New approach to synthesis of nitronyl and imino nitroxides based on SNH methodology

It is shown that SN approach opens new possibilities in the synthesis of polyfunctional nitronyl and imino nitroxides. It is found that the interaction of 4,4,5,5-tetramethyl-4,5-dihydro-1Himidazol-3-oxide-1-oxyl lithium salt Li1 with 3,6-diaryl-1,2,4-triazines leads to formation of the corresponding triazines bearing nitronyl nitroxide or imino nitroxide substituent at position 5 of the heterocycle. The reaction of Li1 with pyridazine-N-oxide gives rise to nitroxide with buten3-ynyl substituent 5. Spin-labeled 5 could be readily transformed by the use of 1,3-dipolar and nucleophilic addition reactions, as well as oxidative coupling, that gives a large group of new paramagnets: 2-(1H-pyrazol-5-yl)vinyl-, 2-ethynylcyclopropyl-, 2-(3-(ethoxycarbonyl) isoxazol5-yl)vinyl-, 1-(pyrrolidin-1-yl)but-3-ynyl-substituted nitronyl nitroxide and a diradical – 2,2′((1E,7E)-octa-1,7-dien-3,5-diyne-1,8-diyl)bis(4,4,5,5-tetramethyl-4,5-dihydro-1H-imidazol-3oxide-1-oxyl). The new nitroxides were characterized by X-ray single crystal data, ESR and static magnetic susceptibility measurements.


Introduction
The increasing application of polyfunctional nitronyl nitroxides (NNs) 1 in the field of molecular magnetism stimulates the development of the chemistry of these compounds. 2The latter includes developing new methods for the targeted synthesis of kinetically stable NNs and methods for their isolation in the form of individual phases, as well as searching for new and general approaches to the synthesis of polyfunctional NNs.A possible pathway involves using a certain synthone that preserves the paramagnetic fragment.NN H1 is such a candidate synthone.Under the action of KOBu t , LiNPr i 2 , or LiN(SiMe 3 ) 2 it forms stable paramagnetic organometallic derivatives M1 {pK a (H1) ≈ 21.9 3 }, which upon further reaction with electrophilic reagents yield the products of C(2)-functionalization. 4 In the course of our study of nucleophilic substitution of hydrogen atom in π-deficient azine-N-oxides under the action of organolithium derivative Li1 we found an efficient method for production of hetaryl-substituted NNs based on S N H methodology (Scheme 1). 5

Syntheses
According to generally accepted views the S N H reactions occur in two stages. 6At the first stage a nucleophilic fragment adds to a compound having electron-withdrawing substituents forming a σ Н -adduct, and at the second stage these intermediates aromatize into the corresponding S N H products.Two types or aromatization exist, viz., oxidative and eliminative, thus leading to two types of S N H reactions: S N H (АО) and S N H (АЕ).The first process requires an oxidizing reagent.
The elimination-type aromatization requires the presence of an auxiliary group with propensity to anionic stabilization in the (hetero)aromatic substrate or in the nucleophilic moiety ("vicarious" nucleophilic substitution 7 ).We found that interaction of highly electrophilic derivatives of 1,2,4-triazine (3,6-diphenyl-1,2,4-triazine 2a or 3-(4-ethylphenyl)-6-phenyl-1,2,4-triazine 2b) with Li1 results in formation of the products of S N H reactions 3a and 4a or 3b and 4b, respectively, as the main products (Scheme 2).Formation of imino nitroxides 4a,b demonstrates that the presence of the nitronyl nitroxide group steers the aromatization of the σ Н adducts into the eliminative route.It is important that in this case the auxiliary group is in the nucleophilic moiety.

