Highly Regio- and Stereoselective Diels-Alder Cycloadditions via Two-Step and Multicomponent Reactions Promoted by Infrared Irradiation under Solvent-Free Conditions

Infrared irradiation promoted the Diels-Alder cycloadditions of exo-2-oxazolidinone dienes 1–3 with the Knoevenagel adducts 4–6, as dienophiles, leading to the synthesis of new 3,5-diphenyltetrahydrobenzo[d]oxazol-2-one derivatives (7, 9, 11 and 13–17), under solvent-free conditions. These cycloadditions were performed with good regio- and stereoselectivity, favoring the para-endo cycloadducts. We also evaluated the one-pot three-component reaction of active methylene compounds 20, benzaldehydes 21 and exo-2-oxazolidinone diene 2 under the same reaction conditions. A cascade Knoevenagel condensation/Diels-Alder cycloaddition reaction was observed, resulting in the final adducts 13–16 in similar yields. These procedures are environmentally benign, because no solvent and no catalyst were employed in these processes. The regioselectivity of these reactions was rationalized by Frontier Molecular Orbital (FMO) calculations.

In this context, and as part of our ongoing research into the use of infrared irradiation as the energy source to promote organic reactions, we herein describe a convenient and versatile synthesis of the new substituted tetrahydrobenzo [d]oxazol-2-one derivatives 7, 9, 11 and 13-17, starting from the exo-2-oxazolidinone dienes 1-3 and the Knoevenagel adducts 4-6 (Tables 1 and 2), as the dienophiles, in the Diels-Alder cycloadditions promoted by infrared irradiation, under solvent-free conditions. Moreover, we also carried out an evaluation of how the reactivity and stereoselectivity of these cycloadditions are affected by the structural modifications in the diene, as well as in the Knoevenagel adducts, such as the replacement of the cyano group by the ethoxycarbonyl group (Tables 1 and 2). In addition, we studied the one-pot three-component reactions to obtain the same cycloadducts starting from methylene active compounds 20a-c, benzaldehydes 21a-d and exo-2-oxazolidinone diene 2 via a cascade Knoevenagel/Diels-Alder process under similar reaction conditions.
We explored synthetic access to the tetrahydrobenzo [d]oxazol-2-one derivatives 7-11 and 13-18, in search of infrared irradiation as a viable promoter of the Diels-Alder cycloadditions, in a two-step synthesis, starting from the exo-heterocyclic dienes 1-3 and the Knoevenagel adducts 4-6.
Initially, the unsubstituted exo-heterocyclic diene 1 was evaluated in terms of reactivity and regioselectivity in the Diels-Alder additions toward derivatives 4a-e, which bear activating substituents such as ethoxycarbonyl (R 1 ) and cyano (R 2 ) groups. Thus, a mixture of diene 1 and olefin 4a (1:1.2 mol-equiv., respectively) was irradiated with an infrared lamp [33] at 50 °C for ca. 3.5 h, under solvent-free conditions, leading to the total conversion of 1 to afford 7a, judging by the 1 H NMR analysis of the crude reaction mixture, as a single regioisomeric product in 73% yield. This high regioselectivity contrasts with that observed for the thermal Diels-Alder reaction of 1 with monosubstituted dienophiles, such as methyl vinyl ketone and methyl propiolate, in which the para/meta regioisomeric ratios were lower (from 1:1 up to 8:2) [17].
