Fischer–Tropsch synthesis of liquid hydrocarbons over mesoporous SBA-15 supported cobalt catalysts

Abstract

The influence of cobalt loading (10–30 wt% Co) and pore size of SBA-15 support on the physico-chemical and catalytic performance of mesoporous Co/SBA-15 catalysts for the Fischer–Tropsch synthesis (FTS) reaction (T = 245 °C, P = 290 psig, H₂/CO = 2, and GHSV = 2000 h⁻¹) has been investigated. Catalysts were characterized by N₂ adsorption–desorption, X-ray diffraction (XRD), electron microscopy, and temperature-programmed reduction (TPR). The dispersion of Co/SBA-15 decreased and the extent of cobalt reduction increased with increasing either the cobalt loading or pore size of SBA-15. A maximum CO conversion was found for the sample with 20 wt% Co loading. More methane and less C₅₊ hydrocarbons were produced over less reducible 10 wt% Co loaded sample. The 20Co/SBA-15 catalysts with larger pores led to larger cobalt crystallite size, lower dispersion and higher reducibility. CO conversion increased with the increase of pore size in the range studied. The 20Co/SBA-15 catalysts with larger cobalt crystallite size showed higher C₅₊ selectivity for the FTS. Finally, at comparable Co loading, CO conversion of Co/SBA-15 catalysts were to be about 2 times greater than a Co/SiO₂ sample, with only a minor difference in product selectivity.

---

1. Introduction

Fischer–Tropsch synthesis (FTS) is a key industrial process to catalytically convert syngas (a mixture of CO and H₂) to liquid hydrocarbons via a surface polymerization reaction.^1–5 In comparison to traditional petroleum-derived liquid hydrocarbons, the FTS liquid hydrocarbons are free of sulfur and aromatics pollutants,^6–8 which make them ideal candidates for the synthesis of clean transportation fuels. Syngas can be produced from various carbonaceous sources,^9–12 such as natural gas, coal, biomass, shale gas, or municipal solid waste, through steam reforming, partial or auto-thermal oxidation, or gasification processes. Therefore, the FTS process is of prominent interest to both academia and industry.

The Fischer–Tropsch (FT) reaction was shown to be catalyzed by certain transition metals including Ru, Fe, and Co.¹ Ru-based catalysts are highly active, but the Ru resource is scarce and expensive and, thus, is not commercially used.⁷ Co-based FTS catalysts are usually preferred because these catalysts are more active per weight of metal, more stable toward deactivation by water (a by-product of the FTS reaction), have lower activity for the competing water–gas-shift (WGS) reaction, and produce a higher fraction of linear long-chain paraffins and less oxygenates than Fe-based catalysts.^13–18

Supported cobalt catalysts are well-known for their activity and selectivity in the reaction of FTS.^19,20 In order to achieve a high density of surface-active sites (Co⁰), cobalt precursors are usually dispersed on porous carriers, such as SiO₂, Al₂O₃, SiC, TiO₂ and so on.^13–20 Generally, the number of Co⁰ active sites available for FTS depends on Co loading amount in the catalyst, the sizes of the Co particles (or dispersion), and their reducibility.^21,22 High activity of Co catalyst can be obtained as cobalt oxide is well dispersed on the support and highly reduced at desirable temperatures. FTS turnover frequencies (TOFs) were found to be independent of the cobalt dispersion and support identity over the accessible dispersion range (0.01–0.12) at typical FTS conditions.²³ Iglesia et al.²¹ showed that at conditions favoring chain growth, i.e., high C₅₊ selectivity, the FTS rates per total Co atoms increase linearly with increasing metal dispersion regardless of the nature of the support used. In general, high productivity of hydrocarbons for supported cobalt-based FTS catalysts requires small cobalt crystallites at high cobalt surface densities.²⁴ However, the smaller the cobalt particle size, the stronger the interaction between cobalt and support, which decreases both the reducibility and activity for FTS.²⁴ Ideally, highly active Co-based catalysts should be prepared by achieving high dispersions of highly reducible Co species at cobalt loadings as high as possible.¹³ In particular, the cobalt loading is very high at 33 g per 100 g support for the current commercial cobalt-based catalyst chosen for slurry bubble column reactor.²⁵ Furthermore, high cobalt loading is necessary to stabilize the catalyst against irreversible deactivation due to re-oxidation of Co⁰ to cobalt oxide occurred during FTS.²⁶ Therefore, the support with high accessible surface area is necessary to disperse such high cobalt content.

SBA-15,^27–29 a silica-based periodic mesoporous material, is one of the most attractive catalyst supports, with high hydrothermal stability and a large surface area of 400–1000 m² g⁻¹, which allows for a better dispersion of a large number of catalytically active species as compared with conventional amorphous silicas.^30–33 It also has a hexagonal array of a uniform tubular channel, with a pore diameter ranging from 5 to 30 nm.²⁷ The ordered mesoporous silicas possessing a uniform pore size distribution should allow for a better control on the cobalt particle size, and thus, on the catalytic performance. In addition, the reducibility is favored for the SiO₂ supported Co catalyst because the strength of interaction between the cobalt and support is lower than the other commonly used support such as Al₂O₃, TiO₂ and so on.^13,34 Thus, SBA-15 is suitable for use as a potential support for the synthesis of a commercial FTS cobalt catalyst with high activity.

The increases of Co dispersion always corresponds to the decrease of reducibility for the supported cobalt due to the larger interaction surface area of Co and support.^14,35 Thus, the efficient control of cobalt dispersion and reducibility in the preparation of SBA-15-supported catalysts is of great importance to improve the FTS catalytic performance, such as activities and selectivities. Although a few studies were conducted on the effects the cobalt particle size, cobalt loading and the pore size of support on the FTS activity and selectivity, controversy persists, because these observations result from complex interplay among many factors.^3,8,20 These factors involved not only the re-adsorption probability of liquid hydrocarbon in the confined space and the diffusion situation, but also the effect of changes in the site density, such as the changes in loading, reducibility and the particle size of active metal.^3,8 Hence, it is difficult to evidence the influence of the support porosity on activity and selectivity, since the metal dispersion or the metal particle size also depends on the pore size distribution. Therefore, more insights are expected to prepare catalysts with a fixed pore size with various metal loading, or a fixed metal loading with different pore sizes.

This work aims to address the sole effect of cobalt loading or pore size of SBA-15 on the cobalt dispersion and its reducibility, as well as on the formation of liquid hydrocarbon fuels in FTS. Both periodic SBA-15 and commercial amorphous silicas (as the control) are used as catalytic support. The catalysts are characterized by N₂ adsorption, X-ray diffraction (XRD), transmission electron microscopy (TEM), temperature-programed reduction (TPR).