Scheme 2
The possibility of the discussed reactions involving azines was far from obvious, as the nucleophilicity of the carbanionic center in Li1 is substantially reduced.This is the most likely reason for the failure to perform the reaction of Li1 with mono-(quinoline) and diazines (quinoxaline, pyrimidine) to obtain the corresponding hetaryl-substituted NNs.In these cases only the initial H1 and the known diradical 1-1 could be isolated from the reaction mixture.
When using pyridazine-N-oxide as the electrophilic reagent in these transformations 2imidazoline-3-oxide-1-oxyl containing in its structure a butenynyl substituent was obtained.
Pyridazine-N-oxide reacts with Li1 in THF at -78 °C with formation, from XRD data, of exclusively E-isomer 5 in 68% yield.The enyne-substituted NN 5 is a "stable" at ambient conditions dark-blue crystalline solid that decomposes only upon heating to ~94-98 °C (Scheme 3).It should be noted that the works of Okusa et al. 8 and Igeta et al. 9 describe the opening of the pyridazine cycle in N-oxide under the action of Grignard reagents (ArMgBr), where the reactions also proceed mostly with formation of l-arylbut-l-en-3-ynes.

Scheme 3
With the development of an efficient method for production of 5 bearing reactive synergistic groups the latter could be used as the basic paramagnetic building block for production of new NNs.Let us first consider the reactions of 1,3-dipolar cycloaddition.It was found that 5 slowly reacts with CH 2 N 2 in ether to form a mixture of two main products, pyrazolylvinyl-substituted NN 6a and a compound with 2-ethynylcyclopropyl substituent 6c.The regioselectivity of CH 2 N 2 addition to triple bond is readily explained by taking into account the strongly π-electronwithdrawing and unsaturated character of the nitronyl nitroxide group.
1,3-Dipolar cycloaddition of CH 2 N 2 to activated double bond probably produces pyrazolines that easily loose a molecule of nitrogen to form a mixture of two geometric isomers of cyclopropane 6c.In certain cases we also isolated amide 6b as a minor product.Its possible origin is the reaction of 6a with methyl isocyanate, 10 which is formed in the samples of Nnitroso-N-methyl urea upon storage (Scheme 4). 11Regioselectivity was also observed in the interaction of 5 with N-oxide of nitrile 7 generated in situ from ethyl chlorooximidoacetate in the presence of Et 3 N 12 that produced a disubstituted isoxazole 8.

Scheme 5
At the same time nitroxides 5 and 9 behave quite differently in the reactions with secondary amines.Thus, while 9 in reaction with pyrrolidine forms exclusively the product of transaddition, 14 for 5 the nucleophilic attack proceeds at the α-carbon of the double bond to produce a mixture of optical isomers 12 (Scheme 6).

Scheme 6
The sensitivity of 5 to the presence of nucleophilic reagents in the reaction mixture probably precluded its oxidative coupling in classic variants (Cu(OAc) 2 , Py, 60-70 °C15 and O 2 , CuCl, Py 16 ), as terminal alkyne 5 in these conditions rapidly decomposes.On the other hand, in the presence of a sterically hindered amine and CuCl 17 the reaction proceeded smoothly to produce diradical 13 in 50% yield.

