High-Field (3.4 T) Electron Paramagnetic Resonance, 1H Electron-Nuclear Double Resonance, ESEEM, HYSCORE, and Relaxation Studies of Asphaltene Solubility Fractions of Bitumen for Structural Characterization of Intrinsic Carbon-Centered Radicals

Petroleum asphaltenes are considered the most irritating components of various oil systems, complicating the extraction, transportation, and processing of hydrocarbons. Despite the fact that the paramagnetic properties of asphaltenes and their aggregates have been studied since the 1950s, there is still no clear understanding of the structure of stable paramagnetic centers in petroleum systems. The paper considers the possibilities of various electron paramagnetic resonance (EPR) techniques to study petroleum asphaltenes and their solubility fractions using a carbon-centered stable free radical (FR) as an intrinsic probe. The dilution of asphaltenes with deuterated toluene made it possible to refine the change in the structure at the initial stage of asphaltene disaggregation. From the measurements of samples of bitumen, a planar circumcoronene-like model of FR structure and FR-centered asphaltenes is proposed. The results show that EPR-based approaches can serve as sensitive numerical tools to follow asphaltenes’ structure and their disaggregation.


Introduction
Because of their ability to precipitate, deposit, and hence interrupt the continuous production of oil from underground reserves, petroleum asphaltenes are considered the most annoying constituent in diverse petroleum systems (petroleum dispersed systems, PDS). Asphaltenes are defined as being soluble in aromatic solvents such as benzene and toluene and insoluble in light alkanes such as n-pentane and n-heptane [1,2]. The phase changes, viscosity, stability, and interfacial properties of PDS are strongly affected by asphaltenes [3][4][5]. It was also stated, for example, that asphaltenes are the most chemically reactive PDS fraction [6]. For this reason, it is necessary to have an understanding of the association of these molecules, their properties, and their stability in PDS.
Currently, there is still ongoing discussion in the literature about the structure, properties, and intra-and intermolecular interactions of asphaltenes and their aggregates. It is known that asphaltenes molecules comprise fused-ring structures, naphthenic rings, and some aliphatic chains, as well as minor amounts of heteroatoms, such as nitrogen, sulfur, and oxygen, and trace metals, such as vanadium and nickel [2][3][4][5][6][7][8][9][10][11]. It is agreed that PDS asphaltenes have similar characteristics, are monomeric, and have a molecular weight of between 200 g/mol and 1500 g/mol [5,12,13]. However, when it comes to the models and mechanisms of asphaltenes aggregates, their stability, their solubility, and even their core

Materials and Methods
The EPR method is based on the resonant (selective) absorption of microwave energy by PCs in an external magnetic field, B 0 . The EPR spectra observed in petroleum components are due to the presence of magnetic moment for unpaired electrons and are affected by several types of interactions between the unpaired electron and its environment [38]: (1) the Zeeman interaction between the unpaired electrons and B 0 , (2) the spin-orbital interaction, (3) the electron-nuclear interaction, (4) the interaction with other unpaired electrons (spin-spin interaction), (5) and the interaction with the environmental bath (spin-lattice interaction).
The most significant contribution is the Zeeman interaction determined by the expression: where h is Planck's constant, ν is the microwave frequency, β is the Boron magneton, g is the spectroscopic splitting factor, and B 0 is the value of the induction of the magnetic field.
As was already pointed out in Introduction, it was found that the value of the FR g-factor, which is close to the g-factor of a free electron, depends on the chemical environment of the unpaired electron. Its deviation from the value of 2.0023 depends directly on the content of heteroelements in asphaltenes and has an inverse dependence on the aromaticity factor of asphaltenes [13,21]. The conventional (continuous wave, CW) and pulsed EPR spectra were registered with Bruker ElexSys 580/680 EPR spectrometers. CW EPR experiments were conducted in the X-band (ν mw = 9.6 GHz) and W-band (ν mw = 94 GHz) frequency ranges, with a magnetic sweep field from 5 mT to 1.5 T for the X-band and 3300 mT to 3500 mT for the W-band. The experimental configurations (modulation A = 0.1 mT, time of integration τ int = 82 ms, and power of mw-source P mw = 100 µW) were tuned to prevent the saturation or distortion of EPR signals.
