Origin of Long-Range Hyperfine Couplings in the EPR Spectra of 2,2,5,5-Tetraethylpyrrolidine-1-oxyls

Abstract

Cyclic nitroxides with several bulky alkyl substituents adjacent to the nitroxide group are known to demonstrate a much higher stability to bioreduction than their tetramethyl analogues. Among these so-called “sterically shielded” nitroxides, the pyrrolidine derivatives are the most stable. The EPR spectra of some sterically shielded pyrrolidine-1-oxyls were reported to show one or two large additional doublet splittings with a hyperfine coupling (hfc) constant (ca. 0.2–0.4 mT). To determine the origin of these hfc, a series of 2-R-2,5,5-triethyl-3,4-bis(hydroxymethyl)-pyrrolidine-1-oxyls with methylene groups stereospecifically enriched with deuterium were prepared, and their CW EPR spectra were studied. In addition, these nitroxides were investigated using quantum chemical calculations on the UB3LYP/def2-TZVP level and NBO analysis. The apparent constants were assigned to hfc with γ-hydrogen in the side chain, with the contribution of the NBO orbital ^βπ*(N–O) to the natural localized molecular orbital ^βσ(C–H) playing the major role. This interaction is efficient if the ethyl substituent is in the pseudoaxial position of the ring and the CH₂–CH₃ bond is codirected with (parallel to) N–O. The apparent constant a_H increases with the Boltzmann population of this conformation.

Introduction

Nitroxides have found plenty of applications in various fields of science and technology.¹⁻³ A broad majority of commonly used nitroxides have a cyclic structure with four methyl substituents adjacent to the N–O group. Recently, nitroxides with bulkier alkyl substituents, e.g., ethyls, instead of methyls attracted much attention. These so-called “sterically shielded” nitroxides demonstrate a much higher stability against chemical reduction to diamagnetic compounds with components of biological systems than their tetramethyl analogues.⁴⁻⁶ To date, sterically shielded pyrrolidine-1-oxyls demonstrate the highest resistance to reduction among all known nitroxides.⁷⁻⁹ They also demonstrate relatively high stability to oxidative destruction in the presence of rat liver microsomes.¹⁰ The EPR spectra of some of these nitroxides, e.g., structures 1a–c and 2 (Figure 1), reveal one or two large additional doublet splittings with a hyperfine coupling (hfc) constant (ca. 0.2–0.33 mT).^7,9,11 Similar spectral features were earlier observed for some cyclic nitroxides with a five-membered saturated ring and bulky substituents adjacent to the N–O· group, i.e., oxazolidine 3(12,13) and imidazolidine 4(14) nitroxides (Figure 1). To the best of our knowledge, a large hfc was never observed in the EPR spectra of conventional tetramethyl nitroxides.

Figure 1.

Structures of nitroxides 1–6.

To assign the splittings in the EPR spectra of 3 and 4, their partially deuterated analogues were used. Replacement of four methylene hydrogens to deuterium in geminal ethyl groups in position 2 or 5 in 4 or in methylene groups next to the spiro node in the cycloalkane ring in 3 resulted in the disappearance of additional splitting,^13,14 unambiguously showing that the unpaired electron has a large hfc with one of the γ-hydrogens in the side chain. This assignment for 4 was supported by the UB3LYP/def2-TZVP level DFT calculations. Large couplings reflect the difference between the Boltzmann populations of different conformations. The calculations predicted only one large a_H constant per pair of geminal ethyl groups for all low-energy conformers, assuming that all of these conformers contribute to the apparent average hfc constant. This result did not allow to assign the large constants neither to a definite atom(s) nor to definite ethyl group(s). Recently, another pattern of the hyperfine structure was described for newly synthesized sterically shielded nitroxides 5 and 6 with only one large splitting of ca. 0.38–0.40 mT on a hydrogen atom.^11,15

Complicated hyperfine structure and line broadening in the EPR spectra of sterically shielded nitroxides makes them inefficient in EPRI and OMRI applications, where radicals with a minimal number of narrow lines are required for better resolution and enhancement.¹⁶⁻¹⁸ Broadened spectral lines can also hamper analysis of molecular dynamics using spin labeling and EPR.⁶ A better understanding of the factors affecting the hyperfine structure in the EPR spectra of sterically shielded nitroxides could help in the molecular design of better structures.

The original method we used for the preparation of various sterically shielded nitroxides, such as 1, 2, 5, and 6, implies the addition of ethynylmagnesium bromide to corresponding cyclic nitrone with the subsequent hydrogenation of the terminal acetylene group.^9,11,15 This method perfectly fits for the preparation of 3,4-bis(hydroxymethyl)-2,2,5,5-tetraethylpyrrolidine-1-oxyls with varying numbers and positions of deuterium atoms. A set of partly deuterated analogues of 1a were prepared. The CW EPR spectra of these nitroxides and the corresponding ethynyl-substituted precursors were investigated. DFT calculations were performed for 1a–c and 5 and the obtained canonical orbitals were analyzed using the NBO method to reveal the relationship between the structure and the observed hfc constants.

