Periselectivity in the Aza-Diels–Alder Reaction of 1-Azadienes with α-Oxoketenes: A Combined Experimental and Theoretical Study

Abstract

Inversions in the periselectivity of formal aza-Diels–Alder cycloadditions between α-oxoketenes generated by a thermally-induced Wolff rearrangement and 1-azadienes were observed experimentally as a function of the α-oxoketene and the 1-azadiene, as well as the reaction temperature and time. Some unexpected inversion in the diastereoselectivity was observed, too. These variations in selectivities were fully rationalized by computational modeling using density functional theory (DFT) methods.

1. Introduction

α-Oxoketenes [1,2,3,4,5,6] were first employed in organic synthesis in 1902 with the pioneering work of Ludwig Wolff on 2-diazo-1,3-dicarbonyl compounds [7]. Since, α-oxoketenes have revealed as versatile chemical intermediates for synthetic applications, ranging from the preparation of simple β-oxocarbonyl derivatives [8] to the total synthesis of complex natural products [4,5]. However, and with few exceptions [9,10], α-oxoketenes are not stable species because of their propensity to react with a number of chemicals, including themselves [11], at significant rates at ambient temperature; thus, they must be generated in situ immediately before use. The most common and practical methods for the generation of α-oxoketenes are the thermal decomposition of dioxinones with concomitant extrusion of acetone [12], and the thermal or photochemical Wolff rearrangement of 2-diazo-1,3-dicarbonyl compounds extruding nitrogen gas [13,14]. α-Oxoketenes 1 are electrophilic species that most commonly react with imines through formal aza-Diels–Alder cycloadditions involving the α-oxoketenes as 4π reaction partners (1-oxadienes) and the imines as the 2π reaction partners to give 1,3-oxazin-4-ones of type 4 (Scheme 1) [1,2,3,4,5,6,15]. However, when α,β-unsaturated imines 2 are employed, another formal aza-Diels–Alder process becomes possible, involving the α,β-unsaturated imines as 4π reaction partners (1-azadienes) and the α-oxoketenes as the 2π reaction partners to give hydropyridin-2-ones of type 3 [16]. Periselectivity, the selectivity in formation of the pericyclic products, in the formal aza-Diels–Alder cycloadditions between α-oxoketenes and some electron rich 2-azadienes was recently examined and found to be dependent on kinetic and thermodynamic factors, as well as the nature of the α-oxoketene employed [11]. For the present study, the periselectivity in the reactions of α-oxoketenes with α,β-unsaturated imines as prototypical 1-azadienes was further investigated and rationalized by a combination of experimental and computational studies.

2. Results and Discussion

We previously reported the formal aza-Diels–Alder cycloaddition between the α,β-unsaturated imine 2a and the five-membered cyclic α-oxoketene 1a derived from the thermal Wolff rearrangement of 2-diazodimedone to give the spiro hydropyridin-2-one 3a in 24% yield after 15 min of reaction at 160 °C (Scheme 2a) [16]. Re-examination of the temperature dependence of the reaction outcome revealed that the other pericyclic product, namely the 1,3-oxazin-4-one 4a, is actually formed when the reaction is performed at 140 °C with a 2:1 3a/4a ratio after 15 min, which allowed isolating 3a in 38% yield and 4a in 21% yield. A similar reaction conducted at 150 °C for 30 min afforded only 3a in 22% isolated yield (3a/4a > 25:1). Stimulated by these observations, some complementary experiments were performed to further illustrate the possible change in periselectivity in the aza-Diels–Alder cycloadditions between α-oxoketenes and 1-azadienes as a function of the reaction temperature (Scheme 2b–e). For instance, the reaction between α-oxoketene 1a and the enantiopure 1-azadiene 2b derived from (–)-perillaldehyde and allylamine at 140 °C for 15 min afforded the corresponding diastereomeric 1,3-oxazin-4-ones 4b in 59% (dr = 1:1) while it was previously shown to afford the spiro hydropyridin-2-one 3b after 15 min at 200 °C [16]. Similarly, it was found earlier that spiro hydropyridin-2-one 3c could be obtained efficiently from the 2-methyl-1-azadiene 2c from a reaction at 160 °C [16], while a mixture of the two possible pericyclic products 3c and 4c was obtained at 140 °C. Some degree of thermodynamic control of the reaction could be demonstrated experimentally in this case: heating a toluene solution of 4c at 160 °C for 15 min produced 25% of 3c together with 20% of 2-methylcinnamaldehyde and unidentified products. Surprisingly, the reactions of α-oxoketene 1a with the 3-chloro 1-azadiene 2d, led exclusively to the corresponding 1,3-oxazin-4-one 4d with no detectable amount of 3d, at 140 °C or 200 °C. The influence of the nature (phenyl vs. methyl) of the substituent at position 4 of the 1-azadiene on periselectivity was briefly evaluated through the reaction of α-oxoketene 1a with 1-azadienes 2e (R = Ph) and 2f (R = Me) at 140 °C for 15 min. It was earlier found that the reaction with 1-azadiene 2e provided the spiro hydropyridin-2-one 3e in 70% yield [16]. The same reaction conducted with azadiene 2f provided only the 1,3-oxazin-4-one 4f albeit in low yield.