To get detailed information about the electron-nuclear and spin-lattice interactions, pulsed EPR experiments were conducted. An electron-spin-echo (ESE) signal was detected with the following pulse sequence: with a π/2 pulse duration of 16-32 ns and time delay τ = 240 ns or longer. Spectra simulations were conducted in the Easyspin package for MatLab [47]. To follow ESE decay and extract the time of transverse (spin-spin, phase-memory) relaxation, T 2 , the same pulse sequence was applied by increasing the time interval τ between the pulses, with a step of 4 ns or longer. To measure the longitudinal (spin-lattice) relaxation time, T 1 , the inversion-recovery three-pulse sequence was used: with a π/2 pulse duration of 16-32 ns and time delay τ = 240 ns or longer. Electron-spin-echo envelope modulation (ESEEM) was implemented using a wellknown two-pulse Hahn sequence: π/2 − τ − π, where the length of π was 32 ns and τ was equal to 200 ns. In this method, the integral intensity of ESE was recorded, depending on the time interval, τ, between two pulses at a fixed magnetic field. This parameter was increased to the required value (before the loss of information on nuclear modulations). To gain the sensitivity and spectroscopic resolution, the smallest possible step, t = 4 ns, was chosen, and the ESE was integrated only at its peak, with an integration time of t = 4 ns. Further spectral analysis of the obtained results involved a Fourier transform using the program OriginPro.
In the case of the presence of anisotropic electron-nuclear interactions caused by magnetic nuclei, a modulation of the ESE decay might be observed [40]. It offered an opportunity to identify an FR nuclear environment and estimate the values of their hyperfine interactions with electron spin. A four-pulse sequence, as shown in Figure 1a, was con-In the case of the presence of anisotropic electron-nuclear interactions caused by magnetic nuclei, a modulation of the ESE decay might be observed [40]. It offered an opportunity to identify an FR nuclear environment and estimate the values of their hyperfine interactions with electron spin. A four-pulse sequence, as shown in Figure 1a, was conducted in the form of a stimulated echo sequence, by inverting a 180-degree pulse inserted after the second π/2 pulse. The application of a π pulse led to a mixing of the nuclear frequencies and offered a possibility to apply hyperfine sublevel correlation spectroscopy (HYSCORE) techniques [40]. Figure 1. (a) Pulse sequence to acquire two-dimensional signals from electron-nuclear interactions (HYSCORE), where NP stands for nuclear polarization, NC stands for nuclear coherences, and EC stands for electron coherences; (b) pulse sequence to obtain Mims ENDOR spectra.
The ability to detect ESE offered an opportunity to obtain ENDOR spectra by using a Mims pulse sequence with an RF pulse of about 18 μs (the pulse length was optimized for 1 H nuclei) inserted between the second and third microwave pulses (Figure 1b). RF frequency in our setup could be swept in the range of 1-200 MHz. More details on the pulsed W-band EPR and ENDOR measurements for PDS with the corresponding references were given in papers [18,19].
Asphaltenes were extracted from the Zuzeevsk bitumen (Tatarstan, Russia). Some properties of the initial sample are listed in Table 1. Asphaltenes (Ainit) were precipitated from the raw material by the addition of 40 mL·g −1 of the petroleum ether (b.p. 40-70 °C). Precipitated asphaltenes were washed in a Soxhlet apparatus with petroleum ether until the filtrate became colorless. Then the asphaltenes with a filter were washed out with benzene, which was then evaporated.
The A1 and A2 fractions of asphaltenes were obtained as follows [30]. One gram of Ainit asphaltenes was first completely dissolved with 28 mL toluene. Petroleum ether (b.p. 40-70 °C) was then added in the amount of 52 mL. The resulting solution was kept Figure 1. (a) Pulse sequence to acquire two-dimensional signals from electron-nuclear interactions (HYSCORE), where NP stands for nuclear polarization, NC stands for nuclear coherences, and EC stands for electron coherences; (b) pulse sequence to obtain Mims ENDOR spectra.
The ability to detect ESE offered an opportunity to obtain ENDOR spectra by using a Mims pulse sequence with an RF pulse of about 18 µs (the pulse length was optimized for 1 H nuclei) inserted between the second and third microwave pulses (Figure 1b). RF frequency in our setup could be swept in the range of 1-200 MHz. More details on the pulsed W-band EPR and ENDOR measurements for PDS with the corresponding references were given in papers [18,19].