2. Experimental

2.1 Catalyst preparation

The SBA-15 molecular sieves were synthesized using Pluronic triblock copolymer P123 (EO₂₀–PO₇₀–EO₂₀, M_AV = 5800, Aldrich) as the structure-directing agent and tetraethyl orthosilicate (TEOS) as silica source.²⁷ First, the triblock copolymer was dissolved in a solution of water and HCl under stirring, and then the required amount of TEOS was added to the above solution at 36 °C and kept under stirring for 20 h. Then, the gel mixture was transferred into polypropylene bottles and heated at designed aging temperature for certain time in static. The synthesis of different pores of SBA-15 depends on the aging temperature (100 or 120 °C) and aging time (24 or 48 h). After the synthesis, the solid obtained was filtered, exhaustively washed with distilled water until neutral pH, dried at 80 °C and finally calcined in a flow of air at 500 °C for 6 h to remove the organic template.

Co catalysts supported on SBA-15 were prepared by wetness impregnation method using a solution of cobalt(II)nitrate dissolved in ethanol in excess with respect to the pore volume of the SBA-15 support (liquid/solid ratio of 4 cm³ g⁻¹), followed by slow evaporation of the solvent in a rotary evaporator at 323 K and vacuum until dryness. The Co/SBA-15 catalysts samples were dried in an oven in 100 °C overnight then calcined in the flow of air at 450 °C for 6 h. The catalysts with different cobalt loading were labeled as xCo/support, and x referred to the weight percentage of Co loading. The 20Co/SBA-15 catalysts with different pore sizes were labeled as 20Co/SBA-15(P_i) with i standing for 1, 2, 3, and 4. The Co/SiO₂ catalyst was prepared by the above-mentioned wetness impregnation method.

2.2 Catalyst characterization

2.2.1 BET surface area and pore size distribution. The textural properties including, specific Brunauer–Emmett–Teller (BET) surface area, pore volume and pore diameter were measured using nitrogen absorption–desorption isotherms at −196 °C over an Autosorb-1 system (Quantachrome Instruments, Boynton Beach, FL, USA). Prior to the experiments, the samples were outgassed at 300 °C for 4 h. The BET surface areas were obtained for adsorption data in a relative pressure range from 0.05 to 0.30. The total pore volumes (TPV) were calculated from the amount of N₂ vapor adsorbed at a relative pressure close to unity (0.99), assuming that the pores are filled with the condensate in the liquid state. The pore size distribution curves were evaluated from the desorption branches of the isotherms using the Barrett–Joyner–Halenda (BJH) formula.

2.2.2 Hydrogen temperature-programed reduction (H₂-TPR). The reduction behavior of the supported oxidized cobalt phases was studied by TPR in a ChemBET Pulsar TPR/TPD equipment (Quantachrome Instruments, Boynton Beach, FL, USA). The catalyst samples were first purged in a flow of nitrogen at 300 °C for 60 min to remove traces of moisture, and then cooled to 40 °C. The TPR of 200 mg of each sample was performed using 4% hydrogen in argon gas mixture with a gas flow rate of 70 cm³ min⁻¹. The samples were heated from 40 to 900 °C with a heating rate of 10 °C min⁻¹. The H₂ consumption rate was monitored in a thermal conductivity detector (TCD) calibrated previously using the reduction of CuO as standard.

2.2.3 X-ray diffraction (XRD). The crystalline structure of SBA-15 support and Co/SBA-15 catalysts were characterized by small-angle X-ray diffraction (Rigaku SmartLab XRD, Rigaku Americas, The Woodlands, TX, USA), using Cu Kα (λ = 0.15418 nm) radiation, operated at 40 kV and 44 mA. Diffraction intensities were recorded from 0.4 to 4° at the rate of 0.12° min⁻¹ with a sampling width of 0.02°. The phase composition and crystalline structure of supported cobalt catalysts were characterized by high angle X-ray diffraction (Utima III XRD, Rigaku Americas, The Woodlands, TX, USA) using Cu Kα (λ = 0.15418 nm) radiation, operated at 40 kV and 44 mA. Diffraction intensities were recorded from 10 to 90° at the rate of 0.5° min⁻¹ with a sampling width of 0.02°. The observed diffraction peaks were assigned by reference to JCPDS data.

The average crystallite sizes of Co₃O₄ in the different catalysts were estimated from the Scherrer formula of eqn (1)³⁶ using the most intense reflection at 2θ = 36.9°, where d_(Co3O4) is the average crystallite size of Co₃O₄, λ = 0.15418 nm is the wavelength of the target Cu Kα, θ_B is the Bragg angle. B is the line broadening by reference to a standard so that B² = B_M² − B_S², where B_M is the full width at half maximum (FWHM) of the diffracted plane of the most intense reflection at 2θ = 36.9°, and B_S is the full width at half maximum (FWHM: 0.1° 2θ) of the standard material in radians.

The average Co₃O₄ crystallite size d_(Co3O4) in the calcined samples were then converted to the corresponding mean cobalt metal diameters in reduced catalysts by considering the relative molar volumes of Co⁰ and Co₃O₄ according to eqn (2).¹³

d_(Co⁰) = 0.75 × d_(Co3O4) (2)

Then, the Co⁰ metal dispersions D_(Co⁰) can be calculated from the average Co⁰ crystallite sizes d_(Co⁰) according to eqn (3),^37,38 which assumes a spherical geometry of the metal particles with uniform site density of 14.6 atoms per nm².

D_(Co⁰) = 96/d_(Co⁰) (3)

2.2.4 Electron microscopy. The support and catalyst morphologies were observed using a JEOL 6500 F field emission scanning electron microscope (FE-SEM) (JEOL USA Inc., Peabody, MA, USA), operating at the accelerating voltage of 5 kV. Before the analysis, the surfaces were coated by a sputter with a thin layer (5 nm) of Pt to avoid the charging effects. Transmission electron microscope (TEM) was used to further characterize the morphology of the catalyst and to study the location of the active phase. The samples were dispersed in ethanol and sonicated for 20 min and then deposited over a Formvar copper grid to be observed in the microscope. The analyses were carried out using a JEOL 2100 electron microscope (JEOL USA Inc., Peabody, MA, USA) operating at 80 kV or 200 kV.

