In the modern oil refining industry, hydrotreating is among the largest-tonnage processes for producing environmentally friendly fuels. Hydrotreating enhances the commercial quality of products due to removal of sulfur and nitrogen compounds and hydrogenation of polyaromatics [1, 2]. However, the deteriorating quality of raw materials and the necessary reprocessing of recyclable petroleum products suggest an urgent need to improve the hydrotreating performance, primarily by using more active catalysts [3].

On an industrial scale, hydrotreating has been conventionally catalyzed by γ-Al2O3-supported Co(Ni)Mo sulfides [1]. There are two major approaches for enhancing the performance of hydrotreating catalysts: by adjusting the properties of the active Co(Ni)MoS phase (such as dispersion, sulfidation degree, and the content of type 2 species); and by using advanced supports that have advantages compared to conventional ones (such as a hierarchical pore structure, a greater number of Brønsted acid sites, and modification with additional elements) [4, 5].

The properties of the active catalyst phase—a component directly involved in the catalytic reaction—can be improved by varying the composition of precursors. For this purpose, organic components (e.g., oxyacids and glycols) have been added to impregnating solutions to promote the embedding of Co(Ni) promoters in the active sulfide phase [6–8]. Furthermore, heteropoly compounds (HPCs), in particular Keggin-type structures, have been demonstrated as promising active phase precursors. HPCs are a class of inorganic multinuclear complexes based on atoms of oxygen, some transition metals (e.g., Mo, W, V, Nb, and Ta), and nonmetals (e.g., Si, As, B, and P) in the highest oxidation state. This class is distinguished by an extraordinary variety of structural types and properties; for example, phosphorus-containing heteropolyacids are precursors of highly active oxide catalysts for oxidative dehydrogenation and of sulfide catalysts for hydrotreating [4, 10]. HPCs are commonly marked by the variability of the molecular structure of anions in response to changes in the pH, concentration, or composition of the solution [11]. The anion structure in the impregnating solution has a major effect on the active phase. For example, in hydrodesulfurization (HDS) and hydrogenation, the catalyst activity has been enhanced by adding vanadium to an HPC precursor of the active phase [12].

A catalyst support for hydrotreating can be modified by incorporating additional elements (e.g., B, Si, P, La, Ti, Zr, K, or Nb) and, thus, affecting both the acid properties of the support and the interaction between the active phase and the support [13–16]. Another promising approach is to use composite supports consisting of several types of materials [17, 18]. For example, the addition of halloysite aluminosilicate nanotubes (HNTs) to an oxide support has been shown to improve the strength and textural properties of the catalyst [19]. Due to the developed charge-bearing surface, the external and internal surfaces of halloysite consist of a negatively charged silica and a positively charged alumina, respectively. The advantages of this material include environmental friendliness, a large pore diameter (10–30 nm) [20], and the ability to selectively modify both the external or internal surfaces. HNTs have proven efficient as supports of sulfide catalysts in hydrotreating of petroleum diesel-range hydrocarbons with the high catalytic activity being provided by the stabilization of active CoMoS species in the HNTs [21].

The purpose of the present study was to synthesize and characterize HNT-supported vanadium-containing CoMo sulfide catalysts, and test them in HDS of dibenzothiophene (DBT). Specifically, we investigated the use of HNTs as a functional component of catalyst supports, as well as the formation of active particles as Keggin-type phosphorus- and vanadium-containing heteropolyacids were loaded from impregnating salt solutions.

EXPERIMENTAL

Initially, Keggin-type PMoV heteropolyacids (specifically, H4PMo11V1O40) were synthesized in accordance with the following sequential procedure described in [13]. V2O5 (≥99. 6%, Sigma Aldrich) was dissolved in H2O2 (a 37% solution, Reaktiv, Russia) in order to prepare peroxovanadium compounds spontaneously decomposing to form a H6V10O28 solution. This solution was stabilized by adding H3PO4 (85% H3PO4, Vekton, Russia) to prepare a H9PV14O42 solution. The H9PV14O42 solution was introduced into a boiling aqueous H3PO4 + MoO3 suspension (MoO3: ≥99%, Vekton). As the suspension evaporated, MoO3 was gradually dissolved to form H4PMo11V1O40. The resultant solution was concentrated by evaporating and used to prepare an active phase precursor [22].

The γ-Al2O3 support was a 2-mm dia. Alumac 3 extrudate (Alumac, France).

An HNT-based support (Al2Si2O5[OH]4, Sigma-Aldrich) was also prepared by extrusion using Pural SB boehmite (Sasol) as a binder and nitric acid (65 wt %, Reaktiv) as a peptizer, with an HNT to boehmite ratio of 70 : 30 (w/w). The extrudates were dried at 80–120°C for 6 h, then calcined at 550°C for 4 h. After the drying and calcination, the support extrudates were about 1 mm in diameter and 1.5–2 mm long.

