The Participation of 3,3,3-Trichloro-1-nitroprop-1-ene in the [3 + 2] Cycloaddition Reaction with Selected Nitrile N-Oxides in the Light of the Experimental and MEDT Quantum Chemical Study

Abstract

The regioselective zw-type [3 + 2] cycloaddition (32CA) reactions of a series of aryl-substituted nitrile N-oxides (NOs) with trichloronitropropene (TNP) have been both experimentally and theoretically studied within the Molecular Electron Density Theory (MEDT). Zwitterionic NOs behave as moderate nucleophiles while TNP acts as a very strong electrophile in these polar 32CA reactions of forward electron density flux, which present moderate activation Gibbs free energies of 22.8–25.6 kcal·mol⁻¹ and an exergonic character of 28.4 kcal·mol⁻¹ that makes them irreversible and kinetically controlled. The most favorable reaction is that involving the most nucleophilic MeO-substituted NO. Despite Parr functions correctly predicting the experimental regioselectivity with the most favorable O-CCCl₃ interaction, these reactions follow a two-stage one-step mechanism in which formation of the O-C(CCl₃) bond takes place once the C-C(NO₂) bond is already formed. The present MEDT concludes that the reactivity differences in the series of NOs come from their different nucleophilic activation and polar character of the reactions, rather than any mechanistic feature.

1. Introduction

Organic compounds containing an isoxazoline motif as a core unit are known to provide an extensive variety of biological activities such as antibacterial, antifungal, anticancer, anticonvulsant, anti-inflammatory, antiviral, and antidepressant [1,2]. One of the most successful approaches to obtain these valuable skeletons, because of its simplicity, efficiency and high selectivity, is the [3 + 2] cycloaddition (32CA) reaction of nitrile oxides (NOs), participating as the three-atom-component (TAC), with an appropriate alkenes (Scheme 1) [3,4,5].

The application of the Molecular Electron Density Theory (MEDT) to the study of 32CA reactions has allowed classifying these reactions into four different types depending on the electronic structure of TACs [6,7]. Thus, NOs, which usually present a zwitterionic electronic structure, participate in zwitterionic-type (zw-type) 32CA reactions, which demand favorable nucleophilic/electrophilic interactions between the reagents to take place easily experimentally. Substitution may change the parent electronic structure of the simplest TACs and, therefore, the reactivity of actual experimental TACs.

Aryl NOs are one of the most common types of experimentally used NOs. The theoretical study of the 32CA reactions of benzonitrile oxide with several alkenes such as N-vinyl pyrrol [8], carvone [9], tomentosin [10], 3-nitroprop-1-ene [11] and methyl acrylate [12] allowed confirming the zwitterionic reactivity of aryl NOs in zw-type 32CA reactions, in which, interestingly, the C-C single bond was always formed prior to the O-C one.

Trihalomethylated organic molecules also present valuable potential in pharmacy. Since the trihalomethyl group greatly withdraws electrons, its presence in compounds can hugely improve molecular properties such as bioactivity, lipophilicity and metabolic stability. Moreover, increased lipophilicity leads to an increase in therapeutic efficiency or the ability to penetrate the blood-brain barrier [13,14]. Trifluoromethylated compounds are the most commonly used molecules of this type [15], however, in recent years, trichloromethylated compounds are gaining importance as sedative-hypnotic agents [16] and antitubercular components [17], for instance.

The biological activity of these compounds can be additionally stimulated by the presence of the nitro group [18,19]. The NO₂ group gives interesting potential possibilities for different-type functionalization in amines, nitrosocompounds, oximes, nitronates and many other [20,21,22,23]. Consequently, structures containing CX₃ and NO₂ functional groups are very attractive materials for further applications. The synthesis of these type of compounds is relatively easy using the corresponding 3,3,3-trihalo-1-nitroprop-1-enes as precursors. In the literature, some examples of the application of 3,3,3-trichloro-1-nitroprop-1-ene (TNP) as component of cycloaddition processes are available [24,25].

The stability of cycloadducts obtained via 32CA reactions involving TNP is very variable. In some cases, the presence of the trichloromethyl CCl₃ group strongly stabilizes the heterocyclic systems, compared to the analogs substituted with other functional groups. For example, 2,3,3-triphenyl-4-nitroisoxazolidine [26] is very unstable; it easily decomposes via retro-cycloaddition process even at 60–80 °C. However, structurally similar 2,3,3-triaryl-4-nitro-5-trichloromethylisoxazolidines melt at 106–195 °C with no decomposition (Scheme 2) [27,28].

In other cases, the presence of the CCl₃ group is not enough to stabilize the corresponding cycloadduct. For instance, diaryldiazomethanes react with TNP yielding unstable 5-nitropyrazoline systems, which surprisingly converted very fast into the corresponding 4-dimethylvinylpyrazolines via hydrochloride extrusion (Scheme 3) [24,25]. It should be underlined at this point that the HCl extrusion from molecular systems including the CCl₃ group is very rare and was unobserved earlier in five-membered heterocycles.

