Solving the puzzling competition of the thermal C²–C⁶ vs Myers–Saito cyclization of enyne-carbodiimides

Introduction

The thermal cyclizations of enediynes [1-6], enediallenes [7-10], bisallenes [11], enyne-allenes [12-17], and enyne-carbodiimides [18-25] have received great interest over the last two decades due to their value for organic synthesis and their interesting mechanistic facets [26]. Like for enyne-allenes, the cyclization mode of enyne-carbodiimides 1 (Scheme 1) can be switched between C²–C⁶ and Myers–Saito reaction channels depending on the substitution at the alkyne terminus, as disclosed by Schmittel [18] and Wang [19]. In 1998, Schmittel et al. [18] reported that the thermolysis of enyne-carbodiimide 1a in toluene/1,4-cyclohexadiene yielded the Myers–Saito product 3a, whereas the thermolysis of enyne-carbodiimide 1b produced the C²–C⁶ product 6b. In 1999 Wang et al. [19] stated that the thermal cyclization of enyne-carbodiimides 1c furnished C²–C⁶ (Schmittel) products 6c in p-xylene under reflux conditions. Based on the experimental findings it was postulated that the cyclization mode was shifted to the C²–C⁶ channel if more sterically demanding substituents than hydrogen were attached to the alkyne terminus, because the resulting strong repulsion of ortho-substituents in diradical 2b,c should disfavor the Myers–Saito pathway. At that time, diradical intermediates were invoked for both pathways based on analogy and computational results [18]. To obtain a deeper understanding of this switch in selectivity, we decided to amalgamate computations and new mechanistic experiments to resolve the puzzle.

For the computational part of this study, we chose first to interrogate enyne-carbodiimides 7a–c (Scheme 2), the synthesis and thermolysis of which have been reported by Wang and coworkers [19]. The methyl (7a) and tert-butyl groups (7b) at the alkyne terminus are ideal to vary the steric influence and the n-propyl group in 7c offers a realistic option for an intramolecular γ-hydrogen abstraction via a six-membered transition state if the diradical is thermally generated. From the computations (vide infra) we conclude that the thermal cyclization of enyne-carbodiimides 7 follows a concerted mechanism along the C²–C⁶ (Schmittel) channel and the decision between the two reaction channels for 7 is not necessarily guided by the free activation energy of the two initial steps, but by the inability of the Myers–Saito diradical to compete in follow-up reactions, either inter- or intramolecular, with the C²–C⁶ (Schmittel) channel.

Results and Discussion

The proper description of organic reactions involving diradical intermediates and transition state structures remains a difficult task in quantum chemistry [27-35]. Singlet diradicals have two electronic configurations that can only be treated accurately by costly multi-reference techniques. An alternative is to apply the unrestricted broken symmetry (U–BS) density functional theory (DFT) approach, which is often in use for larger systems due to its good results [36-49]. Hence, we applied the U–BS DFT approach for the diradical species and the transition states (TSs) connecting them while the restricted method was used for closed shell species. The BLYP functional was chosen for computation because the benchmark studies done by Schreiner showed good applicability of BLYP for the enyne-allene system [3]. Because nitrogen atoms are present in the cumulene core, the enyne-carbodiimide system is expected to be polar and that is why the 6-31+G* basis set was used. All computations were performed in the Gaussian 09 [50] program package. As the reactions take place in a nonpolar solvent (p-xylene), calculations were carried out in vacuum. The nature of all the stationary points was confirmed by frequency analysis. The intrinsic reaction coordinate (IRC) was followed in mass-scaled Cartesian coordinates to validate the TSs. The stability of the wave functions was tested with the stable=opt keyword.