CoPMoV catalysts were prepared by single incipient wetness impregnation of the support with a combined solution of H4PMo11V1O40, CoCO3∙mCo(OH)2∙nH2O (≥98%, Vekton), and citric acid (98%, Reaktiv). After the impregnation, the samples were dried at 60 and 80°C for 2 h each and at 110°C for 6 h. The content of active metals was chosen based on published reports (18 wt % MoO3 [18]); the amount of Co was calculated so that the Co to Mo molar ratio would remain at 0.50. The amount of metals in the synthesized catalysts was controlled using an EDX800HS X-ray fluorescence analyzer (Shimadzu, Japan). Prior to catalytic tests, the catalyst was ground, and only 0.25–0.50 mm particles were used in the test.

The textural properties of the supports and oxide catalysts were characterized by low-temperature nitrogen adsorption on a Quantachrome Autosorb-1 porosimeter. The specific surface area of the catalysts was evaluated by the Brunauer–Emmett–Teller (BET) method at a relative partial pressure (P/P0) of 0.05–0.3. The total pore volume and diameter distribution were calculated using a Barrett–Joyner–Halenda (BJH) model.

The oxide catalysts were examined by temperature-programmed reduction (TPR) using a TPDRO 1100 automated high-precision catalyst characterization system (Thermo Scientific, USA) equipped with a thermal conductivity detector (TCD). Immediately before analysis, the samples were dried in argon at 140°C for 2 h. The TPR was carried out in an N2/H2 mixture (5 vol % H2, 25 mL/min) under heating from 25 to 1000°C at a rate of 10°C/min.

The oxide catalysts were further characterized by backscattered Raman spectroscopy on an inVia microspectrometer (Renishaw, UK) equipped with a charge-coupled detector (CCD), an argon laser (λ = 532 nm), and a 1800 lines/mm grating (with a spectral resolution of 1 cm–1). The excitation source was focused to a spot 2 μm in diameter with a power of 1–5 mW. All Raman spectra were recorded at room temperature in the range of 100 to 3300 cm–1; however, due to an intense background signal of alumina supports, this paper only presents the spectral range of 600–1200 cm–1, typical of Mo–O stretching vibrations.

Sulfide catalysts were characterized by two-step TPR. First, the samples were sulfided in an H2+H2S flow (10 vol % H2S, the balance being hydrogen) at 400°C for 4 h. The sulfided catalysts were then exposed to TPR in an N2/H2 mixture (5 vol % H2, 25 mL/min) under heating from 25 to 600°C at a rate of 10°C/min.

The catalytic performance of mixed catalysts was tested using a fixed-bed flow-type laboratory reactor setup. The catalyst loading was 0.9 mL (0.25–0.50 mm particles). The catalysts were sulfided in the gas phase (H2S/H2 atmosphere, 10:90 v/v) at 400°C and 1 MPa for 2.5 h.

Given that DBT and naphthalene derivatives account for a major portion of sulfur compounds and polyaromatics in crude oil, we used these derivatives as model compounds in our laboratory tests of the hydrotreating catalysts. The model mixture had the following composition: 0.86 wt % DBT and 3.0 wt % naphthalene in toluene, with 1.0 wt % n-hexadecane as an internal standard. No toluene derivatives were detected during the catalytic test.

The catalytic activity in HDS and hydrogenation was investigated under the following conditions: 300–340°C, 3.0 MPa H2, feed gas hourly space velocity (GHSV) 20–60 h–1, and an H2 to feedstock volumetric ratio of 600 nL/L. The hydrotreating was carried out for at least 8 h after stable reagent conversion was reached.

The composition of liquid products was measured on a Crystal 5000 chromatograph equipped with a flame ionization detector (FID) and a 30 m×0.5 mm×0.5 μm column with an OV-101 nonpolar phase (dimethylpolysiloxane as a stationary phase).

The catalytic activity in HDS and hydrogenation was derived from the apparent reaction rate constant calculated by the pseudo-first-order equation:

$$k = - {F \over W}{\rm{ln}}\left( {1 - x} \right),$$
((1))

where F is the flow rate of DBT or naphthalene as a reagent (mol/h); W is the MoO3 weight (g); and x is the reagent conversion (%).