Crystal structures and magnetic properties
Nitroxides 3а,b, 4a,b, and 5 upon crystallization from a mixture of CH 2 Cl 2 with n-heptane formed solid phase as very thin flakes adhering to each other.For compounds 3а, 4a, 4b, and 5 single crystals could be selected and analyzed by XRD.The latter demonstrated that the structure of radicals 3а, 4a, and 4b is characterized by a large value of the dihedral angle α between the median plane of the paramagnetic fragment and the plane of the heterocycle: the angle α is equal to 58.9° in 3a and 69.6° in 4a, while in two crystallographically independent radicals 4b the value of α reaches 89.7 and 86.0° (Figure 1).The twist of the heterocycles is caused by the presence of vicinal phenyl substituent in 3а and 4a,b, with the angle between its plane and the plane of the triazine moiety being in the range 39.9-57.3°,and the phenyl cycle remote from the paramagnetic substituent being twisted by a smaller angle of 13.4-18.5°.Breaking of the conjugation between the paramagnetic moiety and the heterocycle imparts a non-typical for the family of NNs brown-reddish color to crystals of 3а and 3b, which is preserved after dissolution of these paramagnetics, e.g., in CH 2 Cl 2 .An important parameter for the crystal structure of nitroxides is the intermolecular distance between the paramagnetic centers, the oxygens in the N-O groups.In the structures of 3а and 4a,b these distances exceed 4.0 Å, which implies for them a typical behavior for a weakly coupled spin system.Indeed, the experimentally found values of effective magnetic moment (µ eff ) for 3а and 4a,b in the range 300 to 15 K are constant and close to 1.73 B.M. In the case of 3a the value of µ eff increases up to 1.88 B.M. with decreasing the temperature down to 2 K, indicating the domination of weak ferromagnetic interactions in solid state, for 4a,b the value of µ eff somewhat decreases upon cooling, pointing to a weak antiferromagnetic coupling between the unpaired electrons of the paramagnetic centers, while for 3b it remains unchanged (see Supporting Information).
The breaking of the conjugation between the nitronyl nitroxide moiety and the π-system of the substituent mentioned above for 3а and 4a,b is not present in 5, since according to XRD data the angle α (the angle between the planes of ONCNO and -CH=CH-C≡CH) for it is only 5.7º, and the distance C1-C8 of 1.435(2) Å is substantially shorter than single C-C bonds (Figure 2).For nitroxide 5 the bond lengths C8-C9, C10-C11, and C9-C10, equal to 1.317( 2 1.422(2) Å, respectively, correspond to data for conjugate systems С=C-C≡C. 18The distances in the N-O groups are practically equal, -N2-O2 1.278(1) and N1-O1 1.277(1) Å.Short intermolecular contacts O2…O2' (3.168 Å) are realized in solid 5 that couple the radicals into exchange-coupled dimer.In full agreement with this the experimental dependence µ eff (T) is well described within the Bleaney-Bowers model 19 for two-center S = 1/2 exchange clusters taking into account inter-dimer interactions in the molecular field approximation, 20 g = 2, J = -39 K, and nJ' = -1.3K. (see Supporting Information).
Unequivocal determination of the structures of nitronyl nitroxides 6a and 6b required their XRD study.The necessary single crystal samples were obtained by slowly evaporating solutions of 6a and 6b in the mixture of CH 2 Cl 2 with n-heptane.It was found that the structure of 6a is built by a packing of dimers forming via N-H…O hydrogen bonding of the radicals between the NH group of the pyrazolyl cycle and the oxygen of one of the NO groups (O…O distance is equal to 3.778 Å) (Figure 3).This allowed to determine the position of the NH fragment in the pyrazolyl ring and thus to demonstrate the regiospecificity of the reaction of CH 2 N 2 addition to triple bond of the enyne 5.Although two crystallographically independent molecules of 6a are present in the structure of 6a, their geometric parameters are practically identical.The largest difference is found for the angle between the plane of the pyrazolyl ring and the ONCNO fragment: in one half of the nitroxides it is equal to 9.4°, and in another part 10.4°.The lengths of the N-O bonds fall in the range 1.270(5)-1.291(5)Å. O…O contacts (3.778 and 3.781 Å) were found between dimers that lead to chains in the structure.In radical 6b the value of the angle between the plane of the pyrazole cycle and the NCN fragment of the imidazoline cycle is somewhat higher than in 6a and is equal to 15.4°.The nitroxides are linked into chains due to N-H…O type hydrogen bonds between the amide groups (Figure 4).The experimentally found value of µ eff (1.73±0.01B.M.) for 6b at 75 K corresponds to theoretical value for a monoradical.Upon cooling of the samples below 25 K a decrease in the value of µ eff is observed indicating weak exchange interactions of antiferromagnetic character between the paramagnetic centers (see Supporting Information).
After recrystallization, nitroxide 6c was always isolated from mother solutions in the form of oil.It crystallized only as a trinuclear complex with copper(II) hexafluoroacetylacetonate [(Cu(hfac) 2 ) 3 (6c) 2 ]•C 7 H 16 , isolated as violet crystals by slowly evaporating the mixture of nheptane containing equivalent amounts of Cu(hfac) 2 and 6c (see Supporting Information).
Crystals of diradical 13 were grown form a mixture of ethyl acetate with n-heptane.According to XRD data bond lengths in N-O groups are typical for nitroxides and fall in the range 1.26-1.30Å, the lengths of the C10-C11 and C10A-C11A bonds correspond to triple bonds (1.20(2) Å), and C8-C9 and C8A-C9A (1.29(2) and 1.32(2) Å) bonds -to double bonds (Figure 5).In diradical 13, similar to enyne 5, single bonds C7-C8, C7A-C8A, and C11-C11A are significantly shorter than 1.54 Å typical for single bonds.The angle between the planes of the NCN fragments of the imidazoline cycles in diradical 13 is equal to 48.3º.The diradical nature of 13 is supported by its value of the effective magnetic moment (µ eff ), which in the range 75-300 K is practically constant and equal to 2.45 B.M. (see Supporting Information).Since the intermolecular distances between the oxygens of the nitroxide groups are rather large (more than 4.46 Å), the µ eff (T) dependence displayed by 13 is determined by intramolecular antiferromagnetic exchange between the unpaired electrons.Indeed the experimental dependence µ eff (T) is well described within the Bleaney-Bowers model 19 for isolated two-center S = 1/2 exchange clusters with g = 2.0 and J = -16 K.