Given the importance of the isoxazoline framework and the biological activities originated by the presence of the CCl₃ and NO₂ groups, we decided to shed light on the participation of TNP 2 in the 32CA reactions with a series of substituted aryl nitrile N-oxides 1a–e of different nucleophilic character yielding isoxazolines 3a–e (Scheme 4), as a continuation of our comprehensive study about the reactivity of this type of TAC. In particular, we performed an experimental study of the reaction course and structural analysis of reaction products, as well as theoretical investigations regarding the molecular mechanism of cycloadduct formation and nature of relevant structures along the reaction paths, in the framework of MEDT as a modern view of organic chemical reactivity [29]. Our aim is to check on the stability of isoxazolines including the CCl₃ group and the effect of substitution on the reactivity of aryl NOs, providing significant experimental and theoretical insight to help future designs of syntheses of trichloromethyl-functionalized isoxazoline analogs with interesting potential applications.

2. Results and Discussion

2.1. Experimental Study

First, the precursors for the five aforementioned NOs 1a–e were prepared by conversion of the corresponding aromatic aldehydes into the respective oximes via the well-known [30] reaction involving hydroxylamine hydrochloride Subsequent chlorination of the corresponding oximes to yield the hydroximinoyl chlorides precursors was carried out with N-chlorosuccinimide in N,N-dimethylformamide [31]. Next, the preparation of the TNP 2 was carried out via the Henry condensation between chloral hydrate and nitromethane [32]. The obtained trichloronitropropanol was converted into the expected nitroalkene via a two-step protocol consisting of esterification and acetic acid extrusion stages [33].

The considered 32CA reactions can in principle yield two regioisomeric cycloadducts (Scheme 4). The optimization of the reaction conditions was performed for the model reaction between 4-methylbenzonitrile N-oxide 1a and TNP 2. For the generation of NO 1a, different-type protocols were tested. The relatively highest yield of the expected product was obtained (temp: 25 °C, time: 24h, molar ratio hydroxamoyl chlorides/nitroalkene 1:1,25). Regardless of the reaction conditions (

Table 1. Optimization of reaction conditions for model 32CA reaction between TNP 2 and nitrile N-oxide 1a.

Method	Base	Temp. [°C]	Time [h]	Molar Ratio Hydroxamoyl Chlorides/Nitroalkene	Yield [%]
					Isoxazoline	Furoxan
1	K2CO3	25 (r.t.)	24	1:1.25	40.0	1.0
		66 (reflux)	6	1:1.25	40.0	8.1
2	TEA	25 (r.t.)	24	1:1.25	44.4	0.9
		66 (reflux)	6	1:1.25	43.5	5.0
1	K2CO3	25 (r.t.)	24	1:2	9.8	0.8
		66 (reflux)	6	1:2	25.6	7.5
2	TEA	25 (r.t.)	24	1:2	20.8	0.7
		66 (reflux)	6	1:2	47.6	10.6

), only one cycloadduct along with certain amounts of the respective furoxane (which is a secondary product of the dimerization of NOs) [34] was detected by comprehensive HPLC analysis of the reaction and post-reaction mixtures. Under optimized conditions, the yield of nitroisoxazoline 3a is about 45%, together with the furoxan mentioned above, whose yield significantly depends on the reaction conditions (Table 1). The obtained 2-isoxazoline molecular system was isolated via simple fractional crystallization and identified based on the spectral data.

For post-analysis, HR-MS data were analyzed first. It was found that the isolated compound (3a or 4a) gave a pseudo-molecular ion 320,9601mDa [M-H]⁻ (see Supplementary Materials) which corresponds with the proposed C₁₁H₈N₂O₃Cl₃ molecular formula. Bands typical for the NO₂ group [35], -N-O- [11,36], and -C=N- [11] moieties in the heterocyclic ring as well as C-Cl bond [35] vibrations were detected in its IR spectrum. Information about stereochemistry of the isolated compound was provided by ¹H NMR spectroscopy. Protons associated with the isoxazoline ring form AX spin system. The values of J _{4,5 isoxazoline ring} coupling constants are found in the range of coupling constants typical for isoxazolines substituted at the 4- and 5- positions with trans configuration of the substituents.

The single crystal X-ray diffraction experiment confirmed the formation of a heterocyclic compound with the structure presented in Figure 1. The heterocyclic 5-membered cycle is almost flat (RMSD 0.038 Å) and is rotated by 20.6(3)° relative to the aromatic ring, showing no or little electronic coupling. The torsion angle along the atoms N2-C1-C2-C4 (−128.5(2)°) shows the ac arrangement of the heterocyclic substituents. Weak hydrogen C-H…O/N/Cl bonds stabilize the crystal structure in a centrosymmetric triclinic space group P-1. No π-stacking is observed in the crystal network; instead, pnictogen, chalcogen and halogen atoms participate in short Cl…Cl, O…O/N contacts (Table S1 in Supplementary Materials). The crystal contains both enantiomers R,R and S,S.

