Al2O3 catalysts used for mixed alcohol synthesis from synthesis gas

Applied Catalysis A: General 187 (1999) 187–198

Structures and performance of Rh–Mo–K/Al2 O3 catalysts used for mixed alcohol synthesis from synthesis gas Zhong-rui Li, Yi-lu Fu ∗ , Ming Jiang Department of Chemical Physics, University of Science and Technology of China, Hefei 230026, PR China Received 18 January 1999; received in revised form 18 May 1999; accepted 21 May 1999

Abstract A series of rhodium-modified Mo–K/Al2 O3 catalyst samples was prepared by varying the rhodium loading between 0 and 1.0 wt.% and maintaining molybdenum and potassium contents as constants. The structures of the samples were characterized by techniques of XRD, LRS, TPR, XPS and EXAFS and correlated to the catalytic properties of the samples for alcohol synthesis from synthesis gas. It was found that, in the oxidic rhodium-modified samples, a strong interaction of the rhodium modifier with the supported K–Mo–O species occurs. This interaction facilitates the sulfidation and reduction of the supported oxo-molybdenum and leads to a decrease in the size of the molybdenum species and stabilization of the cationic rhodium species on the samples during sulfidation. Upon sulfidation, the sulfided molybdenum species in the rhodium-free sample is mainly present as large patches of MoS2 -like slabs with their basal sulfur planes interacting with the support surface. With the modification of rhodium to the samples, the supported MoS2 -like species becomes highly dispersed, as revealed by the decrease in the average size of the sulfided molybdenum species. The interaction of the rhodium species with the molybdenum component may cause the basal planes of the MoS2 -like species to become oriented perpendicular to the support surface due to favorable bonding of the MoS2 edge planes to the support through Mo–O–Al bonds. In comparison with the sulfided sample free of rhodium, the properties of the rhodium-modified samples for alcohol synthesis from synthesis gas are much improved. It most probably results from the synergic interaction of the rhodium with the molybdenum species that gives rise to the appearance of the catalytically active surfaces or sites. The co-existence of cationic and metallic rhodium stabilized by this interaction may be responsible for the increased selectivity for the formation of C2+ alcohols. ©1999 Elsevier Science B.V. All rights reserved. Keywords: Sulfided Rh–Mo–K/Al2 O3 catalysts; Structure characterizations; Mixed alcohol synthesis

1. Introduction From the points of view of fundamental research and practical applications, mixed alcohol synthesis from synthesis gas is an important subject in C1 chemistry, to which increasing attention has been paid. This ∗ Corresponding author. Fax: +86-551-3631-760 E-mail address: [email protected] (Y.-l. Fu)

is because it offers a possibility for producing clean fuel from coal, natural gas and hydrocarbon wastes via gasification. Mixed alcohols can also be used as an adding stock to increase the octane number of traditional fuels and as an important basic material in chemical industry. In this aspect, molybdenum-based catalysts have been given special interest because of their sulfur-resistant property. For CO hydrogenation over sulfided and reduced molybdenum-based cata-

0926-860X/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 9 9 ) 0 0 2 2 7 - 6

188

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

lysts promoted by alkali components, the proportion of methanol to C2+ alcohols in the reaction products is rather high [1–7]. From a practical point of view, increasing the selectivity of C2+ alcohols is an important issue and has been extensively studied. Murchison et al. [1] reported that, over the K–MoS2 /C catalysts, the increase of the H2 S content in the feed is effective for reducing the methanol level in the products, but the large amount of sulfur compounds in the system is undesirable. It was found that addition of 3d transition metal components (e.g. cobalt and nickel) to reduced or sulfided molybdenum-based catalysts can increase the selectivity of the reaction to C2+ alcohol products [1–3,8,9]. Storm et al. investigated the incorporation of rhodium into reduced K/Co/Mo/Al2 O3 catalysts and found that the activity and selectivity to the production of alcohols are improved [10–12]. As revealed by several authors [13,14], rhodium species in catalysts are capable of catalyzing dissociation, insertion and hydrogenation of CO depending upon the status of the rhodium species, properties of alkali promoters and supports as well as reaction conditions. The results obtained from infrared spectroscopy of the oxidic Mo–Rh/Al2 O3 catalysts reveal that molybdenum preferentially covers the large particles of metallic rhodium on alumina to form a mixed oxide surface phase that chemisorbs little CO at room temperature [13]. After reduction of the catalysts, rhodium and molybdenum components form mixed metallic particles on the catalyst. Its composition is sensitive to the environment: when the atmosphere is non-reducing, molybdenum concentrates on the surface, and when hydrogen is adsorbed, rhodium appears to enrich on the surface [12].The chemical interaction between rhodium and molybdenum in rhodium-modified catalysts gives rise to a different catalytic behavior from either rhodium or molybdenum component alone. Lowenthal et al. [15] correlated the methanol dehydration property of potassium-doped alumina and Rh–Mo/Al2 O3 samples to the strength and amount of Lewis acid determined by TPD of NH3 adsorption. They found that the addition of rhodium diminishes surface acidity of Mo/Al2 O3 . Sudhakar et al. reported that the rate of CO hydrogenation over Mo–Rh/Al2 O3 , in comparison with that over Rh/Al2 O3 , increases by a factor of 10 or more and the product distribution shifted to oxygenates [16]. By using electron

