Novel Molybdenite-Based Nanopowder Catalysts for Hydrodesulfurization

Novel bulk single-component sulfi de catalysts were prepared under the conditions of solid-phase dispersion of МoS2 molybdenite at various mechanical treatment times and various amounts of polar and nonpolar liquid microadditives. The chemical degradation of the samples in the air was found to lead to the formation of surface sulfate anions that shield catalytically active Mo sites. Indirect correlations of the hydrodesulfurization ability of MoS2 powders with the concentration of sulfate anions on their surface, and with the aqueous pH in the powder suspensions, including the dielectric permittivity of the organic dopants, can serve as reference indicators of high catalytic activity in the model reaction of dibenzothiophene hydrogenolysis. The study identifi ed the most active sample able to run for an extended time during multiple cycles without losing its catalytic properties. Also, the paper discusses the dibenzothiophene conversion routes, the product composition, the probable structure of active sites in the catalysts, and the desulfurization degree of diesel fuel components.


DOI: 10.1134/S0965544121070033
The recent increasing trend in the hydrotreatment of petroleum fuels has been to search for approaches to creating novel oil refi ning catalysts [1]. One potentially effective solution to this problem involves the use of submicron-scale fi ne-dispersed catalytic systems as the most suitable for restructuring and "unfolding" of oil globules in slurry processes.
Of particular interest are direct methods for the preparation of bulk sulfi de catalysts based on purifi ed mineral molybdenite (MoS 2 ) as a dimensional precursor of the active component. Typically, MoS 2 is subjected to mechanical grinding in planetary ball mills [2] in an organic solvent medium. During mechanical activation (MA), the deformation and fracturing of large solid fragments is known to be signifi cantly facilitated in liquid media with a high wetting energy. This is due to the forces of electrostatic interaction between liquid dipoles and ions on the crystal surface [3].
A prior study describes grinding of molybdenite in a butanol medium in a planetary ball mill under strong shock impacts on the substrate at high accelerations (g) of 300 to 500 m 2 /s [4]. The nanocrystallites obtained in this manner had a disadvantage: due to carbonization, they lost their catalytic activity as early as after the fi rst service cycle in a model reaction of dibenzothiophene (DBT) hydrodesulfurization (HDS) and in hydrotreatment of petroleum feedstock. The authors of the referenced study provide no explanation for this.
It should be noted that researchers have traditionally attempted to fi nd conditions to prepare samples with the highest possible defect level [4]. These attempts are consistent with classical approaches to the preparation of MA materials with pronounced catalytic properties [5]. The highest-shock-impact mode has generally been used to achieve "high-defect" grinding of molybdenite [4]. However, it is well known that MoS 2 , as a d-transition metal dichalcogenide with a layered structure, can form intercalation compounds (i.e., compounds inserted between adjacent layers) and exfoliate (split into single nanocrystals).
Previously, we reported on the development of a one-step direct method for preparing MoS 2 catalytic nanosystems under dry molybdenite grinding conditions [6]. The method was implemented in a vibratory mill in the presence of small amounts of methanol or water at low grinding media accelerations (g ≤ 50 m 2 /s). Under shear stresses, the MoS 2 layers are mostly exposed to longitudinal fragmentation. This leads to their layerby-layer displacement, mixing, and grinding in the plastic strain mode, followed by the formation of MoS 2 nanocrystallites. The temperature of dopant removal from the MoS 2 samples (the dopant being represented by methanol or water) proved to exceed the boiling points of these dopants by 70-100°C. This fi nding suggests that the dopants are chemically and structurally bound with molybdenite's interlayer space. Unfortunately, the data available from published reports on the structure and properties of such systems are insuffi cient.
The purpose of the present study was to prepare powder HDS catalysts in one step by direct grinding of molybdenite in the presence of small amounts of polar and nonpolar liquids, to measure their physicochemical properties, as well as to determine their catalytic activity in a model DBT conversion reaction and in the hydrotreatment of diesel components.