Likewise, the other 32CA reactions involving NOs 1b–e were explored. In all cases, only 3-aryl-4-nitro-5-trichloromethylisoxazolines (3b–e) were isolated and identified by spectral analyses.

2.2. Theoretical Study

To explain and understand the above experimental outcomes, a MEDT study of these 32CA reactions was carried out. This study is divided in the following sections: (i) analysis of the electronic structures of the reagents; (ii) analysis of the Conceptual DFT (CDFT) reactivity indicators; (iii) study of the potential energy surfaces associated with the five considered reactions; and (iv) characterization of the molecular mechanism associated with the 32CA reactions of NOs 1b and 1e with TNP 2.

2.2.1. Analysis of the Electronic Structures of the Reagents

One appealing procedure that provides a straightforward connection between the electron density distribution and the chemical structure is the quantum chemical analysis of Becke and Edgecombe’s Electron Localization Function (ELF) [37]. Thus, to gain insight about the reactivity of the reagents involved in the aforementioned 32CA reactions, their electronic structure was characterized through the topological analysis of the ELF and Natural Population Analysis (NPA) of their charge distribution. Herein, only the results for NOs 1b and 1e, and TNP 2 are commented on (Figure 2), while those for the three remaining NOs are given in Table S4 in Supplementary Material.

The topological analysis of the ELF (Figure 2) reveals a similar electronic structure for the five NOs. All of them present two V(O1) and V’(O1) monosynaptic basins integrating a total average of 5.67 ē, a V(O1,N2) disynaptic basin integrating 1.69 ē, two V(N2,C3) and V’(N2,C3) disynaptic basins integrating a total average of 6.00 ē, and a V(C3,C3′) disynaptic basin integrating an average of 2.48 ē. These features allow constructing Lewis-like structures with an O1 non-bonding region, an O1-N2 single bond, an N2-C3 triple bond and a C3-C3′ single bond. The absence of any reactive pseudoradical or carbenoid center and the presence of a multiple bond indicates that these NOs have a zwitterionic electronic structure just as the simplest acetonitrile oxide 5 [38] and, consequently, they will participate in zw-type 32CA reactions [39].

NPA shows (Figure 2) that all the NOs, regardless of the substitution, also present a similar charge distribution. In all of them, the O1 oxygen is negatively charged by ca. 0.4 e, while the N2 and C3 atoms are positively charged by ca. 0.2 e. This picture contrasts with Huisgen’s representation of NOs as 1,2-zwitterionic species [40]. On the other hand, the ethylene moiety at TNP 2 is negatively charged by −0.22 e at C4 and −0.06 e at C5, which might be unexpected taking into account the electron-withdrawing character of both CCl₃ and NO₂ groups.

2.2.2. Analysis of the CDFT Reactivity Indices of the Reagents

CDFT [41,42] is a powerful tool to predict and understand the reactivity in organic reactions as it allows the characterization of the electrophilic and nucleophilic behaviors of molecules both qualitative and quantitatively. Thus, to discern any reactivity changes associated with the aryl substitution at NOs 1a–e and characterize their behavior towards TNP 2, the global CDFT reactivity indices of every reagent, namely the electronic chemical potential μ, chemical hardness η, global electrophilicity ω and global nucleophilicity N, were computed and analyzed (

Table 2. B3LYP/6-31G(d) electronic chemical potential μ, chemical hardness η, electrophilicity ω and nucleophilicity N indices, in eV, of NOs 1a–e and TNP 2.

Compound	μ	η	ω	N
1b	−3.42	4.91	1.19	3.24
1a	−3.69	4.97	1.37	2.95
1c	−3.85	5.02	1.47	2.76
1d	−4.12	4.96	1.71	2.52
1e	−5.04	4.03	3.15	2.07
2	−5.84	5.22	3.27	0.67

). These global indices were calculated at the B3LYP/6-31G(d) computational level to make proper classifications based on the electrophilicity and nucleophilicity scales, which were established at that level [41,42,43,44].

In polar reactions there is a flux of electron density from the species with higher electronic chemical potential [43] to the ones with lower electronic chemical potential. The electronic chemical potentials of NOs 1a–e, which range by between −3.42 (1b) and −5.04 (1e) eV, are considerably higher than that of TNP 2, −5.84 eV. Consequently, if the 32CA reactions between NOs 1a–e and TNP 2 are polar (see later), the electron density will flux from the former to the latter species in a forward electron density flux (FEDF) fashion [44,45].

The chemical hardness η of the NOs range by between 4.03 (1b) to 5.02 (1c) eV; only the η of 1e significantly differs from the rest, being a softer species. TNP 2 is the hardest species with η = 22 eV.