The influence of the nature of the α-oxoketene on periselectivity was also evaluated experimentally. The six-membered cyclic α-oxoketene 1b derived from the thermal Wolff rearrangement of 2-diazocycloheptan-1,3-dione was reacted with 1-azadiene 2g at 160 °C to afford cleanly the 1,3-oxazin-4-one 4g, while the same reaction mixture at 200 °C for a prolonged time previously afforded the spiro hydropyridin-2-one 3g in 53% yield (Scheme 3a) [16]. The aza-Diels–Alder reaction of the acyclic α-oxoketene 1c with the 1-azadiene 2h was also examined and found to lead to the 1,3-oxazin-4-one 4h when conducted at 140 °C for 5 min, while the same reaction performed at 200 °C for 15 min resulted in the exclusive formation of iso-3h (Scheme 3b). Significantly, the relative stereochemistry in iso-3h is opposite to the stereochemistry that has been described so far in this and other series of cycloadditions of α-oxoketenes with 4π reaction partners [11,16,17]. Ultimately, it was shown that α-oxoketene 1d derived from the Wolff rearrangement of methyl 2-diazoacetylacetate reacted with the 1-azadiene 2i at 160 °C to give the 1,3-oxazin-4-one 4i in 43% yield, and no reaction at higher temperature was attempted.

These new experimental data indicate that periselectivity in the aza-Diels–Alder cycloadditions between α-oxoketenes and 1-azadienes is generally controlled by competitive kinetic and thermodynamic factors. However, there are some marked differences in reactivity within relatively homogeneous series of substrates, which called for rationalization to enable further developments and applications. To do so, the problem was tackled through computational modeling using density functional theory (DFT) methods. The α-oxoketenes 1b,c and 1e and the model 1-azadienes 2j–m were selected for this part of the work. All calculations were performed using the B3LYP-D3/6-311++G(d,p) level of theory. Free Gibbs energies were generally computed at 298 K corresponding to the temperature of isolation of the products, and additionally at the reaction temperature for kinetically meaningful steps (see the Supplementary Materials for details of the calculations).

First, we examined the reactions between the α-oxoketene 1e and the 1-azadiene 2j to afford either the spiro hydropyridin-2-one 3j or the 1,3-oxazin-4-one 4j (Figure 1a,b). As expected, formation of 3j was found largely thermodynamically favored when compared to the formation of 4j [Δ(ΔG) = −56.1 kJ/mol] with barriers computed at Δ(ΔG^≠) = 12.1 kJ/mol at 298 K and 18.9 kJ/mol at 413 K, which is fully consistent with the experimental results in Scheme 2a. However, the actual aza-Diels–Alder thermodynamic product of the reaction was computationally identified as iso-3j (the diastereomer of 3j) with a relative free Gibbs energy determined at −75.8 kJ/mol, that is more stable than 3j by a tiny but meaningful −2.4 kJ/mol; modeling its formation indicated a relative activation energy of ΔG^≠ = 73.9 kJ/mol at 298 K and 100.5 kJ/mol at 413 K plausible with its formation under the actual reaction conditions (Figure 1c). Because the diastereoselectivity in the model molecule iso-3j has never been reported experimentally in cycloadditions of α-oxoketene 1e with any 4π reaction partner [11,16,17], a competing process is certainly at hand. Actually, the formation of iso-3j through TS6 was found kinetically outcompeted by the cyclodimerization of α-oxoketene 1e (Figure 1c) that occurs with a significantly lower relative activation energy through TS7 to give dimer-1e, a molecule itself amenable to further thermodynamically driven transformations under the reaction conditions [11].