EXPERIMENTAL
The catalysts were prepared by grinding purifi ed mineral molybdenite (a commercial fi ne-dispersed MoS 2 powder, DMI-7), specifi cally by mechanical activation in a vibratory mill in the presence (or absence, for comparison purposes) of dopants. The dopants were represented by organic fl uids such as methanol, ethanol, butanol, hexane, and water, in amounts of 25-200 μL per 3 g of molybdenite. The molybdenite was dispersed under the following conditions: grinding media acceleration g ≤ 50 m 2 /s; frequency 16 Hz; amplitude 2 mm; ball diameter 20 mm; predominant shear stresses; MA time (τ MA ) equals 0.5, 5.0, 8.0, 12.0, and 16.0 h. The following exemplary notation was used for the catalyst samples: *MoS 2 + 100CH 3 OH (8), which indicates that the sample was mechanically activated (as denoted by the asterisk) in the presence of 100 μL of methanol for 8 h.
The specifi c surface area of the prepared catalysts was measured using the Brunauer-Emmett-Teller (BET) method on a Sorptometer-M instrument.
The surface morphology of the catalysts was identifi ed by transmission electron microscopy (TEM) on CM-30 (Philips Netherlands) and JEM-2010 (JEOL Japan) instruments.
X-ray diffraction (XRD) of the prepared samples was performed using a D8-Discover diffractometer (Bruker Germany) in a 2θ range of 8°-46° in monochromated CuK α radiation. The coherent scattering regions (CSR) of the samples were used to measure the nanocrystallite dimensions (in terms of the basal face length, L). The interplanar distances, also referred to as d-spacings (D 002 ), were calculated using the Debye-Scherrer equation [7]. The contribution of microstrains was derived from variations in D 002 and internal elastic microstrains (Δd/d). The crystallographic defects and microstrains were evaluated by variations in the unit cell parameters (c/a).
X-ray photoelectron spectra (XPS) of the MoS 2 samples were recorded on a SPECS photoelectron spectrometer using AlK* radiation (h* = 1486.6 eV). The binding energy scale (E b ) was pre-calibrated according to the peak positions of the ground energy levels for gold (Au4f 7/2 , 84.0 eV) and copper (Cu 2 p 3/2 , 932.67 eV). The powdered samples were deposited onto a biadhesive conductive copper tape. For calibration, the C1s line (E b = 284.8 eV) of carbon present on the catalyst surface was used. In MoS 2 , peaks at 162.3 eV (sulfi de) and 169.2 eV (partially oxidized sulfur) correspond to the binding energies of S2p, whereas a peak at 229.2 eV (typical Mo 4 + state) is attributed to Mo3d. The C1s peak is located at 284.8 eV.
The depth variations in the atomic concentrations of elements were measured by ion etching of the sample surface. The etching was carried out using a SPECS IQE 11/35 ion gun with argon ion energy of 1.05 kV and a current density of 6.8 μA/cm 2 . The chemical state of the elements on a sample's surface was identifi ed by recording individual spectral bands. The lines of S2p, S2s + Mo3d, C1s, and O1s were recorded [8].
IR spectra of the powders were recorded on a SPECORDM-80 spectrophotometer using KBr tablets.
The aqueous acidity of the powder suspensions was determined potentiometrically using a HANNA Instruments microprocessor-based laboratory pH meter.
The activity of the MA-MoS 2 samples was evaluated in a model DBT conversion reaction (S init = 500 ppm) from the residual sulfur concentration (S res ) in the HDS product and the DBT conversion rate constants (k, h -1 ) [9]. The experiments were carried out in a 100 mL Autoclave Engineers Bolted Closure reactor at 340°C, a hydrogen pressure of 3.4 MPa, and a stirring speed of 600 rpm, in 80 mL of a hexadecane solution for a reaction time of 0.5-16 h. The initial concentration of DBT (0.17 g) in a sulfur equivalent amounted to S init = 500 ppm. The weight of the catalyst sample was 0.64 g.
