Hydrodesulfurization Catalysts from Aluminium-Containing Waste

Abstract

A possibility of using aluminum hydroxide, a product of processing of the adsorbent that lost its sorption activity in purification of a polyethylene solution cyclohexane, was demonstrated in the synthesis of aluminum kaolin borate supports for hydrodesulfurization catalysis in oil refining. The influence of typical impurities V, Ti and Na in the composition of the spent adsorbent on the physicochemical and catalytic properties of supports and catalysts was determined. Alumina-kaolin-borate catalyst carriers with the Ti content of approx. 0.2% have been prepared by atomic layer deposition (ALD) of TiCl₄ onto the granules. A positive effect of molecular layering with TiO₂ has been revealed by temperature programmed reduction and oxidation in combination with hydrodesulphurization activity testing.

Graphical Abstract

1 Introduction

As a part of the global efforts related to circular economy, a significant attention is given to recycling of spent catalysts. Hydrodesulfurization is one of the key processes in oil refining, being typically carried out in the presence of Ni-Mo, Ni-W and Co-Mo sulfide catalysts, supported on refractory oxides such as γ-alumina under with hydrogen pressure and temperature varying between 0.5 and 4 and even 8 MPa and 300–400 °C respectively depending on the feedstock [1,2,3]. Oxide catalyst precursors, in the form of trilobe or pentalobe extrudates or hollow cylinders with diameters of 1.5–2 mm and lengths of 3–5 mm, are preliminarily subjected to sulfidation.

In industry, Al-Co-Mo and Al-Ni-Mo hydrodesulfurization catalysts are most widely used [4], when desulfurization of various feedstock with simultaneous removal of nitrogen and oxygen is required along with dearomatization [1,2,3,4,5,6]. There is a substantial amount of literature available on the influence of preparation methods [7], catalyst composition, type of the support (e.g. alumina, titania, zirconia, silica) [8], reducibility of the metal [9], its dispersion, structure of the active sites in the oxide and sulfided forms [7,8,9,10,11,12,13,14].

The mechanism of hydrodesulfurization on sulfided Al-Co-Mo and Al-Ni-Mo was studied using various model substances such as thiophene, benzothiophene [15], and dimethyldibenzothiophene [7], emphasizing differences in the catalytic HDS behavior of cobalt and nickel. Using toluene as an example, a correlation was established between the hydrogenating ability and the Co/Mo ratio in crystallites, and the decisive role of mixed Co-Mo centers decorating “CoMoS” nanocrystals in the sulfided catalysts was noted [16].

In addition to the influence of operation parameters (temperature and flow rate [14]), presence of various additives including compounds containing phosphorus [11, 12] and boron [13, 17] was demonstrated to affect hydrodesulfurization of real oil fractions. Moreover, as already indicated acidity of Al₂O₃-TiO₂ and Al₂O₃-MgO carriers [14] can also have an impact. Of large importance for improving hydroprocessing catalysts is the uniform distribution of the active phase precursors and diminishing interactions with the support surface [18], as strong interactions of the hydrogenating metals (Ni, Co, Mo) with γ-Al₂O₃ lead to lower amounts of the MoS₂ phase active in hydrodesulfurization [19].

In the context of circular economy and utilization of waste, it is tempting to apply alumina-containing industrial waste for preparation of environment-friendly hydrodesulfurization catalysts suitable in HDS of oil refining streams. One of such hazardous wastes generated in large quantities, is spent alumina adsorbent (SAA) applied for purification of the waste cyclohexane solution, i.e. the reaction medium for polymerization of ethylene into polyethylene with organometallic compounds as catalysts. After completion of operation, SAA contained deposits of deactivated polymerization catalysts organometallic compounds of VC₁₆O₅H₁₄ and TiC₁₆O₄H₁₄, as well as chlorine-containing substances. When processing it into a fresh adsorbent for purifying exhaust gases from a naphtha reformer, SAA granules were successively washed with an excess of 1% NaOH solution, deionized water and calcined at 400° C [20]. This is followed by activation in 8–9% NaOH solution at 40 °C, with subsequent drying at 150–180 °C.

Previously, the authors have considered utilization of the spent alumina adsorbent by processing SAA into fresh pseudo-boehmite, which was then applied for synthesis of sorbents for removal of chlorinated compounds [20, 21]. The authors made some preliminary attempts to use the alumina-containing industrial waste SAA for preparation of environment-friendly hydrodesulfurization catalysts suitable in HDS of oil refining streams [20, 22,23,24].

Based on these studies, it was concluded that impurities of vanadium and titanium compounds were not completely removed when treating SAA with an alkaline solution [23]. On the other hand, presence of titania in combination with a more traditional alumina support has been demonstrated for catalytic transformations relevant to biomass valorization via hydrodeoxygenation [25].

The main aim of this work was assessment of utilization of spent alumina adsorbents as a carrier for NiMo catalysts exploring especially the role of titania introduced to the support by atomic layer deposition (ALD). ALD is a method based on the sequential use of a gas-phase chemical process, in particular alternating reactions of TiCl₄ with surface hydroxyl groups of Al₂O₃ and hydrolysis, leading to deposition of a thin film. The resulting catalysts were tested in a pilot unit with real refinery feeds.

