low (entry 3). The unsymmetrical alkyne, 4-methyl-2-pentyne (2e), produced a mixture of 4ea, 5ea, and 6ea in 54% total yield with the ratio of 72:17:11 for the regioisomers (entry 4). When 1-phenyl-1-propyne (2f) was used as a substrate, the corresponding regioisomers of 4fa and 5fa were formed in the ratio of 53:47 in 34% total yield (entry 5). Regioselectivity of this reaction is assumed to be derived from the electronic effect of 2f.^9c In contrast, good steric control of the regioselectivity was observed when 1-trimethylsilyl-1-hexyne (2g) was applied to the reaction, in which the single regioisomer 4ga was obtained in 42% yield (entry 6). Reaction with a bulky terminal alkyne, tert-butylacetylene (2h), also afforded the single regioisomer 4ha in 53% yield (entry 7), whereas all three regioisomers of 4ia, 5ia, and 6ia were detected by using n-butylacetylene (2i) (entry 8); both terminal alkynes 2h and 2i produced significant amounts of alkyne cyclotrimerization byproducts, even in the presence of excess amount of azobenzene in the case of 2i.¹⁴ We further tested the effect of various additives for synthesizing 4aa: ether, iodoarene, nitrile, amide, and triarylphosphine (1 equiv with respect to 3a) were tolerated in the reaction, an improved scope compared to the original Ti system (Table S4).¹⁵

Table 2.

Scope and Limitations of Alkynes ^a

entry	alkyne	yield [%](4:5:6)b
1	R1 = R2 = Et (2b)	74
2	R1 = R2 = nPr (2c)	92
3	R1 = R2 = Ph (2d)	16c
4	R1 = Me, R2 = iPr (2e)	54 (72:17:11)
5	R1 = Me, R2 = Ph (2f)	34 (53:47:0)
6d	R1 = SiMe3, R2 = nBu (2g)	42 (100:0:0)
7	R1 = H, R2 = tBu (2h)	53 (100:0:0)
8e	R1 = H, R2 = nBu (2i)	17 (53:35:12)

^aReaction conditions: 2, 0.988 mmol; 3a, 0.165 mmol; C₆D₅Br, 0.5 mL.

^bNMR yield based on 3a using 1,3,5-trimethoxybenzene (internal standard).

^cIsolated yield.

^dFor 40 h.

^eThe reaction was performed using 1 equiv of 2i and 6 equiv of 3a.

We next examined the effect of the para-substituents of azobenzenes. The coupling reaction with 4,4′-dimethoxyazo-benzene (3b) or 4,4′-dimethylazobenzene (3c) afforded the corresponding pyrroles 4ab and 4ac in 66% and 69% yield, whereas reaction with 4,4′-dibromoazobenzene (3d) or 4,4′-bis(trifluoromethoxy)azobenzene (3e) resulted in the formation of 4ad or 4ae in 91% and 94% yield, respectively (Table S5). Hammett analysis of the catalytic reaction using 3a–e revealed that the reaction rate was accelerated by the electron-withdrawing substituents (ρ = 1.97, Figure S38), being in good accordance with the tendency observed for the Ti-catalyzed system.^9d,15 In addition, a good positive correlation was observed between the initial reaction rate and reduction potentials of azobenzenes, indicating that the reduction potential of azobenzenes is significant to determine the reaction rate (Figure S40).¹⁵

To gain insight into a reaction mechanism, we carried out a kinetic study for pyrrole formation under the optimized reaction conditions where the catalyst was in situ generated by treating VCl₃(THF)₃ with 1a and analyzed the resulting data by variable time normalization analysis.¹⁶ The obtained rate law is provided in eq 1. The reaction obeyed a second-order with respect to the concentration of VCl₃(THF)₃, which was in sharp contrast to the half-order rate dependence on the catalyst concentration for the Ti-catalyzed system,^9d suggesting the involvement of a dimeric vanadium species for azobenzene N═N double-bond cleavage. For the substrates, we found a first-order rate dependence for the concentration of 5-decyne, while the dependency of the concentration of azobenzene was slightly deviated from a first-order rate, being 0.8.

$$\text{Rate}\propto [{\text{VCl}}_3(\text{THE}{)}_3{]}^2[5-\text{decyne}][\text{azobenzene}{]}^{0.8}$$ (1)

During the catalytic reaction, an imidovanadium(III) species is assumed to be responsible for the N═N bond cleavage, giving a bis(imido)vanadium(V) species. We thus examined an in situ generated “V(═NPh)Cl” species for the catalytic reaction. V(═NPh)Cl₃ was treated with an organosilicon reducing reagent, Si-Me₄-DHP,¹⁷ in the presence of 2a and 3a in C₆D₅Br, giving “V(═NPh)Cl”. The stoichiometry of this reaction was clarified by the observation of 2 equiv of ClSiMe₃ and 1 equiv of 2,3,5,6-tetramethylpyrazine (Me₄-pyrazine) by ¹H NMR analysis (Figure S44).¹⁵ Subsequent treatment of the reaction mixture at 145 °C for 20 h resulted in the formation of 4aa in 91% yield, indicating that a similar imidovanadium(III) species may be generated from VCl₃(THF)₃ and 1a in the reaction mixture.

