High loading of short WS₂ slabs inside SBA-15: promotion with nickel and performance in hydrodesulfurization and hydrogenation

Abstract

Layered nanoslabs of a WS₂ phase with a well-defined hexagonal crystalline structure, average slab length of 3.6 nm, and stacking number of 3.2 were inserted into the nanotubular channels of SBA-15, an ordered pure silica material (surface area of 800 m²/g, uniform mesopore diameter of 6.5 nm) at loadings up to 60 wt.%. Sonication of a slurry containing SBA-15 in a W(CO)₆–sulfur–diphenylmethane solution yielded an amorphous WS₂ phase inside the mesopores. By sulfidation with 1.5% dimethyldisulfide in toluene under a hydrogen flow at 593 K and 5.4 MPa, the amorphous phase was transformed into hexagonal crystalline WS₂ nanoslabs (as shown by XRD, HRTEM, and selected area electron diffraction (SAED)). The WS₂ nanoslabs were distributed exclusively inside the mesopores in a uniform manner (HRTEM, quantitative microanalysis), without blocking the pores (N₂-sorption), and were oriented with their edge planes toward the support surface. This study constitutes the first report of such a combination of high loading of a well-defined crystalline catalytic phase into the nanotubular channels of mesoporous silica without blocking them. The first well-resolved HRTEM images of the well-defined crystalline catalytic phase (WS₂) inside the SBA-15 nanotubes are presented. A Ni component was introduced into the WS₂/SBA-15 composite by impregnation from an aqueous solution of nickel acetate. It increased the catalytic activity up to a Ni/W ratio of 0.4. In the hydrodesulfurization (HDS) of dibenzothiophene and the hydrogenation (HYD) of toluene, the activity of the optimized NiWS/SBA-15 catalyst was 1.4 and 7.3 times higher, respectively, than that of a sulfided commercial CoMo/Al₂O₃. This finding illustrates the excellent potential of high loading NiWS/SBA-15 catalysts for deep hydrotreatment of petroleum feedstocks.

Introduction

The potential applications of ordered mesoporous silicas with pore walls of uniform width in the range 1–5 nm and controlled uniform pore diameters in the range 2.5–25 nm [1], [2], [3] as periodic hosts for the preparation of mesoporous catalysts with chemically functionalized surfaces have been widely investigated in the past decade [4], [5], [6]. Much less attention has been paid to the preparation of catalytic mesoporous silicas with inclusions of nanocrystals of well-defined catalytic phases 7. This neglect is indeed surprising in the light of the numerous examples in catalytic practice in which efficient active sites created at the surfaces of such phases (due to optimized geometrical arrangement of the atoms) do not have molecular analogs with similar performance. The best known of these phases are TiO₂ anatase as a photocatalyst and a basis for vanadium oxide catalysts in selective oxidations [8], [9], tetragonal ZrO₂ as a basis for superacidic sulfated zirconia [10], [11], [12], layered Mo(W)S₂ slabs with Co(Ni)-ions decorating their edge planes in hydrotreating catalysis [13], [14], [15], acidic zeolite catalysts 16, and heteropoly compounds in acidic and redox catalysis 17. The potential advantage of mesoporous silica hosts for the preparation of catalytic phase materials lies in the high dispersion of the catalytic phase at high loadings that, depending on the nature of the host, may surpass 30–60 wt.%—a combination that cannot be achieved with conventional supports having textural porosity, such as silica gel or precipitated aluminas.

To take full advantage of ordered mesoporous silicas for the preparation of included catalytic phase dispersions, three main conditions must be fulfilled:

• the entire catalytic phase must be located inside the pore system of the mesoporous host;

• the nanoparticles must have a well-defined optimized crystal structure; and

• there must be minimal blocking of the host's pore system by the catalytic phase dispersions at high loading.

The few attempts that have been made up to the present to prepare high-loading (>30 wt.%) catalytic phase dispersions inside mesoporous silica hosts [18], [19], [20], [21], [22], [23] have shown that simultaneous accomplishment of these three goals is a very complicated problem.