The catalyst placed into the reactor was held under a nitrogen flow at 150°C for 1 h, then during the activation step in an H 2 + H 2 S (9 : 1) atmosphere at 390°C. The catalyst weight was 0.64 g, and the volume of DBT-containing n-hexadecane was 80 mL. After the reaction, the liquid phase and the powder were separated by decantation. The hexadecane was tested for residual sulfur (S res ) by X-ray fl uorescence (XRF) on a Spectroscan SW-D3 instrument. The composition of the model reaction product was identifi ed by chromatography-mass spectrometry (GC/MS) on a DFS magnetic GC/MS instrument (Thermo Scientifi c Germany).
The catalyst identifi ed during the experiments as the most active was subsequently used for the hydrotreatment of the mixed diesel fuel (DF) components. The hydrotreatment was carried out in a stationary autoclaving mode, using mixed DF from the Ryazan refi nery (90 wt % straight-run and 10 wt % from catalytic cracking) as a feedstock. The DF initially contained 2.08 wt % sulfur, 184 ppm nitrogen, and 48.9 wt % aromatic hydrocarbons, and had a density of 0.8574 g/cm 3 . Before being placed into the reactor, the DF was diluted in n-hexadecane at 1 : 40.2 to correspond to S init = 500 ppm, as in the model reaction for DBT.

RESULTS AND DISCUSSION
Figures 1a-1d provide electron micrographs of the initial molybdenite DMI-7 (Fig. 1a), of the methanolcontaining nanocrystalline sample prepared by grinding MoS 2 without dopants (Fig. 1b), and of the sample before the model reaction (Fig. 1c) and after the model reaction (Fig. 1d).
The Fourier transform of areas 1 and 2 in the methanolcontaining sample (Fig. 1c) visualizes a typical set of planes (002) and (103) in the MoS 2 hexagonal structure with d-spacings of 0.6153 and 0.6217 nm [10]. The dispersion of *MoS 2 + 100CH 3 OH(8) increases after the model reaction (Fig. 1d). The low-oriented single-layer and multilayer MoS 2 structures with a linear dimension L ≥ 20 nm and a stacking number of 10 to 20 contain multiple defects and agglomerates (Figs. 1b and 1c). The stacked nanocrystallites are distinguished by angular shifts and exfoliation after the reaction (Fig. 1d). Figure 2 illustrates the XRD patterns of the samples, whereas Table 1 presents the results of their processing, including the basal face length of MoS 2 nanocrystallites (L for CSRs), the d-spacings (D 002 ), the internal elastic stresses (Δd/d), and the microstrains in unit cells (c/a) induced by MA.
In an inert medium without dopants (samples nos. 1-3), an increase in MA time in the series 0…4…8 h reduces the average basal face length of the resultant MoS 2 nanocrystallites from 50 to 20 and 12 nm, respectively (see Table 1 and Fig. 2). Simultaneously, the  According to published reports, in supported catalysts such as Ni(Co)-Mo(W)-S/γ-Al 2 O 3 , the optimal size of MoS 2 nanocrystallite stacks on the support surface is 4-5 nm with an average of 1.5 ± 0.4 layers per stack [11]. However, in the case of bulk catalysts prepared under thermal decomposition of ammonium thiomolybdate followed by grinding of the resultant MoS 2 in a solvent medium, the average length of ordered CSRs (L) is 20-30 nm, with up to 10-15 layers per nanocrystallite stack [12]. In the prepared catalyst samples, the number of single layers per stack is 10-20. The length of basal planes exceeds 15-20 nm. In a previous study by the same researchers, where molybdenite was ground in a liquid medium with butanol excess, the prepared catalyst powder had a specifi c surface area of 81 m 2 /g [4].
The MoS 2 powders examined in the present study were prepared in one step by mechanical grinding of purifi ed mineral molybdenite. They are also nanocrystallites with an average size of L < 50 nm (Table 1). However, they have unexpectedly low specifi c surface areas (S sp = 1.5-2.2 m 2 /g), in contrast to the previously described systems [4,11,12]. Below is a probable explanation.