The aza-Diels–Alder reactions of α-oxoketene 1e with the 3-substituted 1-azadienes 2k (Figure 2) and 2l (Figure 3) were next examined in silico for comparison with the experimental results in Scheme 2c,d. As above, formation of the 1,3-oxazin-4-ones 4k and 4l was found kinetically preferred to the formation of the corresponding spiro hydropyridin-2-ones of 3k and 3l, with two competing reaction paths identified for the formation of 4l (Figure 3a). This is in agreement with the experimentally observed formation of both 1,3-oxazin-4-ones 4c and 4d from reactions at 140 °C. However, at 160 °C, the methyl-substituted spiro hydropyridin-2-one 3c (Scheme 2c) corresponding to the model molecule 3k (Figure 2b) was obtained experimentally, while somehow unexpectedly the chloro-substituted spiro hydropyridin-2-one 3d (Scheme 2d) corresponding to the model molecule 3l (Figure 3b) could not be obtained even at 200 °C. The rationalization of this experimental fact came from the comparison of the relative activation energies for the formation of 3k, 3l, and dimer-1e at 200 °C: TS11 leading to 3k competes favorably with the formation of dimer-1e through TS7, though not by much with Δ(ΔG^≠) = −3.5 kJ/mol at 473 K, while TS17 leading to 3l competes unfavorably with the formation of dimer-1e with Δ(ΔG^≠) = 3.4 kJ/mol at 473 K.

To conclude with the modeling of reactions of the cyclic five-membered α-oxoketene 1e with 1-azadienes, the influence of a 4-phenyl or a 4-methyl substituent on the 1-azadiene was investigated to rationalize the results in Scheme 2e. Comparison of the energy profiles of the reactions α-oxoketene 1e with the model 1-azadienes 2j (Figure 1) and 2m (Figure 4) essentially shows that the formation of the model spiro hydropyridin-2-one 3m bearing a methyl substituent is significantly slower than the reaction leading to the model spiro hydropyridin-2-one 3j bearing a phenyl substituent with Δ(ΔG^≠) = 12.4 kJ/mol at 298 K and 34.8 kJ/mol at 413 K (compare TS4 with TS21). The relative increase in the activation energy induced by the 4-methyl substituent is significant and precludes the formation of the corresponding spiro hydropyridin-2-one 3m at the benefit of the α-oxoketene cyclodimer dimer-1e for kinetic reasons. Altogether, the selected DFT model allowed a faultless rational of the periselectivity in the reactions of α-oxoketene 1e with 1-azadienes, which was found to be governed by a combination of kinetic and thermodynamic factors.

Calculations with the cyclic six-membered α-oxoketene 1b (Figure 5) paralleled the previous findings and were also found in full agreement with the experimental results, with the actual thermodynamic diastereomer of the spiro hydropyridin-2-one product being computationally identified as iso-3n, that is not the model molecule corresponding to the experimentally observed diastereomer 3g from the reaction at 200 °C (Scheme 3a). At 200 °C, there are two orders of magnitude between the kinetic constants estimated for the formation of 3n via TS25 and iso-3n via TS27 (circa k = 60 s⁻¹ and 0.56 s⁻¹, respectively), but the formation of iso-3n remains plausible after 45 min at 200 °C. It should, however, be considered that the ratio Int10/Int11 is higher than 10⁴:1 according to the model, which precludes the formation of iso-3n at a significant rate under the examined experimental conditions (relative rates in the magnitude of 10⁶:1). Additionally, the cyclodimerization of α-oxoketene 1b to give dimer-1b through TS28 is also in competition with the formation of iso-3n at 200 °C. On this occasion, again, the selected DFT model allowed for a complete rationalization of the periselectivity in the reactions of α-oxoketene 1b with 1-azadienes, which was found to be governed by a combination of kinetic and thermodynamic factors, as for α-oxoketene 1e.

The calculations in the acyclic series were complicated by the existence of the additional s-trans conformer of the α-oxoketene 1c (Figure 6a) but again were in full agreement with the experiments. As expected, 4o was identified as the kinetic aza-Diels–Alder product in the model reaction between α-oxoketene 1c and 1-azadiene 2j with ΔG^≠ = 70.7 kJ/mol at 298 K and 102.5 kJ/mol at 473 K. As above, the true thermodynamic cycloadduct was identified as iso-3o with notably a more pronounced difference in stabilization energy between the two diastereomers 3o/iso-3o than in the previous cases for 3j/iso-3j (Figure 1) and 3n/iso-3n (Figure 5). In addition, the difference between the activation energies leading to one or the other diastereomer is reduced in this series with Δ(ΔG^≠) = 10.3 kJ/mol at 298 K and 10.9 kJ/mol at 473 K for TS33/TS35. This is to be compared with Δ(ΔG^≠) = 17.0 kJ/mol at 298 K and 17.1 kJ/mol at 473 K for TS4/TS6, and Δ(ΔG^≠) = 18.8 kJ/mol at 298 K and 18.4 kJ/mol at 473 K for TS25/TS27. The difference between the estimated kinetic constants corresponding to TS33 and TS35 is now reduced to a single order of magnitude (circa k = 0.49 s⁻¹ and 0.031 s⁻¹ for TS33 and TS35 at 200 °C, respectively), allowing for both processes to compete at 200 °C. In this case, the cyclodimerization of the α-oxoketene 1c (via TS36) was found not to compete kinetically with the aza-Diels–Alder processes examined. Altogether, the calculations indicate that in the case of the acyclic α-oxoketene 1c, and only in this case, the actual thermodynamic product of the reaction iso-3o can be obtained, which was demonstrated experimentally with the isolation of iso-3h (Scheme 3b).