Asphaltenes were extracted from the Zuzeevsk bitumen (Tatarstan, Russia). Some properties of the initial sample are listed in Table 1. Asphaltenes (Ainit) were precipitated from the raw material by the addition of 40 mL·g −1 of the petroleum ether (b.p. 40-70 • C). Precipitated asphaltenes were washed in a Soxhlet apparatus with petroleum ether until the filtrate became colorless. Then the asphaltenes with a filter were washed out with benzene, which was then evaporated.
The A1 and A2 fractions of asphaltenes were obtained as follows [30]. One gram of Ainit asphaltenes was first completely dissolved with 28 mL toluene. Petroleum ether (b.p. 40-70 • C) was then added in the amount of 52 mL. The resulting solution was kept in a dark place for 24 h and then filtered. Precipitated material was washed in a Soxhlet apparatus with toluene until the solvent became colorless and was dried and weighed until there was no change in mass. Obtained fraction was called A1. Then the petroleum ether was added to the supernatant in the amount of 228 mL and precipitated for 24 h. Using the operations described above, the fresh precipitated material was separated, washed with toluene in a Soxhlet apparatus, and dried. The A2 fraction of asphaltenes was recovered by evaporating the remaining solvent. According to the results of various investigations reviewed in paper [5], subfraction A1 has a high molecular weight and aromaticity factor, has a low H:C atomic ratio, and contains highly condensed aromatic compounds ("island-like"). Subfraction A2 has a high H:C ratio and a relatively low aromaticity factor ("archipelago-like"). The toluene-to-heptane solvent transfer-free energies of steam-cracked tar and petroleum asphaltenes are consistent with the observation that more alkane-soluble subfractions will be dominated by smaller molecules [49,50]. The results of papers [10,[51][52][53] suggest that asphaltene polydispersity plays a major role in determining asphaltene precipitation and deposition tendencies observed in different oils.

Results and Discussion
As an example, EPR spectrum for A2 in CW mode is shown in Figure 2. It was due to the presence of two types of PCs. The first one (with a 16-line pattern for the powder spectrum) belonged to the VO 2+ complex with a g-factor of axial symmetry (nuclear spin I = 7/2 for 51 V nuclei, g || = 1.963, g ⊥ = 1.985, A || = 468 MHz, A ⊥ = 150 MHz). A detailed analysis of signal from VO 2+ complexes was given in our papers [20,44]. Another one was a single line from FR. in a dark place for 24 h and then filtered. Precipitated material was washed in a Soxhlet apparatus with toluene until the solvent became colorless and was dried and weighed until there was no change in mass. Obtained fraction was called A1. Then the petroleum ether was added to the supernatant in the amount of 228 mL and precipitated for 24 h. Using the operations described above, the fresh precipitated material was separated, washed with toluene in a Soxhlet apparatus, and dried. The A2 fraction of asphaltenes was recovered by evaporating the remaining solvent. According to the results of various investigations reviewed in paper [5], subfraction A1 has a high molecular weight and aromaticity factor, has a low H:C atomic ratio, and contains highly condensed aromatic compounds ("island-like"). Subfraction A2 has a high H:C ratio and a relatively low aromaticity factor ("archipelago-like"). The toluene-to-heptane solvent transfer-free energies of steam-cracked tar and petroleum asphaltenes are consistent with the observation that more alkane-soluble subfractions will be dominated by smaller molecules [49,50]. The results of papers [10,[51][52][53] suggest that asphaltene polydispersity plays a major role in determining asphaltene precipitation and deposition tendencies observed in different oils.

Results and Discussion
As an example, EPR spectrum for A2 in CW mode is shown in Figure 2. It was due to the presence of two types of PCs. The first one (with a 16-line pattern for the powder spectrum) belonged to the VO 2+ complex with a g-factor of axial symmetry (nuclear spin I = 7/2 for 51 V nuclei, g|| = 1.963, g⊥ = 1.985, A|| = 468 MHz, A⊥ = 150 MHz). A detailed analysis of signal from VO 2+ complexes was given in our papers [20,44]. Another one was a single line from FR. A usual way to fit the EPR line shape of FR at X-band is to make it a sum or convolution of Gauss and Lorentz, where ГG ≈ ГL (the line width (Г) for Gaussian (G) and Lorentzian (L) absorption line shapes are connected with the full width at half maximum (FWHM) via ГG = (2ln2) −1/2 ·FWHM and ГL = (3) −1/2 ·FWHM (see [45] for details) and the isotropic g-factor (giso ≈ 2.0038(3)). Measurements in the X-band show that the parameters of FR spectra do not depend on fractioning.