2.3 Catalytic tests

The Fischer–Tropsch synthesis was performed in a down-flow half-inch fixed-bed stainless steel tubular reactor. Typically, 0.5 g of catalyst diluted with quartz sand was loaded in the catalyst bed. The catalysts were reduced in situ at atmospheric pressure by passing a flow of 50% H₂/50% N₂ through the reactor. During reducing process, the temperature of reactor was increased to 400 °C at a heating rate of 1 °C min⁻¹ and maintained at this temperature for 6 h. After the reduction step, the temperature was lowered to 120 °C under the flow of 50% H₂/50% N₂ and then the reactor pressure was slowly increased up to 290 psig by introducing the reactant syngas mixture (64% H₂ : 32% CO : balanced N₂, N₂ used as internal standard). Then, the temperature in the catalyst bed was increased from 120 to 245 °C at a controlled heating rate of 1 °C min⁻¹ in order to avoid instability of the system induced by the highly exothermic FTS reaction. Once the reaction temperature of 245 °C was achieved, the reaction was led to proceed during a period of 10–15 h to ensure steady state of the catalyst activity. The time on stream after achieving steady state is 120 h.

CO conversion was calculated according to eqn (4), where F_in and F_out (mol s⁻¹) are the total molar flow rates of the reactor inlet syngas and outlet gas, respectively, where X_(i),in and X_(i),out are the molar fraction of component i in the reactor inlet syngas and outlet gas.

The product selectivity was calculated according to eqn (5), where n_j represents the number of carbon atoms contained in product j, and X_j represents the molar fraction of product j. The C₅₊ selectivity was calculated by subtracting the amount of C₁–C₄ hydrocarbons and CO₂ in the product gas mixture from the total mass balance.

The consumption rate of CO, rCO (mol s−1 molCo0−1) was calculated according to eqn (6), where nCo0 (mol) is the mole of Co0 of the reduced catalyst in this reaction.

The formation rate of CO2, rCO2 (mol s−1 molCo0−1) was calculated according to eqn (7), where XCO2,out is the mole fraction of CO2 in the reactor outlet gas, and nCo0 (mol) is the mole of Co0 of the reduced catalyst in this reaction.

The reaction rate of Fischer–Tropsch synthesis (FTS), rFTS (mol s−1 molCo0−1) was calculated according to eqn (8).

2.4 Product analysis

The outlet gaseous products were composed of C₁–C₄ ranged hydrocarbons, H₂, CO, CO₂, and N₂ were analyzed by using an online Agilent 7890 GC instrument, equipped with one flame ionization detector (FID) and two thermal conductivity detectors (TCDs). C₁–C₄ ranged hydrocarbons were analyzed by using a HP-Plot Al₂O₃ capillary column (50 m × 530 μm ID) with a FID and N₂ carrier. CO, CO₂, and N₂ were analyzed by using molecular sieve-packed column with a TCD and He carrier. H₂ was analyzed by using molecular sieve-packed column with a TCD and N₂ carrier.

During the reaction the reactor outlet gas passed sequentially through a hot trap (T = 120 °C, P = 290 psig) to collect waxes and water, and a cold trap (T = −5 °C) to collect residual liquid oil and water. A small fraction of the wax product collected from hot trap at 120 °C was first dissolved in CS₂ (0.7 wt% of wax) and then analyzed off-line using a gas chromatograph coupled to a mass spectrometer (GC-MS) equipped with an Agilent 6890 series gas chromatograph system, an Agilent 7863B series injector, a 5973 mass selective detector, and a flame ionization detector (FID). An OV-101 capillary column (FID, N₂ carrier) was used with the temperature programmed (2 °C min⁻¹) from 50 to 300 °C.

The liquid oil product collected from cold trap at −5 °C were analyzed off-line using a gas chromatograph coupled to a mass spectrometer (GC-MS) equipped with an Agilent 6890 series gas chromatograph system, an Agilent 7863B series injector, a 5973 mass selective detector, and a flame ionization detector (FID). An Agilent HP-5 capillary column was used for analyzing hydrocarbons, and an Agilent DB-WAXetr (50 m × 0.32 mm ID, 1.0 μm) capillary column was used for analyzing aqueous phase product that including oxygenated compounds and water, while water was quantified according to an external standard method. Helium was the carrier gas with a flow rate of 1 mL min⁻¹. The temperatures of the injector and detector were both kept at 250 °C. The oven was maintained at 40 °C for 5 min, ramped to 250 °C by 10 °C min⁻¹, and then held at this temperature for 10 min. For the mass spectrometer (MS), temperatures of the transfer line and electron impact (EI) source were 250 and 200 °C, respectively.

The hydrocarbon chain length distribution is given by the Anderson–Schulz–Flory (ASF) distribution.⁸ The chain growth probability, α, can be calculated according to eqn (9):⁸

where W_n is the mass fraction of the species with carbon number n. From the slope of the plot of against n, the value of α can be obtained.

3. Results and discussion

3.1 Support and catalyst characterization

3.1.1 Nitrogen adsorption–desorption. Upon N₂ adsorption, SBA-15 and Co/SBA-15 produce irreversible type IV isotherms with a H1 hysteresis loop (Fig. 1a) that is typical of mesoporous materials with 1D cylindrical channels.³⁵ The N₂ adsorption–desorption isotherm for calcined siliceous SBA-15 exhibited a sharp inflection at a relative pressure in the range of 0.7–0.8 (Fig. 1a), which demonstrated a good quality of SBA-15 material with uniform mesopores.^39,40 The shape of the N₂ adsorption–desorption isotherms of Co-supported SBA-15 catalysts containing ca. 10, 20, and 30 wt% Co as shown in Fig. 1a, was similar to that of the SBA-15 support, which indicated that the mesoporous structure of SBA-15 was largely maintained upon cobalt impregnation.¹³

Fig. 1 Nitrogen adsorption–desorption isotherms obtained at −196 °C for (a) SBA-15, 10Co/SBA-15, 20Co/SBA-15, and 30Co/SBA-15, (b) 20Co/SiO₂ and 20Co/SBA-15, (c) 20Co/SBA-15(P1), 20Co/SBA-15(P2), 20Co/SBA-15(P3), and 20Co/SBA-15(P4), and (d) the corresponding pore size distribution curves calculated from N₂ desorption–desorption isotherms of (c).