2 Experimental

2.1 Materials

Details of the preparation of the supports have been reported previously [20]. Here the most pertinent details are given. The waste aluminium oxide adsorbent (SAA) was supplied from the Shurtan gas chemical complex, and had the chemical composition: and while kaolin was procured from the Angren quarry—"Angren Kaolin" LLC.

To modify the adsorbents, the following materials were used: boric acid 99.6% H₃BO₃ (Ekotec) and TiCl₄ 99.9% – SKT Taurus.

The following materials have been used for modification of adsorbents: NaOH from AO Navoiyazot, and Na₃VO_4.12H₂O, chemically pure, from Chem Craft. Other reagents for preparation of the hydrotreating catalysts are listed below: (NH₄)₄Mo₇O₂₄ × 4H₂O—(Uzbek Factory of Refractory and Heat Resistant Metals, Chirchik), phosphoric acid 98.5% H₃PO₄ (Ammofos OJSC, Almalyk), Ni(NO₃)2 × 6H₂O and Co(NO₃)₂ × 6H₂O (JSC Ural Factory of Chemical Reagents).

2.2 Synthesis of Catalysts

Waste processing consisted of preliminary dissolution of crushed SAA by heating in a solution of nitric acid with the formation of aluminum nitrate, precipitation of AlOOH by adding an ammonia solution at pH = 7–8, followed by maturation of the precipitate at 20–27 °C, filtering and washing, and finally drying the powder pseudoboehmite AlOOH. The grinded SAA (30% by weight with the particle size between 0.04 and 0.1 mm) and kaolin (10% by weight) were added to AlOOH powder during preparation of the carrier, named sample S1_akb, when peptization of the mixture of dry ingredients was carried out by stepwise addition of 3 wt% nitric acid and 2.5 wt.% boric acid solutions. The peptized paste was extruded into trefoil-shaped granules with a circumferential diameter of 4 mm and a length of 3–5 mm. After drying in air, the S1_akb granules were calcined for three hours at 550 °C. A sample of the AKB-Ti(0.18) support (hereinafter S2_ald) was obtained by applying TiCl₄ vapor to S1_akb granules using the ALD method [20]. Here and further in the sample designations, the number in brackets after the symbol of a chemical element corresponds to its content in wt%. The degree of ALD was monitored by absorption bands of surface hydroxyl groups in the near-infrared (NIR) spectrum at 4500 and 7300 cm^–1, which decreased due to interaction with TiCl₄. In the IR spectrum of S1_akb dehydrated at 100 °C, a set of narrow bands 3745; 3730; 3710; 3552 cm^–1 from OH groups was associated with aluminum, silicon and boron atoms, along with a wide band in the region of 3333–3400 cm^–1 from incompletely removed H₂O molecules. Subsequently Ti–OH bonds were formed upon exposure to water vapours and hydrolysis of Ti-Cl bonds followed by exposure again to dry nitrogen at 200 °C to remove the excess of water vapours. Formation of Ti–OH bonds was accompanied by an increase in the intensity of the bands at 4500 and 7300 cm^–1 in the NIR spectrum, as well as at 3670 and 3720 cm^–1 in the IR spectrum. The ALD cycles have been repeated to deposit the desired layers of titania. Formation of the titania layer on the surface of the granules was controlled by appearance in the Raman spectrum of an intense band at ca. 142 cm^{– 1}. The intense band at 142 cm^–1, along with low-intensity bands at 196, 396, 512 and 639 cm^–1, is characteristic of the crystalline anatase phase. The band in the region of 142 cm⁻¹ appeared in the Raman spectra of TiO₂ nanopowders with the nanocluster sizes of 5 to 30 nm [26] not being present in the spectrum of the original S1_akb carrier.

The Ni(3.1)Mo(8.3)P(0.9)/AKB-Ti(0.18) catalyst (hereinafter referred to as C-1_ald) was prepared by single impregnation of the trilobes of the S2_ald support with a combined aqueous solution of (NH₄)₄Mo₇O₂₄ × 4H₂O and Ni(NO₃)₂ × 6H₂O, stabilized with phosphoric acid, according to moisture absorption. Ni(3.5)Mo(11.2)P(1.6)/AKB and Ni(0.7)Co(2.5)Mo(8.2)P(0.9)/AKB-Ti(0.18) catalysts (hereinafter referred to as C-2_akb and C-3_ald) were obtained by impregnating the supports: first with an aqueous solution of (NH₄)₄Mo₇O₂₄ × 4H₂O, stabilized by an appropriate amount of orthophosphoric acid, and after heat treatment at 300 °C with an aqueous solution of nickel nitrate or a mixture of nickel and cobalt nitrates. Ni(3.8)Mo(11.0)P(0.35)/AKB, Ni(4.4)Mo(14.4)P(3.1)/AKB and Co(3.1)Mo(8.2)P(0.26)/AKB-Ti(0.18) catalysts (hereinafter referred to as C-4_akb, C-5_akb and C-6_ald) before applying aqueous solutions of nickel or cobalt nitrate were pre-impregnated twice with an aqueous solution of (NH₄)₄Mo₇O₂₄ × 4H₂O, stabilized with H₃PO₄ with an intermediate heat treatment at 300 °C. As a result, after the final heat treatment at 500 °C, the materials presented in Table 1 were obtained. For comparison purpose, Table 1 shows also characteristics of the industrial catalyst Co(3.4)Mo(8.83)P(0.3)/Al₂O₃ designated Cind.