Electron paramagnetic resonance
Figure 6a shows the experimental and simulated ESR spectra for 5.The experimental spectrum was taken in a degassed toluene solution at room temperature, concentration about 10 -5 M, the estimated accuracy of determination of the hyperfine constants is 0.005 mT (half of the modulation amplitude of 0.01 mT), g-value was measured using solid DPPH as the reference.Modeling of spectra yielded: A N1 = 0.749 mT, A N2 = 0.726 mT, A 12H = 0.021 mT, A H1 = 0.146 mT, A H2 = 0.120 mT, A H3 = 0.067 mT, g iso = 2.0065.The values of the nitrogen hyperfine couplings show that the two nuclei are not completely equivalent in solution due to a relatively rigid extended π-system encompassing the imidazoline moiety and the substituent.Furthermore, ESR spectra show rather pronounced alternating linewidth effect due to modulation of nitrogen couplings, 21 most probably because of "wiggling" the substituent with respect to the imidazoline ring.Experimental and simulated ESR spectra for 3a are shown in Figure 6b.Modeling yielded: A 2N =0.717 mT, A 12H =0.017 mT, A Na =0.086 mT, A Nb =0.022 mT, A Nc =0.012 mT, g iso =2.0065.The dominant structure of the spectrum comes from two equivalent imidazoline nitrogens, one nitrogen from the substituent, and 12 protons from the four guarding methyl groups.Two other nitrogens have rather small hyperfine coupling constants, which are not reliably determined from simulations.The spectrum for 3b (not shown) is almost identical, as the two radicals differ only in a remote ethyl group in the substituent that has practically no effect on the spectrum.Modeling yielded: A 2N =0.716 mT, A 12H =0.018 mT, A Na =0.086 mT, A Nb =0.019 mT, A Nc =0.011 mT, g iso =2.0065.
Figure 7a shows ESR spectrum for 4a (bottom trace) and its modeling (top trace) that yielded the following parameters: A N1 =0.855 mT, A N2 =0.431 mT, A 12H =0.019 mT, A Na =0.054 mT, A Nb =0.012 mT, A Nc =0.007 mT, g iso =2.0059.The characteristically looking spectrum is dominated by two imidazoline nitrogens with the ratio of hfc's approximately 1:2, with an additional nitrogen from substituent and 12 protons from the guarding methyls.Two further nitrogens with minor couplings were introduced into simulations, but these constants are too small to be reliable.As with 3a/b, the spectrum for 4b is very similar, as it is not sensitive to the remote ethyl.The extracted parameters are as follows: A N1 =0.856 mT, A N2 =0.431 mT, A 12H =0.019 mT, A Na =0.054 mT, A Nb =0.012 mT, A Nc =0.008 mT, g iso =2.0060, with two smaller couplings being not very reliable.Figure 7b shows the spectra for 6a.The dominant structure comes from two slightly nonequivalent imidazoline nitrogens, due to locked structure of the radical in solution with hindered relative motion of the imidazoline and the substituent, and two bridge protons.The spectrum is not resolved, but the standard substructure of 12 protons from the four guarding methyl groups reproduces the widths of the lines.Further nuclei with small hfc's could have been introduced in modeling, but the spectrum does not warrant this as it lacks finer details that need to be reproduced in simulations.The parameters are: A N1 =0.732 mT, A N2 =0.762 mT, A 12H =0.016 mT, A Ha =0.137 mT, A Hb =0.114 mT, g iso =2.0066.
Figure 8a shows ESR spectra for 8. Modeling (the top spectrum) yielded: A N1 =0.735 mT, A N2 =0.725 mT, A 12H =0.019 mT, A Ha =0.122 mT, A Hb =0.139 mT, A Hc =0.026 mT, A Na =0.036 mT, g iso =2.0066.The dominant structure comes from two nearly equivalent imidazoline nitrogens (this produces the alternating resolved/unresolved substructures of the lines in the quintet), two bridge protons, and 12 methyl protons.Figure 8b shows the rather weird looking spectrum for 12 that was well reproduced in modeling (top trace).The spectrum has a peculiar substructure of the lines of the 12321 quintet from two imidazoline nitrogens -a quartet of lines with two outermost lines showing a resolved substructure, while the two inner lines are unresolved.The dominant structure comes from two equivalent imidazoline nitrogen (the modeling invariably produced equal values for two independently varied couplings), one proton and one nitrogen with rather large couplings, a standard set of 12 methyl protons, and two slightly different protons, most likely methylene ones (it is their "difference" that produces the specific look of the spectrum).The obtained parameters are: A 2N =0.744 mT, A 12H =0.022 mT, A Ha =0.115 mT, A Na =0.146 mT, A Hb =0.037 mT, A Hc =0.026 mT, g iso =2.0066.Spectra for 6c in Figure 9a show the dominating structure from two equivalent imidazoline nitrogens and 12 methyl protons, but reproduction of the width of the lines (having a resolved substructure) required introduction of further nuclei with small hfc's consistent with the chemical structure of the radical (three independent protons and two equivalent protons in the cyclopropane ring).The modeling reproducibly gives equivalent coupling for the two imidazoline nitrogens, and yielded the following parameters: A 2N =0.747 mT, A 12H =0.021 mT, A Ha =0.044 mT, A Hb =0.028 mT, A Hc =0.023 mT, A 2H =0.046 mT, g iso =2.0064.Finally, the Figure 9b shows the spectra for diradical 13 and its modeling that faithfully reproduces the experiment.The spectrum corresponds to fast exchange limit with couplings to four equivalent nitrogens of two imidazoline moieties A 4N =0.365 mT and the peak-to-peak width of unresolved lines ∆H pp =0.091 mT, g iso =2.0064.The diradical is rather stable in degassed toluene solution, and after about ten hours, when the temperature of solution was varied from 183K to 353K (approximately melting and boiling points of toluene) and the sample was then cooled down back to room temperature, gave the same spectrum as the freshly prepared sample, both in terms of shape and intensity.No signs of monoradical spectra were traced, and there were no changes to the spectrum itself during the extended temperature experiment apart from the expected transformation from non-isotropic to isotropic spectrum upon melting of the matrix, and additional line broadening due to accelerated spin exchange at elevated temperatures.The diradical is thus kinetically and magnetically stable in degassed toluene solution.

