High-spin intermediates of the photolysis of 2,4,6-triazido-3-chloro-5-fluoropyridine

Introduction

High-spin nitrenes are highly reactive intermediates formed during photolysis or thermolysis of aromatic polyazides. Both these processes are widely used in modern science and technology [1-6]. When aromatic polyazides contain nonequivalent azido groups, these groups can undergo selective photolysis, providing important information about the selective cleavage of chemical bonds in organic molecules with light. The direction and selectivity of such processes can be monitored with EPR spectroscopy allowing the reliable identification of isomeric high-spin nitrenes [7-10]. Thus, previous EPR studies have shown that irradiation of triazide 1a with light at λ = 313 nm selectively gave quintet dinitrene 4a as the major intermediate product (Scheme 1) [7,8]. Most recently, a similar selectivity was observed during the photolysis of triazide 1b where quintet dinitrene 4b was the major intermediate product (Scheme 1) [9]. By contrast, the photolysis of triazide 7 occurs selectively on the azido group located on the γ-phenyl ring, yielding dinitrene 9 as a single quintet intermediate (Scheme 1) [10]. All these photochemical studies became possible owing to extensive EPR investigations of various high-spin nitrenes in the past two decades [11-31]. Nowadays, the EPR spectral identification of high-spin nitrenes is based on comparison of their zero-field splitting (ZFS) parameters derived from experimental EPR spectra and calculated by quantum chemistry methods. The most accurate theoretical estimations are obtained at the PBE/DZ level of theory, which overestimates the experimental ZFS values of nitrenes by only about 10% [32]. This accuracy of calculations is sufficient for reliable identification of isomeric dinitrenes 4a–c and 5a–c formed during the photolysis of triazides 1a–c (Scheme 1). Although these dinitrenes show nearly the same D_Q values (≈0.21 cm⁻¹), the E_Q values of 4a–c and 5a–c differ significantly (E_Q4 ≈ 0.055 cm⁻¹, E_Q5 ≈ 0.040 cm⁻¹), thus allowing one to unambiguously discriminate such isomers [29-31].

A new level of complexity is the EPR identification of high-spin intermediates formed during the photolysis of asymmetric triazides such as 11 (Scheme 2). Theoretically, the photolysis of triazide 11 may give three triplet nitrenes 12–14, three quintet dinitrenes 15–17, and septet trinitrene 18. So far, no attempt was undertaken to discriminate such structurally alike isomers as quintet dinitrenes 16 and 17.

In the present work, we report on matrix isolation and EPR studies of high-spin intermediates formed during the photolysis of asymmetric triazide 11, providing the first information about selective photochemical decomposition of three nonequivalent azido groups in monocyclic aromatic compounds.

Results and Discussion

Brief UV irradiation (2 min, λ = 260–320 nm) of triazide 11 in the argon matrix leads to the appearance of just two strong EPR signals of triplet nitrenes at 6985 and 7217 G as well as weak Y₂-transitions of quintet dinitrenes at 3077 and 3337 G. Upon more extended irradiation, new signals in the 30–7000 G region appear due to gradual accumulation of high-spin nitrenes. The intensities of these signals reached their maximum values after 45 min of irradiation and then gradually decayed upon further irradiation. The EPR spectrum recorded after 45 min of irradiation is shown in Figure 1b.

The EPR spectral simulations show that the products of the reaction are two triplet nitrenes with D_T = 1.026 cm⁻¹, E_T = 0 cm⁻¹ and D_T = 1.122 cm⁻¹, E_T = 0.0018 cm⁻¹, two quintet dinitrenes with D_Q = 0.209 cm⁻¹, E_Q = 0.039 cm⁻¹ and D_Q = 0.215 cm⁻¹, E_Q = 0.0545 cm⁻¹, and a septet trinitrene with D_S = −0.1021 cm⁻¹, E_S = −0.0034 cm⁻¹ in a 14:12:20:5:1 ratio (Figure 1a). Surprisingly, the ZFS parameters of quintet and septet nitrenes formed in the reaction were very close to the magnetic parameters of previously studied quintet dinitrenes 4c (D_Q = 0.210 cm⁻¹, E_Q = 0.039 cm⁻¹) and 5c (D_Q = 0.209 cm⁻¹, E_Q = 0.0542 cm⁻¹) and septet trinitrene 6c (D_S = −0.1011 cm⁻¹, E_S = −0.0043 cm⁻¹) [31]. The EPR spectra of such nitrenes, including the complete assignment of all transitions, have been reported previously [29-31]. Quintet and septet nitrenes formed during the photolysis of triazide 11 display very similar EPR spectra.