All the aza-Diels–Alder processes described herein are stepwise processes, occurring first with the formation of the N–C bond to produce a zwitterionic intermediate and then cyclization to the product with formation of the C–C or O–C bond. These processes should thus preferably be referred to as formal aza-Diels–Alder cycloadditions. The physical reason at the origin of this behavior is that the LUMO of all the α-oxoketenes are essentially located over the ketene carbonyl groups, not over the two or four sp² hybridized atoms involved in the cycloaddition, and directed in the mean plane of the α-oxoketenes, not perpendicular to it as in standard Diels–Alder cycloadditions (Figure 7). Actually, the LUMO+1 of the α-oxoketenes are mostly distributed over the α-oxo carbonyl groups and the ketene C=C bonds and oriented perpendicular to the molecular plane [11], but with a significant LUMO/LUMO+1 gap of 0.01453 Ha or 38.1 kJ/mol on average, and no transition states accounting for the symmetry-allowed concerted transformations were located computationally during this work.

In the course of the computational investigations summarized herein, some alternative kinetically, and for the most also thermodynamically, disfavored reaction paths leading to other cycloadducts were identified; this includes β-lactams resulting from formal [2+2] cycloadditions between the C=C bond of the ketene and the C=N bond of the 1-azadiene, and 1,3-oxazin-2-ylidenes resulting from a formal [4+2] cycloaddition between the C=O bond of the ketene and the 4π system of the 1-azadiene.

3. Experimental Section

All reagents were purchased from commercial sources and used without further purification unless otherwise noted. All compounds were weighed and handled in air at room temperature. Microwave-assisted heating was performed using a professional Anton-Paar (Austria) Monowave 300 system in specific sealed tubular reaction vessels. The reactions were monitored by thin layer chromatography (TLC, 60 F₂₅₄) visualized by UV lamp (254 nm or 365 nm). Flash chromatography was performed on silica gel (particle size 40−63 µm). Petroleum ether refers to the fraction with bp = 40–60 °C. NMR spectra were recorded at 300 MHz (¹H) and 75 MHz (¹³C) at 298 K in CDCl₃ using as internal standards the residual non-deuterated signal for ¹H-NMR (δ = 7.26 ppm) and the deuterated solvent signal for ¹³C-NMR spectroscopy (δ = 77.16 ppm). DEPT-135 experiments were used to determine the multiplicity of the ¹³C resonances. Chemical shifts (δ) are given in ppm, coupling constants (J) are given in Hz, and the classical abbreviations are used to describe the multiplicity of the ¹H resonances. Copies of all NMR spectra are provided as Supplementary Materials for this article. High-resolution mass spectra were recorded in triplicate at the Spectropole (https://fr-chimie.univ-amu.fr/spectropole/). 2-Diazo-1,3-dicarbonyls were prepared as previously reported in similar yields and purity [18].

General procedure for the aza-Diels–Alder reactions: a solution of a primary amine (1 equiv) and an α,β-unsaturated aldehyde (1 equiv) in 2–3 mL of anhydrous toluene (ca. 0.4 M) under an argon atmosphere in a microwave dedicated sealed tube containing a Teflon coated magnetic stirring bar was irradiated at 140 °C for 15 min (ramp up time = 2 min), cooled down to 55 °C by an air flow over 5−6 min, and concentrated directly in the reaction vessel to afford the corresponding imine quantitatively as verified by ¹H-NMR analysis. To this material placed under an argon atmosphere was added the diazo compound (1 equiv) and anhydrous toluene (2–3 mL, ca. 0.4 M), the reaction vessel was sealed, and this mixture was irradiated at the set temperature for the time indicated (after a temperature ramp-up time of ca. 2 min). The resulting reaction mixture was cooled down to 55 °C by an air flow over 5−6 min, concentrated, analyzed by NMR to determine the products ratio, and directly purified by flash chromatography when isolation was attempted.