W-band spectra in CW and ESE modes are presented in Figures 3-5. As seen in them, in contrast to the X-band, in the W-band EPR, the lines from FR are not symmetric and their line widths become larger. FR signals in the W-band can be ascribed to single paramagnetic center of axial symmetry, with g‖ ≈ 2.0045(3) and g⊥ ≈ 2.0029 (3). As seen in Figure   300 320 340 360 380 400 A usual way to fit the EPR line shape of FR at X-band is to make it a sum or convolution of Gauss and Lorentz, where Γ G ≈ Γ L (the line width (Γ) for Gaussian (G) and Lorentzian (L) absorption line shapes are connected with the full width at half maximum (FWHM) via Γ G = (2ln2) −1/2 ·FWHM and Γ L = (3) −1/2 ·FWHM (see [45] for details) and the isotropic g-factor (g iso ≈ 2.0038(3)). Measurements in the X-band show that the parameters of FR spectra do not depend on fractioning.
W-band spectra in CW and ESE modes are presented in Figures 3-5. As seen in them, in contrast to the X-band, in the W-band EPR, the lines from FR are not symmetric and their line widths become larger. FR signals in the W-band can be ascribed to single paramagnetic center of axial symmetry, with g ≈ 2.0045(3) and g ⊥ ≈ 2.0029 (3). As seen in Figure 5, even high-resolute EPR spectrum does not allow obtaining information on the FR structural changes during the asphaltene fractioning or dissolution. This may lead to the Nanomaterials 2022, 12, 4218 6 of 13 incorrect conclusion that the possibilities of EPR are extremely limited. More-elaborated EPR techniques, as will be shown below, should be applied in studies on petroleum systems. Nanomaterials 2022, 12, x FOR PEER REVIEW 6 of 14 5, even high-resolute EPR spectrum does not allow obtaining information on the FR structural changes during the asphaltene fractioning or dissolution. This may lead to the incorrect conclusion that the possibilities of EPR are extremely limited. More-elaborated EPR techniques, as will be shown below, should be applied in studies on petroleum systems.     Nanomaterials 2022, 12, x FOR PEER REVIEW 6 of 14 5, even high-resolute EPR spectrum does not allow obtaining information on the FR structural changes during the asphaltene fractioning or dissolution. This may lead to the incorrect conclusion that the possibilities of EPR are extremely limited. More-elaborated EPR techniques, as will be shown below, should be applied in studies on petroleum systems.     . W-band ESE EPR for A1, A2, and Ainit fractions. Region of FR signal is marked (see Figure 5 for details).
Nanomaterials 2022, 12, x FOR PEER REVIEW 6 of 14 5, even high-resolute EPR spectrum does not allow obtaining information on the FR structural changes during the asphaltene fractioning or dissolution. This may lead to the incorrect conclusion that the possibilities of EPR are extremely limited. More-elaborated EPR techniques, as will be shown below, should be applied in studies on petroleum systems.    The decay of the transverse magnetization (ESE decay, T 2 -curve) in the X-band contains modulations (ESEEM, [54]) were caused by anisotropic electron-nuclear interactions (Figure 6a), although the modulation amplitude (modulation depth) was at least an order of magnitude less than that for the vanadyl complexes (cf with the results in paper [40]). Therefore, the measurements were demonstrated for T = 50 K to increase the signal-to-noise ratio. A Fourier-transformed signal in the frequency domain is shown in Figure 6b. It allows for identifying the nuclear environment of FR. As seen in Table 2, the obtained signals belonged to interactions with hydrogen ( 1 H) nuclei in the high-frequency region and unresolved signals from 14 N, 51 V, or even 13 C in the low-field part. In the W-band, the modulation was not registered, although the values of T 2 in particular did not change with the mw frequency (T 2 = 400-450 ns for T = 300 K). Similar to the EPR results (Figure 4), ESEEM did not depend on the type of extracted asphaltenes (sub)fraction. The decay of the transverse magnetization (ESE decay, T2-curve) in the X-band contains modulations (ESEEM, [54]) were caused by anisotropic electron-nuclear interactions (Figure 6a), although the modulation amplitude (modulation depth) was at least an order of magnitude less than that for the vanadyl complexes (cf with the results in paper [40]). Therefore, the measurements were demonstrated for T = 50 K to increase the signal-tonoise ratio. A Fourier-transformed signal in the frequency domain is shown in Figure 6b. It allows for identifying the nuclear environment of FR. As seen in Table 2, the obtained signals belonged to interactions with hydrogen ( 1 H) nuclei in the high-frequency region and unresolved signals from 14 N, 51 V, or even 13 C in the low-field part. In the W-band, the modulation was not registered, although the values of T2 in particular did not change with the mw frequency (T2 = 400-450 ns for T = 300 K). Similar to the EPR results (Figure 4), ESEEM did not depend on the type of extracted asphaltenes (sub)fraction.  Table 2).   Table 2). A comprehensive analysis of the mutual impact of FR and VP on various asphaltene (sub)fractions as extracted from the electronic relaxation times measurements was con-Nanomaterials 2022, 12, 4218 8 of 13 ducted in paper [30]. Because of the spin-spin (FR-VO 2+ ) interaction, it may lead to the shortening of FR relaxation times. It was derived from the relaxation data that FR and VO 2+ are located on the distance of 0.8-1.0 nm and form complexes that can be destroyed by dissolution.