Results and Discussion

Synthesis

The synthesis of 2,2,5-triethylpyrrolidine 7a (stereoisomeric mixture) from 2-aminobutanoic acid, diethyl ketone, and dimethyl fumarate and its conversion into nitrone 8a was described earlier⁷ (Scheme 1). The isotope-labeled compound 7b was prepared from 3-pentanone-(2,4-D₂). To prevent reverse exchange of deuterium for protium, mobile protons in 2-aminobutanoic acid were replaced for deuterium via iterative dissolution in D₂O evaporation procedures. The solvent mixture (DMF–toluene) was heated to reflux with D₂O in the Dean–Stark apparatus prior to the use in the synthesis of 7b.

Scheme 1. Synthesis of Nitrones 8a,b (in Analogy to Ref (7)).

Reduction of the diastereomeric mixture 7b with LiAlH₄ followed by oxidation with H₂O₂/Na₂WO₄ afforded nitrone 8b (Scheme 1). α-Alkylnitrones are known to demonstrate significant CH acidity and nitrone-ene-N-hydroxyamine tautomerism.¹⁹ Proton exchange in the alkaline solution of 8a,b was used to replace methylene hydrogens of the 2-ethyl substituent and hydrogen atom in position 3 of the heterocycle by deuterium (Scheme 2). The rate of exchange was remarkably different for these three hydrogens. For example, the intensity of the signals of diastereotopic methylene hydrogens in the ¹H NMR spectrum of 8a in 5% NaOD in D₂O at 20 °C decreased by 95 and 50% on 14 days, while the signal of the hydrogen atom in position 3 of the heterocycle showed a 13% initial intensity (see Figure S4, Supporting Information). The difference in the H-D exchange rates for diastereotopic protons is a well-known phenomenon (see, e.g., ref (20)). The exchange procedures were finished after the enrichment of the methylene group at the nitrone carbon with D reaching 95% in 8c,d.

Scheme 2. H-D Exchange in 8a,b.

We reported earlier that nitrone 8a can form an insoluble precipitate upon treatment with the Grignard reagent in THF, and this may impede addition to nitrone.⁷ To prevent the precipitate formation, the hydroxymethyl groups were protected with 2,2-dimethoxypropane. The addition of ethynylmagnesium bromide to 5-R-2,2-diethyl-3,4-bis(((2-methoxypropan-2-yl)oxy)methyl)-3,4-dihydro-2H-pyrrole 1-oxides (9a–c) was shown to afford a single isomer (racemate) with the ethynyl group in the cis-position to the neighboring hydroxymethyl group, and the structures of the corresponding nitroxides 10a–c were confirmed by single-crystal X-ray analysis data¹¹ (Scheme 3). Here, we used trimethylsilyl chloride to protect the hydroxymethyl groups in nitrones 8a–d. The resulting crude nitrones 11 without purification were treated with ethynylmagnesium bromide. Subsequent quenching and oxidation with air oxygen in the alkaline solution of methylene blue in analogy to the previously reported procedure¹¹ afforded nitroxides 10a and 12a–c. The ¹H NMR spectrum of the 10a sample prepared using TMS protection (see Figure S8, Supporting Information) showed no trace of other isomers and corresponded to the previously published data.¹¹

Scheme 3. Synthesis of Nitroxides 10a and 12a–c.

Nitroxides 10a and 12a–c were subjected to hydrogenation on Pd/C using ¹H₂ or ²H₂. Subsequent oxidation with air oxygen gave nitroxides 13a–d and 14, differing in the number and position of deuterium atoms in the molecule (Scheme 4).

Scheme 4. Synthesis of Nitroxides 13a–d and 14.

EPR Studies

In the hyperfine structure of the isotropic EPR spectrum of a nitroxide, a splitting on the nitroxide nitrogen (a triplet for ¹⁴N) is usually complemented with additional couplings with other magnetic nuclei, e.g., hydrogens. The coupling constants a_H may be positive or negative depending on the nitroxide structure and conformation. In a broad majority of nitroxides, fast conformational motions lead to dynamic averaging of hydrogen hyperfine couplings, making the resulting a_H small. As a result, multiple unresolved lines in the EPR spectra of conventional nitroxides are usually represented as inhomogeneously broadened triplet. If a conformation with high a_H prevails, e.g., due to steric interactions in the molecule, the resulting Boltzmann average a_H may be high enough to produce a significant visible splitting of the nitroxide triplet spectral lines. Examples of such behavior are given in the introduction. The deuterium nucleus has a higher nuclear spin (s = 1) but 6.5-fold lower hfc constant than ¹H. The partial deuteration of nitroxides 3 and 4 was previously used to suppress large splitting on the ¹H nucleus (a_H ≈ 0.2 mT) in the EPR spectra because smaller triplet splitting a_D ≈ 0.03 mT was no longer resolved and only contributed to the apparent line width.