The structure of compound 7a was established by spectroscopic analysis. The spectrum of High Resolution Mass Spectrometry (HRMS) showed exactly the expected mass (m/z 388.1423); while the IR spectrum showed two carbonyl absorption bands (C=O) at 1757 and 1713 cm −1 and a cyano group absorption at 2362 cm −1 . The 1 H and 13 C NMR spectral data are consistent with the tetrahydrobenzo [d]oxazol-2-one skeleton. It is interesting to note the large difference in the chemical shifts (δ) of the diastereotopic CH 2 protons at the C-4 position of the cyclohexene ring, since H-4β appeared at 2.65 ppm as a ddd (J = 17.1, 4.8, 1.5 Hz) due to the geminal, vicinal and homoallylic couplings, respectively; while the signal due to H-4α appeared at 3.13 ppm as a dddd (J = 17.1, 11.4, 4.2, 2.1 Hz). The large difference in the δ value for these protons could be ascribed to the anisotropic effect of the phenyl groups at N-3 and C-5. Decoupling and Nuclear Overhauser Effect (NOE) experiments provided additional support for the structure: H-4α showed a three-bond coupling with H-5 ( 3 J 4-5 = 11.4 Hz); while the signal of protons H-5 (3.46 ppm) and H-4α (3.16 ppm) were enhanced when H-4β (2.65 ppm) was irradiated ( Figure 1). Interestingly, when the reaction was carried out under thermal (50 °C) and solvent-free conditions, the reaction time was longer and the yield lower (Table 1, entry 2). In an attempt to further improve the yield, under thermal conditions (50 °C), benzene and tetrahydrofuran were used as solvents, without yielding better results (Table 1, entries 3 and 4).
Comparing the reaction times (Table 1, entry 1 vs. entries 2-4), it appears that under infrared irradiation the reaction was substantially faster (~3.5 h) and the yield was higher (73%). As for the regioselectivity, it was comparable in both cases, only affording regioisomer 7. Analysis of the crude reaction mixture by 1 H NMR did not show evidence of regioisomer 8.
To assess the effect of the substituent R 3 in the aromatic ring of the dienophiles on the reactivity and the regioselectivity, several analogues using both electron-poor and electron-rich substituents in 4b-e, were used. When 4b, bearing an electron-releasing group, was irradiated in the presence of 1, the conversion rate slightly decreased (Table 1, entry 5), giving 7b in a lower yield (50%), together with recovered dienophile 4b (50%), However, with the use of dienophile 4d, containing an electron-withdrawing group, a higher yield of the corresponding adduct 7d was obtained. The reactivity trend of the Diels-Alder cycloaddition of dienophiles 4a-e with 1 ( Table 1, entries 4-8) met the expectations of a normal electron-demand process [34].
Cycloadduct 7e was isolated as yellow crystals (EtOAc/hexane, 8:2) and its para regiochemistry (as considered for the relative orientation in the cyclohexene ring between the nitrogen atom and the electron-withdrawing groups of the dienophile) was confirmed by X-ray crystallography ( Figure 2). The X-ray structure shows that the aryl groups in N-3 and C-5 are almost perpendicular to the heterocycle and to the cyclohexene ring, respectively, presenting the following consistent torsion angles: −59.7(2)° for C(3a)-N(3)-C(8)-C(9) and −129.20(13)° for C(4)-C(5)-C(12)-C(13). Complementarily, with the aim of exploring the scope and limitations of the process, as well as of detecting the effect on the cycloadducts induced by the change of the substituents R 1 and R 2 in the dienophiles, the ethoxycarbonyl group in 4 (R 1 = CO 2 Et) was replaced by a CN group and the cyano group (R 2 = CN) by an ethoxycarbonyl group, to produce a series of benzylidenemalononitriles 5a-d (R 1 = R 2 = CN) and diethyl 2-benzylidenemalonates 6a-c (R 1 = R 2 = CO 2 Et), respectively. The reactions were performed under identical conditions to those used for 1 and 4a. The reaction of diene 1 with these two series of analogous dienophiles 5a-d and 6a-c yielded cycloadducts 9a-d and 11a-c, respectively. The fact that in both cases the product was a single para regioisomer indicates a similar behavior in the reactions. It is noteworthy that 1 H NMR analysis (300 MHz) of the crude mixtures did not give evidence of the presence of the corresponding regioisomers 10 and 12.