Table 1 Physicochemical characteristics of supports and catalysts

The limited liquid uptake by the supports required the use of concentrated solutions to obtain catalysts of the calculated composition. Stable solutions of (NH₄)₄Mo₇O₂₄ × 4H₂O were obtained in the pH range = 1.3 ÷ 1.5, when, in addition to hydrated MoO₄²⁻ ions, [NiH₆Mo₆O₂₄]^4– heteropolyanions were formed, which coexisted with anions of the [CoH₆Mo₆O₂₄]^4– type, NO₃^– and PO₄^{– 3} anions and cations Co²⁺, Ni²⁺, NH₄⁺ [22]. The choice of H₃PO₄ as a reagent to prevent precipitation in the impregnating solutions was related to its impact on the structure of active phases namely formation of heteropolycompounds of transition metals [23].

The elemental analysis of the samples has been performed with scanning electron microscope (SEM) EVOMA 10 combined with EDX. The total content of titanium, vanadium, nickel and molybdenum in the obtained samples was also determined by photocolorimetry and of sodium by the flame photometry. The textural properties have been determined by mercury porosimetry (Carlo Erba) operating up 200 MPa.

Mechanical strength was determined by cutting the granule crosswise using a device with a special knife. The breaking load was calculated according to:

$${\text{P}}={{\text{ml}}}_{1}/{{\text{l}}}_{2}$$

(1)

where P is the breaking load, kg; m – the cargo weight, kg; l1 – distance from the beginning of the scale to the center of the load, cm; l2 is the distance from the beginning of the scale to the middle of the knife, cm. To compare the catalysts, the specific strength value was used calculated as the average value of the breaking load for 30 measurements divided by the granule diameter (mm).

XRD was done using Empyrean diffractometer. Assignment of the reflexes in the diffraction patterns was carried out using JCPDS card file. Crystallite sizes were calculated using the Scherrer equation:

$${\text{D}}=\mathrm{n \lambda }/\mathrm{ \beta Cos \theta }$$

(2)

where D is the average crystallite/ grain size in the powder (nm), n is a coefficient depending on the shape of the particles (n is taken equal to 0.9, similar to [7, 27]), λ = 1.54 is the radiation wavelength (Cu), β is the physical broadening of the line in the diffraction pattern in radians (on a scale of 2 θ).

Electronic diffuse reflectance spectra (ESDR) of the catalysts were recorded in the region of 50–10 kcm ⁻¹, on a Hitachi-330 spectrophotometer. The samples were first ground to a powder state, then pressed in a special cuvette followed by recording the spectra using MgO as a standard. Potential distortion of the spectra by species containing Ni, Co, Mo due to the support influence was taken into account [22]. The data are presented below using the Kubelka–Munk function F(R∞),

$${\text{F}}\left({{\text{R}}}_{\infty }\right)={\left(1-{{\text{R}}}_{\infty }\right)}^{2}/2 {{\text{R}}}_{\infty }$$

(3)

where R∞ is the reflectance at infinite depth [11].

The strength and concentration of acid–base centers on the surface of the samples calcined at 350 °C without contact with air were determined from the electronic spectra of adsorbed indicators [22]. Indicators, characterized by their respective pK_a value in parentheses below, triphenylcarbinol (-6.3), benzalacetophenone (-5.6), dicinthalacetone (-3.3), benzolazo diphenylamine (+ 1.5) and bromophenol blue (+ 3.8), exhibited specific absorption bands in the visible ESDR region upon interactions with surface aprotic and proton centers of varying strengths. The appearance in the ESDR spectrum of adsorbed triphenycarbinol of a band at 22.7–23.2 kcm⁻¹ and a band at 19.0 kcm⁻¹ in the spectrum of adsorbed dicinthalacetone indicated their adsorption on strong acidic sites of the Brønsted type (protic acidity). The presence of a band at 21.0 kcm⁻¹ in the spectrum of adsorbed dicinthalacetone indicated the presence of Lewis-type centers (aprotic acidity). To determine the concentration of surface centers with a given pK_a value, two spectra were obtained. The first of them characterized the state of the molecules of the corresponding indicator directly on the surface of the original sample, while the second corresponded to the spectrum of the indicator on the surface after poisoning the acid sites with a given amount of n-butylamine. The concentration of acid sites was determined by:

$${{\text{C}}}_{{\text{acid}}}={\text{F}}\left({{\text{R}}}_{\infty }\right) {{\text{C}}}_{{\text{but}}}/\mathrm{ F}\left({{\text{R}}}_{\infty }\right)-{\text{F}}{\left({{\text{R}}}_{\infty }\right)}_{{\text{but}},}$$

(4)

where C_acid – concentration of acid sites in mmol/g, C_but – concentration of butylamine in mmol /g, F(R∞) and F(R∞)_but are Kubelka–Munk functions at the maximum of the band of the acid form of the corresponding indicator, respectively, before and after poisoning of the surface acid sites with butylamine.