The electrophilicity ω indices [46] of the NOs range by between 1.19 (1b) and 3.15 (1e) eV. According to the electrophilicity ω scale [47] 1a and 1b are classified as moderate electrophiles with ω values below 1.5 eV, 1c lays on the borderline between moderate and strong electrophiles, while 1d and 1e are strong electrophiles. Please note that 1e is actually a very strong electrophile with a ω value above 3.0 eV. On the other hand, the nucleophilicity N indices of NOs range from 2.07 (1e) to 3.24 (1b) eV. Within the nucleophilicity scale [48] 1b is classified as strong nucleophile, 1a lays on the borderline between moderate and strong nucleophiles, while 1c–e are classified as moderate nucleophiles.

Given that TNP 2 is classified as a very strong electrophile with ω = 3.27 eV and as a very poor nucleophile with N = 0.67 eV, and considering the zwitterionic nature of the NOs, it is expected that the zw-type 32CA reaction involving 1b should be the most favorable one, and that the reactivity should decrease in the order 1b > 1a > 1c > 1d > 1e. The global indices also suggest a favorable reaction between moderate-to-strong nucleophiles and a strong electrophile, in agreement with the computed activation Gibbs free energies (see later).

By approaching a non-symmetric electrophilic/nucleophilic pair along a polar or ionic process, the most favorable reactive channel is that associated with the initial two-center interaction between the most electrophilic and nucleophilic centers of the reagents [49]. Thus, to characterize the most nucleophilic and electrophilic centers of these species, the nucleophilic P_k⁻ Parr functions of NOs 1a–e and the electrophilic P_k⁺ Parr functions of TNP 2 were analyzed [50]. Parr functions were calculated at the ωB97X-D/6-311G(d,p) level.

Analysis of the nucleophilic P_k⁻ Parr functions at the reactive sites of NOs 1a–e clearly indicates that the most nucleophilic center is the O1 oxygen atom with values ranging from 0.39 (1b) to 0.53 (1e) (Figure 3 and Figure S1 in Supplementary Materials). Interestingly, NO 1e, with the strong electron-withdrawing nitro NO₂ group, has the strongest nucleophilically activated O1 oxygen, while NO 1b, with the strong electron-donating methoxy MeO group, has the weakest, yet strong, activation. The C3 carbon is not nucleophilically activated in any case. On the other hand, the C4 carbon adjacent to the trichloromethyl group at TNP 2 is more than twice as electrophilically activated, P_k⁺ = 0.32, as the C5 carbon near the NO₂ group, P_k⁺ = 0.13. Consequently, the most favorable reaction path along a polar 32CA reaction between NOs 1a–e and TNP 2 would correspond to the O1-C4 interaction, which agrees with the experimental regioselectivity (Scheme 4). In addition, the null nucleophilic activation of the C3 carbon of these NOs accounts for the full regioselectivity experimentally found.

2.2.3. Study of the Potential Energy Surfaces Associated with the 32CA Reactions between NOs 1a–e and TNP 2

Due to the non-symmetry of the reagents and the linear structure of NOs 1a–e, the 32CA reactions between NOs 1a–e and TNP 2 can only take place along two regioisomeric reaction paths depending on the relative position of the R-aryl substituent of NOs 1a–e with respect to the nitro NO₂ group of TNP 2; namely A, associated with the O1-C4 two-center interaction leading to ortho isoxazolines 3a–e, and B, associated with the C3-C4 two-center interaction leading to meta isoxazolines 4a–e (Scheme 5). Analysis of the potential energy surfaces allowed finding only one TS for each reaction path, indicating that these reactions follow a one-step mechanism. Electronic energies in gas phase and in solvent are given in Table S6 in Supplementary Materials, full thermodynamic data are gathered in Table S7 in Supplementary Materials, while relative enthalpies and Gibbs free energies are graphically represented in Figure 4.

The activation enthalpies of the considered reaction paths range by between 8.4 (TS-o-b) and 12.6 (TS-m-e) kcal·mol⁻¹, while the reaction enthalpies are found in the narrow range of −43.8 and −44.6 kcal·mol⁻¹. After inclusion of room temperature and entropies to enthalpies, the relative Gibbs free energies are by between 13.6 and 15.8 kcal·mol⁻¹ higher as a consequence of the unfavorable entropies associated with these bimolecular processes. Thus, the activation Gibbs free energies reach values of 22.8–26.8 kcal·mol⁻¹, while formation of cycloadducts is exergonic by 28.3–28.9 kcal·mol⁻¹.

Some appealing conclusions can be drawn from these energy profiles: (i) in agreement with the experimental outcomes, the five 32CA reactions are ortho regioselective, as the meta TSs are found between 1.2 and 2.3 kcal·mol⁻¹ above the ortho ones in the Gibbs free energy profiles (Figure 4); (ii) the activation Gibbs free energies decrease in the order NO₂ > Cl > F > Me > MeO, in complete agreement with the analysis of the CDFT indices (see Section 2.2.2); (iii) these 32CA reactions are kinetically and thermodynamically favorable, and their strong exergonic character makes them irreversible, formation of the corresponding cycloadducts being therefore under kinetic control; (iv) the strong exergonic character of the reactions points out the very high stability of cycloadducts presenting the trichloromethyl CCl₃ group. Please note that the corresponding analog of 3b without CCl₃ is 1.8 kcal·mol⁻¹ less stable; and finally, (v) the substitution at benzonitrile N-oxide has an important role increasing the reaction rates. Please note that the activation Gibbs free energies of the reaction of unsubstituted benzonitrile N-oxide with TNP 2 is 23.5 kcal·mol⁻¹, which means a reaction rate 3.5 times lower.