The best yields of the tetrahydrobenzo [d]oxazol-2-one derivatives 9 and 11 corresponded to the reactions between exo-heterocyclic diene 1 with benzylidenemalononitriles 5a-d ( Table 1, entries 9-12). In contrast, with the reactions between 1 and the ethyl (E)-2-cyano-3-phenylacrylates 4a-e, the corresponding yields of derivatives 7a-e were lower (Table 1, entries [1][2][3][4][5][6][7][8]. When the sterically more demanding diethyl 2-benzylidenemalonates 6a-c were used, the yields of adducts 11a-c were the lowest of all (Table 1, entries [13][14][15]. The reactivity trend found for the Knoevenagel dienophiles can also be explained by the higher electron-withdrawing effect of the cyano group in comparison with the ethoxycarbonyl group [35]. In accordance with previous reports, the regiochemistry of these cycloadditions mainly depends on the electron-donating effect of the nitrogen atom of the heterocycle ring of the diene [16]. However, the exclusive formation of the para regioisomer in our case contrasts with the tendency of the exo-heterocyclic diene 1 to produce a mixture of para/meta regioisomers [17]. This is probably due to the fact that the dienophiles used in the present work are geminally substituted by two electron-withdrawing groups, which enhance the reactivity and, consequently, the regioselectivity [36].

Diels-Alder Cycloaddition with Dienes 2 and 3
In order to evaluate the effect of the substituent in the exo-heterocyclic diene on the reactivity and selectivity in the course of the Diels-Alder reaction, dienes 2 and 3, bearing a methyl group in the double bond, were added to dienophiles 4-6.
The endo/exo ratios of adducts 13a-e/14a-e were determined by integration of the double signals of the methyl groups C-16 in the 1 H NMR spectra of the crude mixtures ( Table 2, entries 1-5). The separation of these mixtures was achieved by column chromatography on silica gel using hexane as eluent. The structural elucidation of the main products 13a-e was made on the basis of their spectroscopic data (NMR, HRMS and IR). All the data are consistent with the substituted tetrahydrobenzo [d]oxazol-2-one skeleton of 13a-e. The 1 H NMR spectrum of 13a shows the presence of ten aromatic protons at 7.26-7.51 ppm, a quartet integrating for two protons (OCH 2 CH 3 ) at 4.09 ppm, and two overlapped signals attributed to H-7 and H-5 protons at 3.49-3.54 ppm. The proton H-4α appears as a doublet of doublets of doublets (J = 17.1, 11.1, 1.8 Hz) at 2.93 ppm; while the proton H-4β appears as a doublet of doublets (J = 17.1, 5.4 Hz) at 2.62 ppm. There is a signal at 1.35 ppm as a doublet integrating for three protons (H-16) and at 1.11 ppm (OCH 2 CH 3 ) as a triplet.
The 13 C NMR spectrum of 13a displays signals for two carbonyl groups at 164.7 ppm (CO 2 Et) and 154.0 ppm (C-2), ten signals for vinyl and aromatic carbons at 138.0-120.0 ppm, one signal corresponding to the cyano group at 118.2 ppm, and seven signals at 62.9, 52.2, 40.4, 37.0, 27.0, 15.9 and 13.7 ppm for sp 3 carbon atoms. The attributions of the signal were supported by 2D experiments such as Heteronuclear Multiple-Quantum Coherence (HMQC) and Heteronuclear Multiple-Bond Coherence (HMBC).
The relative configuration at C-5, C-6 and C-7 of 13a was determined by NOE experiments (Figure 3), where an enhancement of the signals of protons H-5 and H-4α was observed when the signal of H-4β was irradiated. Likewise, when H-5 was irradiated, an NOE effect was observed for H-4β and H-13. The irradiation of H-4α induced an NOE effect on the signals of H-4β, H-9 and H-13. An enhancement of the signals of protons H-5 and H-7 was observed when the signal of H-16 was irradiated. These data support a syn relationship between H-4β, H-5 and H-16 protons, and justify assigning the structure of the compound 13a as the endo cycloadduct. The assignment of the stereochemistry of compound 13a was confirmed by X-ray crystallography ( Figure 4). The phenyl and ethoxycarbonyl groups at the stereogenic C-5 and C-6 centers have a trans diequatorial orientation. The torsion angle C(12)-C(5)-C(6)-C(6a) of 48.42(18)° supports the gauche conformation for the phenyl and ethoxycarbonyl groups. Meanwhile, the methyl group at the stereogenic C-7 center has a pseudoaxial orientation. Therefore, in the solid state, the carbocyclic six-membered ring adopts a half-chair conformation. It is likely that the presence of an electron-donating methyl group in 2 greatly polarizes the π-system of the diene, giving rise to the major para regioisomers 13a-e/14a-e. The endo preference might be due to both steric and electronic factors which favor the endo transition state (vide infra).