Raman spectra were obtained on an InVia Raman spectrometer (Renishaw). The measurements were carried out at room temperature. A Cobalt CW 532 nm DPSS laser RL532C100 (diode-pumped solid state laser) with a wavelength of 532 nm and a nominal energy of 100 mW was used as an excitation source.

Extraction of soluble substances from the catalysts was carried out first with distilled water and then with a 12.5 M aqueous ammonia solution. The extraction procedure was carried out three times with intermediate calcination at 500 °C. The content of the elements Co, Ni, Mo, Ti P, Si and Al extracted from the catalysts with water or an aqueous ammonia solution was determined by absorption spectroscopy using an SP 9 atomic adsorption spectrophotometer PYE UNICAM.

The temperature programmed reduction was performed with Auto Chem 2910 HP system, equipped with the thermal conductivity detector. First, a sample (100 mg) was heated in the quartz reactor from the ambient temperature to 500 °C at the rate of 5 °C/min under the nitrogen flow rate of 50 ml/min, followed by holding at this temperature for 30 min. After cooling to the room temperature in the nitrogen flow, 10 vol.% of hydrogen (20 ml/min) in argon was used for the catalyst reduction increasing temperature with the ramp 10 °C/min.

Prior to oxygen chemisorption, the catalyst sample (1 g) was heated to 500 °C under 10 vol.% of hydrogen in argon following by holding at the highest temperature for 30 min. Upon cooling to 30 °C, oxygen pulses on 0.2 ml were conducted recording the amounts of unreacted oxygen at the reactor outlet.

The catalysts were tested in hydrotreating of three types of hydrocarbon feedstock:

• 1) naphtha—crude gasoline fraction with a boiling point (T_boil) of 30–90° C, including mainly n-paraffins, iso-paraffins C₃-C₆ and naphthenes, as well as traces of benzene derivatives C₆-C₇, along with olefins (< 1%) and sulfur containing compounds (S_ini) – 0.15%;

• 2) crude kerosene fraction with the boiling point T_boil. of 120—315 °C, including mainly n-paraffins, iso-paraffins C₃-C₉ of ca. 38% and naphthenes—about 25%, polyaromatic hydrocarbons C₃-C₉ (PAH)—18%, as well as impurities of benzene derivatives C₆-C₈ and olefins—no more than 1%, and finally sulphur containing compounds with S_initial—0.25%, including mercaptans—0.02%;

• 3) deasphalted residue with T_boil. above 500–590 °C, containing predominantly paraffin-naphthenic hydrocarbons, PAH—33.1%.

Among impurities the following should be mentioned: resinous-asphaltene substances 1.27%, sulfur (S_initial) -1.32% and nitrogen—0.1%. The iodine value, which reflects the amount of unsaturated hydrocarbons, was 0.5 g I₂/100 g. Activity of the catalysts in hydrodesulfurization and hydrogenation of polyaromatic compounds in the fed was evaluated in a continuous fixed-bed reactor (Scheme 1) under conditions typical for industry. After loading into the reactor, the catalysts were sulfided at 400 °C for 4 h in a flow of 10 vol. % H₂S in H₂ under atmospheric pressure [23].

Scheme 1

Schematic of a high-pressure pilot plant for testing activity of hydrotreating catalysts. 1 – burette; 2 – piston pump; 3 – hydrogen cylinder; 4 – pressure gauge; 5 – reactor; 6—catalyst bed; 7 – porcelain balls; 8 – thermocouple; 9 – temperature regulator; 10-reactor heating; 11- cooler; 12- high pressure separator; 13 – low pressure separator; 14 –sample receiver; 15 – H₂S absorber; 16 – gas meter

Testing of catalytic properties was carried out at a volumetric feed rate of 2 h⁻¹, hydrogen to hydrocarbon feedstock ratio of 300 m³/m³, 320 °C and pressure (MPa): 3.0 for naphtha, 4.0 for crude kerosene fraction and 6.0 MPa for deasphalted residue. The change in the content of polyaromatic compounds was assessed based on the results of structural group analysis using the R-D-M method (refractive index-density-molecular weight).

3 Results and Discussion

3.1 Texture and Surface Properties of Materials

SEM analysis of the elemental composition of S1_akb and S2_ald carriers revealed mainly the presence of Al, Si and O (Fig. 1 on the left), and impurity elements (V, Ti, Cl and others) were detected only at individual points. Titanium was unambiguously detected only when scanning the surface of granules of the S2_ald support or the near-surface layer (Fig. 1 in the center). When scanning the surface of the catalyst granules, Ti practically did not appear in the spectra (Fig. 1 on the right). The average results of SEM analysis of transition elements and phosphorus at arbitrary points in the volume and on surface of the catalysts, as well as physicochemical characteristics are presented in Table 1.

Fig. 1

SEM analysis of the S2_ald carrier in the main body of the granules (left) and in the near-surface layer (center); catalyst Ni(0.7)Co(2.5)Mo(8.2)P(0.9)/AKB-Ti(0.18) – on the right

As can be seen from Table 1, introduction of a catalytically active phase decreased the surface area of S1_akb supports, while for S2_ald only minor changes were observed.

The mechanical crushing strength along the diameter of the granules decreased slightly after ALD modification of the carrier with titanium dioxide.