The multistep impregnation of the hexagonal 1D mesoporous silica SBA-15 with an aqueous YEu-nitrate solution (Y/Eu=32.3) yielded a material in which most of the Y₂O₃ phase, at loadings up to 22 wt.%, was spread as an amorphous monolayer coating the internal pore surfaces [7], [18]. In a high-loading material (35 wt.% Y₂O₃), Y₂O₃ was present in three different forms: in addition to the oxide monolayer, oxide nanoparticles were detected inside the pores (HRTEM) and large cubic Y₂O₃ crystals (18 nm (XRD)) outside the silica particles 18. The blocking extent (BE) of the silica mesopores for this high-loading material was relatively high (i.e., BE≈0.65). BE was calculated by us as BE=1−NSA, where NSA, the normalized surface area, was defined as $\text{NSA}\text{=}\text{SA}{}_{\mathrm{<mtext>composite</mtext>}}\text{/}\text{SA}{}_{\mathrm{<mtext>pure</mtext><mspace\; sp="0.3"\; width="2px"\; linebreak="nobreak"\; is="true"></mspace><mtext>silica</mtext>}}\text{(1−y)}$, in which y is the weight fraction of the guest component in the composite material 23. The multistep impregnation of cubic silica MCM-48 with aqueous Fe-nitrate facilitated the synthesis of a high-loading composite material (42.5 wt.% Fe₂O₃) in which the entire host phase was located inside the silica pore system, as shown by HRTEM 19. The pore BE (0.35) was significant, but relatively low due to the 2D pore system, and the host phase existed in the form of disordered iron oxide nanoparticles with less strong linking of the FeO₆ octahedra relative to the well-defined hematite phase (EXAFS). One-step impregnation of MCM-41 with an aqueous solution of the heteropolyacid H₃PW₁₂O₄₀ yielded a composite with 33 wt.% loading that contained both grafted heteropolyacid anions (FTIR) and crystals of the heteropolyacid phase (XRD) with a high pore BE≈0.7 20. The catalytic activity of this composite in the cracking of 1,3,5-triisopropylbenzene increased with increasing heteropolyacid loading and passed through a sharp maximum at 23 wt.%. The multistep grafting with aluminum butoxide of MCM-41, in which the pores had been expanded by addition of mesitylene to the surfactant, facilitated the synthesis of a composite with 38 wt.% alumina loading and no pore blocking 21. The catalytic activity of the host amorphous alumina phase in the alkylation of phenol with methanol was about five times that of the reference material—bulk alumina 21—due to a substantially higher population of acidic pentahedral aluminum atoms, as shown by ²⁷Al FAM(II)-MQMAS NMR 22. Ultrasonication of a slurry of wide-pore (expanded) MCM-41 in a Mo(CO)₆–decaline solution yielded a composite with 45 wt.% MoO₃ loading, in which the guest phase was spread as an amorphous monolayer coating the internal pore surfaces (HRTEM, XPS, MAS NMR) with a minimal pore BE of 0.07 23.

It is possible to control the distribution mode of the catalytic phase (spreading or nanoparticle formation) at high loadings by varying the nature of the guest precursor and the method of insertion. However, to date, the fulfillment of all three conditions delineated above in an optimized host/guest catalytic composite has not yet been achieved with any system. The main problem is to combine the formation of a well-defined nanocrystalline catalytic phase inside the mesopores with uniform distribution of the nanocrystals in the pore volumes and high accessibility of the nanocrystals to the reacting molecules (i.e., low BE).

The purpose of this study was to address the above-described problem by investigating, as a model system, the insertion of a layered WS₂ phase into SBA-15 mesopores. The new catalyst was designed on the basis of available knowledge about the sulfide catalyst Mo(W)S₂. It is known that the active sites of the sulfide catalyst are located in the edge planes of Mo(W)S₂ slabs [13], [14], [15]. It is generally agreed that the shorter the slabs, the higher the fraction of edge planes and hence the higher the catalyst activity. Recently, it was found that the stacking degree of supported Mo(W)S₂ slabs plays an important role in the catalytic performance of sulfide catalysts; i.e., it determines the activity of the catalyst in the hydrogenation (HYD) of aromatics [24], [25]. The orientation of the Mo(W)S₂ slabs relative to the support surface also seems to have an influence on the catalytic activity [14], [26]. The underlying premise of the current study was therefore that the highly ordered porosity of the SBA-15 support would provide a way of controlling the structure (length and stacking degree) and orientation of the metal sulfide nanocrystals at high loading.

It has previously been shown that ultrasonication of a solution of Mo(CO)₆ in decalin in the presence of dissolved oxygen yielded a Mo phase spread as an amorphous monolayer coating the internal pore surfaces of MCM-41 23. The amorphous Mo monolayer was formed by interaction between coordinatively unsaturated Mo species (produced by ultrasonically induced decarbonylation) and surface silanols and siloxane bridges [23], [27]. In our catalyst system, the amorphous WS₂ phase was prepared directly by ultrasound irradiation of a solution of W(CO)₆ in diphenylmethane in the presence of dissolved elemental sulfur under argon at 90 °C 28. To avoid spreading of the WS₂ phase, an excess of sulfur was used during ultrasonic deposition of WS₂ on the SBA-15 support. Further thermal treatment of this amorphous WS₂ phase in a hydrogen sulfide/hydrogen atmosphere converted it to layered WS₂ nanocrystals. The location, structure, and orientation of the WS₂ slabs were studied by HRTEM, XRD, SAED, N₂-sorption, and qualitative microanalysis (EDS). The promotion of these slabs with Ni and the catalytic performance of the final product in the HYD of toluene and the HDS of dibenzothiophene (DBT) were also investigated.