To obtain more-detailed information on the FR-VP structure, we used HYSCORE. X-band HYSCORE for FR is shown in Figure 7. The left part (quadrant) of the spectrum consists of intense but structureless signals that are located along the diagonal with a wide distribution of hyperfine value, from almost 0 MHz to 15 MHz (with two intense peaks at 6.4 MHz and 12.8 MHz). It was probably due to the wide distribution of distances between FR and other nuclei, for example forming a VP skeleton, specifically 14 N and 51 V. Thus, FR, at least with respect to 14 N and 51 V, did not have a definite localization in the VP structure. In the right quadrant of Figure 7, it is possible to observe the signal from hydrogen 1 H with a split of 4.9 MHz. This pattern suggests that the FR has a quite definite structure with respect to hydrogen and that therefore 1 H-electron hyperfine values might give essential information on the changes in asphaltenes structures. The low-frequency part of the right quadrant can be ascribed to an interaction with 51 V and 13 C (see Table 2). In the W-band, HYSCORE experiments were not conducted, because, as was pointed out above, the ESEEM signal was too weak. A comprehensive analysis of the mutual impact of FR and VP on various asphaltene (sub)fractions as extracted from the electronic relaxation times measurements was conducted in paper [30]. Because of the spin-spin (FR-VO 2+ ) interaction, it may lead to the shortening of FR relaxation times. It was derived from the relaxation data that FR and VO 2+ are located on the distance of 0.8-1.0 nm and form complexes that can be destroyed by dissolution.
To obtain more-detailed information on the FR-VP structure, we used HYSCORE. Xband HYSCORE for FR is shown in Figure 7. The left part (quadrant) of the spectrum consists of intense but structureless signals that are located along the diagonal with a wide distribution of hyperfine value, from almost 0 MHz to 15 MHz (with two intense peaks at 6.4 MHz and 12.8 MHz). It was probably due to the wide distribution of distances between FR and other nuclei, for example forming a VP skeleton, specifically 14 N and 51 V. Thus, FR, at least with respect to 14 N and 51 V, did not have a definite localization in the VP structure. In the right quadrant of Figure 7, it is possible to observe the signal from hydrogen 1 H with a split of 4.9 MHz. This pattern suggests that the FR has a quite definite structure with respect to hydrogen and that therefore 1 H-electron hyperfine values might give essential information on the changes in asphaltenes structures. The low-frequency part of the right quadrant can be ascribed to an interaction with 51 V and 13 C (see Table 2). In the W-band, HYSCORE experiments were not conducted, because, as was pointed out above, the ESEEM signal was too weak. Now consider how the nuclear environment of paramagnetic centers changed by exploiting the abilities of the W-band ENDOR. The ENDOR spectra can be registered both for the FR and for vanadyl porphyrins. VP ENDOR was discussed in detail in [18,19,21]. In this paper, we investigate FR ENDOR. For that, the duration of the mw pulse was chosen to be 44 ns ( Figure 1b); thus, the pulse spectrum excited all FRs but did not affect the VP spectrum. In the FR ENDOR, only an interaction with 1 H was observed; no interaction with 13 C, 51 V, or 14 N was detected. As seen in Figure 8, the 1 H ENDOR spectrum is typical for the unoriented (powder) samples. Now consider how the nuclear environment of paramagnetic centers changed by exploiting the abilities of the W-band ENDOR. The ENDOR spectra can be registered both for the FR and for vanadyl porphyrins. VP ENDOR was discussed in detail in [18,19,21]. In this paper, we investigate FR ENDOR. For that, the duration of the mw pulse was chosen to be 44 ns ( Figure 1b); thus, the pulse spectrum excited all FRs but did not affect the VP spectrum. In the FR ENDOR, only an interaction with 1 H was observed; no interaction with 13 C, 51 V, or 14 N was detected. As seen in Figure 8, the 1 H ENDOR spectrum is typical for the unoriented (powder) samples.