Taking into account that these 32CA reactions take place through favorable kinetic control, the Eyring-Polanyi equation [51] was used to estimate the composition of the reaction mixture.

$$k=\frac{k{k}_{B}T}{h}{e}^{-\frac{\mathsf{\Delta }{G}^{‡}}{RT}}$$

(1)

From this equation, the relative reaction rate constants k_rel can be obtained as:

$$k_{rel}=e^{-\frac{\mathsf{\Delta }\mathsf{\Delta }{G}^{‡}}{RT}}$$

(2)

were $\mathsf{\Delta }\mathsf{\Delta }{G}^{‡}$ is the difference between the relative activation Gibbs free energies of two TSs, R the gas ideal constant, and T the reaction temperature.

The expected percentage of ortho and meta isoxazolines for each reaction are reported in

Table 3. Calculated composition of the reaction mixture for the 32CA reactions of NOs 1a–e with TNP 2.

	3	4
Cl (d)	91.4	8.6
MeO (b)	98.0	2.0
NO2 (e)	88.7	11.3
F (c)	97.1	2.9
Me (a)	98.1	1.9

. Considering the ωB97X-D/6-311G(d,p) Gibbs free energies in tetrahydrofuran (THF) and the reaction temperature, the ortho and meta isoxazolines should be obtained in an average 94.7:5.3 relation; while the reactions of 1a–c present a similar regioselectivity, those involving 1d,e are slightly less regioselective. This result indicates that while the ortho isoxazolines 3a–e are expected to be the major isomers, formation of the meta isoxazolines 4a–e should be negligible, in clear agreement with the experimental outcomes (Scheme 4). This result emphasizes that the extent of regioselectivity cannot be predicted simply by ΔΔG^‡ between regioisomeric TSs.

The optimized geometries of the regioisomeric TSs involved in the 32CA reactions of NO 1b with TNP 2 in THF are displayed in Figure 5, while the rest are given in Figure S1 in Supplementary Materials. The distances do not significantly change with the substitution of the TAC. Thus, at the more favorable ortho TSs, the average O1-C4 distance is 2.074 Å, while the average C3-C5 distance is 2.234 Å. At the meta TSs, the average distance between the O1 and C5 atoms is 2.133 Å, while that between the C3 and C4 atoms is 2.184 Å. Some information can be extracted from these geometrical data: (i) the more favorable ortho TSs are geometrically more asymmetric, Δd = 0.155 Å, than the meta ones, Δd = 0.051 Å, which can be considered almost symmetrical; (ii) the O-C distance is always shorter than the C-C one; (iii) taking into account the C-C and C-O bond formation distances characterized from previous ELF topological analyses [6], it can be concluded that neither O-C nor C-C bond formation has started yet at any of these TSs; and (iv) the heterocyclic 5-membered cycle in NO 1b is rotated by 11.7° relative to the aromatic ring and the torsion angle along the atoms N2-C1-C2-C4 is −117.3°, which are 8.9° and 11.2° lower than the respective experimental X-ray diffraction data (see Section 2.1).

The polar character of these zw-type 32CA reactions was evaluated by computing the global electron density transfer (GEDT) [52,53] at the TSs. The GEDT values in THF range by between 0.11 and 0.15 e. These values indicate that the corresponding zw-type 32CA reactions have somewhat polar character. The ortho TS are only a hundredth more polar than the meta ones. The GEDT increases in the order NO₂ > Cl > F > Me > MeO, in agreement with the increased nucleophilic character in the same order (see Section 2.2.2). This trend in the polar character of the reactions follows the same trend as the computed activation energies, emphasizing the relation between the polarity and reactivity in zw-type 32CA reactions; [39] i.e., the stronger the polar character, the more favorable the reaction. Consequently, the different nucleophilic character of NOs 1a–e is responsible for their different reactivity.

2.2.4. Characterization of the Bond Formation Process

The Bonding Evolution Theory (BET) [54], which combines the topological analysis of the ELF with catastrophes theory [54], allows characterizing the electron density rearrangement along a reaction path and, thus, enables a description of the bonding changes associated with a given molecular mechanism [55,56].