A decrease in the mechanical strength and the specific surface area after impregnation of the catalytically active phase occurred due to a decrease in the diameter of small pores in the final catalysts. Mo, P, Co, Ni as well as Al containing compounds were deposited in the pores due to partial leaching of the support in the impregnating solutions [28]. The numbers in the parentheses after the symbol of an element indicate its concentration in weight percent. No aprotic acid sites were found on the surface of all the supports and catalysts studied, in contrast to CoMo/γ-Al₂O₃P(2)Ce(1.75) catalysts [12]. ALD modification with titania was accompanied by a redistribution of acid sites on the surface of the S1_akb support. Rather strong proton centers with pKa <—5.6 disappeared at the expense of the concentration of weaker acid sites. Brønsted acid sites were found on the surface of the catalysts, the concentration of which increases with increasing concentrations of molybdenum and phosphorus, decreasing with increasing molar ratio Co + Ni/Mo.

3.2 XRD

The diffraction pattern of the S2_ald carrier, described in [20] and presented in Fig. 2, did not illustrate specific TiO₂ reflections corresponding to the interplanar distances d = 3.52 Å (anatase) or d = 3.25 and 2.487 Å (rutile) due to the low concentration and high dispersion of deposited titanium dioxide using the ALD method. According to SEM data [20], titanium was located in the outer layer, while diffraction patterns were taken in the form of a compressed powder of crushed S2_ald granules. Only broad peaks from γ-Al₂O₃ and narrow peaks from α-SiO₂ were observed, as in the diffraction pattern of the unmodified S1_akb support. Wide peaks from kaolinite with d = 7.17; 4.46; 3.57 and 2.48 Ǻ, characteristic of Angren kaolin, did not appear in the studied samples due to its destruction during calcination.

Fig. 2

Diffractograms: 1)S2_ald, 2) C-2_akb, 3) C-4_akb; 4) C-5_akb; 5) C-1_ald, 6) C-6_ald; 7) C-3_ald; 8) C_ind (industrial catalyst). – γ—Al₂O₃; – SiO₂; – CoMoO₄; – NiMoO₄; – MoO₃

Similar diffraction patterns were obtained on a titanium-containing support for C-3_ald and C-1_ald materials. Intense peaks of SiO₂ and broad peaks of γ-Al₂O₃ partially or completely overlapped the main peaks, corresponding to probable compounds: MoO₃, NiO, CoO, Co₃O₄, NiAl₂O₄, CoAl₂O₄, NiMoO₄ and CoMoO₄. This made interpretation of the diffraction patterns challenging, but could indicate a high dispersion of hydrogenating metal compounds in the catalysts supported on S2_ald with a moderate amount of phosphorus introduced during impregnation [29]. The diffraction patterns of the synthesized nickel molybdenum catalysts differed significantly from the catalyst (1 mmol Ni, 3.1 mmol Mo per 1 g Al₂O₃) obtained by impregnation of alumina with an aqueous solution of both nickel hexahydrate and ammonium molybdate tetrahydrate, where rather narrow peaks of NiMoO₄ and NiAl₂O₄ were present, as well as a weak broad NiO peak [6].

Calculations using the Scherrer equation (peak 2 θ = 28.52°) for an industrial sample of C_ind, not containing SiO₂ and TiO₂, with similar amounts of Co, Mo and P in the composition C-6_ald, revealed similar sizes of CoMoO₄ crystallites – 3.55 nm and 3.05 nm, respectively. The size of NiMoO₄ crystallites (peak 2 θ = 29.2°) in the C-4_akb catalyst was of the same order—3.21 nm. The NiMoO₄ crystallite sizes increased slightly as the phosphorus content increased above 1%, up to 3.33 nm in the C-2_akb catalyst. The sizes of Co(Ni)MoO₄ crystallites coincided with the average particle sizes of about 3.1 nm in calcined catalysts of the composition Co(2.89–3.71)Mo(11.95)P(0.86–4.45) on γ-Al₂O₃, determined by transmission electron microscopy (TEM) [16]. At higher concentrations of nickel, molybdenum and phosphorus in the C-5_akb catalyst, the calculated sizes of NiMoO₄ crystallites already reached 14.5 nm. The appearance of a single peak 2 θ = 23.4° from MoO₃ [9] with particle sizes of 7.17 and 58.6 nm in the catalysts containing more than 1% phosphorus indicated poor dispersion of molybdenum oxide. A slight increase in the background in the angle range 2 θ = 10.0–13.80° was probably due to the presence of Ni polymolybdate, including [Mo₇O₂₄]⁻⁶ ions [7].