In the chemistry of metal nanopowders, the differences in BET parameters (Table 1) are usually attributed to agglomeration and conglomeration of nanoparticles and to the generation of new bulk hierarchical structures [13]. The energy saturation in nanostructures is known to be eliminated by relaxation of internal stresses in crystallites as well as by mutual collapses of the exposed faces of nanoparticles. On the other hand, the chemical degradation of nanosystems is caused by the secondary conversion of unstable and metastable surface compounds during their interaction with environmental media. The products may have various adhesion and cohesion strengths. These processes may occur simultaneously, so they are diffi cult to separate. Therefore, the S sp values provided in Table 1 are, in fact, "apparent" values corresponding  to agglomerated nanocrystallites. The causes of the agglomeration are described below (Table 2).
For heterogeneous systems, any chemical interaction is well known to begin at the interface. Therefore, for the purposes of evaluating the catalytic activity of MA molybdenite samples, the chemical state of their surface is important to know. Table 2   For the samples of Group III, prepared in the presence of 100 μL of alcohols and hexane, the variation trends are different. The quantities of sulphate anions on the surface of molybdenite powders with n-butanol and n-hexane are almost equal (19-20%). So are the respective SO 4 2-/ S 2ratios (0.24-0.25). In contrast, the CH 3 OH-doped sample (S-2) and the C 2 H 5 OH-doped sample (S-5) have obvious differences: 34.4 and 23.7% SO 4 2-, respectively.
On the other hand, for the hexane-containing sample of Group IV (S-7-Ar), the surface affi nity of SO 4 2anions is as pronounced as for the ethanol-containing MoS 2 (S-5-Ar). Sulfate anions are present in amounts of 8.6% and 8.8 %, respectively.
Catalysts S-2-1 and S-5-1, obtained after the model DBT HDS reaction, are classified as Group V. As compared to S-2 and S-5, the presence of sulfate anions declines by a factor of 11 and 5, respectively. Moreover, their bonds with the catalyst surface become so weak that, after subsequent irradiation with Ar + ions (Group VI), the surface sulfate anions are not detected at all (0%).
The IR spectra of *MoS 2 + 100CH 3 OH(8) before and after the reaction (spectra 1 and 2, respectively, in Fig. 3) agree with, and add to, the above XPS data. The surface of the initial powder (Fig. 3, 1) exhibits a bending band of weakly adsorbed water near 1630 cm -1 , in addition to bands that can be attributed to the sulfate ion (SO 4  4 2-. Although the catalyst's spectrum after the reaction (Fig. 3, 2) is much less congested, it indicates the presence of molybdenum oxides on the surface. Some additional points should also be noted. According to published data, sulfate anions or their precursors can primarily exist on the powder surface in covalently bound forms [14]. At the same time, due to interaction with air moisture (Fig. 3, 1), trace amounts of acids may also be present on the surface (Table 2). Obviously, this very presence is directly responsible for the above-mentioned agglomeration of powder nanoparticles and for the low S sp values (see Table 1). When labile compounds ingress from the powder surface into the aqueous medium, corresponding acids are readily formed. The addition of barium chloride to the aqueous phase of powder suspensions leads to a curdy precipitate of barium sulfate, which represents an analytical test for the presence of sulfate anions [15] and sulfuric acid. It is not only sulfate anions that can act as H 2 SO 4 precursors. Precursors can also be other readily hydrolysable S-Oand O=S-containing compounds able to pass from the surface of catalyst particles into the aqueous phase, thus causing low pH values (Table 3) and the agglomeration of powder particles.  achieve a required HDS degree) extends the MA grinding time from 8 to 12 h. It is only under this condition that the residual sulfur content (S res ) decreases from 50 to 7 ppm. Based on the fi ndings illustrated in Figure 4, most experiments in this study were carried out on the samples with τ MA = 8 h ( Table 3). The Table 3 data on samples S-1, S-2, S-3, and S-4 prepared with τ MA = 8 h (see also Fig. 3) show that a decrease in the amount of methanol dopant from 200 μL to 100, 50, and 25 μL affects the S res in the DBT HDS products according to the following trend: 50...3...25... 