The septet product formed in the reaction can be safely assigned to trinitrene 18 since only this trinitrene has a septet spin state. According to EPR spectral simulations (Figure 1a), this trinitrene is formed in ca. 2% yield. Due to different substituents in positions 3 and 5 of the pyridine ring, trinitrene 18 has different angles Θ₁, Θ₂, Θ₃ between the nitrene C–N bonds and different Mulliken spin populations on the nitrene units (Table 1). Theoretically, such asymmetric trinitrenes should show rather large E_S values [30]. Nevertheless, the E_S value of trinitrene 18 is even smaller than that of C_2v symmetric trinitrene 6c [31]. DFT calculations show that trinitrene 18, on comparison with C_2v symmetric trinitrenes 6a-c, has slightly different orientations of the principal magnetic axes due to the presence of only one heavy chlorine atom. Thus, the principal axes D_XX^SS and D_XX^Tot of C_2v symmetric trinitrene 6a were directed along the line connecting the nitrene units in positions 2 and 6 of the pyridine ring and were parallel to the D_ZZ^SO axis of the spin–orbit tensor D^SO connecting two heavy chlorine atoms in positions 3 and 5 of the pyridine ring [30]. In the case of trinitrene 18, the D_ZZ^SO axis of the spin-orbit tensor D^SO almost coincides with the C(3)–Cl bond (Figure 2). As a result, the principal axis D_XX^Tot in trinitrene 18 is turned away from the line connecting the α-nitrene units by about 4°.

Previous studies of nitrenes 6a–c have shown that the most accurate theoretical evaluations of D_S and E_S by the DFT approach were obtained by using the PBE and BLYP functionals in combination with the DZ or EPRII basis sets [32]. Extensive DFT calculations performed in the present work show that the best agreement between the experiment and theory is observed at the BLYP/EPRII level of theory, overestimating the experimental D_S and E_S values of trinitrene 18 by only 6% (Table 2). The breakdown of D_S and E_S into D^SS, D^SO, E^SS, and E^SO reveals that the BLYP/EPRII calculations are especially good in estimations of the spin–orbit coupling contributions E^SO to the small parameters E_S of septet trinitrenes (Table 3).

By analogy with previous studies of dinitrenes 5a–c [29-31], the quintet molecule with D_Q = 0.215 cm⁻¹ and E_Q = 0.0545 cm⁻¹ can safely be assigned to dinitrene 15 with Θ = 115.7°. Thus, for instance, quintet dinitrene 5a showed D_Q = 0.202 cm⁻¹ and E_Q = 0.0554 cm⁻¹ [30], and its difluoro-derivative 5b displayed D_Q = 0.213 cm⁻¹ and E_Q = 0.0556 cm⁻¹ [29]. The most accurate theoretical estimations of the magnetic parameters in dinitrene 5a–c were obtained by using the PBE/DZ calculations [32]. Such calculations are also the most accurate for estimations of D_Q in dinitrene 15 (Table 2). The theoretical value of D_Q = 0.2226 cm⁻¹ obtained from the PBE/DZ calculations overestimates the experimental D_Q value in dinitrene 15 by just 3.5%. On the other hand, as in the case with trinitrene 18, the BLYP/EPRII calculations predict the most realistic value of E_Q = 0.0572 cm⁻¹ in dinitrene 15 (Table 2). According to EPR spectral simulations, dinitrene 15 is formed in the photolysis of triazide 11 in only ca. 10% yield.