The EPR spectra of nitroxides 1a, 10a, 12a–c, 13a–d, and 14 are shown in Figure 2, and the parameters of the spectra are listed in Table 1. The hyperfine structure of the spectra of the nitroxides shows clear dependence on the amount and position of ¹H and ²H atoms in the structure. In general, it follows the pattern previously described for 4:¹⁴ each pair of geminal ethyls adjacent to the nitroxide group can produce one additional large splitting. Replacement of all methylene hydrogens in geminal pairs of ethyls for deuterium suppresses the apparent large splitting (cf. 1a and 13b,d), but the lines remain inhomogeneously broadened because the unresolved hfc with deuterium contributes to the line widths. In our previous investigation, neither the experimental data nor high-level DFT calculations allowed assignment of the large a_H in the EPR spectra of 4a–c to a specific ethyl group of the geminal pair because low conformational barriers implied fast exchange between the conformations (ethyl group rotation and ring puckering).¹⁴ The above procedure of the synthesis of nitroxides 12, 13, and 14 afforded stereospecifically deuterated compounds, and this gives an opportunity to separate the influence of hydrogen atoms in different positions of the molecule on the hyperfine structure of the EPR spectrum. The spectra of partly deuterated nitroxides 13c and 13d are nearly identical triplets with minor differences in the line widths. In contrast, the spectrum of 14 shows additional splitting, a_H = 0.23 mT. Therefore, replacement of ¹H for ²H in the methylene moiety of the ethyl group trans to the neighboring hydroxymethyl one produces a negligible effect, while protons of the cis-ethyl group show a large hfc. Obviously, even if the conformational barriers are low, a hydroxymethyl group at the asymmetric center at position 3 of the ring brings strong asymmetry to the population of the conformations of neighboring ethyl groups. This feature makes nitroxide 1a fundamentally different from 3, 5, and 6, where conformations are degenerate and protons of both geminal substituents may equally contribute to the apparent large constant. As a practical consequence, it is sufficient to replace the protons of one particular methylene group in each geminal ethyl pair in 1a to remove the large splitting on γ-hydrogens in the EPR spectrum.

Figure 2.

Experimental (red) and calculated (black) EPR spectra of nitroxide shown on the right side. The parameters of simulation are listed in Table 1.

Table 1. Parameters of the EPR Spectra and Simulations.

					line width
no.	g	aN (mT)	a1 (mT)	a2 (mT)	Gauss (mT)	Lorenz (mT)
1a	2.0057 ± 5 × 10–5	1.53 ± 2 × 10–3	0.23(H)	0.23(H)	0.21	0.05
13a			0.23(H)	0.035(D)	0.19	0.05
13b			0.23(H)	0.035(D)	0.18	0.05
13c			0.035(D)	0.035(D)	0.16	0.05
14			0.035(D)	0.23(H)	0.19	0.03
13d			0.035(D)	0.035(D)	0.14	0.05
10a			0.22(H)		0.16	0.02
12a			0.034(D)		0.12	0.03
12b			0.22(H)		0.14	0.03
12c			0.034(D)		0.13	0.03

Comparison of the EPR spectra of various sterically shielded nitroxides with a five-membered saturated ring (see, e.g., refs (11) and (15)) shows that the value of a_H may vary over a wide range, varying with minor structural changes. For example, in 1a–c, the increase of the substituent R by one methyl group leads to the growth of one of the two a_H from 0.224 to 0.331 mT (R = i-Pr), but the addition of one more methyl group (R = t-Bu) results in the disappearance of one of the two a_H.¹¹ Quantum chemical calculations were used to find the general criteria for the appearance of large hfc splittings and to assign these hfc to specific hydrogen atoms (see below).

Quantum Chemical Calculations

The hfc constants on hydrogens for 1a (R = Et), 1b (R = i-Pr), 1c (R = t-Bu), and 5 were calculated at the UB3LYP/def2-TZVP level. It is known that special basis sets must be used for the precise prediction of the hfc constant on nitroxide nitrogen (a_N(iso)) and solvent effects with hydrogen bonding of the nitroxide group must be taken into account. However, these precautions are not needed for the estimation of hfc with hydrogens (a_H(iso)).^21,22 The observed ¹H hfc constants are averaged over the Boltzmann distribution of conformations, and conformations with lower energies have a greater statistical weight. To find the lowest-energy conformations, we used a systematic rotor search method at the MMFF94 level. Since the parameters of the force field for the nitroxide moiety were not available, the parameters for ketones of the same topology were used, whose geometry and electronic properties are very similar. It should be noted that the most favorable conformations found for 1a and 1c coincide with the conformations obtained from the X-ray analysis data.^7,11

The calculated hfc constants on hydrogen atoms for the lowest-energy conformation are shown with color in Figure 3. The hydrogen atoms with the highest a_H are shown in red. The hfc on hydrogens of the methyl group were not taken into account because fast rotation across the CH₂–CH₃ bond should lead to averaging. In agreement with the experimental data, the calculation predicted for 1a two large isotropic Fermi contact couplings on two ethyl groups in the cis-position to the neighboring hydroxymethyl group. The predicted values of a_H are in good agreement with the experimental data (Table 2). It should be noted that introduction of bulkier substituents (1b and 1c) produces significant changes in the geometry of the molecule. Nevertheless, the structures 1a–c and 5 demonstrate one common feature: the large a_H is predicted on the methylene hydrogen of the ethyl group in pseudoaxial position with the CH₂–CH₃ bond nearly parallel to the N–O axis. This conformation favors overlapping of the smaller lobe of one of the sp³-hybridized C–H bond orbitals of the methylene moiety of the ethyl group with the nonbonding orbital of the nitroxide group.