47 ppm, respectively. Moreover, a difference in the methanol amount as small as 50 μL leads to an eight-fold drop in the residual sulfur content in the HDS product (S res = 25 ppm for S-3 vs. S res = 3 ppm for S-2). This suggests a threshold type of activity growth for S-2. Such effects indicate the high structural sensitivity of molybdenite to the amount of methanol doped during MA and to resulting changes in the catalytic activity of the samples. This may be due to the layered structure of the substrate. Table 3 also clearly shows that samples 1.3 and S-2 provide the highest DBT conversion degree, as indicated by the lowest S res (7 and 3 ppm, respectively). However, the reaction rate constants for these two samples differ by a factor of nearly two (0.35 and 0.65, respectively). These samples were prepared using different MA times (12 vs. 8 h) and different amounts of methanol dopant (200 vs. 100 μL). The pH values, which refl ect the amounts of sulfate anions and other acid-forming S-O structures on the powder surface, are 2.8 vs. 2.1, respectively. The latter value is the lowest among the entire series of samples under study. The reaction product for S-2 is represented by biphenyl (BPh) and cyclohexylbenzene (CHB) in amounts of 43% and 57%, respectively (see Table 3 and Fig. 5, reaction route 1). The HDS product for sample 1.3 contains BPh and CHB in almost equal amounts (49 and 50%, Table 3). For the catalytic system *MoS 2 + 200CH 3 OH(12) ( Table 3, Table 3), which is signifi cantly different from the S-2 case. The replacement of methanol with ethanol, butanol, and hexane (S-5, S-6, and S-7, respectively) leads to the preferential conversion of DBT to BPh in amounts of 91-94%. The and BCH is bicyclohexyl. c Aqueous pH in powder suspension. d DBT conversion rate constant, k, h -1 . e The notation "-" refers to lack of data. f 0.01 mL of H 2 SO 4 was taken instead of powder catalyst. catalysts in Table 3 are listed in descending order of the dielectric permittivity (ε) of the dopants used for their preparation. In accordance with such an approach, the reaction mechanism changes towards an increasing contribution of the cracking ("direct") route (Fig. 5,  route 1). However, the DBT conversion degree remains low (S res = 20-47). The highest DBT HDS rate constant (0.65 h -1 ) was achieved for S-2.
An interesting result was obtained in the presence of protic H 2 SO 4 in amounts that ensure pH 2.1 (a value typical of the most active catalyst S-2). In contrast to S-2, the hydrodesulfurizing effect in the model reaction provided by the protic acid catalysis only reaches S res = 470 ppm (Table 3, sample 1.4, S init = 500 ppm).
For a more visible representation of the correlation between the XRD and XPS data (Tables 1 and 2) and the catalytic activity of the samples (Table 3), Fig. 6 presents a polynomial curve plotted for the most active samples (Table 3, S res < 30 ppm). This curve indicates the relationship between the residual sulfur and the BPh content in the product.
The curve in Fig. 6 shows that the high HDS ability of the bulk sulfi de catalysts tested in this study leads to biphenyl formation in amounts less than 50%. This corresponds to the cracking route of DBT conversion. In other words, the presence of biphenyl in the DBT HDS product in an amount not exceeding 50% is not only a marker of the "direct" reaction route, but also points to the high HDS ability of the catalyst examined in the study.  pH values in the aqueous phases of the CH 3 OH-doped nanopowder suspensions (τ MA = 8 h, Table 3) and the DBT HDS degree in their presence. The reliability of linear approximation (Fig. 7) is R 2 = 0.9927. The hexane-, water-, and methanol-doped samples with τ MA = 12 h do not fall within the linear correlation area (Fig. 7). It is worth noting that the pH values in the aqueous phases of the powder suspensions lie in the strong-acidic region of pH 2.1-2.6. For S res < 50 ppm, the amount of methanol (25 to 200 μL) and the HDS ability of the sample do not exhibit pronounced correlations with corresponding pH. However, the aqueous pH in powder suspensions can be considered as a predictor of the powder's catalytic activity.
A reasonable question is, to what extent the sulfate anions on the MA molybdenite powder's surface are responsible for the pH values. The answer is more or less provided in Fig. 8a Recent publications have noted a positive effect of S-O-containing compounds in sulfi de catalysts on their activity in HDS processes. However, the true cause of this effect has remained unclear. The above-described fi ndings of the present study are consistent with the conclusions made by other researchers for catalytic systems with a different composition prepared by different methods [16].