microscopy, Foley et al. observed mixed rhodium and molybdenum particles on the surface of the catalyst. XPS results indicate that some amount of rhodium appears in cationic form [17]. So far, sulfided molybdenum-based catalysts modified by rhodium used for alcohol synthesis have seldom been reported in literature. In the present work, a series of sulfided Rh–Mo–K/Al2 O3 catalyst samples with different rhodium loadings has been prepared and investigated for the synthesis of mixed alcohols from synthesis gas. The interaction between the rhodium and molybdenum components and its effect on reducibility, sulfidability, and dispersion of the supported molybdenum species are elucidated, based upon the characterizations obtained by using a variety of techniques, as mentioned below. The correlation between the catalyst structures and their catalytic properties for alcohol synthesis are discussed. 2. Experimental 2.1. Sample preparation The oxidic Mo–K/Al2 O3 sample was prepared by a sequential pore volume impregnation method. ␥-Al2 O3 support (BET surface area: 270 m2 g−1 ) was first impregnated with K2 CO3 solution, followed by drying at 393 K for 12 h and calcining in air at 573 K for 1 h. Then the material was impregnated with (NH4 )6 Mo7 O24 ·4H2 O solution by drying at 393 K for 12 h and calcining in an oxygen flow of 40 ml min−1 at 773 K for 24 h. The resulting sample was further calcined in air at 1073 K for 12 h. Rhodium–modified samples were prepared by impregnating the obtained Mo–K/Al2 O3 sample with RhCl3 solutions, followed by drying at 393 K for 12 h and calcining in air at 773 K for 2 h. The sulfided samples were obtained by heat-treating the oxidic ones in a flow of CS2 /H2 mixed gas (8.7% CS2 ) of 30 ml min−1 at 673 K for 6 h. The molybdenum content in the samples, expressed as weight ratio of MoO3 /Al2 O3 , is 0.24, the atomic ratio of K/Mo is 0.8, and the rhodium loading varies from 0 to 1.0% of the total weight of the samples. 2.2. Characterization methods The patterns of X-ray diffraction (XRD) were obtained on a D/MAX-␥A rotatory target diffracto-

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

meter using Cu K␣ radiation (λ = 0.15418 nm, 40 kV and 100 mA). The samples were ground into fine powder and packed into sample holders for measurements. Laser Raman spectra (LRS) were recorded on an SPEX-1403 spectrometer with a resolution of 2 cm−1 using the 488.0 nm−1 radiation line from a Spectra-Physics-2020 argon laser. The laser beam intensity and the spectrum slit width were 100 mW and 3.5 cm−1 , respectively. The samples were pressed into pellets for the measurements. Temperature-programmed reductions (TPR) of the oxidic samples were carried out on a conventional TPR setup in a flow of H2 /N2 mixed gas (10% H2 ) of 30 ml min−1 with a heating rate of 10 K min−1 at ambient pressure. The mixed gas was purified by passing through the 105 deoxy agent and 5A zeolite and then introduced to the oxidic samples placed in a tube reactor. The hydrogen consumption was continuously monitored by an on-line gas chromatograph with a thermal conductivity detector. X–ray photoelectron spectra (XPS) were obtained on a XSAM 800 spectrometer with Al K␣ excitation radiation at 1486.6 eV. The Al 2p peak centered at 74.5 eV from the samples was used as an internal standard for binding energy calibration. The spectra were deconvoluted by using a Gaussian function. The surface atomic concentrations of the samples were calculated according to Wagner’s atomic sensitivity factors. The spectra of extended X-ray absorption fine structure (EXAFS) were measured at the beamline of 4WIB of Beijing Synchrotron Radiation Facility (BSRF). The storage ring was operated at 2.2 GeV with a typical current of 50 mA. The fixed-exit Si(111) flat double crystals were used as a monochromator. The spectra were recorded in transmission mode with ionization chambers filled with argon. The K-edge of molybdenum metal positioned at 19,999 eV was used for calibration. Data analysis was performed following a standard procedure [18]. Phase shifts and backscattering amplitudes extracted from the spectra of Na2 MoO4 ·2H2 O and MoS2 standard compounds were used to calculate the structural parameters of the samples. 2.3. Measurements of catalytic activities The catalytic activities of the sulfided Rh–Mo– K/Al2 O3 samples for the mixed alcohol synthesis