The major quintet product of the reaction is the dinitrene with D_Q = 0.209 cm⁻¹ and E_Q = 0.039 cm⁻¹ (38%). This dinitrene can be assigned either to dinitrene 16 or to dinitrene 17. Previous EPR studies have shown that quintet dinitrenes with Θ = 121 ± 1° displayed E_Q = 0.04 ± 0.001 cm⁻¹, and dinitrenes with Θ = 125 ± 1° showed E_Q = 0.035 ± 0.001 cm⁻¹ [21,29-31]. According to DFT calculations, the dipolar angles Θ in dinitrenes 16 and 17 are equal to 121.6° and 125.4°, respectively (Table 1). These data suggest that the dinitrene with E_Q = 0.039 cm⁻¹ can safely be assigned to dinitrene 16. The most realistic theoretical estimation of E_Q in dinitrene 16 is obtained from PBE/DZ calculations of the spin–spin interaction parameter E_Q^SS (Table 3). This fact indicates that the contribution of the spin–orbit interactions to the total parameter E_Q of dinitrene 16 is quite small, as is typical for most of quintet dinitrenes [32].

Although we do not observe diagnostic signals of dinitrene 17 in the experimental EPR spectrum, this dinitrene may be a minor product of the photolysis of triazide 11. Extensive EPR spectral simulations show that five other quintet molecules with E_Q/D_Q = 0.038/0.213, 0.037/0.217, 0.036/0.221, 0.035/0.225 and 0.034/0.229 also display intense Y₂ transitions at 3077 G beside dinitrene 16 with E_Q/D_Q = 0.039/0.209. In EPR spectra of individual molecules, these six quintet species are easily identified owing to different positions of their Z₁, X₁, X₂, Z₂, A₁ and A₂ transitions (See Supporting Information File 1). However, when the mixture of dinitrenes 16 and 17 is formed, isomer 17 becomes detectable in EPR spectra only at relatively high concentrations (>5%). Thus, for instance, in EPR spectra of two quintet molecules with E_Q/D_Q = 0.039/0.209 and 0.038/0.213, the minor component with E_Q/D_Q = 0.038/0.213 becomes visible when its ratio to the major component exceeds 1:6 (14%). Based on these data, we do not exclude that dinitrene 17 with E_Q/D_Q ≈ 0.036/0.221 may be formed in ca. 5% yield along with dinitrene 16 (≈33%) during the photolysis of triazide 11.

Theoretically, several triplet mononitrenes can be formed during the photolysis of triazide 11 (Figure 3). As a rule, triplet pyridyl-2-nitrenes show D_T = 1.03 ± 0.02 cm⁻¹ and E_T ≈ 0 cm⁻¹, while triplet pyridyl-4-nitrenes show D_T = 1.13 ± 0.02 cm⁻¹ and E_T ≈ 0 cm⁻¹ [29-31]. When chlorine atoms are set in ortho-positions to the nitrene center, such triplet pyridylnitrenes show E_T > 0.003 cm⁻¹ [30]. Since one of triplet mononitrenes obtained in the present work shows D_T = 1.026 cm⁻¹ and E_T = 0 cm⁻¹, this nitrene can safely be assigned to 12 (Scheme 2). Another triplet mononitrene shows a D_T = 1.122 cm⁻¹ typical for pyridyl-4-nitrenes and a small parameter E_T = 0.0018 cm⁻¹ that is characteristic for triplet nitrenes with one chlorine atom in an ortho-position to the nitrene center. This mononitrene can safely be assigned to 13. According to EPR spectral simulations, triplet nitrenes 12 and 13 are formed in 27 and 23% yield, respectively. DFT calculations of the parameters D_T^SS agree well with the experiment, overestimating the D_T parameters of nitrenes 12 and 13 by just 2 and 3%, respectively (Figure 3). Unfortunately, as in the case of many other triplet mononitrenes [32], such calculations are too crude for estimations of the parameters E_T, predicting E_T = 0.006 ± 0.001 cm⁻¹ for all isomers of 12–14.