The grinding of MoS 2 in a dispersed liquid medium results in physical adsorption of the solvent on its surface    [4]. Therefore, the fi nal physicochemical properties of MA-MoS 2 can be signifi cantly affected by the dielectric permittivity (ε) of the dispersion medium [17], which, in turn, is a key infl uencing factor for the Coulomb force between point charges on the surface of the material subjected to grinding. This assumption is supported by the above discussions on the data of Table 2 and Table 3.
The specifi city of the MoS 2 systems considered in the present study stems not only from the variety of types and amounts of dopant liquids, but also, as a result, from the non-uniformity of the chemical composition of the nanocrystallite surfaces generated. These surfaces consist of different combinations of ions, Mo oxides, SO 4 2groups, as well as S-H and S-O structural fragments (Fig. 3). Given that the separation degree of induced charges in the prepared powder samples (as listed in Table 3) will be different for each individual ε value, it is reasonable to anticipate different contributions of dopants to the catalytic activity of MA samples. To the best of our knowledge, such an approach to the evaluation of the effects of dopant's trace amounts has never been considered in published reports.   Table 3). This suggests that the MA powder preparation conditions that were implemented ensure free and uniform access for the dopants to the substrate surface. The function in Fig. 9 is symbatic: an increase in the dopant's dielectric permittivity corresponds to an increase in the HDS ability of the MA sample. This plot is consistent with the data in Table 2.
Among the dopants examined, sample S-2 (*MoS 2 + 100CH 3 OH(8)) has the highest HDS ability (see Table 3). The catalytic activity of this sample proved to remain unchanged (S res ≤ 3 ppm) after fi ve (5) service cycles and even after 6-month storage. Importantly, this ability is reproducible without intermediate regeneration of the sample surface, provided that the sample is stored under a solvent bed. Table 4 presents the data on hydrotreatment of mixed DF with a high presence of sulfur, nitrogen, and aromatic hydrocarbons. The initial composition in the presence of *MoS 2 + 100CH 3 OH(8) is provided.
The presence of heteroaromatic nitrogen compounds and polyaromatic condensed structures in the feedstock is well known to signifi cantly complicate the hydrotreatment process [18]. This is due to competition and blockage of catalytically active sites, as well as their poisoning.  Obviously, the DF group composition presented in Table 4 is too hard for hydrotreatment under relatively mild temperature and pressure conditions indicated. Nonetheless, the HDS of DF components over *MoS 2 + 100CH 3 OH(8) resulted in a 8-fold reduction in residual sulfur (S res = 62 ppm). According to published data, the hydrogen excess in hydroprocesses generally amounts to 200-600-1000-1200 times [19]. The experimental conditions implemented in the present study correspond to a four-fold H 2 defi ciency, which serves as evidence of a considerably higher resource provided by the novel catalysts studied here.

CONCLUSIONS
Under the conditions of solid-phase mechanical activation of molybdenite in an inert atmosphere containing small amounts of polar/nonpolar liquids, novel unsupported bulk single-component sulfi de catalysts with high hydrodesulfurization ability were prepared for the fi rst time. It was demonstrated that trace amounts of polar and nonpolar liquid dopants can signifi cantly affect the dispersion, surface chemical state, and catalytic activity of powders, all in accordance with the dielectric permittivity of the organic dopants. The defect level of MA molybdenite nanocrystallites is not a decisive criterion of their catalytic activity. The chemical degradation of the samples in the air was found to lead to the formation of surface sulfate anions that shield the catalytically active sites, potentially Mo ions in various oxidation states. Being responsible for the agglomeration of nanoparticles, the sulfate anions shield active sites, thus contributing to maintaining the initial activity level of nanodispersed molybdenite samples for extended periods (6-12 months). The reaction mechanism of DBT hydrogenolysis is fairly consistent with the cracking route. Even under hydrogen defi ciency in the reactor, the presence of a methanol-containing MoS 2 intercalate (100 μL) ensures a six-fold reduction in the sulfur content of the diluted DF.