The inflection of the adsorption branch of the isotherm occurred at a lower relative pressure (0.65–0.75) for Co-containing samples as compared to the pure mesoporous silica SBA-15, suggesting a decrease of the average pore diameter after cobalt impregnation.¹³ The average pore diameter of the pure silica SBA-15 obtained from N₂ adsorption at −196 °C using the BJH method was 8.0 nm, while it decreased to about 7.1 nm for the 30 wt% Co/SBA-15 catalyst sample (see Table 1). The N₂ adsorption–desorption isotherm for 20Co/SiO₂ catalyst is shown in Fig. 1b. The N₂ adsorption–desorption isotherms and the corresponding pore size distributions calculated from N₂ desorption isotherms for 20Co/SBA-15 catalysts with different pore sizes (4.9, 6.5, 7.8, and 9.7 nm) are presented in Fig. 1c and d, respectively.

Table 1 Properties of the supports and catalysts determined by N₂ adsorption, XRD and TPR characterization

Sample	BET surface area (m^2 g^−1)	Pore size (nm)	Pore volume (cm^3 g^−1)	d(Co3o4)^a (nm)	d(Co^0)^b (nm)	D(Co^0)^c (%)	Reducibility^d (%)
a Average crystalline size of Co3O4 calculated from the Scherrer equation: .b Average cobalt crystallite size estimated from the corresponding d(Co3O4) by applying the molar volume correction: d(Co^0) = 0.75 × d(Co3O4).c Dispersion of Co^0: D(Co^0) = 96/d(Co^0).d Estimated from H2 consumption of TPR profiles below 400 °C.
SiO2	154	23.3	1.00	—	—	—	—
20Co/SiO2	114	15.4	0.88	27.1	20.3	4.7	90
SBA-15	460	8.0	1.85	—	—	—	—
10Co/SBA-15	407	7.8	1.62	13.3	10.0	9.6	48
20Co/SBA-15	384	7.2	1.72	14.9	11.2	8.6	63
30Co/SBA-15	359	7.1	1.36	20.6	15.5	6.2	86
20Co/SBA-15(P1)	457	4.9	0.72	10.5	7.9	12.1	37
20Co/SBA-15(P2)	455	6.5	0.80	13.8	10.4	9.2	52
20Co/SBA-15(P3)	396	7.8	0.85	15.0	11.3	8.5	65
20Co/SBA-15(P4)	354	9.7	0.90	16.0	12.0	8.0	74

---

The chemical composition and textural properties obtained by N₂ adsorption–desorption of the pure silica SBA-15 sample and supported cobalt catalysts samples are given in Table 1. The BET surface area and total pore volume of the siliceous calcined SBA-15 are 460 m² g⁻¹ and 1.85 cm³ g⁻¹, respectively, which are typical for SBA-15 synthesized under similar conditions.^13,27,28 Both the BET surface area and the total pore volume significantly decreased upon Co being impregnated, with greater decrease at higher Co loading (Table 1). This may result from a partial blockage of the SBA-15 pores by cobalt oxide clusters and/or a partial collapse of the mesoporous structure.¹³ As shown in Table 1, SBA-15 gives the BET surface area of 460 m² g⁻¹, which is much larger than that of commercial silica SiO₂ (154 m² g⁻¹). The pore size of SBA-15 (8.0 nm) is much smaller than that of SiO₂ (23.3 nm).

3.1.2 Electron microscopy. Scanning electron microscopy (SEM) images (Fig. 2) revealed that the as-synthesized SBA-15 sample consists of many rope-like domains (Fig. 2b, low magnification) with relatively uniform sizes of ∼1 μm, which aggregated into wheat-like macrostructures (Fig. 2a, high magnification), which agreed with the observation reported by Zhao et al.^27,28 Wheat-like mesoporous SBA-15 can have a long aspect of as much as several hundred micrometers, with a relatively uniform diameter of ∼10 μm that is made up of a bundle of wires (diameter ∼ 300 nm). TEM images showed the highly ordered hexagonal arrangement of the channels along two directions, parallel (Fig. 3c) and perpendicular (Fig. 3d).

Fig. 2 SEM images of as-synthesized hexagonal mesoporous silica SBA-15 at (a) low magnification and (b) high magnification, and TEM images of SBA-15 by the direction of (c) parallel and (d) perpendicular.

Fig. 3 HRTEM images of calcined 20Co/SBA-15 catalysts (a) 20Co/SBA-15(P1), (b) 20Co/SBA-15(P2), (c) 20Co/SBA-15(P3), and (d) 20Co/SBA-15(P1). The insets are the cobalt particle size distributions.

HRTEM images of calcined 20Co/SBA-15 catalysts with different pore sizes (4.9, 6.5, 7.8, and 9.7 nm) are shown in Fig. 3. The highly ordered hexagonal arrangement of the channels along the direction parallel to the c axis. The mesoporous structure of SBA-15 was clearly retained after cobalt impregnation and calcination. The average particle sizes of 20Co/SBA-15 with different pore sizes are 13.8, 8.9, 19.1 and 16.1 nm, respectively. The insets shown in Fig. 3 are the corresponding cobalt particle size distributions. The observed TEM cobalt particles were larger than pore size of SBA-15 support, because the cobalt particles were distributed both inside (small grey particle < pore size) and outside (relative big dark particle > pore size) the pore channels of SBA-15. The discrepancies found between the Co₃O₄ crystallite sizes obtained from XRD (Table 1) and Co₃O₄ particle sizes observed from TEM may be due to the partial agglomeration of cobalt particles during TEM sample preparation.

3.1.3 X-ray diffraction (XRD). XRD patterns of 20Co/SBA-15 and 20Co/SiO₂ are presented in Fig. 4a. Co₃O₄ is the only detectable crystalline cobalt species present after catalyst calcination. Diffraction peaks at 2θ of 19.0, 31.3, 36.9, 44.8, 59.4 and 65.2 indicate that after calcination, cobalt was present in the form of crystalline Co₃O₄ spinel on all of the catalysts,¹³ which agreed with JCPDS no. 42-1467. The only difference between the XRD patterns of the calcined samples was the width of the Co₃O₄ peaks. For all catalysts, the Co₃O₄ crystallite size (Table 1) was calculated from the (311) diffraction at 2θ = 36.9° according to the Scherrer equation.³⁶ The average Co₃O₄ crystallite size was calculated to be 27.1 nm for 20Co/SiO₂, while it was calculated to be 14.9 nm for 20Co/SBA-15. The smaller crystallite size indicated better cobalt dispersion for 20Co/SBA-15 than 20Co/SiO₂.^41–44 HRTEM image (Fig. 4b) further confirms that cobalt was present in the form of Co₃O₄ crystalline phase after calcination. The interplanar spacing d was measured to be 0.287 and 0.208 nm, which was characteristic of Co₃O₄ (220) and (400), respectively.