Recent EPR studies have shown that the best EPR spectral simulations for high-spin nitrenes were obtained only when an additional line-broadening parameter Γ(E) was used in the spin-Hamiltonian calculations [24]. The necessity of the use of this parameter in calculations is due to the presence in matrices of numerous conformational isomers of the starting azides. Upon UV irradiation, these conformers decompose to high-spin nitrenes that slightly differ each from other in the angles N–C–C and Θ and the parameters E. The formation of such high-spin nitrenes affects significantly the line-width and intensity of some lines in the experimental EPR spectra. Similar effects were also observed in EPR spectra of nitrenes 15–18. In order to adequately reproduce the experimental EPR spectrum from Figure 1b, the line-broadening parameters Γ(E) = 75 MHz for dinitrenes 15 and 16 and Γ(E) = 40 MHz for trinitrene 18 were used in the spectral simulations. According to these values of Γ(E), the variations in the angles Θ for dinitrenes 15/16 and trinitrene 18 do not exceed 0.7° and 0.4°, respectively. Despite these very small variations in the angles Θ, all our attempts to theoretically reproduce the experimental EPR spectrum from Figure 1b without the use of the line-broadening parameters Γ(E) were unsuccessful. EPR spectra of nitrenes 15, 16 and 18 simulated for Γ(E) = 0, 40 and 75 MHz are presented in Supporting Information File 1.

The results obtained show that almost all triplet and quintet nitrenes detected in the present study arise from the photolysis of the azido groups located in ortho-positions to the fluorine atom of pyridines. On comparison with the chlorine, the fluorine is a much more electron-negative and less bulky atom. Both of these factors favor strong π– conjugation of the ortho-azido groups with the pyridine ring. On photoexcitation, such azido groups should be more efficiently involved in reorganization of the molecular electronic system and more easily adopt geometries of the locally excited predissociation states [33,34]. DFT calculations show that namely azido groups located in positions 4 and 6 of triazide 11 have high localization of the lowest unoccupied molecular orbital (LUMO) density (see below in Figure 4). Similar localization of the LUMO density has previously been calculated for the azido group set on the γ-phenyl ring of triazide 7, which underwent selective photolysis to form triplet nitrene 8 (Scheme 1) [10]. All these data indicate that of the three azido groups of triazide 11 the azido groups in ortho-positions to the fluorine atom should be the most photoactive.

The photodissociation of the azido groups in triazide 11 can be modeled by computational methods [8,34]. Thus, for instance, the geometry optimizations of the most stable rotamers 11a and 11b in their lowest singlet excited states by using CIS/PM3 or CIS/6-311+G* calculations yield structures 11a-S₁* and 11b-S₁*, in which the azido groups in ortho-positions to the fluorine atom are locally excited (Scheme 3). These calculations confirm that the local excitation of such groups requires lower activation energies [8,34]. According to theory [34-36], the local excitation of the azido group during the photoexcitation results in considerable bending of the N=N=N fragment from about 171° to about 117° and appearance of the σ-type antibonding interactions in the N–N₂ bond. Figure 4 illustrates these effects for the structure 11a-S₁*.

Thus, the first stage of the photolysis of triazide 11 in argon matrices involves the photochemical generation of excited states 11a-S₁* and 11b-S₁*, which then undergo dissociation of the locally excited azido groups to form triplet nitrenes 12 and 13 (Scheme 3). On the second stage, triplet nitrenes 12 and 13 are excited to triplet states 12a-T₁*, 12b-T₁* and 13a-T₁* which then dissociate to quintet dinitrenes 15 and 16. The latter is formed as the major product owing to efficient excitation of the azido groups located in ortho-positions to the fluorine atom of nitrenes 12 and 13. The photoexcitation of the azido group in position 2 of nitrene 12 is a less efficient process leading to the formation of dinitrene 15 in just about 10% yield. The photoexcitation of the azido group in position 2 of nitrene 13 is a still much less efficient process. Quintet dinitrene 17 either is not formed at all or its yield is very low (<6%). Finally, the photodecomposition of the remaining azido groups in dinitrenes 15 and 16 gives septet trinitrene 18 in just about 2% yield. The low yield of trinitrene 18 may be associated with low efficiency of the azido group excitation in quintet dinitrene 16. One also cannot exclude that the photoexcitation of quintet azidonitrenes into predissociation states is a much less efficient process in comparison with the photoexcitation of singlet azides. In contrast to singlet azides, quintet azidonitrenes already have four singly occupied orbitals, and photoexcitation of these species may lead to excited states in which the azido groups are not locally excited.