189

from synthesis gas were measured with a fixed-bed reactor equipped with an on-line gas chromatograph. For each experiment, 0.5 ml of the sample was charged into a stainless steel reactor with an inner diameter of 6 mm. The synthesis gas is composed of CO (30%), H2 (60%) and N2 (10%). The effluent gas was cooled to 273 K and separated into gas and liquid phases at high pressure. The gaseous products were directly analyzed on a chromatograph through a sampling valve, and the liquid ones were collected for a proper period with their volume and weight measured and subsequently analyzed on the same chromatograph by injection. CO, CO2 and H2 in the gas phase and H2 O in liquid products were analyzed on a 2 m TDX-01 column by using a thermal conductivity detector with argon as carrier gas. The hydrocarbons, alcohols and other oxygenated compounds were analyzed on a 2 m Porapak Q column by using a hydrogen flame detector with N2 as carrier gas. The composition of hydrocarbons was calculated using 1.04% CH4 as standard gas, while that of the mixed alcohols was calculated directly from the peak areas by using a standard liquid of mixed alcohols. Under the present experimental conditions, since only trace amounts of CO2 could be detected by our thermal conductivity detector, the activity mentioned below would be referred to as being CO2 -free.

3. Results 3.1. XRD The XRD patterns of the oxidic samples are shown in Fig. 1. The ␥-Al2 O3 support gives rise to the peaks at d value of 0.239, 0.198 and 0.139 nm. The pattern of the rhodium-free sample exhibits the peaks of different K–Mo–O species (Fig. 1 A (a)). The strong peaks at d values of 0.396, 0.277 and 0.195 nm may be attributed to the species related to K0.85 Mo6 O17 according to the standard diffraction data. The species also have peaks which occur at d values of 0.328, 0.239, 0.175, 0.159, 0.144 and 0.138 nm. The other weak peaks may result from several kinds of Kx Moy Oz species (y = 3, 4, 7) formed on the support due to calcination of the samples at high temperature of 1073 K. It is important to note that the main species formed on the support, as determined from peak intensities, is K0.85 Mo6 O17 . For

190

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

Fig. 1. XRD patterns of the oxidic (A) and sulfided (B) samples with different rhodium loadings: (a) 0, (b) 0.5%, (c) 0.8%, (d) 1.0%.

molybdenum species with a potassium component, the bands typical of bulky MoO3 (e.g. 818 and 996 cm−1 ) are absent [19]. According to the literature [18] and previous work performed in our laboratory [20], the bands which appear at 948 and 896 cm−1 can be assigned to the octahedrally and tetrahedrally coordinated surface oxo-molybdenum species (Mo(Oh) and Mo(Td)), respectively. The bands occurred at 933, 906, and 352 cm−1 correspond to the symmetric stretching, asymmetric stretching and bending vibrations of the terminal Mo=O bond in the octahedrally coordinated MoO6 species of polymolybdate phases (e.g. K0.85 Mo6 O17 , as revealed by XRD). The bands at 558 and 220 cm−1 are due to the symmetric stretching and deformation vibrations of Mo–O–Mo in the MoO6 unit. In addition, two bands at 1000 and 378 cm−1 are characteristic of Al2 (MoO4 )3 , a species resulting from the strong interaction of the molybdenum with the support [20]. In comparison with the spectrum of the rhodium-free sample, the band intensities of the rhodium-modified samples remarkably decrease with increasing rhodium loading, but no new band could be observed (Fig. 2(b–c)). This suggests that the sizes of the original

the rhodium-modified samples (Fig. 1 A (b)–(d)), all the peaks arising from K–Mo–O species decrease. No peaks related to the rhodium species can be detected. After sulfidation, the diffractions of the K–Mo–O species observed for the oxidic samples are removed (Fig. 1 B). Besides the strong peaks arising from the support, all the other peaks are weak and the intensities decrease with increasing rhodium loading. The broad peaks at d values of 0.615, 0.236, 0.220 nm can be assigned to the MoS2 species. The diffractions of 0.236, 0.271, 0.157 nm and 0.345, 0.296 nm may be assigned to K0.4 MoS2 and K2 MoS4 , respectively. 3.2. LRS LRS of the oxidic samples are presented in Fig. 2. The spectrum of the Al2 O3 support is essentially featureless in the region of 100 ∼ 1100 cm−1 under study. The supported oxo-molybdenum species on the oxidic rhodium-free sample gives rise to several bands in this region (Fig. 2(a)). Due to the interaction of

Fig. 2. LRS of the oxidic samples with different rhodium loadings: (a) 0, (b) 0.5%, (c) 0.8%, (d) 1.0%.

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

Fig. 3. LRS of the sulfided samples with different rhodium loadings: (a) 0, (b) 0.5%, (c) 0.8%, (d) 1.0%.