The molecular mechanism of zw-type 32CA reactions involving NOs and different types of ethylene derivatives has already been recently studied [2,9,10,12,53,57,58,59,60,61]. As a general outlook, these reactions start with the depopulation of the C≡N triple bond of the NO framework and the C=C double bond of the ethylene derivative, which gives rise to the creation of the corresponding carbon pseudoradical centers [62]. The electron density at the oxygen atom of the NO also experiences some rearrangement. Then, the bond formation takes place either by coupling of two carbon pseudoradical centers in the case of the C-C single bond [52], or by donation of the oxygen non-bonding electron density to the remaining carbon pseudoradical center in the case of the C-O single bond. In polar reactions, the formation of the first single bond takes place between the most nucleophilic center of the nucleophile and the most electrophilic center of the electrophile as characterized by CDFT and the Parr functions. Accordingly, in zw-type 32CA reactions of NOs, this should be the C-O single bond (see Section 2.2.2); however, previous BET studies [57,63] have shown that regardless of the polar character, the C-C single bond forms first in 32CA reactions of NOs.

Thus, the characterization of the bond formation processes in the 32CA reactions of NOs 1a–e with TNP 2 would not only add valuable information about this question, but would also help identifying any remarkable change in the molecular mechanisms which may account for the observed reactivity differences. To do so, BET was applied over the 32CA reactions of NOs 1b,e with TNP 2 in THF. Only the structures involved in the bond formation are reported and discussed herein. ELF attractor positions for the selected structures involved only in the most favorable reaction of NO 1b are displayed in Figure 6.

Analysis of the bonding changes along the 32CA reactions of NOs 1b,e with TNP 2 shows very similar bond formation processes, regardless of the substitution of the TAC. This behavior is in agreement with the similar electronic structures of these NOs (see Section 2.2.1). At the two TSs, both the N2-C3 and C4-C5 bonding regions have been depopulated with respect to the separated reagents (see Section 2.2.1) to 4.29 (TS-o-b) and 4.40 (TS-o-e) ē, which can already be considered a double bond and an overpopulated single bond, respectively, but no pseudoradical center is observed yet. Further approach of the two molecular frameworks leads to the creation of pseudoradical centers at the three carbon atoms involved. At S1, the two C3 and C5 pseudoradical centers reach populations of ca. 0.5 and 0.8 ē, just before merging each other at S2 to give rise to the formation of the first C3-C5 single bond, at a C-C distance of ca. 1.97 Å, with an electron population of 1.48 (S2-b) and 1.33 (S2-e) ē. At S3, right before the formation of the second O1-C4 single bond, the non-bonding electron densities associated with both the O1 oxygen and the C4 carbon reach ca. 5.74 and 0.19 ē. At S4, formation of the O1-C4 single bond takes place at an O-C distance of ca. 1.76 Å by donation of ca. 0.6 ē of the O1 oxygen electron density to the C4 pseudoradical center, whose population has slightly decreased to ca. 0.2 ē. The new O1-C4 bond has a population of ca. 0.8 ē. At the final isoxazolines 3b,e, the O1-C4 and C3-C5 bonds reach ca. 1.34 and 2.02 ē. As formation of the second O1-C4 single bond occurs when the first C3-C5 formed bond has already reached 78% (1b) and 83% (1e) of its final population at 3b,e, it can be considered that these polar zw-type 32CA reactions follow a two-stage one-step mechanism [64]. Indeed, the evolution of the forming bond distances is considerably synchronous (

Table 4. Electron populations of the most relevant ELF valence basins and O1-C4 and C3-C5 distances for TS-o-b, isoxazoline 3b and the structures S1-b–S4-b involved in the formation of the O1-C4 and C4-C5 single bonds along the more favorable regioisomeric path A of the 32CA reaction between NO 1b and TNP 2. Populations are given in average number of electrons, ē, and distances in angstroms, Å.

	TS-o-b	S1-b	S2-b	S3-b	S4-b	3b
d(O1-C4)	2.051	1.865	1.856	1.760	1.751	1.428
d(C3-C5)	2.253	1.999	1.988	1.862	1.850	1.512
V(O1,N2)	1.40	1.31	1.30	1.23	1.23	1.07
V(N2)	2.09	2.44	2.46	2.58	2.60	2.80
V(N2,C3)	2.17	1.81	1.80	1.75	1.74	1.67
V′(N2,C3)	2.12	1.90	1.89	1.81	1.81	1.60
V(C4,C5)	3.45	2.47	2.45	2.29	2.27	2.03
V(O1)	2.83	2.73	2.72	2.65	2.62	2.47
V′(O1)	2.82	2.89	2.89	2.59	2.56	2.42
V′′(O1)				0.51
V(C4)		0.26	0.27	0.18
V(C3)		0.51
V(C5)		0.79
V(C3,C5)			1.32	1.56	1.57	2.01
V(O1,C4)					0.79	1.33

and

Table 5. Electron populations of the most relevant ELF valence basins and O1-C4 and C3-C5 distances for TS-o-e, isoxazoline 3e and the structures S1-e–S4-e involved in the formation of the O1-C4 and C4-C5 single bonds along the more favorable regioisomeric path A of the 32CA reaction between NO 1e and TNP 2. Populations are given in average number of electrons, ē, and distances in angstroms, Å.