Fig. 4 (a) XRD patterns of calcined 20Co/SiO₂ and 20Co/SBA-15 catalysts, and (b) HRTEM images of 20Co/SBA-15 catalyst.

The small-angle XRD pattern (Fig. 5a) for as synthesized (after calcination in air at 500 °C for 6 h) mesoporous silica (SBA-15) mainly showed one peak at 2θ = 0.7° that was indexable as (100) reflection associated with p6mm hexagonal symmetry.^27,28 After loading with cobalt catalyst (Fig. 5a), the mesoporous structures of SBA-15 were well preserved, confirming the hexagonal SBA-15 is thermally stable.

Fig. 5 (a) Small-angle XRD patterns of SBA-15 and Co/SBA-15 catalysts with different cobalt loadings, (b) high-angle XRD patterns of Co/SBA-15 with different cobalt loadings, (c) small-angle and (d) high-angle XRD patterns of Co/SBA-15 catalysts with different pore sizes of SBA-15 support.

The high-angle XRD patterns for the Co/SBA-15 catalysts with different cobalt loading (10–30 wt%) are shown in Fig. 5b. Diffraction peaks at 2θ of 19.0, 31.3, 36.9, 44.8, 59.4 and 65.2 indicated that after calcination, cobalt was present in the form of crystalline Co₃O₄ spinel on all of the catalysts,¹³ which agreed with JCPDS no. 42-1467. The small-angle and high-angle XRD patterns of Co/SBA-15 catalysts with different pore sizes of SBA-15 support are presented in Fig. 5c and d. The only difference between the XRD patterns of the calcined samples was the width of the Co₃O₄ peaks. For all catalysts, the Co₃O₄ crystallite size (Table 1) was calculated from the most intense reflection peak (311) at 2θ = 36.9° according to the Scherrer equation.³⁶

3.1.4 Hydrogen temperature programmed reduction (H₂-TPR). The H₂-TPR profiles have been normalized per weight of cobalt in the catalyst to facilitate the discussion. The influence of Co loading on the reduction behavior for Co/SBA-15 catalysts calcined at 450 °C is shown in Fig. 6a. One main reduction peak that can be deconvoluted to two peaks close to each other with temperature maxima at about 333–374 °C (T1_max) and 356–398 °C (T2_max), respectively, was observed. The first peak could be assigned to reduction of Co₃O₄ to CoO, and the second one to be the subsequent reduction of CoO to Co⁰.^13,14,45,46 The calculated T2_max/T1_max deconvoluted peak area ratio were 0.47, 0.26, and 0.20 for 10, 20, and 30Co/SBA-15 catalysts, respectively. The T2_max/T1_max deconvoluted peak area ratio decreased with Co loading, indicating a higher reduction degree of CoO to metallic Co with increasing the average diameter of Co₃O₄ particles. Besides these main reduction peaks, two broad reduction features in the temperature range of 410–600 °C and 600–850 °C were also observed, suggesting the presence of surface Co–support species with different degrees of interactions between them.^13,23

Fig. 6 Temperature programmed reduction profiles of calcined cobalt catalysts.

As shown in Fig. 6a, the relative ratio of the species reducing at high temperature (600–800 °C) to the overall reduction pattern decreased and the maximum temperature for these species shifted to lower temperature, which indicates a lower strength of interaction between cobalt and SBA-15 support, with increasing Co loading. Such a high reduction temperature may be ascribed to the reduction of cobalt silicate species formed during the TPR experiments by the reaction of highly dispersed CoO with the siliceous SBA-15.¹³ Actually, cobalt silicate species were reported to be reduced at temperatures above 450 °C,⁴⁷ while bulk Co₃O₄ became completely reduced at temperatures below 500 °C.^48,49 Therefore, most of the cobalt in the calcined 10Co/SBA-15 sample strongly interacted with the SBA-15 support, and 30Co/SBA-15 sample had a weaker interaction between cobalt and SBA-15 support.

The TPR spectra of 20 wt% Co/SBA-15 catalyst with different pore sizes are shown in Fig. 6b. One main reduction peak that can be deconvoluted to two peaks close to each other with temperature maxima at about 349–371 °C (T1_max) and 367–388 °C (T2_max), respectively, was observed. This peak resulted from the reduction of Co₃O₄ to CoO, and subsequently to Co⁰.^13,14,45,46 The different cobalt loading, the varied Co₃O₄ crystallite size and dispersion. Thus, the reduction peak temperature varied with different loading. The calculated T2_max/T1_max deconvoluted peak area ratio were 0.40, 0.50, 0.16 and 0.30 for 20Co/SBA-15 catalysts with pore sizes of 4.9, 6.5, 7.8, and 9.7 nm, respectively. Two broad reduction features in the temperature of 450–600 °C and 600–850 °C were also observed, suggesting the presence of surface Co–support species with different degrees of interaction between them. As the pore size of SBA-15 increases, the broad peaks (600–850 °C) shifted to lower temperature, and the relative contribution of species reducing at the high temperature to the overall reduction pattern decreased, which indicated a lower strength of interaction between cobalt and SBA-15 support. All these indicated that the smaller pore size of Co/SBA-15 catalysts were more difficultly reduced.

Alternatively, two reduction peaks of SiO₂ supported cobalt catalyst were observed (Fig. 6c), which can be deconvoluted to five peaks. The first peak located at 383 °C corresponds to the reduction of Co₃O₄ to CoO, and the second peak at 406 °C indicates the reduction of CoO to Co.^13,45,46 The temperature of two stage reduction of Co₃O₄ of 20Co/SiO₂ was more higher (∼10 °C) than that of 20Co/SBA-15. In addition, the third, fourth, and fifth peaks at 498, 550, and 576 °C, respectively, were observed, which indicates more difficult reduction process possibly due to stronger cobalt oxide–SiO₂ support interaction.

3.1.5 Relationship of cobalt reducibility with cobalt crystallite size, cobalt dispersion and SBA-15 pore size. The extent of Co reduction estimated from the amount of H₂ consumed of TPR profiles below 400 °C is given in Table 1. Fig. 7 shows that the cobalt reducibility increases with increasing the sizes of Co₃O₄ crystallites, and the cobalt reducibility decreases with increasing the cobalt dispersion on SBA-15 support. The extent of reduction of Co/SBA-15 samples prepared from cobalt nitrate increased from 48 to 86% when increasing the Co loading from 10 to 30 wt% (Table 1). In a previous characterization study, Khodakov et al.³⁵ found that the hydrogen reduction properties of supported cobalt oxide particles depended on the size of the Co₃O₄ crystallites, with larger particles being easily reduced. This finding is in agreement with our observed increase of reducibility with increasing Co₃O₄ crystallite size in the xCo/SBA-15 series with different Co loadings.