In contrast, the reactions of diene 2 with dienophiles 6a-c (R 1 = R 1 = CO 2 Et), under the same experimental condition, provided single diastereoisomers 17a-c in low yields ( Table 2, entries 11-13), due in part to the self-dimerization of diene 2 to adduct 19, isolated as a by-product [17]. These results indicate that dienophiles 6a-c are less reactive and more stereoselective that dienophiles 4 and 5.
It appears that these reactions are sterically sensitive, since the use of the more hindered dienophiles 6a-c afforded the corresponding products 17a-c in the poorest yields, although with a better stereoselectivity.
As shown in Tables 1 and 2, the reaction times for diene 2 were similar to those employed for diene 1. This is rather unexpected as previously mentioned [17], since the electron-releasing effect of the methyl substituent of diene 2 should increase the reactivity in Diels-Alder additions according to Alder's rule. This behavior is also presumably due to the steric effect.

Multicomponent Reactions
In recent years, the development of multicomponent reactions in order to produce biologically active compounds has been accelerated and thus has become a very important area of research in organic and medicinal chemistry.
As an attempt to obtain compounds 13-18 more efficiently, we turned our attention to a one-pot procedure. Our synthetic strategy was based on the knowledge that the dienophiles 4 and 5 are accessible through a simple Knoevenagel condensation between compounds 20a-b and benzaldehydes 21a-d [27][28][29], followed by a subsequent Diels-Alder cycloaddition with diene 2, to generate cycloadducts 13-16.
Initially, in this multicomponent approach, a mixture of ethyl 2-cyanoacetate (20a), benzaldehyde (21a) and diene 2 was reacted in a 1:1:1 (mol-equiv.) ratio under infrared irradiation and solvent-free conditions. After 35 min, this reaction led to the desired mixture of tetrahydrobenzo [d]oxazol-2-ones 13a/14a (65:35), albeit in moderate yield (55%), along with some amount of 4a and 19 (40% and 5%, respectively). It is worth noting that the regio-and stereoselectivity was similar ( Table 3, entry 1) to those found in the previous methodology ( Table 2, entry 1). The Knoevenagel adduct 4a was detected from the crude mixture by H 1 NMR analysis, which supports the idea that its initial formation was accomplished before the intermolecular Diels-Alder reaction with diene 2 took place, to give the corresponding adducts.
A similar behavior was observed for the analogous substrates 21b-d with 20a and diene 2, since the cycloadduct mixtures 13b/14b, 13c/14c and 13d/14d were obtained in comparable yields (40-64%) to those obtained via the two-step procedure, confirming the efficiency of the multicomponent approach (Table 3,

entries 2-4).