To describe the obtained ENDOR data, we took a simplified model of the electronic structure of FR. We supposed that the nuclei of the environment affect the electron spin only through the interaction between magnetic dipoles, which is anisotropic in nature. To simplify the calculations, we neglected the admixture of the Pauli (contact) interaction, assuming that such an interaction corresponds to energies above 2 MHz. In this case, it was possible to fit the ENDOR spectrum with good accuracy and determine the contributions to the spectrum of protons located at different distances from the FR (Figure 9). We carried out the spectrum deconvolution procedure with additional conditions for non-negativity and the smoothness of the distribution. When calculating the spectrum, we used convolution with the intensity of the spectrum because of the blind spot, the maximum suppression of which for all spectra was at a frequency of 2.1 MHz. Moreover, 32 amplitude points were used in the distribution of the contribution of protons, depending on the distance. As can be seen in Figure 9, the deconvolution results (thin black lines) quite well describe the experimental spectra. To describe the obtained ENDOR data, we took a simplified model of the electronic structure of FR. We supposed that the nuclei of the environment affect the electron spin only through the interaction between magnetic dipoles, which is anisotropic in nature. To simplify the calculations, we neglected the admixture of the Pauli (contact) interaction, assuming that such an interaction corresponds to energies above 2 MHz. In this case, it was possible to fit the ENDOR spectrum with good accuracy and determine the contributions to the spectrum of protons located at different distances from the FR (Figure 9). We carried out the spectrum deconvolution procedure with additional conditions for nonnegativity and the smoothness of the distribution. When calculating the spectrum, we used convolution with the intensity of the spectrum because of the blind spot, the maximum suppression of which for all spectra was at a frequency of 2.1 MHz. Moreover, 32 amplitude points were used in the distribution of the contribution of protons, depending on the distance. As can be seen in Figure 9, the deconvolution results (thin black lines) quite well describe the experimental spectra. For all asphaltene (sub)fractions, a peak was observed in the region of 0.75 nm (Figure 9, right panel). It corresponds to the nearest protons in aromatic cores where FR was localized. The initial fraction was characterized by a wide density distribution of interacting protons around FR, up to distance of 1.5 nm. The distance distributions for subfractions A1 and A2 differed in the vicinity of 0.45 nm. For the A1 there were practically no protons in this region, which most likely was due to the prevalence of aromatic cores. In fraction A2, there were a number of protons at this distance because of either alkane chains or the boundaries of the aromatic rings. When dissolved, both fractions showed fewer protons at a distance of 1.25 nm than the initial asphaltenes fraction. This might be at- For all asphaltene (sub)fractions, a peak was observed in the region of 0.75 nm (Figure 9, right panel). It corresponds to the nearest protons in aromatic cores where FR was localized. The initial fraction was characterized by a wide density distribution of interacting protons around FR, up to distance of 1.5 nm. The distance distributions for subfractions A1 and A2 differed in the vicinity of 0.45 nm. For the A1 there were practically no protons in this region, which most likely was due to the prevalence of aromatic cores. In fraction A2, there were a number of protons at this distance because of either alkane chains or the boundaries of the aromatic rings. When dissolved, both fractions showed fewer protons at a distance of 1.25 nm than the initial asphaltenes fraction. This might be attributed to the "washing out" of hydrogen-enriched compounds from asphaltenes.