Fig. 7 Relationship between Co₃O₄ crystallite size, cobalt dispersion and its reducibility.

Characterization results (Table 1) showed that the catalyst porosity played an important role in the structure and reducibility of supported cobalt species. The size of supported Co₃O₄ crystallites measured by XRD was found to depend on pore sizes, as smaller Co₃O₄ particles being observed in narrow pore SBA-15 (Table 1). The dependence of Co₃O₄ crystallite sizes on catalyst pore diameters suggested preferential localization of Co₃O₄ particle in silica pores (Table 1). It was shown that narrow pore size distribution in periodic mesoporous silicas could stabilize supported nanoparticles.^50,51 Zhang et al.⁵⁰ reported the procedure to prepare nanosized ZnS particles in ordered mesoporous silicas. Iwamoto et al.⁵¹ showed that MCM-41 materials could stabilize nanoparticles of iron oxides. Therefore, the sizes of Co₃O₄ particles in mesoporous silica could be controlled by pore sizes of SBA-15.

Fig. 8 shows that the sizes of Co₃O₄ crystallites and the cobalt reducibility depended on the average pore diameters in SBA-15; larger Co₃O₄ crystallites were found in SBA-15 with wider pores. Differences in Co₃O₄ crystallites sizes result in difference in cobalt reducibility. As shown in Fig. 8, small cobalt particles in narrow pores were more difficult to reduce than larger particles in wide pores. It is known that reducibility of Co₃O₄ crystallites depends on their sizes; the ease of reduction increasing from smaller to larger particles.³⁵ It was suggested that the interaction between metal and support in smaller particles was much stronger than in larger particles and this interaction was likely to stabilize small oxidized particles and clusters in silica.^52–54 The reducibility of Co species is higher in larger pore of SBA-15, because larger pore SBA-15 contain larger, and thus easier to reduce, Co₃O₄ crystallites.^13,35 Therefore, SBA-15 porosity influences the structure of Co supported species in two ways: first, the size of encapsulated Co species is controlled by the pore diameter of the SBA-15 support; second, cobalt reducibility is affected by SBA-15 porosity, and the reducibility of cobalt species being much larger in wide pore SBA-15 supports.

Fig. 8 Dependences of Co₃O₄ crystallite sizes and extents of Co reduction at 400 °C on the different pore diameters of SBA-15.

3.2 Fischer–Tropsch synthesis

3.2.1 Activity of Co/SBA-15 catalysts.
3.2.1.1 Effect of cobalt loading. The catalytic activity of the different Co/SBA-15 catalysts was studied in a fixed-bed reactor under FTS reaction conditions: P = 290 psig, T = 245 °C, H₂/CO = 2, GHSV = 2000 h⁻¹. The activity and selectivity values reported here corresponded to the period of steady-state behavior. The catalytic results are summarized in Table 2. CO conversion had a maximum at a cobalt loading of 20 wt%. CO conversion increased from 18.3% (10Co/SBA-15) to 35.6% (20Co/SBA-15), then decreased to 22.4% (30Co/SBA-15). In general, the activity of reduced Co catalysts should be proportional to the concentration of surface Co⁰ sites.²¹ As observed in Fig. 9, the steady-state FTS reaction rate calculated on the basis of reduced cobalt with different cobalt loading increased as a function of Co⁰ dispersion. The maximum activity obtained thus reflected the inverse trend of cobalt dispersion and direct trend of extent of reduction observed when increasing Co loading. However, the maximum concentration of surface Co⁰ sites was achieved for the 20 wt% Co sample exhibiting a balanced of good dispersion and high reducibility. It can also be seen from Table 2 that at similar cobalt loading (20 wt%), CO conversion based on the mesoporous SBA-15 support is about 2 times than that based on amorphous silica. This higher activity can be attributed to a higher dispersion of cobalt on the high surface area mesoporous support (Table 1) and a larger reducibility of cobalt oxide (Table 1 and Fig. 6c).

Table 2 FTS results for Co/SBA-15 catalysts^a

Catalyst	CO conversion (%)	Selectivity (wt%)			Hydrocarbon distribution (wt%)			α^d
		CO2	ROH^b	RH^c	C1	C2–4	C5+
a Reaction conditions: P = 290 psig, T = 245 °C, H2/CO = 2, GHSV = 2000 h^−1.b ROH for alcohol.c RH for hydrocarbon.d Chain growth probability obtained from the ASF plot in the C8–C14 hydrocarbons range.
20Co/SiO2	18.8	2.3	10.0	87.7	18.4	19.6	62.0	0.80
10Co/SBA-15	18.3	1.8	7.8	90.4	20.9	17.7	61.4	0.82
20Co/SBA-15	35.6	1.6	6.3	92.1	20.3	16.9	62.8	0.83
30Co/SBA-15	22.4	1.1	5.6	93.3	20.1	16.0	63.9	0.82
20Co/SBA-15(P1)	27.6	1.0	8.2	90.8	21.1	16.7	62.2	0.81
20Co/SBA-15(P2)	34.5	1.7	6.3	92.0	20.5	16.8	62.7	0.81
20Co/SBA-15(P3)	40.3	0.6	6.0	93.4	19.5	17.0	63.5	0.82
20Co/SBA-15(P4)	43.8	0.4	7.0	92.6	19.2	16.2	64.6	0.83

---

Fig. 9 FTS reaction rate based on reduced cobalt with different cobalt loading as a function of Co⁰ dispersion.

3.2.1.2 Effect of pore size of SBA-15 support. Catalytic data show that CO conversion were much lower on Co catalysts with 20 wt% loading supported by narrow pore materials. As shown in Table 2, CO conversion increased from 27.6% to 43.8% as the pore size of SBA-15 increased from 4.9 nm [20Co/SBA-15(P1)] to 9.7 nm [20Co/SBA-15(P4)] in the range studied. As observed in Fig. 10, the steady-state FTS reaction rate calculated based on reduced cobalt with different pore size of SBA-15 support increased as a function of Co⁰ dispersion. Catalyst characterization revealed that narrow pore catalysts contained smaller cobalt crystallites (Table 1). Thus, the activity of the catalysts seems to be affected by the size of cobalt crystallites. Xiong et al.²⁴ found that CO conversion of Co/SBA-15 catalysts with high cobalt loading (30 wt%) prepared by impregnation method increased from 19.3% to 30.9% as SBA-15 pore size increased from 3.7 to 9.0 nm. They showed that the catalyst with larger pore led to larger cobalt cluster size, lower dispersion and higher reducibility. The larger pores created more adsorbed CO of the bridge-type on FTS based on diffuse reflectance infrared Fourier transform spectroscopy (DRIFS) study. It was also suggested that FTS was structure sensitive and favored sites to which CO was strongly coordinated. It was suggested that partially reduced cobalt species did not catalyze FTS and that small particles could be easily re-oxidized by water or other reaction products at the conditions of FTS.³⁵ Our data are consistent with the suggestion that lower activity of small cobalt crystallites could be attributed to their lower reducibility.