On the other hand, the multicomponent reaction between malononitrile 20b, benzaldehydes 21a-d and diene 2 produced fairly good yields of cycloadducts 15a-d/16a-d. However, in the presence of diethyl malonate (20c), no domino Knoevenagel condensation/Diels-Alder cycloaddition was observed at all. When the reaction temperature was increased to 80 °C, compound 19 was obtained instead of the expected adducts 17/18. These results revealed that the dimerization of 2 is also promoted by IR irradiation to yield 19. The structure of the latter was established by spectroscopic data and corroborated by the study of X-ray diffraction ( Figure 5). Previously, we observed the dimerization of diene 2 under thermal conditions (xylene, 120 °C, 10 h) [17]. Comparing the NMR data of these compounds, we found that there were notable differences in chemical shifts, as well as in the difference in their melting points (196-198 °C and 243-244 °C), which suggests that this dimer corresponds to different diastereoisomer. This result can be attributed to the probable influence of infrared radiation as a source of energy. The higher reactivity of ethyl 2-cyanoacetate (20a) and malononitrile (20b) in comparison with diethyl malonate (20c), which successively leads to the Knoevenagel condensation and Diels-Alder reaction under infrared irradiation conditions, may be explained in terms of the difference of acidity constants of the activated methylene: 20c (pK a = 13) [37], 20b (pK a = 11) [38] and 20a (pK a = 9) [39]. This acidity can affect the formation of the Knoevenagel products and, consequently, the final adduct. In addition, these results also suggest that the steric hindrance generated by the ethoxycarbonyl group seems to play a role in controlling the domino reactions and therefore in providing acceptable yields.

Diels-Alder Regioselectivity and FMO Theory
The regioselectivity of the Diels-Alder additions of dienes 1-3 to dienophiles 4-6 was rationalized in terms of the FMO theory [34]. The geometries of dienes 1 and 2 were previously calculated [17], while the geometries of diene 3 and dienophiles 4-6 were calculated using the B3LYP/6-31G** method [40][41][42] without any symmetry constraints calculation, and employed as the starting point for the ab initio molecular orbital calculations, using the RHF/6-31G** basis set [43]. It is noteworthy that for derivatives 6, the cis ethoxycarbonyl group to the aryl ring adopts a preferential non-coplanar conformation, leaving the trans acrylate moiety in conjugation with the aromatic substituent. This conjugation is also observed for derivatives 4. This is probably due to the fact that in this conformation the aryl ring is maintained coplanar to the acrylate conjugated π-system, giving rise to a higher stability.
By using the same basis set, the energies of the FMO were calculated for both dienes and dienophiles (Table 4). Since, in the entire series lower, energy gaps were calculated for the interaction between HOMO diene -LUMO dienophile (Normal Electronic Demand) than between the opposite interaction LUMO diene -HOMO dienophile (Inverse Electronic Demand), as illustrated by some examples in Table 5, it is then expected that the reaction is conducted under the former interaction. Table 4. Ab initio 6-31G** calculations of energies (eV) and coefficients (C i ) of the frontier molecular orbitals for dienes 1-3 and dienophiles 4-6 a .   As expected, the methyl group attached to the diene moiety in dienes 2 and 3 induced an increase of the energy of the HOMO, with respect to the energy of the unsubstituted diene 1. Hence, the reactivity of dienes 2 and 3 should be higher than that of diene 1, as observed for the cycloadditions with mono substituted dienophiles [17]. Nevertheless, in the case of dienophiles 4-6, the reaction times are very similar for all the dienes (Tables 1 and 2), which indicates a similar reactivity as well. It is likely that other factors are involved, such as the steric hindrance generated between the dienes and the substituents in dienophiles 4-6. Although these factors are not sufficiently important to modify the regioselectivity, which is para in the whole series, the preference for the anti relative configuration between the methyl group and the aryl ring in adducts 13-17 seems to support their existence. Moreover, in spite of the presence of electron-withdrawing groups in the aryl ring of the dienophiles, such as the nitro group, which may induce a higher reactivity and higher selectivity [34,36], there is no correlation between the stereoselectivity and the structure of the dienophiles bearing other substituents. Once again, this suggests the significant effect of the steric repulsions at the transition state, and also seems to be the reason for the formation of the single endo stereoisomer (17) in the case of the more hindered dienophiles 6 ( Table 2, entries 11-13). The stabilizing secondary orbital interactions eventually present at the endo transition state may reinforce this preference.