3.3 ESDR

The ESDR of the pressed powders of the S1_akb and S2_ald carriers were identical; only very weak absorption bands were observed at 510 and 550 nm from the Fe³⁺-O-Fe²⁺ impurity in the kaolin composition [22]. In ESDR of whole S2_ald granules, obvious absorption in the region of 200–345 nm was attributed to the O^–2 → Ti⁴⁺ charge transfer band characteristic of anatase [14, 27]. In the ESDR of catalysts obtained in the “Auto zero” mode relative to the corresponding supports (Fig. 3), absorption bands were observed indicating different coordination of transition metal ions, described in detail in the literature [7, 14, 22,23,24]. A strong charge transfer band at 200–220 nm is attributed to the superposition of absorption bands from isolated MoO₄²⁻ particles with the tetrahedral coordination Mo⁶⁺_Td and polymolybdates of the Mo_Td⁶⁺ – O – Mo_Oh⁶⁺ type. The shift of the edge of this band to the long-wavelength region, observed at higher content of molybdenum and phosphorus, indicated formation of the polymer chains Mo_Oh⁶⁺ – O – Mo_Oh⁶⁺, as well as the bulk phase MoO₃ in the C-5_akb and C-2_akb catalysts [9]. The spectral manifestation of cobalt ions in octahedral coordination (Co _Oh²⁺) included bands in the region of 469 and 510 nm. Bulk NiMoO₄ with the octahedral coordination – Ni _Oh²⁺ corresponded to a system of bands at 417–435 and 735–813 nm, while weak bands at 714 and 769 nm indicated the presence of surface nickel-molybdate structures, and bands in the region of 322 and 333 nm are attributed to the surface associates of the Al-O-Ni–O-Mo- type. A doublet of bands at 595–602 and 628–633 nm from Ni _Td²⁺ and a triplet at 555, 602 and 666 nm from Co _Td²⁺ indicated partial incorporation of group VIII ions into tetrahedral vacancies of alumina. Absence of non-stoichiometric oxides NiO_1+x and CoO_1+x in the catalyst composition follows from a low absorption level in the region of 330–1000 nm in the ESDR spectra.

Fig. 3

ESDR of catalysts: 1) C-2_akb; 2) C-4_akb; 3) C-5_akb; 4) C-1_ald; 5) C-6_ald; 6) C-3_ald; 7) C_ind. A) -catalysts calcined after extraction with water, b) calcined after extraction with ammonia solution

3.4 Raman Spectra

Raman spectroscopy is widely used to identify various transition metal compounds that are formed during synthesis of hydrotreating catalysts, depending on the characteristics of their preparation [29]. In the Raman spectra of the S1_akb and S2_ald carriers, weak bands from Al₂O₃ appeared at 250 cm⁻¹ (Fig. 4) [31]. The anatase form of TiO₂ in the S2_ald spectrum corresponded to a clear band of moderate intensity at 141.4 cm-1 and weak bands at 397, 520 and 638 cm⁻¹ [26, 27]. In the spectra of catalysts on S2_ald, titanium dioxide appeared only as a band at 141.4–142.0 cm⁻¹, since weak high-frequency bands overlapped with bands from molybdenum compounds, in contrast to the NiMo catalyst on a support containing 95% TiO₂ [14]. The set of bands 966–967, 940, 890, 926, 590, 370 and 230 cm⁻¹ in the Raman spectra of impregnating solutions of (NH₄)₄Mo₇O₂₄^×4H₂O stabilized H₃PO₄ (pH of 1.25–1.54) is better corresponding to the spectrum of the Mo₈O₂₆⁴⁻ [7] and P₂Mo₅O₂₃⁴⁻ anions rather than Mo₇O₂₄⁶⁻ [8, 29].

Fig. 4

Raman spectra: 1) S1_akb; 2) S2_ald; 3) C-2_akb; 4) C-4_akb; 5) C-5_akb; 6) C-1_ald; 7) C-6_ald; 8) C-3_ald; 9) C_ind

The Raman spectra of the catalysts contained the main bands from polymolybdate ions in the high-frequency region 880–1023 cm⁻¹, the intensity ratio and the position of the maxima of which varied depending on the concentration of Mo and P, as well as on the Mo/P ratio (Fig. 4). The MoO₄²⁻ bands in the region of 910–916 cm⁻¹, comparable in intensity to the bands of polymolybdates, were observed for catalysts containing no more than 8.3 wt.%. molybdenum (spectra 6–8 in Fig. 4). The bands at 890–1000 cm⁻¹ and 830–970 cm⁻¹ in the Raman spectra were attributed to symmetric and asymmetric stretching vibrations of the terminal Mo = O bond. The band around 310–370 cm⁻¹ corresponded to the bending vibrations of the Mo = O terminal bond, and the bands around 560 and 210 cm⁻¹ were assigned to the symmetric Mo–O-Mo stretching vibrations and the Mo–O-Mo bending vibrations, respectively. The intensity of the band of polymolybdate anions with a characteristic peak in the region of 570–580 cm⁻¹ [19, 29] was at maximum in the Raman spectrum of the C-5_akb catalyst and decreased as the concentration of molybdenum and phosphorus decreased (Fig. 4).

The complex shape of intense bands with peaks in the region of 912–1010 cm^–1 of catalysts on S2_ald support is due to the superposition of a number of signals from polymolybdates, normal molybdates and molybdenum oxide. Presence of MoO₄²⁻ ions in the Raman spectra in Fig. 4 was evidenced by groups of bands close to those published in the literature for NiMoO₄ species (709 and 964 cm^–1) and Al₂(MoO₄)₃ (~ 993, 1004, 1026 cm^–1 – bending vibrations of the Mo = O bond of three MoO₄ units and 822, 888, 915 cm^–1 – stretching vibrations of the Mo = O bond of three MoO₄ units) [6].