Conclusion

In contrast to theoretical expectations, photodissociation of triazide 11 with light at λ = 260–320 nm occurs selectively on the azido groups located in positions 4 and 6 of the pyridine ring to give triplet mononitrenes 12 and 13 as the primary photoproducts. On further irradiation, quintet dinitrenes 15 and 16 and septet trinitrene 18 are formed. The maximum yield of the latter is about just 2%. Preferential photodissociation of the azido groups located in ortho-positions to the fluorine atom of pyridines is associated with strong π-conjugation of these groups with the pyridine ring. On photoexcitation, such azido groups are more efficiently involved in reorganization of the molecular electronic system and more easily adopt geometries of the locally excited predissociation states. Despite the lack of symmetry, trinitrene 18 shows a small parameter of E_S = −0.0034 cm⁻¹ due to a large positive value of the spin-orbit coupling parameter E_S ^SO (≈0.005 cm⁻¹). In overall, the ZFS parameters of 18 are very close to the ZFS parameters of previously studied C_2v symmetric septet pyridyl-2,4,6-trinitrenes. The most accurate theoretical predictions of the ZFS parameters of 18 are obtained at the BLYP/EPRII level of theory. At the same time, modern DFT calculations are still too crude to be used for reliable EPR spectral identification of such structurally alike quintet isomers as dinitrenes 16 and 17. Such quintet isomers are better discriminated based on analysis of their dipolar angles Θ. Among theoretical approaches, the most accurate estimations of the ZFS parameters of dinitrene 16 are obtained from PBE/DZ calculations of its parameters D_S^SS and E_S^SS. The study also demonstrates that successful EPR spectral simulations for nitrenes 15, 16 and 18 are possible only if additional line-broadening parameters Γ(E) of 40–75 MHz along with appropriate values of g, S, D, E and ν₀ are used in the spin-Hamiltonian calculations.

Experimental

Triazide 11 was synthesized according to the literature procedure [37]. X-band EPR spectra were recorded with a Bruker-Elexsys E500 EPR spectrometer with an ER077R magnet (75 mm gap between pole faces), an ER047 XG-T microwave bridge, and an ER4102ST resonator with a TE₁₀₂ cavity. Solid argon matrices doped with triazide 11 were prepared by vacuum co-deposition of two separate molecular beams (Ar and triazide 11 vapor) on the tip of an oxygen-free high-conductivity copper rod (75 mm length, 3 mm diameter) cooled at 5 K. The vapor of 11 was produced by an oven heating the polycrystalline 11 to 60 °C. The matrix-isolated samples were irradiated with a high-pressure mercury lamp, by using a filter passing the light at λ = 260–320 nm, and spectra were recorded at various irradiation times.

The computer simulations of EPR spectra were performed by using the EasySpin program package (version 4.0.0) [38]. The simulations were performed by using matrix diagonalization methods for S = 1, 2 or 3 using the parameters ν₀ = 9.605832 GHz and g = 2.0023 and line widths ΔH = 60 G for S = 1, and 30 G for S = 2 and S = 3.

The geometries of the molecules were optimized at the B3LYP/6-311G(d) level of theory with the Gaussian 03 program package [39]. The nature of the stationary points was assessed by means of vibrational frequency analysis. The spin-Hamiltonian parameters (g, D, E) and orientations of the D tensor were obtained from additional single-point calculations with the ORCA program package [40,41]. DFT calculations of the direct spin–spin (SS) coupling D^SS and spin–orbit coupling (SO) D^SO parts of the D tensors were performed by using the McWeeny–Mizuno and Pederson–Khanna approaches, respectively [42].