Fig. 10 FTS reaction rate based on reduced cobalt with different pore size of SBA-15 support as a function of Co⁰ dispersion.

3.2.2 Product selectivity of Co/SBA-15 catalysts.
3.2.2.1 Effect of cobalt loading. The selectivity to different FTS products obtained for different cobalt loadings is presented in Table 2. It can be seen that the Co/SBA-15 with 10–30 wt% Co loading exhibited a high selectivity toward hydrocarbons (90.4–93.3%), and low selectivity toward alcohols (5.6–7.8%) and CO₂ (1.1–1.8%) reflecting a low activity for the competitive water–gas-shift reaction (WGS: CO + H₂O → CO₂ + H₂) typically observed for Co-based FTS catalysts.¹³ A higher concentration of light product (CH₄, C₂–C₄) among the hydrocarbon fraction was observed for the low cobalt loaded catalyst. High methane selectivity has also been reported for low cobalt loaded catalysts with small particle, high dispersion and low reducibility.³⁷ This effect was ascribed to diffusion limitations for CO in catalyst pores could increase H₂/CO ratio in catalyst pore and thus, methane selectivity.⁵⁵ The chain growth probability, α, derived from the slopes of the Anderson–Shultz–Flory (ASF) plots in the C₈–C₁₄ hydrocarbon range did not change with different Co loadings (Table 2). The formation of C₅₊ hydrocarbons was favored over Co/SBA-15 catalysts exhibiting low cobalt dispersion and high reducibility. As shown in Fig. 11, as the cobalt loading increased from 10 to 30 wt%, the selectivity to C₅₊ hydrocarbons increased from 61.4 to 63.9%.

Fig. 11 Effect of cobalt loading on cobalt dispersion and C₅₊ selectivity.

3.2.2.2 Effect of pore size of SBA-15 support. The selectivity to different FTS products obtained for different SBA-15 pore sizes is also presented in Table 2. The decrease in pore sizes resulted in higher methane selectivity, whereas the chain growth probability, α, seemed not to be affected by variation in catalyst pore diameters. Analysis of literature data shows that higher methane selectivities are observed in FTS when Co catalysts are not completely reduced or contain smaller cobalt particles.²⁴ Thus, higher methane selectivities observed on narrow pore SBA-15 are likely to be attributed to the presence of either unreduced cobalt species or the small cobalt particles, which produce higher relative amounts of methane than large cobalt particles. Table 2 also show that lower C₅₊ selectivity observed on narrow pore samples is likely to be related not only to lower probability of chain propagation of narrow pore silicas, but also to higher methane selectivity. As shown in Fig. 12, as the pore size of SBA-15 increased from 4.9 to 9.7 nm, the selectivity to C₅₊ hydrocarbons increased from 62.2 to 64.6%. This was attributed to the increase of pore size of SBA-15, which decreased the cobalt dispersion, and thus increased the cobalt reducibility.

Fig. 12 Effect of pore diameter of SBA-15 on cobalt dispersion and C₅₊ selectivity.

4. Conclusions

A series of SBA-15 supported cobalt catalysts with different cobalt loading and different pore diameters have been prepared and the role of cobalt loading and pore size on FTS have been investigated. The reduction of the catalysts took place in two stages, with Co₃O₄ reduction to CoO and subsequently to Co⁰. The dispersion of Co/SBA-15 decreased and the extent of cobalt reduction increased with increasing either cobalt loading or pore size of SBA-15. A maximum CO conversion was found for the sample with 20 wt% Co loading. More methane and less C₅₊ hydrocarbons were produced over less reducible 10 wt% Co loading sample. The 20Co/SBA-15 catalysts with larger pore led to larger cobalt crystallite size, lower dispersion, and higher reducibility. CO conversion increased with the increase of pore size in the range studied. The 20Co/SBA-15 catalysts with larger cobalt crystallite size showed higher C₅₊ selectivity for the FTS. Finally, at comparable Co loading, CO conversion of Co/SBA-15 catalysts were about 2 times than a Co/SiO₂ sample, with only minor difference in product selectivity.

Acknowledgements

Financial support provided by the Cooperative State Research, Education, and Extension Service, and U.S. Department of Agriculture, under the Award No. 2012–10008–20302.