Figure 3.

Calculated structures of nitroxides 1a–c and 5. The calculated values of isotropic Fermi contact coupling constants are shown with color. The values of the large constants are given in MHz.

Table 2. Predicted and Calculated Values of the Isotropic hfc Constants for Nitroxides 1a–c and 5.

nitroxide	atom	aH(iso), MHz/Gauss	<aH(iso)>, MHz/Gauss	aH (experimental), Gauss
1a	H′	8.6/3.1	5.7/2.0	2.24
	H″			2.24
1b	H′	9.0/3.2	8.9/3.2	3.31
	H″	8.8/3.1	6.4/2.3	2.52
1c	H′	–2.3/–0.8	–1.3/–0.5	2.55
	H″	6.7/2.4	6.6/2.4
5	H′	11.3/4.0	11.2/4.0	3.64
	H″	3.2/1.1	2.3/0.8

To investigate the orbital interactions, NBO analysis was used, which showed a linear dependence of the isotropic constant of the hyperfine interaction on the natural spin density on hydrogen atoms, both within a single molecule and in comparison of molecules 1a–c and 5 with each other (see the Supporting Information, Figures S37–S41). Interestingly, high values of the hyperfine constants for some hydrogen atoms of methyl groups predicted by UB3LYP/def2-TZVP calculations cannot be explained within the framework of the natural spin density. Presumably, these outliers result from the direct proximity of the hydrogen atoms to the orbitals of the nitroxyl group.

The main process of the emergence of natural spin density on hydrogen atoms can be attributed to the transfer of electronic density from the ^βσ(C–H) orbital to the ^βπ*(N–O) orbital in the β-spin space due to their direct overlap. Another large transfer of electronic density is observed from the ^αn(N) orbital to the ^ασ*(C–H) orbital in the α-spin space, and it should be noted that both effects contribute to an increase in the difference in the α and β populations of the hydrogen atoms. This is clearly demonstrated by the decomposition of natural localized molecular orbitals on the basis of natural bond orbitals (see the Supporting Information, Table S2). In the series of nitroxides, a linear dependence is observed between a_H and the contribution of the NBO orbital ^βπ*(N–O) to the natural localized molecular orbital ^βσ(C–H) (see the Supporting Information, Figure S42).

The above mechanism of the emergence of spin density on hydrogen atoms should be highly sensitive to changes in the geometry of the molecule, mainly on dihedral angles φ (N–C–C–H′) and θ (O–N–C–C) (Figure 4).

Figure 4.

Overlapping of the ^βσ(C–H) and ^βπ*(N–O) orbitals and dihedral angles φ (N–C–C–H′) and θ (O–N–C–C).

Relaxed scanning of the dihedral angle φ (N–C–C–H′) was performed for nitroxides 1a–c and 5 (Figure 5 and Supporting Information, Figures S43–S49). The comparison of the graphs for 1a, 1b, and 1c shows that the conformation with the minimal energy does not always correspond to the highest a_H. Moreover, the value of a_H depends not only on the steric demand but also on the shape of the substituent.

Figure 5.

Relaxed scanning of the dihedral angle φ (N–C–C–H′) for nitroxide 1a.

Boltzmann averaging of the isotropic a_(iso)H constants over the rotation of the ethyl group certainly does not provide a complete account of the entire conformational ensemble, but it gives a qualitative representation. For instance, in the case of 1a and 1b, averaging over rotation at 298 K leads to a decrease in the average values of ⟨a_H′(iso)⟩, while in the case of 1c, it practically disappears. The initial a_H(iso) and averaged data ⟨a_H(iso)⟩ are listed in Table 2.

The estimated values of hfc constants ⟨a_(iso)H⟩ expectedly differ from experimental data a_H, but they adequately reflect the experimentally observed trends for nitroxides of both imidazolidine and pyrrolidine series. The emergence of the large a_H seems to be a general phenomenon for nitroxides with a five-membered saturated ring and large alkyl substituents adjacent to the N–O group. The large a_H should be attributed to γ-hydrogen in the side chain, for which the σ-orbital of the C–H bond overlaps with the nonbonding orbital of the nitroxide group. The apparent a_H is high if a conformation where such an overlapping occurs is predominant.

Conclusions

In this work, we studied the origin of the large additional hfc constants in the EPR spectra of some nitroxides with a five-membered saturated ring. The a_H of 0.15–0.4 mT were observed in the spectra of some sterically shielded nitroxides with large alkyl substituents adjacent to the N–O group. Some of these nitroxides demonstrate high resistance to bioreduction, but the complex structure of their EPR spectra makes their use in a number of applications unfavorable. The large hfc constants were found to result from overlapping of the smaller lobe of the C–H orbital in the α-position of the side alkyl chain and the nonbonding orbital of the nitroxide group. A high Boltzmann population of the conformation, where such an overlapping occurs, is a necessary condition for additional large splitting emergence. Conformational analysis of the nitroxide structure can reveal the hydrogen atoms, which contribute to the a_H, and selective replacement of these hydrogens to deuterium can suppress the undesired splitting.