4a: Following the general procedure, the reaction between ortho-nitrocinnamaldehyde (172 mg, 0.97 mmol), furfurylamine (86 μL, 0.97 mmol), and 5,5-dimethyl-2-diazo-cyclohexan-1,3-dione (161 mg, 0.97 mmol) at 140 °C for 15 min afforded the spiro hydropyridin-2-one 3a (147 mg, 38%) as a yellow oil and the 1,3-oxazin-4-one 4a (80 mg, 21%) as a yellow oil. Rf (AcOEt/petroleum ether 4:6) = 0.45. HRMS (ESI+) m/z: [M+H]⁺ calcd for C₂₂H₂₃N₂O₅⁺ 395.1601, found 395.1604. ¹³C{¹H}-NMR (75 MHz, δ ppm/CDCl₃): 165.8 (C), 161.8 (C), 150.4 (C), 147.8 (C), 142.3 (CH), 133.4 (CH), 131.3 (C), 131.0 (CH), 129.1 (CH), 128.9 (CH), 127.2 (CH), 124.7 (CH), 110.5 (CH), 108.8 (CH), 108.3 (C), 88.7 (CH), 46.2 (CH₂), 41.0 (CH₂), 38.6 (CH₂), 36.3 (C), 30.0 (CH₃), 29.9 (CH₃). ¹H-NMR (300 MHz, δ ppm/CDCl₃): 7.99 (dd, J = 1.6, 7.8 Hz, 1H), 7.59 (ddd, J = 1.6, 7.4, 7.4 Hz, 1H), 7.50–7.43 (m, 2H), 7.33 (dd, J = 0.9, 1.5 Hz, 1H), 7.15 (d, J = 15.7 Hz, 1H), 6.32–6.23 (m, 3H), 5.85 (dd, J = 0.9, 6.7 Hz, 1H), 4.90 (d, J = 15.8 Hz, 1H), 4.40 (d, J = 15.8 Hz, 1H), 2.46–2.25 (m, 4H), 1.15 (s, 3H), 1.13 (s, 3H).

4b: Following the general procedure, the reaction between (–)-perillaldehyde (155 μL, 1.0 mmol), allylamine (75 μL, 1.0 mmol), and 5,5-dimethyl-2-diazo-cyclohexan-1,3-dione (166 mg, 1.0 mmol) at 140 °C for 15 min afforded the 1,3-oxazin-4-one 4b (193 mg, 59%, dr = 1:1) as a yellow oil. Rf (Et₂O/petroleum ether 2:8) = 0.18. HRMS (ESI+) m/z: [M+H]⁺ calcd for C₂₁H₃₀NO₂⁺ 328.2271, found 328.2271. ¹³C{¹H}-NMR (75 MHz, δ ppm/CDCl₃): 165.3 (C), 164.9 (C), 163.0 (C), 162.7 (C), 148.8 (C), 148.6 (C), 133.3 (CH), 133.3 (CH), 132.0 (C), 131.8 (C), 128.9 (CH), 128.2 (CH), 117.0 (CH₂), 116.9 (CH₂), 109.0 (CH₂), 109.0 (CH₂), 107.9 (C), 107.7 (C), 91.7 (CH), 91.3 (CH), 46.0 (CH₂), 46.0 (CH₂), 44.8 (CH₂), 44.6 (CH₂), 41.0 (CH₂), 40.9 (CH₂), 40.6 (CH), 40.2 (CH), 36.1 (C), 36.0 (C), 30.2 (CH₂), 30.0 (CH₂), 29.9 (CH₃), 29.9 (CH₃), 29.8 (CH₃), 29.7 (CH₃), 27.0 (CH₂), 26.7 (CH₂), 24.6 (CH₂), 23.6 (CH₂), 20.7 (CH₃), 20.6 (CH₃). ¹H-NMR (300 MHz, δ ppm/CDCl₃): 5.79–5.66 (m, 4H), 5.46 (br, 2H), 5.15–5.07 (m, 4H), 4.10–4.65 (m, 4H), 4.59–4.45 (m, 2H), 3.31–3.20 (m, 2H), 2.32–2.25 (m, 8H), 2.16–1.90 (m, 8H), 1.86–1.80 (m, 2H), 1.69 (s, 6H), 1.50–1.36 (m, 2H), 1.21–1.14 (m, 2H), 1.10 (s, 6H), 1.08 (s, 6H).