References

1. A. Y. Khodakov, W. Chu and P. Fongarland, Chem. Rev., 2007, 107, 1692.
2. Y. Zhang, L. Ma, T. Wang and X. Li, RSC Adv., 2015, 5, 45426.
3. Y. Liu, J.-F. Chen and Y. Zhang, RSC Adv., 2015, 5, 29002.
4. O. O. James, B. Chowdhury, M. A. Mesubi and S. Maity, RSC Adv., 2012, 2, 7347.
5. J. C. Park, D. H. Chun, J.-I. Yang, H.-T. Lee, S. Hong, G. B. Rhim, S. Jang and H. Jung, RSC Adv., 2015, 5, 44211.
6. Q. Yan, Y. Lu, C. Wan, J. Han, J. Rodriguez, J. Yin and F. Yu, Energy Fuels, 2014, 28, 2027.
7. E. de Smit and B. M. Wechhuysen, Chem. Soc. Rev., 2008, 37, 2758.
8. Q. Zhang, J. Kang and Y. Wang, ChemCatChem, 2010, 2, 1030.
9. Y. Lu, B. Cao, F. Yu, J. Liu, Z. Bao and J. Gao, ChemCatChem, 2014, 6, 473.
10. J. Hu, F. Yu and Y. Lu, Catalysts, 2012, 2, 303.
11. S. Luo, L. Zeng, D. Xu, M. Kathe, E. Chung, N. Deshpande, L. Qin, A. Majumder, T. Hsieh, A. Tong, Z. Sun and L. Fan, Energy Environ. Sci., 2014, 7, 4104.
12. J. Wang, G. Cheng, Y. You, B. Xiao, S. Liu, P. He, D. Guo, X. Guo and G. Zhang, Int. J. Hydrogen Energy, 2012, 37, 6503.
13. A. Martínez, C. López, F. Márquez and I. Díaz, J. Catal., 2003, 220, 486.
14. L. Yu, X. Liu, Y. Fang, C. Wang and Y. Sun, Fuel, 2013, 112, 483.
15. A. R. de la Osa, A. de Lucas, L. Sánchez-Silva, J. Díaz-Maroto, J. L. Valverde and P. Sánchez, Fuel, 2012, 95, 587.
16. S. Sun, K. Fujimoto, Y. Yoneyama and N. Tsubaki, Fuel, 2002, 81, 1583.
17. W. Ma, G. Jacobs, D. E. Sparks, M. K. Gnanamani, V. R. R. Pendyala, C. H. Yen, J. L. S. Klettlinger, T. M. Tomsik and B. H. Davis, Fuel, 2011, 90, 756.
18. P. R. Karandikar, Y. J. Lee, G. Kwak, M. H. Woo, S. J. Park, H. G. Park, K. S. Ha and K. W. Jun, J. Phys. Chem. C, 2014, 118, 21975.
19. J. Zhang, J. Chen, J. Ren, Y. Li and Y. Sun, Fuel, 2003, 82, 581.
20. J. S. Girardon, E. Quinet, A. Griboval-Constant, P. A. Chernavskii, L. Gengembre and A. Y. Khodakov, J. Catal., 2007, 248, 143.
21. E. Iglesia, S. L. Soled and R. A. Fiato, J. Catal., 1992, 137, 212.
22. B. G. Johnson, C. H. Bartholomew and D. W. Goodman, J. Catal., 1991, 128, 231.
23. E. Iglesia, Appl. Catal., A, 1997, 161, 59.
24. H. Xiong, Y. Zhang, K. Liew and J. Li, J. Mol. Catal. A: Chem., 2008, 295, 68.
25. G. Jacobs, Y. Y. Ji, B. H. Davis, D. Cronauer, A. J. Kropf and C. L. Marshall, Appl. Catal., A, 2007, 333, 177.
26. P. J. van Berge, J. van de Loosdrecht, S. Barradas and A. M. van der kraan, Catal. Today, 2000, 58, 321.
27. D. Zhao, J. Feng, Q. Huo, N. Melosh, G. H. Fredrickson, B. F. Chmelka and G. D. Stucky, Science, 1998, 279, 548.
28. D. Zhao, J. Sun, Q. Li and G. D. Stucky, Chem. Mater., 2000, 12, 275.
29. M. Shakeri, R. J. M. Klein Gebbink, P. E. de Jongh and K. P. de Jong, Angew. Chem., Int. Ed., 2013, 52, 10854.
30. G. Prieto, A. Martínez, R. Murciano and M. A. Arribas, Appl. Catal., A, 2009, 367, 146.
31. H. Li, J. Li, H. Ni and D. Song, Catal. Lett., 2006, 110, 71.
32. O. González, H. Pérez, P. Navarro, L. C. Almeida, J. G. Pacheco and M. Montes, Catal. Today, 2009, 148, 140.
33. A. Y. Khodakov, R. Bechara and A. Griboval-Constant, Appl. Catal., A, 2003, 254, 273.
34. Y. Wang, M. Noguchi, Y. Takahashi and Y. Ohtsuka, Catal. Today, 2001, 68, 3.
35. A. Khodakov, A. Griboval-Constant, R. Bechara and V. L. Zholobenko, J. Catal., 2002, 206, 230.
36. Y. Lu, F. Yu, J. Hu and J. Liu, Appl. Catal., A, 2012, 429–430, 48.
37. R. C. Reuel and C. H. Bartholomew, J. Catal., 1984, 85, 63.
38. R. D. Jones and C. H. Bartholomew, Appl. Catal., 1988, 39, 77.
39. Z. Luan, M. Hartmann, D. Zhao, W. Zhou and L. Kevan, Chem. Mater., 1999, 11, 1621.
40. Z. Luan, E. M. Maes, P. A. W. van der Heide, D. Zhao, R. S. Czernuszewicz and L. Kevan, Chem. Mater., 1999, 11, 3680.
41. G. L. Bezemer, J. H. Bitter, H. P. C. E. Kuipers, H. Oosterbeek, J. E. Holewijn, X. Xu, F. Kapteijn, A. Jos van Dillen and K. P. de Jong, J. Am. Chem. Soc., 2006, 128, 3956.
42. J. P. den Breejen, P. B. Radstake, G. L. Bezemer, J. H. Bitter, V. Frøseth, A. Holmen and K. P. de Jong, J. Am. Chem. Soc., 2009, 131, 7197.
43. G. Prieto, A. Martínez, P. Concepción and R. Moreno-Tost, J. Catal., 2009, 266, 129.
44. Ø. Borg, P. D. C. Dietzel, A. I. Spjelkavik, E. Z. Tveten, J. C. Walmsley, S. Diplas, S. Eri, A. Holmen and E. Rytter, J. Catal., 2008, 259, 161.
45. P. Arnoldy and J. A. Moulijn, J. Catal., 1985, 93, 38.
46. B. Viswanathan and R. Gopalakrishnan, J. Catal., 1986, 99, 342.
47. B. Sexton, A. Hughes and T. Turney, J. Catal., 1986, 97, 390.
48. S. Bessell, Appl. Catal., A, 1993, 96, 253.
49. G. Sewell, C. O'Connor and E. van Steen, Appl. Catal., A, 1995, 125, 99.
50. W. H. Zhang, J. L. Shi, H. R. Chen, Z. L. Hua and D. S. Yan, Chem. Mater., 2001, 13, 648.
51. M. Iwamoto, T. Abe and Y. Tachibana, J. Mol. Catal. A: Chem., 2000, 155, 143.
52. A. Y. Khodakov, A. Griboval-Constant, R. Bechara and F. Villain, J. Phys. Chem. B, 2001, 105, 9805.
53. R. Bechara, D. Balloy, J. Y. Dauphin and J. Grimblot, Chem. Mater., 1999, 11, 1703.
54. A. Y. Khodakov, J. Lynch, D. Bazin, B. Rebours, N. Zanier, B. Moisson and P. Chaumette, J. Catal., 1997, 168, 16.
55. R. B. Anderson, W. K. Hall, A. Krieg and B. Seligman, J. Am. Chem. Soc., 1949, 71, 183.

---