Satisfactory prediction of the hfc constant value can be achieved via Boltzmann averaging over the rotation of the ethyl (alkyl) group in a low-energy conformation. These predictions can help in the molecular design of the new structures with desired properties. Minor structural changes may change the Boltzmann population of conformations and produce significant changes of a_H. The possibility of creating functional spin probes based on this principle deserves further study.

Experimental Section

Compounds 1a–c, 10a, and 5 were prepared according to the literature protocols.¹¹ The ¹H NMR spectra were recorded at 300, 400, or 500 MHz as indicated next to each NMR analysis. The NMR spectra of nitroxides were recorded from their amines formed under reduction by the Zn/TFA system: to the solution of the nitroxide (15 mg) in CD₃OD (250 μL), Zn powder (100 mg) was added, and the resulting mixture was heated to gentle reflux under stirring. After that, TFA (100 μL) was added dropwise, and the mixture was heated under stirring maintaining the reflux (15 min), filtered into an NMR tube, and diluted to the required volume with CDCl₃ or acetone-d₆. The IR spectra were acquired on an FT-IR spectrometer in KBr or neat (for oils) and are reported in wavenumbers (cm^–1). The HRMS spectra were recorded on a double-focusing, high-resolution mass spectrometer equipped with high-performance toroidal ESA. Reactions were monitored by TLC carried out using UV light 254 nm, 1% aqueous permanganate, and 10% solution of phosphomolibdic acid in ethanol and/or Dragendorff’s reagent as visualizing agents. Column chromatography was performed on silica gel 60 (230–400 mesh).

Dimethyl 2,2-Di(1-²H₄)ethyl-5-ethyl-3,4-pyrrolidinedicarboxylates (7b)

To prepare isotope-containing starting materials, 3-pentanone was repeatedly stirred with 5% solution of NaOD in D₂O until methylene proton signals disappeared (2.2%, residual protons; see the Supporting Information, Figure S1) in the ¹H NMR spectrum, dried with Na₂CO₃, and distilled; 2-aminobutanoic acid was dissolved in D₂O and solvent was distilled off twice; the mixture of DMF and toluene with D₂O was placed into the Dean–Stark apparatus and stirred under reflux for 4 h. To the mixture of DMF (25 mL) and toluene (25 mL) in the Dean–Stark apparatus, 2-aminobutanoic acid-D₃ (5.3 g, 0.05 mol), dimethyl fumarate (7.2 g, 0.05 mol), and 3-pentanone-D₄ (27 mL, 0.25 mol) were added, and the reaction mixture was stirred under reflux for 45 h (until the amino acid precipitate was completely dissolved). The solvents were distilled off in vacuum, the residue was dissolved in ethyl acetate (100 mL), and the solution was washed with 5% aqueous NaHCO₃. H₂O (200 mL) was added, 10% aqueous NaHSO₄ solution was added dropwise until the water layer pH = 4, and the organic layer was extracted with 1% aqueous NaHSO₄ twice. Acidic extracts were collected and basified with NaHCO₃ and extracted with ethyl acetate. The extract was dried with Na₂SO₄, and the solvent was distilled off in vacuum to give 9.42 g (69%) of yellow oil, a mixture of isomers 7b, which was used for the next step without further purification. IR (neat): ν_max: 1733 (C = O). ¹H NMR (400 MHz; CDCl₃, δ), isomer A (55%): 0.80 (s, 3H), 0.85 (s, 3H), 0.91 (s, 3H), 0.91 (t, J_t = 7.4 Hz, 3H), 1.18–1.31 (m, 1H), 1.46–1.58 (m, 1H), 1.61–1.72 (m, 1H), 2.94 (m, 1H), 3.11–3.19 (m, 2H), 3.63 (s, 3H), 3.64 (s, 3H); isomer B (45%): 0.76 (s, 3H), 0.91 (s, 3H), 0.92 (t, J_t = 7.4 Hz, 3H), 1.31–1.42 (m, 1H), 1.46–1.58 (m, 2H), 3.11–3.19 (m, 1H), 3.24 (dt J_d = 5.3 Hz, J_t = 8.4 Hz, 1H), 3.49 (m, 1H), 3.61 (s, 3H), 3.62 (s, 3H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₂₁²H₄NO₄ 275.2029; found 275.2031.

