In Situ Allene Formation via Alkyne Tautomerization to Promote [4+2]-cycloadditions with a Pendant Alkyne or Nitrile

Abstract

We have explored the net-[4+2]-cycloadditions of a variety of aryl (or alkenyl) alkynes. Tautomerization via base-catalyzed alkyne to allene isomerization produces a transient allene, which undergoes stepwise cyclization with not only a pendant alkyne but with a nitrile as well. The operative mechanisms for these reactions were studied by DFT and compared to the slower, thermal cyclization of the precursor alkyne.

Graphical Abstract

Allenes are involved in a number of cycloisomerization reactions, such as the Garratt–Braverman,¹ Myers–Saito,² and Schmittel³ cyclizations. These processes, depending on the substituents present, often undergo cyclization to benzenoid products via diradical intermediates.⁴ Recently, Liu and coworkers reported a variant that can be viewed as a cyano-Schmittel cyclization using, for example, o-propynylbenzonitriles such as 1 as the substrates, which undergo in situ alkyne to allene isomerization (cf. 2) and cyclization to pyridine-containing products such as 3 (Figure 1a).⁵ A related example is seen in the form of a minor product in a reaction reported by Ollis and coworkers in which the intermediate allene 5 gave rise to the toluene derivative 6 (Figure 1b).⁶ The involvement of an alkyne or nitrile, allene, and ene (C=C of an arene or alkene)⁷ in a net-[4+2]-cycloaddition process could be more broadly classified as a tridehydro-Diels–Alder reaction (TriDDA), given that the operative substrate possesses a total of three sp-hybridized atoms that ultimately reside within the newly formed six-membered ring.⁸

Figure 1.

(a) Cyano-Schmittel cyclization via base catalyzed alkyne to allene isomerization. (b) Cyclization of an alkenyl allene with a pendant alkyne. (c) Current work exploring the net-[4+2]-cycloaddition of allenyl arenes with a pendant alkyne or nitrile.

We have explored an analogous cyclization motif in which the allene substrate, bearing a pendant alkyne or nitrile (8, Figure 1c), was generated in situ. Each of our substrates 7 has a three-atom linker with a central methanesulfonamide. The alkyne to allene isomerization (cf. 7 to 8) was envisioned to set up cyclization to 9.

The first set of reactions we examined was with substrates having the combination of a pendant nitrile and an in situ-generated arylallene (Figure 2a). In order to effect the in situ alkyne to allene isomerization, the non-nucleophilic base DBU was used [as in the analogous pentadehydro-Diels–Alder (PDDA)⁸ and cyano-Schmittel⁵ reactions]. The influence of electronic factors can be seen by comparing the cyclization of the benzonitrile and aniline substrates vs. their phenyl counterpart (cf. 7b and 7c vs. 7a). The enhanced acidification by the electron-withdrawing cyanophenyl substituent presumably lowers the barrier for alkyne to allene isomerization, leading, qualitatively, to a faster reaction.⁹ Electron-donation from the amino substituent, in contrast, slows the rate of cyclization. Ortho substitution on the aryl motif resulted in formation of a single regioisomer, 9d, the isomeric, dearomatized product in which the methyl-bearing carbon participated in the cyclization was not detected. Furthermore, furan and indole were found to be viable heteroaromatic partners in the cyclization, giving rise to 9e and 9f in good yield. In several of these reactions, progress was monitored by in situ ¹H NMR analysis in CDCl₃. Evidence for an intermediate allene species was not detected, suggesting that the initial tautomerization was the rate-limiting step and that the allene, once formed, cyclized rapidly.

Figure 2.

Net-[4+2]-cycloadditions of nitrile-containing substrates with (a) aryl allenes or (b) alkenyl allenes.

The aryl component could be replaced by a simple alkene, as shown in the reactions of 10 and 12 (Figure 2b). The disubstituted alkene in the former cyclizes cleanly to afford, following tautomerization, the o-phenylpyridine derivative 11 in excellent yield. Notably, the trisubstituted alkene 12 cyclizes with the in situ-generated allene to give the non-aromatic product 13 in 74% yield. In this case the quaternary carbon blocks the usual rearomatization process.