Experimental Section
General.
Pyridazine-1-oxide, 22 2,3-bis(hydroxyamino)-2,3-dimethylbutane sulphate monohydrate 23 and ethyl 2-chloro-2-(hydroxyimino)acetate 24 8 were synthesized as described in literature.THF was distilled from sodium benzophenone ketyl in a recycling still.Other reagents and solvents from commercial sources were of the highest purity available and were used as received.The reactions were monitored by TLC using "Silica gel 60 F 254 aluminum sheets, Merck".Chromatography was carried out with the use of "Merck" silica gel (0.063-0.100 mm for column chromatography) for column chromatography.C, H, and N elemental analyses were carried out by the Chemical Analytical Center of the Novosibirsk Institute of Organic Chemistry.The melting points were determined on a Boethius apparatus and were not corrected.Infrared spectra (4000-400 cm -1 ) were recorded with a Bruker VECTOR 22 instrument in KBr pellets.UV/Vis spectrum of diradical 13 was registered with Hewlett Packard 8453 spectrophotometer.X-Band CW ESR spectra were recorded in dilute degassed toluene solutions at room temperature on a Bruker EMX spectrometer and modeled in free package Winsim v.0.96 as described earlier. 25The magnetochemical experiment was run on an MPMS-5S ("Quantum Design") SQUID magnetometer at temperatures from 2 K to 300 К in a homogeneous external magnetic field of up to 49.5 kOe.The molar magnetic susceptibility χ was calculated using Pascal's additive scheme including diamagnetic corrections.
Stage 2. To a cooled at about 5 °C mixture of 4,4,5,5-tetramethylimidazoline-1,3-diol (16 g, 0.1 mol), CHCl 3 (150 ml) and water (150 ml) NaIO 4 (34.2 g, 0.16 mol) was added in small portions under stirring over the period of 40 min.After further stirring for 30 min the organic layer was separated.The aqueous layer was filtered (from NaIO 3 ) and extracted with CHCl 3 (3×30 ml).The organic phases were combined, dried over Na 2 SO 4 and filtered through a layer of Al 2 O 3 (2×15 cm).The solution was diluted with 30 ml of n-heptane and evaporated (pressure 70 mm Hg, bath at room temperature) until the onset of product crystallization, after which the flask was kept at about -10 °C for 20 h.The residue was filtered off and washed on filter with cold nhexane to give 13.8 g (88%) of 0.5-3 mm dark-red crystals covered with small yellowish crystals visible in a microscope.The product, which according to TLC and IR data contained a minute amount of 1,3-dihydroxy-4,4,5,5-tetramethylimidazolidyn-2-one and 2-iodo-4,4,5,5-tetramethyl-4,5-dihydro-1H-imidazol-3-oxide-1-oxyl, was used in synthesis without additional purification, because in the course of the differentiating procedures the content of admixtures further increased.To prepare a reference sample nitroxide H1 was crystallized from n-hexane, and darkred flakes were manually selected from the deposited crystals, mp.(E)-2-(But-1-en-3-ynyl)-1-hydroxy-4,4,5,5-tetramethyl-4,5-dihydro-1H-imidazol-3-oxide-1oxyl (5).To a vigorously stirred solution of H1 (157 mg, 1.0 mmol) in 3 ml of absolute THF a 2.0 M solution of LDA (or 1.06 M of LiN(SiMe 3 ) 2 ) in THF (1.1mmol) was added at -78 °C under argon atmosphere.The reaction mixture was stirred at -78 °C for 20 min.To the obtained bright-red solution of the lithiated derivative Li1 a solution of pyridazine-N-oxide (106 mg, 1.1 mmol) in 4 ml of THF was added at -78 °C under argon atmosphere, and the reaction mixture turned into a greenish-brown suspension.The cooling bath was removed; the reaction mixture was stirred for 2 h and concentrated in vacuo.The obtained residue was put through a column with SiO 2 using ethylacetate as the eluent.The blue-colored fraction was evaporated until dry under reduced pressure, the obtained residue was recrystallized from a mixture of n-heptane and CH 2 Cl 2 .Yield 141 mg (68%), blue-green crystals, mp.94-98 °С (with decomposition), R f 0.65 (EtOAc)