191

under oxidizing conditions, rhodium is easily dispersed on Al2 O3 up to a saturation concentration of 10% of the support area and the excess rhodium is aggregated as three-dimensional Rh2 O3 particles. In the present work, all the samples were activated in air at a much higher temperature (1073 K) and the interaction between rhodium and the support is much stronger than that between rhodium and SiO2 in Rh/SiO2 calcined at a lower temperature. Therefore, the reduction peak of Rh/Al2 O3 occurs at a much higher temperature (∼750 K) (Fig. 4(a)). Careres et al. studied the reducibility of aluminasupported molybdenum catalysts and reported that there are two H2 consumption peaks located in the regions of 773 ∼ 873 K and 973 ∼ 1073 K, respectively [24]. They interpreted them as two-step reductions of Mo6+ , i.e. Mo6+ → Mo4+ and Mo4+ → Mo0 . For the Mo–K/Al2 O3 sample (Fig. 4(b)), only one peak is observed at 780 K corresponding to the reduction of Mo6+ to Mo4+ , consistent with the results obtained by Careres et al. [24]. For the rhodium-modified samples, two reduction peaks are observed (Fig. 4(c–e)). The low temperature peaks can be ascribed to the

K–Mo–O species, due to the interaction between rhodium and K–Mo species, have become much smaller. After sulfidation, the oxo-molybdenum species observed for the oxidic samples are not detected, as indicated by the absence of the corresponding bands (Fig. 3). For the rhodium-free sample, the bands at 378 and 406 cm−1 , characteristic of MoS2 , are observed. With the incorporation of rhodium into the samples, the band intensities of the MoS2 species decrease due to the interaction of rhodium with the sulfided molybdenum species (Fig. 3(b–d)). 3.3. TPR TPR profiles of the oxidic samples are shown in Fig. 4. Wong et al. [21] studied the TPR properties of Rh/SiO2 calcined at about 473 K and reported that two reduction peaks are observed at about 353 and 460 K, respectively. They ascribed them to two kinds of oxidic rhodium species in the sample. When the sample is pre-treated at higher temperature, the peak at about 460 K becomes larger, while that at about 353 K becomes smaller. Yao et al. [22,23] reported that,

Fig. 4. TPR profiles of the oxidic samples: (a) Rh(1.0%)/Al2 O3 , (b) Mo–K/Al2 O3 , (c) Rh(0.5%)–Mo–K/Al2 O3 , (d) Rh(0.8%)– Mo–K/Al2 O3 , (e) Rh(1.0%)–Mo–K/Al2 O3 .

192

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

Table 1 Relative concentrations of surface Mo, S and Rh species on sulfided samplesa Sample (wt.%Rh)

Relative atomic concentration (%) Mo Mo4+

0 0.5 0.8 1.0 a

S Mo6+

S2−

229.0

MoS2+x 230.0

Mo5+ 231.3

232.7

58 61 65 69

15 14 11 7

11 11 13 15

16 14 11 9

Rh 169.0

Rh0 307.1

Rh+ 308.0

Rh2+ 309.1

Rh3+ 310.3

46 40 37 36

– 46 35 44

– 43 39 34

– 10 25 21

– 1 1 1

S6+

162.0

S2 163.0

S0 164.0

22 38 44 46

20 12 8 4

12 10 11 14

2−

Mo 3d5/2 , S 2p and Rh 3d5/2 binding energies were used for calculations.

reduction of the rhodium species. In comparison with the TPR profile of Rh/SiO2 [21], their positions are shifted to lower temperatures. The high temperature peaks are observed at about 770 K, which are located between the reduction peaks of Rh/Al2 O3 at about 753 K and Mo–K/Al2 O3 at 780 K. They may correspond to the reductions of both rhodium and Mo6+ (to Mo4+ ) species. It can be determined from the peak positions that the reduction of the rhodium species incorporated into the samples, in comparison with that of Rh/Al2 O3 , becomes difficult, while the reduction of the molybdenum species becomes easier. 3.4. XPS The XPS of Mo, S and Rh elements in the samples are shown in Fig. 5. By assuming that the peak shape of the single component is Gaussian-type and the background is linear, the spectra were deconvoluted according to the electronic binding energies presented in Refs. [25,26]. The ratios of the peak areas of M 3d5/2 to M 3d3/2 (M = Mo, Rh) were assumed as 1.5. The S 2s peaks hidden inside the Mo 3d peaks were considered and corrected. For clarity, only the deconvoluted Mo 3d5/2 , S 2p3/2 and Rh 3d5/2 spectra are shown in Fig. 5, those of Mo 3d3/2 , S 2p1/2 and Rh 3d3/2 being omitted. The relative atomic concentrations of the surface species calculated from the deconvoluted results are listed in Table 1. In Table 1, the binding energies of rhodium, which are adopted from Beng’s work [26], only make relative sense since they strongly depend upon the chemical environments. As seen from Fig. 5 and Table 1, four kinds of components can be found for Mo, S and Rh. The molybdenum species can be classified into incom-

pletely reduced Mo6+ and Mo5+ , MoS2 , and MoS2+x in sulfur-rich environments. Sulfur is represented by S6+ , S0 or (S–S)n , S2− and sulfur–molybdenum structures in sulfur-rich environments or (S–S)2− strings [27]. The valence of rhodium varies from 0 to +3. Incorporation of rhodium into Mo–K/Al2 O3 decreases the contents of Mo6+ , MoS2+x , S6+ and S2 2+ species and increases those of Mo5+ , Mo4+ , S0 and S2− . In all the samples, a large fraction of rhodium is presented as cationic species. By taking the surface atomic concentration of molybdenum as 1, the relative surface atomic concentrations of the other elements in the samples are calculated as shown in Table 2. As the detection depth of the XPS technique, the obtained results should represent the information of the surface layer,