4c: Following the general procedure, the reaction between α-methyl-cinnamaldehyde (140 μL, 1.0 mmol), benzylamine (109 μL, 1.0 mmol), and 5,5-dimethyl-2-diazo-cyclohexan-1,3-dione (167 mg, 1.0 mmol) at 140 °C for 15 min afforded a 2:1 mixture of the 1,3-oxazin-4-one 3c and the 1,3-oxazin-4-one 4c (311 mg, 82% total) that we failed at separating by flash chromatography. Some amount of 1,3-oxazin-4-one 4c could be isolated by treatment of the 3c/4c 2:1 mixture by 0.75 equiv NaBH₄ in ethanol followed by standard work-up and flash chromatography eluted with Et₂O/petroleum ether 2:8 because 4c is chemically inert under these conditions. Rf (Et₂O/PE 2:8) = 0.21. HRMS (ESI+) m/z: [M+H]⁺ calcd for C₂₅H₂₈NO₂⁺ 374.2115, found 374.2111. ¹³C{¹H}-NMR (75 MHz, δ ppm/CDCl₃): 165.5 (C), 163.2 (C), 137.4 (C), 135.8 (C), 131.4 (CH), 131.3 (C), 129.0 (CH), 129.0 (CH), 128.5 (CH), 128.5 (CH), 128.3 (CH), 128.3 (CH), 127.8 (CH), 127.8 (CH), 127.3 (CH), 127.2 (CH), 107.7 (C), 93.5 (CH), 46.2 (CH₂), 45.7 (CH₂), 41.1 (CH₂), 36.2 (C), 30.0 (CH₃), 29.8 (CH₃), 13.9 (CH₃). ¹H-NMR (300 MHz, δ ppm/CDCl₃): 7.38–7.19 (m, 10H), 6.37 (s, 1H), 5.65 (s, 1H), 5.32 (d, J = 15.4 Hz, 1H), 3.96 (d, J = 15.4 Hz, 1H), 2.46–2.33 (m, 4H), 1.90 (d, J = 1.2 Hz, 3H), 1.20 (s, 3H), 1.16 (s, 3H).

4d: Following the general procedure, the reaction between α-chloro-cinnamaldehyde (144 mg, 0.86 mmol), benzylamine (94 μL, 0.86 mmol), and 5,5-dimethyl-2-diazo-cyclohexan-1,3-dione (144 mg, 0.86 mmol) at 140 °C for 15 min afforded the 1,3-oxazin-4-one 4d (210 mg, 61%) as a colorless oil. Rf (Et₂O/petroleum ether 4:6) = 0.35. MS (ESI+) m/z: 316 [M+Na]⁺, 332 [M+K]⁺. ¹³C{¹H}-NMR (75 MHz, δ ppm/CDCl₃): 164.8 (C), 161.9 (C), 136.5 (C), 132.9 (C), 129.5 (CH), 129.5 (CH), 129.1 (CH), 128.9 (CH), 128.7 (CH), 128.7 (CH), 128.3 (CH), 128.3 (CH), 128.0 (CH), 128.0 (CH), 127.6 (CH), 127.1 (C), 108.1 (C), 90.3 (CH), 46.2 (CH₂), 46.1 (CH₂), 41.0 (CH₂), 36.3 (C), 29.9 (CH₃), 29.1 (CH₃). ¹H-NMR (300 MHz, δ ppm/CDCl₃): 7.58 (dd, J = 1.7, 7.3 Hz, 2H), 7.41–7.29 (m, 7H), 6.62 (s, 1H), 5.76 (s, 1H), 5.44 (d, J = 15.4 Hz, 1H), 3.99 (d, J = 15.4 Hz, 1H), 2.53–2.29 (m, 5H), 1.19 (s, 3H), 1.13 (s, 3H).

4f: Following the general procedure, the reaction between crotonaldehyde (76 μL, 0.91 mmol), furfurylamine (81 μL, 0.91 mmol), and 5,5-dimethyl-2-diazo-cyclohexan-1,3-dione (152 mg, 0.91 mmol) at 140 °C for 15 min afforded the 1,3-oxazin-4-one 4f (53 mg, 20%) as a colorless oil. Rf (Et₂O/petroleum ether 2:8) = 0.41. HRMS (ESI+) m/z: [M+H]⁺ calcd for C₁₇H₂₂NO₃⁺ 288.1594, found 288.1591. ¹³C{¹H}-NMR (75 MHz, δ ppm/CDCl₃): 165.9 (C), 162.5 (C), 151.0 (C), 142.0 (CH), 133.4 (CH), 124.4 (CH), 110.3 (CH), 108.1 (CH), 107.9 (C), 89.7 (CH), 46.1 (CH₂), 41.1 (CH₂), 38.2 (CH₂), 36.2 (C), 30.0 (CH₃), 29.9 (CH₃), 17.5 (CH₃). ¹H NMR (300 MHz, δ ppm/CDCl₃): 7.31–7.30 (m, 1H), 6.28–6.20 (m, 2H), 5.85–5.73 (m, 2H), 5.56 (d, J = 6.4 Hz, 1H), 4.87 (d, J = 15.7 Hz, 1H), 4.18 (d, J = 15.7 Hz, 1H), 2.37–2.27 (m, 4H), 1.73 (d, J = 6.4 Hz, 3H), 1.13 (s, 3H), 1.12 (s, 3H).