We further examined the reaction of the in situ generated imidovanadium(III) species “V(═NPh)Cl” with 3a in toluene at 80 °C, followed by subsequent addition of PMe₃ to trap a bis(imido)vanadium complex, V(═NPh)₂Cl(PMe₃)₂ (7), in 63% yield (eq 2). In this reaction, half an equivalent of 3a with

(2)

respect to V(═NPh)Cl₃ was consumed. The molecular structure of 7 was confirmed by NMR and single-crystal X-ray diffraction analysis (Figure 1).¹⁸ Complex 7 catalyzed the pyrrole formation in the presence of B(C₆F₅)₃ as a phosphine scavenger in 88% yield (eq 3), clearly indicating that the

(3)

bis(imido)vanadium(V) species is a catalytically relevant species. We also found that complex 7 was generated by treating VCl₃(THF)₃ with 1a in the presence of 3a in C₆D₅Br at 145 °C followed by addition of PMe₃.¹⁵

Figure 1.

Molecular structure of complex 7 with 30% thermal ellipsoids. All hydrogen atoms are omitted for clarity.

On the basis of the kinetic study and the control experiments, we propose a plausible catalytic cycle for the vanadium-catalyzed [2+2+1] coupling of 5-decyne (2a) and azobenzene (3a) (Scheme 2). Reaction of VCl₃(THF)₃ with 1a generates an on-cycle mono(imido)vanadium(III) species A together with a release of 2 equiv of ClSiMe₃ under the reaction conditions. The V(III) species A reductively cleaves the N═N double bond of 3a upon reaction with half an equivalent of 3a, giving a bis(imido)vanadium species B,¹⁹ presumably through a dimerization process similar to that proposed for the Ti system.^9d,e This is consistent with the second-order rate dependence on the concentration of the catalyst and the requirement of the imido ligand for showing high catalytic activity. In addition, formation of B is consistent with the generation of 7 after exposure of VCl₃(THF)₃ to 1a and 3a under the reaction conditions and trapping with PMe₃. One of two V═N moieties in B reacts with alkyne via [2+2] addition to form an azavanadacyclobutene C.²⁰ Next, subsequent insertion of the second alkyne into the V–C bond affords an azavanadacyclohexadiene intermediate D. The formation of azametallacyclohexadienes from imidometal complexes and alkynes was already noted for molybdenum,^21b tungsten,^21b titanium,^21a and zirconium^21c complexes. After the second insertion, reductive elimination from D forms the corresponding pyrrole and mono(imido)vanadium(III) species E, in which the low-valent vanadium center might be stabilized by the coordination of generated pyrrole.²² Odom and co-workers observed the similar reductive elimination from azamolybda- or azatungstacyclohexadiene, giving pyrrole derivatives ^21b. Finally, the pyrrole is dissociated to regenerate a mono(imido)vanadium(III) species A. In fact, no reaction was observed for complex 7 and 3-hexyne (2b) at room temperature in the presence of B(C₆F₅)₃, while we only found the formation of N-phenyl-2,3,4,5-tetraethylpyrrole (4ba) without any intermediates such as C, D, and E upon heating. Based on the kinetic study, the mono(imido)vanadium(III) A is the resting state for generating catalytically active bis(imido) vanadium B via a bimetallic mechanism, and further reaction of B with 1 equiv of alkyne is the rate-determining step in this catalytic reaction. The key differences between this V-catalyzed reaction and the previous Ti-catalyzed reaction is low-valent species as a resting state during this catalytic cycle, which is consistent with the relative reducing ability of Ti(II) and V(III).

Scheme 2.

Plausible Reaction Mechanism for Vanadium-Catalyzed [2+2+1] Coupling of Alkynes and Azobenzenes

In summary, we have demonstrated catalytic [2+2+1] pyrrole formation from alkynes and azobenzenes by VCl₃(THF)₃ upon activation with 1a as an imido source. Control experiments and kinetic studies provided a plausible reaction mechanism, in which bis(imido)vanadium(V) species is assumed to be a catalytic active species. Bimetallic activation of azobenzene by a mono(imido)vanadium(III) species is a key step in the catalytic cycle, as evidenced by the observation of the second-order rate dependence on the concentration of the catalyst. Further development of this vanadium-catalyzed pyrrole formation reaction is ongoing in our laboratories.