Fig. 5. Mo, S and Rh XPS of the sulfided samples: (a) 0, (b) 0.5%, (c) 0.8%, (d) 1.0%.

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

193

Table 2 Relative surface concentrations of elements on the sulfided samples Sample (wt.%Rh)

Mo

Rh

S

K

1Mo4+

1S2−

1S2− / 1Mo4+

0 0.5 0.8 1.0

1 1 1 1

– 0.2 0.3 0.3

3.2 3.5 3.4 3.4

0.6 0.3 0.4 0.5

0.7 0.8 0.8 0.8

1.3 1.8 1.7 1.7

1.7 2.3 2.3 2.3

or more exactly, a few atomic layers of the samples. Here, the amount of molybdenum species in MoS2 and MoS2+x , and that of sulfur species in MoS2 and S2 2− are defined as 1Mo4+ and 1S2− , respectively. 1Mo4+ is similar for all the samples and takes up the major part (>70%) of the molybdenum component, indicating that the molybdenum in the samples mainly exists in the form of MoS2 -like species. With increasing rhodium loading, both 1S2− and the ratio of 1S2− /1Mo4+ are increased. It implies that the molybdenum species become deeply sulfided and reduced. In addition, an enrichment of rhodium on the surface can be determined for rhodium-modified samples. For example, for the sample with rhodium loading of 1.0%, the atomic Rh/Mo ratio on the surface is 4.3 times as much as that in the bulk.

Fig. 6. Fourier transforms of k3 χ (k) for the crystalline MoS2 and the sulfided samples. (a) MoS2 , (b) Mo–K/Al2 O3 , (c) Rh(0.5%)–Mo–K/Al2 O3 , (d) Rh(1.0%)–Mo–K/Al2 O3 .

3.5. EXAFS The magnitude of the Fourier transforms (k3 χ(k), 1k = 17 ∼ 151 nm−1 ) of the samples together with the MoS2 standard compound are shown in Fig. 6. MoS2 is of hexagonal structure with one molybdenum atom surrounded by six sulfur atoms at a distance of 0.241 nm and six molybdenum neighbors at 0.316 nm. In the Fourier transforms of the samples (Fig. 6(b–d)), mainly two peaks are observed at 0.201 and 0.283 nm. They are located at almost the same positions as those for MoS2 standard compound (Fig. 6(a)), indicating that the local structure of the sulfided molybdenum species is similar to that of MoS2 . In comparison with those of MoS2 , however, the magnitudes of Fourier transforms of the Mo–S and Mo–Mo coordinations are significantly decreased and the ratios of the magnitude of Mo–Mo shell to that of Mo–S (mixed with Mo–O) are also lower. The inverse Fourier transforms of the filtered Mo–S and Mo–Mo shell data for the MoS2 standard compound and the samples are pre-

Fig. 7. Fourier-filtered Mo K-edge EXAFS oscillations of Mo–S shell (r, 0.09 ≈ 0.24 nm) (A) and Mo–Mo shell (r, 0.24 ≈ 0.32 nm) (B) for the crystalline MoS2 and sulfided samples. (a) MoS2 , (b) Mo–K/Al2 O3 , (c) Rh (0.5%)–Mo–K/Al2 O3 , (d) Rh (1.0%)–Mo–K/Al2 O3 .

sented in Fig. 7. It can be seen that the amplitudes of the samples decrease with increasing rhodium loading. The low oscillation amplitudes of both Mo–Mo and Mo–S shells in the inverse Fourier transforms of the samples compared to those of MoS2 demonstrate that the sizes of the supported MoS2 -like species are smaller. The structural parameters obtained from the fitting results of the molybdenum coordination shells for the samples are presented in Table 3. For clarity,

194

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

Table 3 Structure parameters from the Fourier-filtered data for the sulfided samples Samples

Bond

R/nm

N

1σ 2 (10−6 /nm2 )a

1E0 /eVb

Mo–K/Al2 O3

Mo–O Mo–S Mo–Mo Mo–O Mo–S Mo–Mo Mo–O Mo–S Mo–Mo Mo–O Mo–S Mo–Mo

0.177 0.241 0.316 0.197 0.243 0.315 0.200 0.243 0.315 0.177 0.241 0.316

0.2 5.1 3.7 1.0 4.0 2.4 1.5 3.4 2.0 4.0 6.0 6.0

0 2 1 18 11 11 60 20 24

2.43 0.65 1.14 2.84 −1.25 1.01 −13.4 −0.54 0.32

Rh(0.5%)–Mo–K/Al2 O3

Rh(1.0%)–Mo–K/Al2 O3 Na2 MoO4 ·2H2 OMoS2

a b

Relative Debye–Waller factor of the sample to that of the standard compounds. Correction of the inner potentials of the samples based upon those of the crystalline MoS2 .