In the computations, we were unable to locate any diradical intermediate along the C²–C⁶ (Schmittel) channel. Rather the results indicate that all enyne-carbodiimides 7a–c follow a concerted intramolecular Diels–Alder (DA) pathway toward the DA products 9_DA (Figure 1), which subsequently rearrange to the isolated final products 8a–c thus avoiding a C²–C⁶ diradical mechanism. As expected due to the strong steric shielding about the tert-butyl group, enyne-carbodiimide 7b exhibits the highest free energy barrier, i.e., 37.8 kcal mol⁻¹ at 138 °C (= the reaction temperature). The free energies of activation for enyne-carbodiimides 7a and 7c are significantly lower at 32.2 and 32.8 kcal mol⁻¹, respectively. Disturbingly, though, the computations predict in all cases a strong kinetic preference for the Myers–Saito (Figure 1) over the concerted C²–C⁶ (DA) pathway. The C²–C⁶ Diels–Alder TS 7a_TS_DA is 3.8 kcal mol⁻¹ higher in energy than the corresponding Myers–Saito TS 7a_TS_MS, while the difference is 4.8 and 4.3 kcal mol⁻¹ for 7b and 7c, respectively. Thus, at this point, the computational results completely contradict the experimental findings. One plausible explanation for the exclusive formation of 8 is that in absence of any efficient hydrogen donor, the kinetically preferred Myers–Saito diradical intermediate 7_INT_MS has to be in a fully reversible equilibrium with the starting material 7, so that eventually product 9 will form. Indeed, at the stage of the Myers–Saito diradical intermediate 7c_INT_MS there are two intramolecular radical quenching processes possible and both exhibit a higher lying barrier than that of the C²–C⁶ cyclization 7c_TS_DA (ΔG^‡ = 32.8 kcal mol⁻¹ – referenced to 7c). The first one invokes a γ–H abstraction at the propyl chain of 7c_INT_MS via the six-membered cyclic TS 7c_TS_γHT (ΔG^‡ = 36.1 kcal mol⁻¹) and the second possibility an intramolecular β–H abstraction via 7c_TS_βHT (ΔG^‡ = 39.9 kcal mol⁻¹). As a consequence, enyne-carbodiimide 7c preferentially cyclizes via the C²–C⁶ pathway to 9c_DA, because the Myers–Saito route with its rate-limiting γ- or β-hydrogen abstraction steps is disfavored on the free energy surface (ΔΔG^‡ = 3.3 and 7.1 kcal mol⁻¹).

After the insightful computational analysis of the reaction mechanism of 7a–c, we chose to use computations in search for an enyne-carbodiimide with a preferred Myers–Saito pathway that would not carry hydrogen at the alkyne terminus. For any suitable candidate the intermediate Myers–Saito diradical should undergo intramolecular hydrogen abstraction at a lower barrier than that of the concerted C²–C⁶ cyclization. In order to realize a facile intramolecular H-abstraction in the Myers–Saito diradical via a 6-membered transition state [26,45], we decided to implement the N,N-dimethylaminomethyl group at the alkyne terminus. Two factors should be favorable for effective hydrogen transfer: (i) the nitrogen atom [51] stabilizes the incipient radical center and (ii) six equivalent hydrogens are available for transfer. Radical stabilization by heteroatoms (with lone pairs) has recently gained interest as a reaction design tool [52,53]. Thus, enyne-carbodiimide 11 with its CH₂NMe₂ group was selected as a target and prepared (vide infra) according to Scheme 3.

Computations at the BLYP/6-31+G* level of theory on the thermal cyclization of 11 (Scheme 4) provided a barrier 11_TS_DA of 29.9 kcal mol⁻¹ for the concerted C²-C⁶ DA cyclization toward 13_DA. Notably, both the Myers–Saito diradical cyclization step via 11_INT_MS (27.1 kcal mol⁻¹) and the follow-up intramolecular hydrogen abstraction 11_INT_MS → 11_INT_γHT are favored kinetically over formation of 13_DA, suggesting that in the thermolysis the Myers–Saito product 12 should prevail. Due to the low free activation energy for the strongly exergonic intramolecular hydrogen abstraction 11_INT_MS → 11_INT_γHT of only 1.5 kcal mol⁻¹, the diradical cyclization step is now rate determining in the Myers–Saito pathway.