The selectivity of DBT HDS was evaluated along two routes: direct HDS and prehydrogenation (HYD). Accordingly, we calculated SelHYD/HDS: the ratio of the total concentration of products obtained by the “direct hydrogenation” route, including tetrahydrodibenzothiophene (THDBT), dicyclohexyl (DCH), and cyclohexylbenzene (CHB), to the concentration of biphenyl (BP), a product obtained by the direct desulfurization of the DBT molecule:

$$Se{l_{{\rm{HYD}}/{\rm{HDS}}}} = {{{C_{{\rm{CHB}}}} + {C_{{\rm{DCH}}}} + {C_{{\rm{THDBT}}}}} \over {{C_{{\rm{BP}}}}}},$$
((2))

where CCHB, CDCH, CTHDBT, and CBP are the concentrations of cyclohexylbenzene, dicyclohexyl, tetrahydrodibenzothiophene, and biphenyl, respectively [23].

The apparent activation energy (Ea) was evaluated based on lnk plotted experimentally as a function of temperature.

RESULTS AND DISCUSSION

Properties of supports and oxide catalysts. This section presents the characterization data on the supports and related oxide catalysts: see Table 1 for the compositions and textural properties, Figure 1 for the adsorption/desorption isotherms, and Figure 2 for the pore size distribution. The isotherms are IUPAC Type IV, typical of mesoporous materials [24]. The alumina-based samples had a Type H1 hysteresis loop with a maximum at P/P0 = 0.6–0.8, indicative of cylindrical pores.

Table 1. Characterization of supports and a series of oxide catalysts
Full size table
Fig. 1.
figure 1

Nitrogen adsorption/desorption isotherms at 77 K for supports and oxide catalysts.

Full size image
Fig. 2.
figure 2

Pore size distribution for supports and oxide catalysts.

Full size image

IUPAC Type H3 hysteresis loops were observed for both the HNT-based support and the HNT-based oxide catalyst. These samples exhibited a bimodal pore size distribution with peaks near 6 nm and 20–30 nm, assigned to the binder and to halloysite cavities, respectively. The loading of active metal precursors decreased both the specific pore areas and pore volumes. The pore size distribution curves also reveal that the peaks shifted towards smaller diameters.

Figure 3 illustrates the TPR profiles of the oxide catalyst samples. The curves display two major peaks. The first peak (at Tmax = 500–520°C), similar for both samples, is attributed to the reduction of octahedral Mo species. The second peak (high-temperature) corresponds to the reduction of tetrahedral Mo species; its shift towards lower temperatures for the HNT-supported sample (850°C vs. 930°C) indicates a lower interaction between the active phase and the support. The 620°C peak is attributed to the reduction of Co species [25].

Fig. 3.
figure 3

TPR profiles of oxide catalyst samples.

Full size image

The oxide precursor state on the catalyst surface was further characterized by Raman spectroscopy (Fig. 4). A major peak at 944 cm–1 with a broad shoulder near 892 cm–1 is assigned to \({{\rm{P}}_{\rm{2}}}{\rm{M}}{{\rm{o}}_{\rm{5}}}{\rm{O}}_{23}^{6 - }\) anions, all in agreement with [26]. The spectrum of CoMo/HNT also has a peak at 995 cm–1, attributed to initial \({\rm{PM}}{{\rm{o}}_{{\rm{12}}}}{\rm{O}}_{40}^{3 - }\) anions. This indicates that the Keggin-type complexes contained in the impregnating solution were partially retained [27].

Fig. 4.
figure 4

Raman spectra of oxide catalyst samples.

Full size image

Properties of sulfide catalysts. To properly determine the active phase properties of the catalysts, the sulfided samples were also examined by TPR (Fig. 5).

Fig. 5.
figure 5

TPR profiles of sulfided catalyst samples.

Full size image

Reduction of sulfided CoMo catalysts causes a break of Mo–S bonds, thus making it possible to evaluate the bond strength and the potential for the generation of coordinatively unsaturated sites (CUS). Sulfided CoMoS/HNT and CoMoS/Al2O3 catalysts are known to exhibit hydrogen uptake between 150 and 300°C [28]. The reduction of the HNT-supported sample started at a markedly lower temperature, thus indicating a lower Mo–S binding energy in this sample. The greater mobility of sulfur atoms might be associated with a decreased interaction between the active phase and the support due to the presence of silica in HNTs, thus increasing the content of multilayer CoMoS species. During the reduction, CUS were generated at the edges and corners of CoMoS species. The amount of hydrogen uptake correlates with the amount of newly-generated CUS. CoMoS/HNT and CoMoS/Al2O3 exhibited similar hydrogen uptakes and, hence, similar CUS numbers.

Table 2 presents the catalytic test data for the sulfided catalysts, specifically the values of DBT and naphthalene conversion under various process conditions. The DBT conversion was 23.3 to 96.2% for CoMoS/HNT and 11.4 to 82.3% for CoMoS/Al2O3.