Table 4 Effect of rhodium on alcohol synthesis over the sulfided samplesa Sample (wt.% Rh)

Conv. (%CO)

Alc.sel.b (%C)

MeOH/C2+ OH ratioc

Alc.STYd (g/l h)

HC.STYd (g/l h)

0.0 0.5 0.8 1.0

3.2 4.4 4.9 5.7

39.4 58.7 61.8 63.7

0.82 0.42 0.40 0.38

19.1 61.8 44.5 56.6

15.8 16.3 18.4 19.5

Reaction conditions: 4.0 Mpa, 600 K, 4800 h−1 , H2 /CO = 2. Selectivity to alcohols. c Molar ratio of methanol to higher alcohols. d Space–time yield. a

b

the crystallographic data, coordination numbers and distances of the Na2 MoO4 ·2H2 O and MoS2 standard compounds, which were used to extract the experimental phases and amplitude functions of the samples, are also collected in Table 3. The errors for the fitted parameters corrected as described by Stern et al. [28] are estimated to be 20% in coordination number N, 1% in distance R, 10% in the Debye–Waller factor 1σ 2 , and 10% in 1E0 . Since the backscattering amplitudes contain an unknown static and thermal disorder and a damping due to photoelectron losses in the shells, the values of the disorder parameter 1σ 2 reported for the samples are given relative to those for standard compounds. The moderate 1σ 2 values for the rhodium-modified samples indicate a locally ordered MoS2 structure, but obviously a higher disorder than that for the MoS2 standard compound due to the decrease in the average particle size of MoS2 slabs in the samples.

3.6. Performance of the samples used for alcohol synthesis The results obtained from CO hydrogenation over the sulfided Rh–Mo–K/Al2 O3 catalysts after initial 48 h are presented in Table 4. Over the rhodium-free sample the activity toward alcohol formation is rather low, methanol being the main product. With rhodium modification, the activity for alcohol formation remarkably increases without noticeable increase of hydrocarbon yields and selectivity to the formation of C2+ alcohols also improves. This suggests that the incorporation of rhodium into the samples promotes the conversion of synthesis gas to higher alcohols. The activity for alcohol formation depends strongly upon space velocity and reaction pressure. The results obtained over the sulfide Rh(0.5%)–Mo–K/Al2 O3 sample at a high pressure of 10.0 MPa with different space velocities after initial 48 h are given in Table 5. It can

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

195

Table 5 Alcohol synthesis on Rh(0.5%)–Mo–K/Al2 O3 at different space velocitiesa GHSV (h−1 )

4800 9600 14400 a b

Conv. (%CO)

11.1 6.4 5.5

Alc.Sel.b (%)

59 71 83

Selectivity to alcohols (%) MeOH

EtOH

PrOH

11 20 26

19 20 28

17 18 19

MeOH/C2+ OH

Alc.STYb (ml/l h)

0.24 0.39 0.47

174 249 389

BuOH 12 13 9

Reaction conditions: 10.0 Mpa, 623 K, H2 /CO = 2. See footnotes in Table 4.

be seen from this table that with increasing space velocity both selectivity and yield of alcohols improve drastically. Moreover, in comparison with the results obtained on the same sample at lower pressure (Table 4), the yield of alcohols is also obviously increased. A durability test of the Rh(0.5wt%)–Mo–K/Al2 O3 sample was conducted for 100 h. The activity vs. reaction time is illustrated in Fig. 8. During the induction period of the reaction, the activity and selectivity to alcohol products are very low, but gradually increase with time on stream and reach a steady state after about 48 h. In contrast, the yield of hydrocarbons

decreases with time and levels off after the induction period. This may imply that active sites of the catalysts for alcohol synthesis are created during the reaction and some of them are converted from the active sites for hydrocarbon synthesis by reactant modification. A study of reaction stability on K–Mo/Al2 O3 catalyst found that some of the sulfur species including those coordinated with Mo atoms are lost during the catalytic reaction process. The Mo–S and Mo–Mo coordination number for the post-reaction sample decrease notably compared to those for the fresh sulfided sample and some highly coordinated unsaturated sites are formed which may be favorable for the formation of catalytic active centers for the synthesis of mixed alcohols [29]. From Fig. 8, it can also be found that the Rh–Mo–K/Al2 O3 samples are thermally stable under reaction conditions.