	TS-o-e	S1-e	S2-e	S3-e	S4-e	3e
d(O1-C4)	2.105	1.932	1.924	1.770	1.761	1.429
d(C3-C4)	2.217	1.972	1.960	1.771	1.761	1.509
V(O1,N2)	1.44	1.34	1.33	1.23	1.23	1.08
V(N2)	2.01	2.39	2.41	2.61	2.62	2.79
V(N2,C3)	2.25	1.83	1.83	1.74	1.73	1.67
V′(N2,C3)	2.15	1.91	1.90	1.78	1.78	1.61
V(C4,C5)	3.43	2.49	2.47	2.23	2.21	2.04
V(O1)	2.82	2.74	2.73	2.64	2.62	2.47
V′(O1)	2.79	2.84	2.85	2.57	2.54	2.40
V′′(O1)				0.53
V(C4)		0.23	0.24	0.20
V(C3)		0.56
V(C5)		0.75
V(C3,C5)			1.33	1.67	1.68	2.02
V(O1,C4)					0.81	1.34

), thus emphasizing that geometrical analyses are not valid to characterize the stage of the bond formation progress.

The most relevant finding from this BET study is that despite the Parr functions accounting for the solely isolation of regioisomeric isoxazolines 3b,e, formation of the C3-C5 single bond involving the least nucleophilic and electrophilic centers of the reagents takes place before the expected more favorable O1-C5 one. This unexpected pattern is also observed in previous BET studies of polar zw-type 32CA reactions of NOs [57,63]. On the other hand, the similar bonding changes taking place along the 32CA reactions of the two most electronically different NOs 1b and 1e with TNP 2 indicates that the computed reactivity changes among the series of NOs 1a–e are mainly due to their different nucleophilic character and polarity of the corresponding reactions, rather than any structural or mechanistic reason.

3. Methods and Procedures

3.1. Experimental

3.1.1. Analytical Techniques

HPLC analyses was done using a Knauer device with a UV-VIS detector (LiChrospher 18-RP 10 µm column, eluent: 80% methanol). M.p. values were measured on the Boetius apparatus and were uncorrected. IR spectra were derived from the FTS Nicolet IS 10 spectrophotometer. UV/Vis spectra were recorded using spectrometer UV-5100 BIOSENS and was determined for the 200–500nm range. NMR spectra were registered on an Agilent 400 MHz spectrometer using CDCl₃ as a solvent. HR-MS spectra were acquired on a UPLC-MS-MS Waters Xevo G2 QTof instrument with ESI ionization.

3.1.2. RTG Analysis

The diffraction intensities were collected at room temperature with the use of SuperNova X-ray diffractometer with Atlas S2 CCD detector using the mirror-monochromatized CuKα radiation (λ = 1.54184 Å). The phase problem was solved by direct methods using the SHELXS-97 program and the structure model refined by the full-matrix least-squares method on F² using the SHELXL-97 program implemented in Olex2.refine package [65,66]. Details on crystal structure refinement and geometry are given in Tables S1 in Supplementary Materials.

CCDC No. 2,115,222 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge via http://www.ccdc.cam.ac.uk/conts/retrieving.html (accessed on 20 August 2021) (or from the Cambridge Crystallographic Data Centre, 12, Union Road, Cambridge CB2 1EZ, UK; fax: +44-1223-336033).

3.1.3. Materials

The components of the cycloaddition were prepared according to procedures described in the literature. In particular, nitrile N-oxides were generated in situ from respective hydroxamoyl chlorides, obtained via chlorination of respective oximes according to known procedures [31]. The 3,3,3-trichloro-1-nitroprop-1-ene was synthetized via three-step protocol starting from nitromethane and chloral hydrate [32]. Commercially available (Sigma Aldrich, St. Louis, MO, USA) chemicals were used as solvents and as components for the preparation of addends.

Cycloaddition between Nitrile N-Oxides and 3,3,3-Trichloro-1-nitroprop-1-ene—General Procedure

Erlenmeyer flask containing 10 cm³ of THF was placed in an ice bath. Next, 0.003 mole of 3,3,3-trichloro-1-nitroprop-1-ene and 0.0024 mole of hydroximinoyl chloride was added and stirred for 10 min. Then 0.0015 mole of K₂CO₃ was dosed in small portions during 30 min period. After that time, the ice bath was removed. Change of mixture color and turbidity was observed. The mixture was left for 24 h with constant stirring. The solvent was evaporated and remaining solid was mixed with diethyl ether and filtrated to remove insoluble side products. The ether was removed under vacuum, and remaining crude product was washed with light petroleum ether and crystallized from ethanol. A white crystalline solid was obtained.