The synthetic strategy (Scheme 3) was to first prepare carbodiimide 10 [24] according to a literature procedure. In the next step, a Sonogashira coupling between 10 and N,N-dimethylprop-2-yn-1-amine afforded the desired enyne-carbodiimide 11 in 24% yield. Thermolysis of 11 (Scheme 5) was carried out in presence of 1,4-cyclohexadiene (1,4-CHD) under toluene reflux condition for 12 h. On heating enyne-carbodiimide 11, the characteristic strong IR absorptions of the carbodiimide unit at 2143 and 2123 cm⁻¹ disappeared, which clearly indicated participation of this moiety during the reaction. Conversion of two methyl groups (NMe₂ in 11, 2.23 ppm) to one (NMe in 12, 2.59 ppm) in the ¹H NMR spectrum suggested the involvement of one methyl group in the cyclization process. In combination with above two findings, we initially identified the Myers–Saito cyclization product 12, the assignment of which was further confirmed by ¹H, ¹³C, ¹H,¹H-COSY NMR and also by IR spectroscopy and mass spectrometry. For 14, a broad singlet at 10.02 ppm that is characteristic for its NH proton was observed in the crude NMR mixture. In summary, the Myers–Saito product 12 was produced as major product, 67% yield, whereas only trace amounts of the C²–C⁶ cyclization product 14 arising from 13_DA were obtained. The experimental finding is thus in full agreement with the computational predictions and supports the hypothesis of the Myers–Saito diradical being a kinetically favored species formed in a reversible side equilibrium with the starting enyne-carbodiimide. The diradical can be trapped in good yield by hydrogen transfer and possibly also other reactions (e.g., addition to unsaturated bonds), if this process has a sufficiently low barrier [23].

Conclusion

Experimental and computational techniques have been used to shed light on the C²–C⁶ vs Myers–Saito competition in the thermal cyclization of enyne-carbodiimides. Although in most cases the C²–C⁶ DA cyclization product is furnished exclusively the results suggest that the Myers–Saito diradical intermediate reversibly forms in a kinetically preferred side equilibrium. This diradical intermediate remains a dead-end route as long as there are no powerful inter- or intramolecular hydrogen transfer pathways available. Considering the computational data, however, also other trapping reactions involving the Myers–Saito diradical should be able to compete with the concerted C²–C⁶ (DA) pathway, opening new venues for the preparation of heterocycles from carbodiimides [54].

Experimental

General methods. Anhydrous solvents were used to perform reactions under inert atmosphere (argon). Potassium was utilized to dry tetrahydrofuran (THF) and toluene. Triethylamine (Et₃N) and dichloromethane (DCM) were dried over calcium hydride. Compounds were purified by flash column chromatography (silica gel, 0.035–0.070 mm). To describe NMR signals, the following abbreviations were utilized: s = singlet, d = doublet, t = triplet, dd = doublet of doublets, tt = triplet of triplets, m = multiplet.