2,2-Di(1-²H₄)ethyl-5-ethyl-3,4-bis(hydroxymethyl)-3,4-dihydro-2H-pyrrole 1-oxide (8b)

A solution of crude 7b (9 g, 0.033 mol) in dry ether (50 mL) was added dropwise to a stirred solution of LiAlH₄ (3 g, 0.079 mol) in dry ether (150 mL). The mixture was stirred at ambient temperature for 2 h, and then the flask was placed into a cold water bath and quenched with 10% aqueous KOH. The organic phase was separated via decantation, the remaining wet precipitate was washed with ether 3 × 50 mL, and the combined extracts were evaporated in vacuum. The residue was dissolved in methanol (100 mL) and mixed with a solution of sodium tungstate (1.02 g, 3.2 mmol) and EDTA disodium salt (1.08 g, 3.2 mmol) in water (50 mL) and hydrogen peroxide 30% (17 mL) was added. The solution was allowed to stand at ambient temperature for 24 h, and then a catalytic amount of MnO₂ (0.2 g, 2.3 mmol) was carefully added to quench remaining H₂O₂. After the oxygen evolution ceased, the solution was evaporated in vacuum. The residue was triturated with CHCl₃ (30 mL) and Na₂CO₃ (5 g), the suspended matter was filtered off, and the solvent was distilled off in vacuum. The residue was triturated with EtOAc/diethyl ether (1:1), and the crystalline yellowish precipitate of 8b was collected; yield 6.55 g (86%), mp 60–63 °C, IR (KBr) ν_max: 3361 (O–H), 3324 (O–H). ¹H NMR (400 MHz; D₂O, δ): 0.72 (s, 3H), 0.80 (s, 3H), 1.09 (t, J_t = 7.5 Hz, 3H), 2.29 (dq, J_q = 7.5 Hz, J_d = 15 Hz, 1H), 2.47–2.58 (m, 1H), 2.70 (dq, J_q = 7.5 Hz, J_d = 15 Hz, 1H), 2.82–2.92 (m, 1H), 3.71–3.95 (m, 4H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₂H₁₉²H₄NO₃ 233.1924; found 233.1922.

General Procedure of Hydrogen–Deuterium Exchange in Nitrones

A solution of nitrone 8a or 8b (1g) in 5% NaOD/D₂O (10 mL) was kept at ambient temperature for 20 days. The conversion was controlled by NMR spectroscopy. On the final step, the reaction stayed at 40 °C for another 5 days. After the exchange was finished, the reaction mixture was neutralized by dry CO₂ bubbled, and the solvent was distilled off in vacuum. To the residue, EtOAc (50–100 mL) was added and heated to the reflux. The precipitate obtained was filtered off; the solution was evaporated in vacuum to give nitrone 8c or 8d correspondingly.

2,2-Diethyl-5-(1-²H₂)ethyl-3,4-bis((²H)-hydroxymethyl)-3,4-dihydro-2H-(4-²H)pyrrole 1-oxide (8c)

¹H NMR (300 MHz; 5% NaOD/D₂O, δ): 0.59 (t, J_t = 7.2 Hz, 3H), 0.68 (t, J_t = 7.3 Hz, 3H), 0.93 (s, 3H), 1.40–1.72 (m, 4H), 2.14 (dd, J_d = 5.3 Hz, J_d = 9 Hz, 1H), 3.25 (AB, J_AB = 10.8 Hz, 1H), 3.37–3.62 (m, 2H), 3.76 (AB, J_AB = 10.8 Hz, 1H).

2,2- Di(1-²H₄)ethyl-5-(1-²H₂)ethyl-3,4-bis((²H)-hydroxymethyl)-3,4-dihydro-2H-(4-²H)pyrrole 1-oxide (8d)

¹H NMR (300 MHz; 5% NaOD/D₂O, δ): 0.58 (s, 3H), 0.67 (s, 3H), 0.93 (s, 3H), 2.14 (dd, J_d = 5.3 Hz, J_d = 9 Hz, 1H), 3.25 (AB, J_AB = 10.8 Hz, 1H), 3.38–3.61 (m, 2H), 3.76 (AB, J_AB = 10.8 Hz, 1H).

2-Ethynyl-2,5,5-triethyl-3,4-bis(hydroxymethyl)-pyrrolidine-1-oxyls (10a and 12a–c)

A solution of trimethylsilyl chloride (1.09 g, 10 mmol) in dry THF (10 mL) was added dropwise to a solution of nitrone 8a–d (1 g, 4.37 mmol) and triethylamine (1.01 g, 10 mmol) in dry THF (35 mL) upon stirring in an ice bath. The reaction mixture was stirred at ambient temperature for 2 h. Then, the solvent was evaporated under reduced pressure, the residue was triturated with dry diethyl ether, and the precipitate was filtered off. The combined filtrate was concentrated under reduced pressure. Then, a solution of ethynylmagnesium bromide (0.6 M in THF, 44 mmol, 73 mL) was added in an argon atmosphere. The reaction mixture was kept at ambient temperature for 10 days. The mixture was quenched carefully with water until the formation of two phases. The organic layer was diluted with ether and separated, and the aqueous layer was extracted with EtOAc (15 mL × 3). The combined extract was evaporated under reduced pressure, the residue was dissolved in methanol (20 mL), and an aqueous solution of NaOH (0.05 g in 30 mL) was added to remove the protective groups. Then, catalytic amounts of methylene blue (2–3 mg) were added, and reaction mixture was bubbled with air for 12 h. Methanol was distilled off under reduced pressure, and the aqueous layer was acidified to pH 3–4 with 1 M H₂SO₄ and extracted with EtOAc (15 mL × 4). The organic extracts were collected and dried with Na₂SO₄, and the solvent was distilled off in vacuum. The residue was subjected to column chromatography on silica gel, eluent hexane/EtOAc (1:2), to give 10a as yellow crystals; yield 0.98 g (88%), mp 112–114 °C, IR (KBr) ν_max: 3509 (O–H), 3444 (O–H), 3220 (C≡C–H), 2102 (C≡C). ¹H NMR (400 MHz; CDCl₃/CD₃OD/CF₃COOH, δ): 0.82 (t, J_t = 7.5 Hz, 3H), 0.84 (t, J_t = 7.5 Hz, 3H), 1.04 (t, J_t = 7.4 Hz, 3H), 1.67 (q, J_q = 7.4 Hz, 2H), 1.86–2.03 (m, 2H), 2.23 (m, 2H), 2.87 (s, 1H), 3.48–3.82 (m, 4H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₂₄NO₃ 254.1751; found 254.1753. Anal. Calcd for C₁₄H₂₄NO₃: C, 66.11; H, 9.51; N, 5.51. Found: C, 66.18; H, 9.65; N, 5.50. nitroxides 12a–c were prepared using similar procedures from 8b–d.