It was quite notable that transformations of 1 H ENDOR were observable also with the dissolution of the A1 fraction, which is supposed to be poorly soluble in toluene. A1 was subjected to partial dissociation by dissolving it in deuterated toluene in proportions of 1:5 and 1:20 by volume. Of course, this degree of dilution was not enough for the complete destruction of the asphaltene aggregates [4,5], but as seen in our results, 1 H ENDOR can serve as a sensitive tool to follow the partial destruction of asphaltene aggregates, along with the electronic relaxation times (cf with [26,30]).
To determine the type of distribution of the concentration of hydrogen nuclei, depending on the distance from the paramagnetic center, it was necessary to take into account the decrease in the intensity of the ENDOR signal over distance. For the decrease in the intensity of the ENDOR signal as r −3 , the calculated proton concentration distribution is shown in Figure 10. In Figure 10, it can be seen that during the partial dissociation of asphaltene aggregates, protons located at a distance of 0.9 nm were replaced by solvent molecules (group 1), and only protons located on the distance of 0.7 nm remained (group 2). Moreover, at a dilution of asphaltenes in ratio of 1:20, the proton concentration peak became narrower, and the dip around r = 0.45 nm became deeper. Unfortunately, the further dissolution of asphaltenes in toluene led to a sharp suppression of the ENDOR signal, hindering further reliable measurements. It can be concluded that the second group of protons belonged to the asphaltene complex itself, whereas the first group belonged to other molecules, ones that are weakly bound to each other. Probably, this is because A1-type molecules stacked with each other (which prevents the protons from approaching FR), whereas A2-type molecules did not or did so to a lesser extent. When asphaltenes were slightly diluted with deuterated toluene, the molecules partially dissociate in the plane and the layers were shifted relative to each other; therefore, protons appeared at a distance of 0.45 nm. Further dilution led to further separation between the layers and a decrease in the number of protons at a distance of 0.45 nm and 0.7 nm.

Conclusions
Based on the described experiments and calculations, this paper constructed a hypothetical model of a near environment for an FR-centered structure in asphaltenes (sub)fractions and peculiarities of various EPR approaches to characterize asphaltenes. In Figure 10, it can be seen that during the partial dissociation of asphaltene aggregates, protons located at a distance of 0.9 nm were replaced by solvent molecules (group 1), and only protons located on the distance of 0.7 nm remained (group 2). Moreover, at a dilution of asphaltenes in ratio of 1:20, the proton concentration peak became narrower, and the dip around r = 0.45 nm became deeper. Unfortunately, the further dissolution of asphaltenes in toluene led to a sharp suppression of the ENDOR signal, hindering further reliable measurements. It can be concluded that the second group of protons belonged to the asphaltene complex itself, whereas the first group belonged to other molecules, ones that are weakly bound to each other. Probably, this is because A1-type molecules stacked with each other (which prevents the protons from approaching FR), whereas A2-type molecules did not or did so to a lesser extent. When asphaltenes were slightly diluted with deuterated toluene, the molecules partially dissociate in the plane and the layers were shifted relative to each other; therefore, protons appeared at a distance of 0.45 nm. Further dilution led to further separation between the layers and a decrease in the number of protons at a distance of 0.45 nm and 0.7 nm.

Conclusions
Based on the described experiments and calculations, this paper constructed a hypothetical model of a near environment for an FR-centered structure in asphaltenes (sub)fractions and peculiarities of various EPR approaches to characterize asphaltenes.

1.
The nearest FR environments in asphaltenes were similar to each other: EPR spectra either in X-or W-band were the same for the (sub)fractions under study.
The wide distribution of distances between FR and other magnetic nuclei, such as 14 N and 51 V, as seen in HYSCORE results and low-modulation amplitudes in ESEEM, suggested that 14 N or 51 V were not obligatory heteroatoms for the nearest environment of FRs. 4.
From 1 H ENDOR, it followed that for an FR of asphaltenes, the highest proton concentrations were observed at distances of 0.8-1.0 nm from FR core. This value grew in a row: A2→ A1→ A init . It follows that fraction A1 consisted of the most symmetrical and homogeneous molecular complexes presumably stacked (which prevents the protons from approaching FR core) on top of one another.

5.
1 H ENDOR can serve as a sensitive tool to follow the partial destruction of asphaltene aggregates, such as by dissolution. By dissolving in toluene, the peak of the highest proton concentrations shifted to the value of <0.7 nm.