Substrates in which the pendant nitrile was replaced with a terminal alkyne were then investigated (Figure 3). These required higher reaction temperatures (i.e., 150 °C) to effect the net-[4+2]-cycloaddition. We hypothesized that DBU was not a sufficiently effective base at promoting the isomerization of substrates 7g-l to the intermediate allene 8-C. Changing the catalyst used for this alkyne to allene isomerization, from DBU to the more basic 1,5,7-triazabicyclo[4.4.0]dec-5-ene (TBD), greatly increased the rate of reaction.¹⁰ Indeed, the phenyl bis-alkyne substrate 7g underwent the requisite isomerization and cyclization at room temperature using TBD. Again, the influence of electronic factors is evident when examining the benzonitrile, aniline, and phthalimide substrates (cf. 7h, 7j, and 7i). In comparison with the parent substrate 7g, the cyano- and phthalimidophenyl substrates (7h and 7i) cyclized faster, while the aminophenyl counterpart 7j required heating at 60 °C to achieve a comparable rate of cyclization. As with 7d to 9d, presence of the ortho methyl group on the aryl moiety in 7l resulted in the formation of a single naphthalene product, 9l.

Figure 3.

Cyclizations involving engagement of an alkyne with a pendant arylallene.

Several ¹H NMR competition experiments were conducted that reveal relevant points. First, a mixture of 7b, 7g, and 7h in d-benzene was treated with TBD (ca. 1 equiv) as the base at ambient temperature. Even at the first time point (4 min), all of 7b had been converted to its product 9b. The cyclization of 7h proceeded with a t_1/2 of ca. 3 min, which was significantly faster than 7g (t_1/2 = 27 h). This supports the qualitative observation that the electron-withdrawing benzonitrile increases the acidity of its propargylic CHs, leading to faster alkyne to allene isomerization. Second, a mixture of 7b and 7h was treated with a sub-stoichiometric amount of TBD (ca. 0.1 equiv) to offer a direct comparison of the initial rates of the allene isomerization of the substrates with a pendant nitrile vs. terminal alkyne. At ambient temperature 7b achieved 10% conversion to give 9b after ca. 6 min, while, in contrast, it took 7h 72 h to reach ca. 10% conversion to give 9h. This comparison demonstrates an interesting and surprisingly large effect that the pendant nitrile has in accelerating the alkyne to allene isomerization, when compared to the analogous pendant alkyne.

The next set of substrates examined (Figure 4) addressed the possibility of cyclization of a different type of substrate motif; namely, one in which a pendant allene might capture a conjugated arylalkyne (cf. 15). It was anticipated that isomerization of the propargyl moiety in a precursor diyne 14 to the allene 15 would be a slower process than for the previous motifs in which the alkyne was further conjugated with π-functionality. The effect of DBU was modest, if any. That is, we observed that the rates of conversion of substrates 14 under purely thermal conditions [classical tetradehydro-Diels–Alder (TetraDDA) reaction]¹¹ were nearly identical to those in the presence of DBU. Isomerization of the terminal alkyne to its allene tautomer at reduced temperature was attempted from 14a using the stronger bases TBD and potassium tert-butoxide. Both lead to poor yields of 9g (32% yield at 60 °C for TBD and 22% at 0 °C → rt for KO^tBu). Nonetheless, the lower temperatures for these reactions indicated that the 14 to 15 to 16 pathway was viable, but that the use of stronger bases came at the expense of significant decomposition.

Figure 4.

Cyclizations involving a pendant enyne in the presence of DBU.