2-Ethynyl-2-(1-²H₄)ethyl-5,5-diethyl-3,4-bis(hydroxymethyl)-(3-²H)pyrrolidine-1-oxyl (12a)

Yellow crystals, yield 0.88 g (80%), IR (KBr) ν_max: 3506 (O–H), 3446 (O–H), 3224 (C≡C–H), 2102 (C≡C). ¹H NMR (400 MHz; CDCl₃/CD₃OD/CF₃COOH, δ): 0.83 (t, J_t = 7.5 Hz, 3H), 0.86 (t, J_t = 7.4 Hz, 3H), 1.04 (s, 3H), 1.68 (q, J_q = 7.4 Hz, 2H), 1.97 (q, J_q = 7.5 Hz, 2H), 2.26 (dd, J_d = 6.6; 3.7 Hz, 1H), 2.86 (s, 1H), 3.48–3.86 (m, 4H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₂₁²H₃NO₃ 257.1955; found 257.1959.

2-Ethynyl-2-ethyl-5,5-di(1-²H₄)ethyl-3,4-bis(hydroxymethyl)-pyrrolidine-1-oxyl (12b)

Yellow crystals, yield 0.95 g (86%), IR (KBr) ν_max: 3508 (O–H), 3444 (O–H), 3222 (C≡C–H), 2104 (C≡C). ¹H NMR (400 MHz; CDCl₃/CD₃OD/CF₃COOH, δ): 0.83 (s, 3H), 0.85 (s, 3H), 1.07 (t, J_t = 7.4 Hz, 3H), 1.83–2.06 (m, 2H), 2.26 (m, 1H), 2.85 (s, 1H), 3.48–3.87 (m, 4H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₂₀²H₄NO₃ 258.2002; found 258.2000.

2-Ethynyl-2-(1-²H₄)ethyl-5,5-di(1-²H₄)ethyl-3,4-bis(hydroxymethyl)-(3-²H)pyrrolidine-1-oxyl (12c)

Yellow crystals, yield 0.89 g (81%), IR (KBr) ν_max: 3506 (O–H), 3448 (O–H), 3222 (C≡C–H), 2104 (C≡C). ¹H NMR (400 MHz; CDCl₃/CD₃OD/CF₃COOH, δ): 0.78 (s, 3H), 0.80 (s, 3H), 1.00 (s, 3H), 2.20 (dd, J_d = 4.0; 6.2 Hz, 1H), 2.86 (s, 1H), 3.43–3.81 (m, 4H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₁₇²H₇NO₃ 261.2206; found 261.2202.

General Procedure of Hydrogenation of Nitroxides 10a and 12a–c

A hydrogenation was performed in a reaction vessel equipped with a magnetic stirrer and a connection line to a gasometer filled with hydrogen/deuterium. Deuterium was prepared by dropwise addition of D₂O (3.5 mL) into a flask containing lithium foil (0.4–0.5 g) under the layer of absolute THF. A solution of nitroxide 10a or 12a–c (0.2 g) in THF (3 mL) was placed into the reaction vessel, the catalyst (Pd/C, 4%, 70 mg) was added to a solution, and the system was purged with hydrogen/deuterium and closed. The mixture was vigorously stirred until hydrogen absorption ceased, then the catalyst was filtered off and washed with THF, and slow stream of air was passed through the solution for 24 h. The solution was evaporated in vacuum, and the residue was subjected to column chromatography on silica gel, eluent hexane/EtOAc (1:1), to give yellow crystals 14 or 13a–d.

2-((1-²H₂-2-²H₂)Ethyl)-2,5,5-triethyl-3,4-bis(hydroxymethyl)-pyrrolidine-1-oxyl (13a)

IR (KBr) ν_max: 3486 (O–H), 3446 (O–H). ¹H NMR (300 MHz; CD₃OD/(CD₃)₂CO/CF₃COOH, δ): 0.90–1.06 (m, 10H), 1.75–1.86 (m, 1H), 1.89–2.01 (m, 5H), 2.32 (m, 2H), 3.64–3.84 (m, 4H). HRMS (EI/DFS) m/z: [M – H]⁺ calcd for C₁₄H₂₃²H₄NO₃ 261.2237; found 261.2241.