4g: Following the general procedure, the reaction between cinnamaldehyde (126 μL, 1.0 mmol), benzylamine (109 μL, 1.0 mmol), and 2-diazo-cycloheptan-1,3-dione (152 mg, 1.0 mmol) at 160 °C for 15 min afforded the 1,3-oxazin-4-one 4g (crude product that required no purification, 379 mg with purity >90% from the NMR analyses, quantitative) as a brown oil. HRMS (ESI+) m/z: [M+H]⁺ calcd for C₂₃H₂₄NO₂⁺ 346.1802, found 346.1807. ¹³C{¹H}-NMR (75 MHz, δ ppm/CDCl₃): 163.2 (C), 160.7 (C), 137.2 (C), 135.3 (CH), 134.7 (C), 128.2 (CH), 128.2 (CH), 128.2 (CH), 128.2 (CH), 127.8 (CH), 127.4 (CH), 127.4 (CH), 126.9 (CH), 126.5 (CH), 126.5 (CH), 121.8 (CH), 106.5 (C), 86.8 (CH), 45.3 (CH₂), 26.9 (CH₂), 21.6 (CH₂), 21.5 (CH₂), 21.1 (CH₂). ¹H-NMR (300 MHz, δ ppm/CDCl₃): 7.35–7.16 (m, 10 H), 6.56 (d, J = 15.9 Hz, 1H), 6.31 (dd, J = 7.5, 15.9 Hz, 1H), 5.61 (d, J = 7.5 Hz, 1H), 5.11 (d, J = 15.5 Hz, 1H), 4.25 (d, J = 15.5 Hz, 1H), 2.45 (br, 2H), 2.19 (br, 2H), 1.68 (br, 4H).

4h: Following the general procedure, the reaction between cinnamaldehyde (126 μL, 1.0 mmol), n-propylamine (82 μL, 1.0 mmol), and 2-diazo-acetylacetone (126 mg, 1.0 mmol) at 140 °C for 5 min afforded the 1,3-oxazin-4-one 4h (213 mg, 78%) as an orange oil. Rf (EtOAc/petroleum ether 2:8) = 0.27. MS (ESI+) m/z: 294 [M+Na]⁺. ¹³C{¹H}-NMR (75 MHz, δ ppm/CDCl₃): 163.5 (C), 158.0 (C), 135.0 (CH), 135.0 (CH), 128.4 (CH), 128.4 (CH), 128.4 (CH), 126.7 (CH), 126.7 (CH), 122.7 (CH), 105.0 (C), 87.0 (CH), 44.7 (CH₂), 21.5 (CH₂), 16.9 (CH₃), 11.1 (CH₃), 10.1 (CH₃). ¹H-NMR (300 MHz, δ ppm/CDCl₃): 7.38–7.24 (m, 5H), 6.63 (d, J = 15.9 Hz, 1H), 6.33 (dd, J = 15.9, 6.6 Hz, 1H), 5.54 (d, J = 6.6 Hz, 1H), 3.65 (ddd, J = 13.9, 7.7, 7.1 Hz, 1H), 2.96 (ddd, J = 13.9, 7.7, 7.1 Hz, 1H), 1.87 (s, 3H), 1.76 (s, 3H), 1.62–1.49 (m, 2H), 0.86 (t, J = 7.5 Hz, 3H).