Lower intensity and blurriness of the characteristic band of polymolybdate compounds in the Raman spectra (from vibrations of the terminal groups M = O) (Fig. 4) was observed in the range of 954–965 cm⁻¹. This along with the subtle bands also corresponding to the crystalline phases of NiMoO₄ (915 and 720 cm⁻¹) and MoO₃ (1000–993, 820, 670 cm⁻¹) [7, 29], indicated higher dispersion of transition metals in C-1_ald compared to C-2_akb.

A wide band from 840 to 940 cm⁻¹, attributed in [7] to tetrahedral MoO₄²⁻ ions present in NiMoO₄ and CoMoO₄ phases, was masked by intense bands of polymolybdate ions (950 cm⁻¹) and large MoO₃ crystals (820 cm⁻¹), identified by XRD (Fig. 2).

Low-intensity bands at about 373 cm⁻¹ in the spectra of the C-2_akb, C-4_akb, C-5_akb catalysts and 196–198 cm⁻¹ of the C_ind catalyst are attributed to the low crystalline structures of NiAl₂O₄ and CoAl₂O₄, respectively [31], which were absent in the case of catalysts based on S2_ald. Formation of Co₃O₄ and NiO was not detected. Specific lines of Co₃O₄ (472, 514,610 and 678 cm⁻¹) and NiO (540 cm⁻¹) clearly appeared in the Raman spectra of oxides, obtained by precipitation from the corresponding nitrates with a NaOH solution.

3.5 Selective Extraction of Active Components

Table 2 presents the data for the elements in aqueous and aqueous-ammonia extracts from the catalysts, as well as the carrier. As mentioned in [29], during extraction with water highly dispersed MoO₃ and transition metal molybdates species are removed from hydrotreating catalysts, being just weakly bound to the surface. During the subsequent alkaline hydrolysis, when under the influence of NH₄OH, -Al-O–Mo, -Al-O–Mo-O-Ni- and similar bonds were destroyed, the structures most active in hydrodesulfurization reactions migrated into the aqueous ammonia solution. The maximum amount of structures weakly associated with the S1_akb support was 7.2% for Mo, and 2.56% for Ni of the weight of the initial C-2_akb catalyst. On the ALD modified support, the amount of structures extracted with water from the C-6_ald catalyst reached 6.3% for Mo ca. 2.7% for Co. The minimum amount of water-soluble compounds was recorded for the industrial catalyst: Mo – 3.73%, and Co –1.72%. A large number of structures extracted with an aqueous ammonia solution were present in the C-5_akb catalyst: 4.94% Mo, as well as in industrial C_ind catalyst: 4.12% Mo. The presence of a certain amount of Ti, Si, P and Al in the extracts may indicate their interactions with molybdenum polyanions in the acidic environment of the impregnation solution with the formation of Al₂(MoO₄)₃ and heteropoly compounds. These elements were practically absent in the extracts of the blank experiment with the S2_ald carrier. Raman spectra (Fig. 4) confirmed the presence of Al₂(MoO₄)₃ [6] in the catalysts under study. The spectroscopic data contradict possible preservation of Mo and P complexes during calcination, in particular: P₂Mo₅O₂₃⁵⁻ (956, 926, 870 cm⁻¹); HP₂Mo₅O₂₃⁵⁻ (936, 882 cm⁻¹); H₂P₂Mo₅O₂₃⁴⁻ (944, 894 cm⁻¹); H₂PMo₁₁CoO₄₀⁵⁻ (1008, 971, 954, 886, 816 cm⁻¹), described in [18].

Table 2 Amount of elements extracted from the catalysts

3.6 TPR

Analysis of the profile of TPR curves of oxide forms of catalysts makes it possible to estimate the strength of the metal-support interactions and predict the transformations of specific structures reduced by hydrogen at low temperatures into catalytically active Co(Ni)MoS type II phases during sulfidation prior to hydrotreating [12]. The obtained data coincided with the published data [24, 30]. Under the selected conditions, the TPR curve of Co₃O₄ had a peak at 340 °C with a small shoulder at about 400 °C. The reduction of NiO occurred at 300 °C, while transformation Ni_Td²⁺  → Ni^o in the NiAl₂O₄ was initiated above 700 °C. The S1_akb and S2_ald carriers did not absorb hydrogen within the experimental sensitivity. Sulfidation of catalysts in hydrogen environment, as well as removal of sulfur, nitrogen and oxygen from hydrocarbon feedstocks, is carried out at temperatures rarely exceeding 400 °C [4, 7, 11, 14,15,16,17, 19], thus position of the first peak in the TPR curves can be considered as the most relevant for predicting catalytic properties. Consumption of hydrogen for the reduction of oxides present in the catalysts based on the S2_ald carrier was detected at 250 – 300 °C (Fig. 5, curves 2–4).