3.2. Computational

DFT calculations were performed using the hybrid ωB97X-D functional [67] which includes long-range exchange (denoted by X) correction as well as the semiclassical London-dispersion correction (indicated by suffix-D). The standard 6-311G(d,p) [68] basis set was used, which includes d-type polarization for second row elements and p-type polarization functions for hydrogen atoms. The Berny method was used in optimizations [69,70]. A similar theory level has been commonly used for the mechanistic research aspects of cycloaddition reactions [71,72,73,74,75] The TSs were characterized through frequency analysis, presenting only one imaginary frequency. The intrinsic reaction coordinate (IRC) paths [76] were computed to find the unique connection between the TSs and the minimum stationary points using the second order González-Schlegel integration method [77,78]. Solvent effects of THF were considered by full optimization of the gas-phase structures at the same computational level using the polarizable continuum model (PCM) [79,80] in the framework of the self-consistent reaction field (SCRF) [81,82,83]. Values of ωB97X-D/6-311G(d,p) enthalpies, entropies and Gibbs free energies in THF were calculated with standard statistical thermodynamics at 298.15 K and 1 atm [68].

The GEDT [52] values were estimated by a natural population analysis (NPA) [84,85] using the equation GEDT(f) = ${\mathsf{\Sigma }}_{q\in f}q$, were q are the atoms of a framework (f) at the TSs. CDFT reactivity indices [41,42] were calculated at the B3LYP/6-31G(d) computational level because the electrophilicity and nucleophilicity scales were established at that level. The global electrophilicity ω index [46] is given by the following expression, $\omega =\frac{{\mu }^2}{2\eta }$, in terms of the electronic chemical potential μ and chemical hardness η. Both quantities may be approached in terms of the one-electron energies of the frontier molecular orbitals HOMO and LUMO, ε_H and ε_L, as $\mu =\frac{\left({\epsilon }_H+{\epsilon }_L\right)}{2}$ and $\eta ={\epsilon }_L-{\epsilon }_H$, respectively [41]. The global nucleophilicity N index, ref. [86] based on the HOMO energies obtained within the Kohn-Sham scheme, ref. [87] is defined as N = ε_H(Nu) − ε_H(TCE), where tetracyanoethylene (TCE) is the reference. Parr functions were calculated at the ωB97X-D/6-311G(d,p) level. All computations were carried out with the Gaussian 16 suite of programs [88].

The topology of the ELF [37] of the ωB97X-D/6-311G(d,p) monodeterminantal wavefunctions was carried out using the TopMod [89] package with a cubical grid of step size of 0.1 Bohr. The analysis of the bond formation processes was performed by applying the Bonding Evolution Theory (BET) along the corresponding reaction paths; 312 (methoxy) and 337 (nitro) nuclear configurations along the IRC paths were considered. GaussView program [90] was used to visualize molecular geometries of all the systems as well as the position of the ELF basin attractors.

4. Conclusions

The zw-type 32CA reactions of a series of aryl-substituted NOs 1a–e with TNP 2 have been both experimentally and theoretically studied within the framework of MEDT. These reactions, carried out at room temperature, during 24 h and using THF as solvent, lead to the formation of single regioisomeric isoxazolines 3a–e with moderate to high yields. Regioisomeric isoxazolines 4a–e were not detected and only little amounts of furoxan, coming from the dimerization of the NOs, were isolated.

The topological analysis of the ELF of NOs 1a–e allows characterizing their zwitterionic structure, just as that of the simplest acetonitrile oxide, thus being able to participate in zw-type 32CA reactions. No relevant change is observed in the electronic structure within this series of TACs upon substitution, suggesting that the substitution would not lead to major changes in reactivity.

Thus, CDFT shows that NOs 1a–e behave as moderate nucleophiles while TNP 2 acts as a very strong electrophile in 32CA reactions of FEDF. According to the nucleophilic character of NOs 1a–e, it is expected that the zw-type 32CA reaction involving 1b should be the most favorable one, and that the reactivity should decrease in the order 1b > 1a > 1c > 1d > 1e. The nucleophilic and electrophilic Parr functions characterize the O1 oxygen of the NOs as their most nucleophilic center and the C4 carbon of TNP 2 as its most electrophilic center. Therefore, the expected more favorable regioisomeric isoxazolines are 3a–e, in agreement with the experimental outcomes.

The activation Gibbs free energies of the five 32CA reactions are very similar, ranging from 22.8 to 25.6 kcal·mol⁻¹, while formation of isoxazolines 3a–e is exergonic by ca. 28.4 kcal·mol⁻¹, thus being irreversible and kinetically controlled. All of them present an ortho regioselectivity, in agreement with the experimental obtaining of only 3a–e. The values of GEDT computed at the TSs confirm the polar character of these zw-type 32CA reactions.

The BET study of the bond formation processes along the 32CA reactions involving NOs 1b,e indicates a similar bonding pattern for both reactions. Consequently, the reactivity differences in the series of NOs 1a–e come from the different nucleophilic activation of the TACs and polar character of the reactions, rather than any mechanistic feature. Interestingly, despite the more favorable two-center interaction between the O1 and C4 atoms, formation of the C3-C5 bond takes place first, through a two-stage one-step mechanism in which formation of the second O1-C4 bond takes place once the C3-C5 bond is practically formed by up to 80%.