The last substrate class examined (Figure 5) was that having the potential to proceed by a double cyclization. Indeed, the dinitrile 17, containing a 1,4-disubstitued benzene core, efficiently cyclized in the presence of DBU to give both the major phenanthroline and linear pyridoquinoline products, 18-maj and 18-min, respectively (Figure 5a). The structures of these were based on analysis of the ¹³C satellite peaks for the central benzenoid ring aromatic proton resonance in the ¹H NMR spectrum of each. In the case of 18-maj these peaks showed a clear doublet (J = 8 Hz) for the H¹²C=¹³CH coupling, whereas each leg of the analogous satellite resonance in 18-min was a singlet. The somewhat surprising preference for formation of the angular rather than linear skeletal motif¹² stands in contrast to that often observed in classical electrophilic cyclizations where steric factors imposed by a meta-substituent often drive the preference (cf. intramolecular Friedel-Crafts, Bischer-Napieralski, or Nazarov reactions). In the thermal cyclizations here, likely involving a final ring closure of a diradical intermediate (cf. DFT results in Figure 6 discussed below) the steric demands are minimal. In a second example, the double cyclization of the furan-containing precursor 19 afforded the dipyridofuran 20 in excellent yield at room temperature (Figure 5b).¹³ These reactions demonstrate the facile synthesis of compounds having a structural motif that could serve as bidentate ligands.

Figure 5.

Double net-[4+2]-cycloadditions of nitrile-containing substrates on (a) a phenyl core and (b) a furan core.

Figure 6.

DFT calculations of the stepwise net-[4+2]-cycloaddition of the model aryl allene II with a pendant nitrile (energies shown in red) or alkyne (energies shown in blue) in comparison to the stepwise and concerted pathways of the analogous TetraDDA reaction of tautomer I, done at the [SMD(o-dichlorobenzene)/(U)MN15/6-31G(d)] level of theory.

We then undertook DFT calculations to provide better understanding of aspects of the reaction. Our study focused on a comparison of the net-[4+2]-cycloaddition of substrates having a pendant nitrile vs. alkyne. We used the model substrate I in which the Ms groups of 7a and 7g, respectively, were replaced by a simple Me substituent to simply the conformational space that needed to be examined [Figure 6, energies for the nitrile (X = N) vs. alkyne (X = CH) analogs are given in red vs. blue font, respectively]. The isomerization of an alkyne to allene using DBU or TBD, which is likely the overall rate-determining step (r.d.s.), has been studied computationally.^10a Thermodynamically, the allene intermediate II of these model compounds were found to be lower in energy than those of the starting alkynes I.

The first bond formation proceeds with a 20.9 kcal mol⁻¹ barrier to give the diradical intermediate III (X = CH). In contrast, the initial cyclization step with a pendant nitrile was found to proceed with a 17.1 kcal mol⁻¹ barrier. The diradical intermediate III then undergoes a facile, low-barrier second bond formation for both the alkyne- and nitrile-derived species to give the dearomatized, cyclic intermediates IV in a highly exergonic step. Incidentally, we explored this final ring-closure for the competitive formations of 18-maj and 18-min by DFT, using the N-Me analogs of the N-Ms compounds studied experimentally. Using the same computational methodology (now at 70 °C) the difference in activation barriers for the competing ring-closures enroute to either 18-maj or 18-min was 0.9 kcal mol⁻¹, well aligned with the observed 4:1 selectivity.

The activation energies for both a stepwise and concerted TetraDDA reaction of the alkyne and nitrile I were also computed.¹⁴ The barriers for I proceeding via any of TS3–TS5 are quite high. Although we do not know the precise barrier for the I to II interconversion, the fact that these TBD-catalyzed processes occur at room temperature clearly indicates that that barrier is substantially lower than those for the thermal TetraDDA reactions of I itself. It is notable that the free energy of the cyclic allenes VI is > 50 kcal mol⁻¹ higher in energy than those of the unstrained isomeric IV.

We have examined the net-[4+2]-cycloaddition of a variety of arylallenes bearing pendant nitriles and alkynes. The rate of cyclization is accelerated by the base-catalyzed alkyne to allene tautomerization, a process whose rate is significantly affected by electronic factors. DFT calculations support the stepwise, diradical nature of the cyclization and are consistent with the higher activation energy of computed and seen experimentally for the thermal TetraDDA process of the arylalkyne itself. Further, we demonstrated that a double net-[4+2]-cycloaddition can be used to efficiently generate compounds containing intriguing structural motifs.