HOMO LUMO
The exclusive para regioselectivity (N-Ar/CO 2 Me or CN groups) observed in all the cycloadditions can be explained on the basis of the coefficient differences for the HOMO diene -LUMO dienophile interactions (Table 4). These latter should generate the greatest perturbation, since the energy gap is smaller than the inverse interactions (LUMO diene -HOMO dienophile ). Indeed, if the largest FMO coefficients become bonded preferentially at the transition state [44][45][46][47], and considering that the relative magnitude of the coefficient of the terminus C-4 is bigger than that of C-1 in the HOMO of dienes 1-3, and that the beta C-1 coefficient is bigger than that of the alpha C-2 in the LUMO of olefins 4-6, a "para" orientation is expected, in agreement with the experimental results. This para regioselectivity supports the idea that the electronic effects also control the course of the reaction, despite the presence of steric interactions generated between the methyl group and the geminal disubstituted carbon at the vicinal carbons in the adducts 13-17.
Therefore, for these cycloadditions, the regio and stereoselectivities can be ascribed to the electronic effects, which are due to the polarization of the π-systems, and to the steric interactions, the latter mainly caused by the polysubstitution of the dienophiles.

General Procedures and Instrumentation
All reactions were carried out under nitrogen in anhydrous solvents. All glassware was dried in an oven prior to use. All commercially available compounds were used without further purification. Tetrahydrofuran and benzene were distilled from sodium benzophenone ketyl under an N 2 atmosphere prior to use. n-Hexane and ethyl acetate were distilled before use. Melting points (uncorrected) were determined with a Fisher-Johns melting point apparatus. 1 H NMR and 13 C NMR spectra were recorded on a Varian Mercury (300 MHz) and Varian VNMR System (500 MHz) instruments, in CDCl 3 as solvent and with TMS as internal reference. High-resolution mass spectra (HRMS) were obtained with a JSM-GCMate II mass spectrometer, and electron impact techniques (70 eV) were employed. X-ray data were collected on Siemens P4 and Oxford Diffraction Xcalibur S single-crystal X-ray difractometers. Thin-layer Chromatography (TLC) analyses were performed using silica plates and were visualized using UV (254 nm) or iodine. The Knoevenagel adducts 4a-e, 5a-d and 6a-c [27][28][29] and the exo-2-oxazolidinone dienes 1-3 [16][17][18][19][20] were prepared by the methods described in the literature. 7a-e, 9a-d, 11a-c, 13a-e/14a-e, 15a-

e/16a-e and 17a-c via a Two-Step Reaction. Method A
A mixture of the Knoevenagel adducts 4a-e, 5a-d, or 6a-c (1.2 equiv.) and the corresponding dienes, 1, 2, or 3 (1 mol-equiv.) was placed in a 25 mL two-necked, round-bottomed flask (equipped with a reflux condenser, a rubber septum and under nitrogen atmosphere), and the mixture was stirred and was irradiated with an infrared lamp [33] at 50 °C for 30 min-6 h under solvent-free conditions until the consumption of the diene (tlc). The reaction mixture was allowed to cool to room temperature, and then purified by column chromatography over silica gel (230-400 mesh) using n-hexane/EtOAc (98:2) as eluent, to afford the corresponding cycloadducts 7a-e, 9a-d, 11a-c, 13a-e/14a-e, 15a-e/16a-e and 17a-c.

General Procedure for the Synthesis of Adducts 13a-d/14a-d and 15a-d/16a-d via a One-Step Reaction. Method B
A mixture of active methylene compounds 20a-c (1 mol-equiv.), benzaldehydes 21a-d (mol-equiv.) and the corresponding diene 2 (1 mol-equiv.), was placed in a 25 mL two-necked, round-bottomed flask (equipped with a reflux condenser, a rubber septum and under nitrogen atmosphere), and the mixture was stirred and was irradiated with an infrared lamp [33] at 50 °C for ~30 min-6 h, under solvent-free conditions, until the consumption of the diene (tlc). The reaction mixture was allowed to cool to room temperature, and then was purified by column chromatography on silica gel (230-400 mesh) using n-hexane/EtOAc (98:2) as eluent, to afford the corresponding cycloadducts 13a-d/14a-d and 15a-d/16a-d.