X-Ray structure determinations
Crystal data for compounds were collected on a Smart APEX CCD diffractometer using graphite-monochromated Mo-Kα (λ = 0.71073 Å).The cell parameters were determined and refined by the least squares method for all reflections.The structures were solved by direct methods and refined by least squares procedures on F 2 .All non-hydrogen atoms were refined anisotropically.Positions of all hydrogen atoms were calculated geometrically and refined as riding on the respective carbon-bonded atoms.All structure solution and refinement calculations were performed with Bruker Shelxtl Version 6.12.The worst situation was with 13, which forms very fine thread-like single crystals.As a result, we determined only the structure of the molecule, and only the unit cell parameters are given for this compound.Compound (3a).

a b 3 .
Figure Structure of 6a (a) and H-bonded dimer (b) in the solid nitroxide 6a.

a b 4 .
Figure Structure of 6b (a) and H-bonded chain (b) in the solid nitroxide 6b.

Figure 6 .
ESR spectra of 5(a) and 3a (b): experimental (bottom trace) and the result of their modeling (top trace).

Figure 7 .
ESR spectra of 4a(a) and 6a (b): experimental (bottom trace) and the result of their modeling (top trace).

Figure 8 .
ESR spectra of 8 (a) and 12 (b): experimental (bottom trace) and the result of their modeling (top trace).

Figure 9 .
ESR spectra of 6c (a) and 13 (b): experimental (bottom trace) and the result of their modeling (top trace).