4. Discussion 4.1. Effects of co-existence of cationic and metallic rhodium and dispersion of rhodium upon catalytic properties for alcohol synthesis

Fig. 8. Durability test of CO hydrogenation over the sulfided Rh (0.5%)–Mo–K/Al2 O3 sample. Reaction conditions: 4.0 Mpa, 623 K, 4800 h−1 and H2 /CO=2.

As described by TPR results, with the incorporation of rhodium into the samples, the reduction of the molybdenum species becomes easier, but the rhodium species is more difficult to reduce than that in Rh/Al2 O3 . XPS results also indicate that after sulfidation a large fraction of the incorporated rhodium in the samples remains in a cationic state. Van Den Berg et al. suggested that oxidic rhodium species may partly cover the surface of molybdenum species to form mixed oxides containing Mo–O–Rh bonds which cannot be easily eliminated by the reduction in H2 at 773 K [14]. It is definite that the cationic rhodium species can be stabilized in the Mo–K/Al2 O3

196

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

catalysts by the interaction of rhodium with molybdenum species. Storm et al. [10] reported that for the reduced Rh/Co/Mo/K/Al2 O3 catalysts the existence of the cationic rhodium species interacting with the molybdenum species is favorable to the formation of C2+ alcohols. Sudhakar et al. [16] and Foley et al. [17] also pointed out that the interaction of rhodium with molybdenum affects the status of the rhodium species and thus CO adsorption. They suggested that the formation of alcohols on rhodium catalysts can be catalyzed by the sites which are less electron-rich than those needed for hydrocarbons, and more of these electron-poor sites can be created by the interaction of rhodium with molybdenum. Van Den Berg et al. [14] investigated the product distribution of CO hydrogenation over a rhodium-containing catalyst, in which the cationic and metallic rhodium species co-exist. The presence of cationic rhodium species decreases the heat of CO chemisorption and thus increases the concentration of surface molecular CO, a species beneficial to the formation of C2+ alcohols. They proposed that hydrocarbon products are formed via CO dissociation and hydrogenation of surface carbon over metallic rhodium species, methanol is formed via CO activation and hydrogenation over cationic rhodium species, and C2+ alcohols are formed via a dual-site mechanism, i.e. alkyl groups are formed on metallic rhodium species and subsequently migrate to rhodium cations where they are inserted into an undissociated CO molecule to produce oxygen-containing products. The reaction results shown in Table 4 demonstrate that the activity and selectivity to the formation of C2+ alcohols improve significantly by the addition of rhodium to the samples. A similar reaction pathway for C2+ alcohol formation as proposed by Van Den Berg et al. may be operative, i.e. a part of alkyl intermediates formed on the molybdenum and metallic rhodium species migrate to the cationic rhodium to produce C2+ alcohols. XPS results also reveal that, upon sulfidation, the rhodium species is enriched on the surface of the catalysts, as indicated by the surface atomic Rh/Mo ratios (Table 2). Therefore, the co-existence of cationic and metallic rhodium species in the catalysts may regulate the selectivity to the formation of alcohols on the rhodium-modified samples. The dispersion of the incorporated rhodium species may also be an important issue related to the proper-

ties of the catalysts for alcohol synthesis. Yao et al. investigated Rh/Al2 O3 catalysts and found that, upon heat treatment of the samples, the aggregated Rh2 O3 species, which could be detected by XRD measurements, is formed [22]. As mentioned above, however, no XRD peaks arising from the rhodium species incorporated into the Mo–K/Al2 O3 samples are observed. This suggests that rhodium aggregation is suppressed by the interaction between rhodium and the K–Mo species. As the data in Table 4 show, the yield of hydrocarbons over the Rh–Mo–K/Al2 O3 samples is much lower than that obtained with the rhodium-free sample. It may also indicate that the supported rhodium species are highly dispersed, since the formation of hydrocarbons needs large rhodium clusters [30]. The results of Arakawa et al. [31] also indicate that selectivity to oxygenates (e.g. ethanol) increases with increasing dispersion of rhodium. Therefore, high dispersion of the rhodium species caused by Rh–Mo interaction may also be an important factor that is responsible for the formation of alcohols.

4.2. Effect of microstructures of the supported sulfided molybdenum species upon catalytic properties for alcohol synthesis The results of LRS and EXAFS reveal that the molybdenum species in the sulfided samples exist mainly in the form of MoS2 -like crystallites (Figs. 3 and 6). Topsøe et al. [32] claimed that the MoS2 crystallites supported on alumina can be present as large patches of a wrinkled, one slab thick MoS2 -like layer, which are stabilized through the weak van der Waals interaction of the basal planes of MoS2 particles with the surface of the support. For the sulfided sample free of rhodium, this might be the case, since the size of the slabs is rather large as indicated by the Mo–Mo coordination numbers (Table 3). Based upon the results obtained from HRTEM measurements, Hayden et al. [33] reported that on alumina support MoS2 crystallites occur in the form of platelets with a ratio of height-to-width between 0.4 and 0.7, and some of them are oriented with their basal planes parallel to the Al2 O3 surface and some oriented at a non-zero angle to the surface. A similar scheme was also visualized by Prins et al. [34]. They suggested that the orientation of the MoS2 crystallites on