2-((1-²H₂-2-²H₂)Ethyl)-2-(1-²H₂)ethyl-5,5-diethyl-3,4-bis(hydroxymethyl)-(3-²H)pyrrolidine-1-oxyl (13b)

IR (KBr) ν_max: 3486 (O–H), 3452 (O–H). ¹H NMR (300 MHz; CD₃OD/CDCl₃/CF₃COOH, δ): 0.61–0.89 (m, 10H), 1.44–1.73 (m, 4H), 2.04 (m, 1H), 3.34–3.65 (m, 4H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₂₁²H₇NO₃ 265.2503; found 265.2505.

2-((1-²H₂-2-²H₂)Ethyl)-2-ethyl-5,5-di(1-²H₄)ethyl-3,4-bis(hydroxymethyl)-pyrrolidine-1-oxyl (13c)

IR (KBr) ν_max: 3486 (O–H), 3448 (O–H). ¹H NMR (500 MHz; CD₃OD/CDCl₃/CF₃COOH, δ): 0.90 (m, 1H), 0.93 (s, 3H), 0.96 (s, 3H), 0.98 (t, J_t = 7.5 Hz, 3H), 1.81–1.94 (m, 2H), 2.23 (m, 2H), 3.62 (m, 2H), 3.76 (m, 2H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₂₀²H₈NO₃ 266.2566; found 266.2570.

2-((1-²H₂-2-²H₂)Ethyl)-2,5,5-tri(1-²H₆)ethyl-3,4-bis(hydroxymethyl)-(3-²H)pyrrolidine-1-oxyl (13d)

IR (KBr) ν_max: 3486 (O–H), 3448 (O–H). ¹H NMR (300 MHz; CD₃OD/(CD₃)₂CO/CF₃COOH, δ): 0.84–1.02 (m, 10H), 2.29 (m, 1H), 3.62–3.82 (m, 4H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₁₇²H₁₁NO₃ 269.2754; found 269.2744.

2,5,5-Tri(1-²H₆)ethyl-2-ethyl-3,4-bis(hydroxymethyl)-(3-²H)pyrrolidine-1-oxyl (14)

IR (KBr) ν_max: 3486 (O–H), 3446 (O–H). ¹H NMR (400 MHz; CD₃OD/CDCl₃/CF₃COOH, δ): 0.96 (s, 3H), 0.97 (t, J_t = 7.4 Hz, 3H), 0.99 (s, 6H), 1.72 (dq, J_d = 14.7 Hz, J_q = 7.4 Hz, 1H), 1.84 (dq, J_d = 14.7 Hz, J_q = 7.4 Hz, 1H), 2.25 (dd, J_d = 7.1 Hz, J_d = 3.9 Hz, 1H), 3.64 (d, J_d = 11.2 Hz, 1H), 3.64 (dd, J_d = 11.4 Hz, J_d = 7.1 Hz, 1H), 3.79 (d, J_d = 11.2 Hz, 1H), 3.79 (dd, J_d = 11.4 Hz, J_d = 3.9 Hz, 1H). HRMS (EI/DFS) m/z: [M]⁺ calcd for C₁₄H₂₁²H₇NO₃ 265.2503; found 265.2502.

CW EPR

The X-band EPR spectra were obtained in diluted and oxygen-free aqueous solutions at room temperature at concentrations of 10^–4 M on a Bruker X-band (9 GHz) spectrometer Elexys E540. Experimental CW EPR settings were as follows: microwave power, 2.0 mW; modulation frequency, 100 kHz; modulation amplitude, 0.03 mT; time constant, 20 ms; sweep time, 21 s; number of points, 1024; number of scans, 1. For determining the isotropic g-values (g_iso), we recorded the X-band CW EPR spectra of mixtures of the investigated radicals with Tempo. The simulations of the solution EPR lines were carried out using different modes (fast motion model) in the software package EasySpin, which is available at www.easyspin.org.

Calculations

The geometry of conformations of the nitroxides was obtained from X-ray crystallography data (CCDC 1811119; CCDC 2209672) for 1a–c and 5 and by systematic rotor search at the MMFF94 level in the Avogadro program²³ for corresponding ketones that form when NO is replaced by CO.

Following optimization was performed at the UB3LYP/def2-TZVP level.²⁴ The isotropic hyperfine splitting (a_iso) constants were calculated as isotropic Fermi contact couplings at the same level of theory.²⁵ The influence of a solvent on the a_iso constants was not taken into account. The results of the UB3LYP calculation did not suffer from the spin contamination effect; the expectation value of S² was within 0.754 and 0.755 for radicals. All calculations were performed with the ORCA 5.0.3²⁶ suite of programs with default settings. NBO analysis was performed with the NBO 7.0.10 program.²⁷ Visualization of the calculation results was performed with the ChemCraft program (https://www.chemcraftprog.com).

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.3c06090.

• Copies of the ¹H NMR, IR, and HRMS spectra of compounds and DFT calculations data (PDF)

The authors declare no competing financial interest.