2-(3-(Dimethylamino)prop-1-yn-1-yl)-4-methyl-N-((phenylimino)methylene)aniline (11). To a solution of N-phenyl-N´-(2-iodo-4-methylphenyl)carbodiimide [55] (10, 1.78 g, 5.11 mmol) and N,N-dimethylprop-2-yn-1-amine (637 mg, 820 μL, 7.66 mmol) in 50 mL of dry NEt₃ were added PdCl₂(PPh₃)₂ (70.0 mg, 100 μmol) and copper(I) iodide (38.0 mg, 200 μmol) at room temperature under nitrogen atmosphere. The reaction mixture was stirred overnight. Then the solvent was removed under reduced pressure and diethyl ether (40 mL) was added. The suspension was filtered through celite and the filtrate was washed with water and dried over MgSO₄. After removal of the solvent, column chromatography over silica gel using (n-hexane/ethyl acetate, 1:1, R_f = 0.28) furnished pure product 11 as a dark-green oil. Yield: 354 mg (1.22 mmol, 24%). T_onset = 119 °C (DSC). ¹H NMR (400 MHz, CDCl₃) δ 2.25 (s, 6H), 2.30 (s, 3H), 3.14 (s, 2H), 7.02 (d, ³J = 8.2 Hz, 1H), 7.07 (dd, ³J = 8.2 Hz, ⁴J = 1.6 Hz, 1H), 7.16 (tt, ³J = 7.4 Hz, ⁴J = 1.2 Hz, 1H), 7.22–7.25 (m, 2H), 7.28 (d, ⁴J = 1.6 Hz, 1H), 7.30–7.34 (m, 2H); ¹³C NMR (100 MHz, CDCl₃) δ 20.7, 44.1, 48.4, 81.7, 92.2, 119.5, 124.2, 124.3, 125.2, 129.4, 130.0, 133.8, 134.1, 135.1, 136.4, 139.3; IR (KBr) : 2976 (w), 2944 (w), 2864 (w), 2825 (w), 2780 (w), 2143 (s, N=C=N), 2123 (s, N=C=N), 1593 (s), 1519 (m), 1488 (s), 1355 (w), 1324 (w), 1284 (w), 1252 (m), 1213 (s), 1034 (m), 908 (s), 822 (m) cm⁻¹; EIMS (70 eV) m/z (%): 289 (29, M⁺), 288 (99, M⁺ − H), 274 (54, M⁺ − CH₃), 246 (47), 245 (100, M⁺ − N(CH₃)₂), 244 (20), 243 (26), 231 (14), 218 (15), 203 (18), 183 (62); ESIMS: calcd. for (C₁₉H₂₀N₃)⁺, 290.2; found, 290.2.

3,7-Dimethyl-1-phenyl-1,2,3,4-tetrahydropyrimido[4,5-b]quinoline (12). To 10 mL of dry degassed toluene was added 2-(3-(dimethylamino)prop-1-yn-1-yl)-4-methyl-N-((phenylimino)methylene)aniline (11, 40.0 mg, 138 µmol) and 1,4-CHD (448 mg, 530 μL, 5.60 mmol). The reaction mixture was allowed to reflux for 12 h. After the solvent was removed under reduced pressure, purification was performed by preparative silica gel TLC using n-hexane/ethyl acetate (2:3, R_f = 0.22). Yield 27.0 mg (93.0 µmol, 67%); white solid; mp 124–126 °C; ¹H NMR (400 MHz, CDCl₃) δ 2.43 (s, 3H), 2.59 (s, 3H), 4.13 (s, 2H), 4.67 (s, 2H), 7.21 (tt, ³J = 6.4 Hz, ⁴J = 2.1 Hz, 1H), 7.29 (dd, ³J = 8.6 Hz, ⁴J = 2.0 Hz, 1H), 7.33 (br s, 1H), 7.39–7.45 (m, 4H), 7.52 (d, ³J = 8.6 Hz, 1H), 7.56 (s, 1H); ¹³C NMR (100 MHz, CDCl₃) δ 21.4, 40.6, 55.8, 73.0, 117.2, 124.3, 125.0, 125.8, 125.8, 126.7, 129.0, 131.0, 132.3, 133.6, 144.3, 145.1, 151.7; IR (KBr) : 2945 (m), 2870 (w), 1630 (m), 1594 (m), 1496 (s), 1488 (s), 1455 (m), 1412 (m), 1343 (s, C-N), 1312 (m), 1296 (w), 1225 (w), 1140 (w), 1123 (m), 1068 (m), 1047 (w), 897 (w), 819 (m), 746 (m), 692 (m) cm⁻¹; EIMS (70 eV) m/z (%): 289 (60, M⁺), 288 (66, M⁺ − H), 245 (100, M⁺ − H − CH₂NCH₃), 244 (11), 91 (17); ESI: calcd. for (C₁₉H₂₀N₃)⁺, 290.2; found, 290.2.