Table 2. Catalytic test data for sulfided catalysts. Reaction conditions: 300–340°C, 3.0 MPa H2, GHSV 20–60 h–1, and H2 : feed = 600 nL/L
Full size table

Under the test conditions (300–340°C, 3.0 MPa H2, GHSV 20–60 h–1, and H2 : feed = 600 nL/L), tetralin was the only naphthalene hydrogenation product identified; no products of its deeper hydrogenation (e.g., cis-/trans-decalin) were detected. The naphthalene conversion was 7.9–45.2% for CoMoS/HNT and 2.3–25.8% for CoMoS/Al2O3.

The highest conversion values (96.2% DBT and 25.8% naphthalene) were achieved at 340°C and GHSV 20 h–1 for CoMoS/HNT. Separate mention should also be made of a noticeable difference between the DBT conversion over CoMoS/HNT (85%) and over CoMoS/Al2O3 (41%) at 340°C and 40 h–1. This impressive catalytic activity of the HNT-supported sample at the higher GHSV might be associated with its better-developed pore system.

Table 3 provides the reaction rate constants and activation energies observed during the catalytic test.

Table 3. Reaction rate constants and activation energies in hydrodesulfurization of DBT and hydrogenation of naphthalene over sulfided CoMo catalysts
Full size table

Over the entire temperature range, CoMoS/HNT exhibited markedly higher rate constants of DBT hydrodesulfurization than CoMoS/Al2O3: 34.8 vs. 22.8 at 300°C, 93.0 vs. 65.1 at 320°C, and 206.1 vs. 133.9 at 340°C. The almost equal values of the apparent activation energy of DBT HDS over both catalyst samples (129.5 and 130.0 kJ/mol) are indicative of similar reaction mechanisms. Moreover, these activation energies, consistent with available published reports (118.6 kJ/mol [29]), serve as evidence that the catalytic activity was investigated in the kinetic region.

Table 4 presents the selectivity of DBT HDS along two routes: direct desulfurization (HDS) and prehydrogenation (HYD).

Table 4. Selectivity of DBT hydrodesulfurization along direct desulfurization (HDS) and prehydrogenation (HYD) routes
Full size table

In the naphthalene hydrogenation, the halloysite-supported catalyst also exhibited greater activity, as indicated by the higher rate constants. The lower apparent activation energy of naphthalene hydrogenation for CoMoS/HNT (81.5 kJ/mol compared to 98.6 kJ/mol for CoMoS/Al2O3) might be associated with the high activity of the active phase’s edge sites in this hydrogenation reaction. The increased number of the edge sites was induced by the presence of silica-enriched surface areas in the HNTs, which weakened the interaction between the support and the active sulfide species. This assumption is confirmed by the TPR data for the sulfided catalysts.

The HNT-supported catalyst exhibited higher selectivity (0.31 compared to 0.21 for CoMoS/Al2O3). In the presence of this catalyst, therefore, the hydrodesulfurization occurred to a greater extent by preliminary hydrogenation of DBT’s aromatic rings to produce THDBT and CHB. The superiority of this sample in terms of SelHYD/HDS is indicative of the difference in the active phase’s properties (specifically in the number and ratio of hydrogenation and desulfurization sites) between the HNT-supported and Al2O3-supported samples.

The TPR characterization of the oxide and sulfide catalysts, the Raman spectroscopy data, and the catalytic test data demonstrated that the differences in the structure and composition of the support affected the properties of both the oxide precursors and the sulfided active phases of the catalysts. This, in turn, weakened the interaction between the active phase and the support, as evidenced by the physicochemical characterization data.

CONCLUSIONS

Sulfided CoMo catalysts supported on γ-Al2O3 and halloysite nanotubes (HNTs) were synthesized by incipient wetness impregnation with solutions of Keggin-type vanadium-containing heteropolyacids. The HNT-supported catalyst exhibited a higher activity in hydrodesulfurization of dibenzothiophene and in hydrogenation of naphthalene than the conventional alumina-supported catalyst. The hydrodesulfurization rate constant at 340°C for the HNT-supported catalyst was 206.1×104 mol g–1 h–1 compared to 133.9× 104 mol g–1 h–1 for its counterpart on the conventional support. This enhanced catalytic activity was induced by a weaker interaction between the active phase and the HNT support and, hence, by a greater number of newly-formed multilayer CoMoS species highly active in hydrogenation reactions.

This study demonstrated the high performance of HNTs as a support for vanadium-containing cobalt–molybdenum catalysts in hydrotreating processes. Thus, further research into the activity of this catalyst type in processing of real-world diesel-range hydrocarbons, including recycled products, holds much promise.