iso-3h: Following the general procedure, the reaction between cinnamaldehyde (151 μL, 1.2 mmol), n-propylamine (98 μL, 1.2 mmol), and 2-diazo-acetylacetone (151 mg, 1.2 mmol) at 200 °C for 15 min afforded the hydropyridin-2-one iso-3h (123 mg, 38%) as a yellow oil. Rf (EtOAc/pentane 1:10) = 0.62. HRMS (ESI+) m/z: [M+H]⁺ calcd for C₁₇H₂₂NO₂⁺ 272.1645, found 272.1645. ¹³C{¹H}-NMR (101 MHz, δ ppm/CDCl₃): 208.1 (C), 170.0 (C), 138.4 (C), 129.3 (CH), 129.1 (2CH), 128.9 (2CH), 127.9 (CH), 108.0 (CH), 59.7 (C), 50.4 (CH), 48.7 (CH₂), 30.0 (CH₃), 22.6 (CH₃), 22.0 (CH₂), 11.4, (CH₃). ¹H-NMR (400 MHz, δ ppm/CDCl₃): 7.31–7.24 (m, 3H), 7.15–7.13 (m, 2H), 6.29 (dd, J = 7.9, 0.9 Hz, 1H), 5.28 (dd, J = 8.0, 5.2 Hz, 1H), 3.65–3.50 (m, 2H), 3.46 (d, J = 5.2 Hz, 1H), 1.71 (q, J = 7.5 Hz, 2H), 1.55 (s, 3H), 1.50 (s, 3H), 1.0 (t, J = 7.5 Hz, 3H). The relative stereochemistry in iso-3h was determined by a NOESY experiment.

4i: Following the general procedure, the reaction between cinnamaldehyde (126 μL, 1.0 mmol), allylamine (75 μL, 1.0 mmol), and methyl 2-diazo-acetylacetate (142 mg, 1.0 mmol) at 160 °C for 5 min afforded the 1,3-oxazin-4-one 4i (124 mg, 43%) as a yellow oil. Rf (EtOAc/petroleum ether 2:8) = 0.37. MS (ESI+) m/z: 308 [M+Na]⁺. ¹³C{¹H}-NMR (75 MHz, δ ppm/CDCl₃): 170.6 (C), 167.9 (C), 139.0 (C), 132.3 (CH), 128.7 (CH), 128.7 (CH), 128.6 (CH), 128.0 (CH), 128.0 (CH), 127.4 (CH), 117.5 (CH₂), 108.0 (CH), 54.7 (C), 51.6 (CH₃), 49.9 (CH), 48.7 (CH₂), 21.4 (CH₃). ¹H-NMR (300 MHz, δ ppm/CDCl₃): 7.32–7.15 (m, 3H), 7.14–7.09 (m, 2H), 6.22 (dd, J = 7.8, 2.2 Hz, 1H), 5.93–5.77 (m, 1H), 5.33–5.21 (m, 3H), 4.26 (dd, J = 15.3, 5.6 Hz, 1H), 4,15 (dd, J = 15.3, 5.9 Hz, 1H), 3.66 (dd, J = 3.8, 2.1 Hz, 1H), 3.32 (s, 3H), 1.55 (s, 3H).

4. Conclusions

The formal aza-Diels–Alder cycloadditions of α-oxoketenes with 1-azadienes were examined in detail. Two distinct pericyclic products were obtained experimentally: 1,3-oxazin-4-ones of type 4 when the α-oxoketenes reacted as 4π partners (1-oxadienes) with the C=N bond of the 1-azadienes as 2π components, and hydropyridin-2-ones of type 3 when the 1-azadienes reacted as the 4π partners with the C=C bond of the ketene groups as the 2π components. Generally, it was shown experimentally that 1,3-oxazin-4-ones 4 are the kinetic products, and that hydropyridin-2-ones 3 can be obtained from reactions conducted at higher temperature, indicating some degree of thermodynamic control. However, some small variations in the structure of the α-oxoketenes or the 1-azadienes have led to sharp differences in the periselectivity, as well as inversion of the diastereoselectivity with an acyclic α-oxoketene. Computational modeling of these aza-Diels–Alder cycloadditions by DFT methods confirmed that the 1,3-oxazin-4-ones 4 are invariably the kinetic products, as well as that both diastereomers of hydropyridin-2-ones 3 are thermodynamically favored with, in all cases, the actual thermodynamic diastereomers computationally identified as iso-3. Experimentally, when possible, the reactions with cyclic α-oxoketenes have led exclusively to hydropyridin-2-ones 3, not their diastereomers iso-3. In these cases, the computational model allowed identifying that formation of the thermodynamic product iso-3 is actually kinetically impeded under the examined reaction conditions, notably by the formation of the α-oxoketenes cyclodimers. This situation is different for reactions with an acyclic α-oxoketene, which led experimentally to the thermodynamic product iso-3. In this case, the model shows that formation of iso-3 is now kinetically allowed under the examined reaction conditions. This difference of behavior between cyclic and acyclic α-oxoketenes results from the existence of the additional and more stable s-trans conformation in acyclic α-oxoketenes. All in all, the factors governing periselectivity and diastereoselectivity in the aza-Diels–Alder cycloadditions of α-oxoketenes with 1-azadienes have been fully identified by a combined experimental and theoretical approach.