Fig. 5

The TPR profiles: 1)C-2_akb; 2) C-6_ald; 3) C-1_ald; 4) C-3_ald; 5) C-4_akb; 6) C-5_akb; 7) C_ind

The maximum hydrogen consumption due to the reduction of octahedrally coordinated ions Ni_Oh²⁺ (Co_Oh²⁺) and Mo_Oh⁶⁺ → Mo⁺⁵ in the structure of the corresponding polymolybdates and dispersed phase MoO₃ was observed in the range of 380–400 °C [7, 24]. The first maximum in the TPR curve of the C-4_akb catalyst of a similar composition on the S1_akb support was observed at 405 °C (Fig. 5, curve 5) and shifted to 430 °C with an increase in the phosphorus content from 0.35 to 1.6% at a similar concentration of active metals. The second maximum in the TPR curves in the region of 480–540 °C was attributed to the reduction of Ni_Oh²⁺(Co_Oh²⁺) in the structure of the corresponding monomolybdates with a simultaneous intensification of the reactions Mo_Oh⁶⁺ → Mo⁺⁵ → Mo⁺⁴. The intensive involvement of Ni_Td²⁺ and Co_Td²⁺ ions in the spinel structures, as well as Mo_Td⁶⁺ in monomolybdates, with a possible transition to the metallic state, into the reduction process was evidenced by high-temperature maxima in the TPR curves in the region of 685–725 °C with a wide shoulder from the continuing hydrogen consumption. The diffraction patterns of the samples, taken after completion of the TPR experiment at 900 °C, contained peaks corresponding to the metal particles Ni^o and Co^o. Peaks from Mo^o were not detected, since complete restoration of molybdate structures required aging the samples in a hydrogen environment at a temperature of at least 915 °C [28].

Calculation of the area under the TPR curves, reflecting the amount of consumed hydrogen in the temperature range 300–500 °C at the first stage of reduction of polymolybdate structures, NiMoO₄ and CoMoO₄, revealed the maximum hydrogen consumption of ca. 2.1 mmol H₂/g for C-4_akb and C-3_ald catalysts (Table 3). Although the multi-peak nature of the TPR curves (Fig. 5) did not allow an accurate determination of hydrogen consumption for reduction of a particular phase, the temperature of the first maximum, along with the total hydrogen consumption in the temperature range typical for the operation of hydrotreating catalysts, made it possible to compare the easiness of hydrogenating elements reduction.

Table 3 Amount of absorbed H₂ in TPR mode and chemisorbed O₂

Considering that sulfidation of oxide forms of hydrotreating catalysts is carried out in hydrogen environment with sulfur-containing compounds at 320–400 °C [12, 14], and hydroprocessing is carried out at 260–400 °C [7, 15, 17], hydrogen consumption for TPR per catalyst weight was calculated for two temperatures of two temperature intervals 250–400 °C and 300–500 °C (Table 3). The amount of oxygen chemisorbed by equal amounts of catalysts pre-treated with hydrogen at 500 °C indirectly indicated dispersion of the reduced metals. C-4_akb catalyst with a NiMoO₄ crystal size of about 3.2 nm, differing from C-2_akb not only in the amount of phosphorus, but also in larger MoO₃ crystals absorbed ca. 1.4 fold more oxygen, which can be attributed to difference in the dispersion of oxides. The negative effect of excess phosphorus was even more pronounced for the C-5_akb catalyst. Reduced catalysts on the S2_ald support with a lower content of hydrogenating elements absorbed more oxygen, which correlated well with the results of estimating the crystal sizes of MoO₃, CoMoO₄ and NiMoO₄ using the Scherrer formula.

3.7 Catalyst Activity

A comparison of the hydrodesulfurization function of catalysts in the hydrotreating of hydrocarbon fractions that differ substantially in the sulfur content and the boiling points (Table 4) revealed high activity of the ALD modified samples. Based on the results of 30 days tests in hydrotreating of deasphalted residue and kerosene, it was shown that stability of the C-3_ald catalyst at the same level as unmodified C-4_akb. Co-Mo and Co–Ni-Mo catalysts with molybdate crystal sizes smaller than 4 nm, containing less than 0.9 wt.% phosphorus and having Brønsted acid sites of medium strength on the surface, removed sulfur with a very similar efficiency from naphtha (> 99.8%), kerosene (> 94%) and deasphalted residue (66.67–72.7%). The C-4_akb catalyst, en par in terms of desulfurization activity with the industrial catalyst, had a better dearomatization performance. The C-4_akb catalyst combined with the demetallization catalyst ZSh-25 (Uzbekistan) and GO-70 (Russia), was in operation at the Fergana Oil Refinery for 6 years providing a required decrease in the concentration of polyaromatic hydrocarbons in the production of base oils from deasphalted vacuum distillation residue.

Table 4 Activity of catalysts during hydrotreating of various raw materials

4 Conclusions

The influence of ALD modification of an aluminum kaolin borate carrier obtained from industrial waste on the physicochemical characteristics of hydrodesulfurization catalysts was studied using XRD, ESDR, Raman spectroscopy as well as TPR. Conditions for the preparation of C-4_akb catalysts, which combine a high hydrodesulfurization function with activity in the hydrogenation of polyaromatic hydrocarbons, have been optimized. The results of determining the activity at the pilot plant revealed certain advantages of the molecular layering of titanium dioxide on the aluminum kaolin borate carrier S1_akb in the processes of hydrodesulfurization of hydrocarbon raw materials, especially mercaptan sulfur—an important parameter in the production of aviation fuel.