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

alumina depends upon their size, i.e. the crystallites grow on the support with large dimensions parallel and small dimensions perpendicular to the basal sulfur planes. For the rhodium-modified samples, the formed MoS2 crystallites may be small enough to interact with the support with their basal planes oriented perpendicular to the support surface. This point may also be interpreted by the fact that the contribution of Mo–O coordination increases with increasing the rhodium loading (Table 3). Based upon the suggestion of Prins et al. [34], the supported oxo-molybdenum species should be fully sulfided under the present conditions used for the sulfidation of the catalysts. In such a case, the contribution of Mo–O coordination observed for the samples may not arise from the oxygen atoms merely connected to molybdenum atoms due to incomplete sulfidation of the samples, but from the interaction of the molybdenum atoms with the surface oxygen atoms of the alumina support, i.e. from the bonding of Mo–O–Al. That would imply that the above-mentioned weak van der Waals interaction of the MoS2 crystallites with the alumina support may not be an appropriate interpretation to the stabilization of the sulfided molybdenum species on the support in the case with the rhodium-modified samples. The tiny sulfided molybdenum species may be stabilized mainly through their edge planes directly interacting with the surface oxygen atoms of the support, due to the favorable bonding between the MoS2 edge planes and the support [33]. If the sulfided molybdenum species interact with the support through their basal planes, instead of the edge planes, the average Mo–S coordination would not change much with decreasing slab size and the contribution of Mo–O coordination would not be so pronounced (Table 3). Therefore, the MoS2 -like crystallites are present as thin, hexagonally shaped slabs with truncated edges. The high-energy edge planes contacting the gas phase, presumably active in the catalytic reactions of alcohol synthesis, can be assigned as the ¨ set of planes, while the truncated edge planes (1010) bonded to the surface of alumina can be assigned as ¨ planes, due to the finest geometrical fit the (21¨ 10) with the most stable surface of alumina (110) [33]. Since the size of the sulfided molybdenum species in the sample may not be uniform, the small crystallites in the rhodium-free sample may also be stabilized through the bonding of Mo–O–Al which gives rise

197

to the minor contribution of Mo–O coordination as shown in Table 3. On the other hand, many authors proposed that the catalytic active sites on sulfided molybdenum-based catalysts are due to the so-called coordinately unsaturated molybdenum (Mo(CUS)) sites, on which chemisorption of probe molecules can occur [34–36]. They are formed on the surface of MoS2 crystallites during sulfidation or even induced by the stream of the reactants. At a reaction temperature of 573 K, H2 molecules adsorbed on the Mo(CUS) sites possibly dissociate and then, by spillover or other ways, migrate to the basal planes of the MoS2 crystallites, where they bond with sulfur atoms to form SH species. A multilayer growth of MoS2 crystallites perpendicular to the basal planes is unfavorable to H2 adsorption. For the oxidic Rh–Mo–K/Al2 O3 samples, the XRD and LRS results clearly demonstrate that the incorporation of rhodium improves the dispersion of the oxo-molybdenum species (Figs. 1 and 2). The sulfidation of the species gives rise to a dispersion of the supported MoS2 crystallites higher than that in the rhodium-free sample, as also revealed by LRS and EXAFS results (Figs. 3 and 6). Accordingly, the average size of the MoS2 crystallites is reduced, which will facilitate the creation of Mo(CUS) sites. In addition, as a result of reduction in the size of the MoS2 crystallites, migration of the adsorbed H2 molecules from the Mo(CUS) sites to the MoS2 surface to form the SH species becomes easier, which may be responsible for the high selectivity to the formation of C2+ alcohols (Table 3).

5. Conclusions With the incorporation of rhodium into the oxidic Mo–K/Al2 O3 samples, a strong interaction of the rhodium modifier with the oxo-molybdenum components occurs. This improves the sulfidability and reducibility of the supported molybdenum species. After sulfidation of the samples, the cationic and metallic rhodium species co-exist due to the interaction of rhodium with the sulfided molybdenum species. The sulfided molybdenum species in the rhodiumfree sample are present as large patches of MoS2 -like slabs which may be oriented with their basal planes parallel to the surface of the support. With the modifi-

198

Z.-r. Li et al. / Applied Catalysis A: General 187 (1999) 187–198

cation of rhodium to the catalysts, the sulfided molybdenum species becomes highly dispersed. The interaction of rhodium with the molybdenum species may cause the basal planes of the MoS2 -like species to become oriented perpendicular to the support surface due to the favorable bonding between the MoS2 edge planes and the support. The activity for alcohol synthesis over the rhodium-modified catalysts is much higher than that obtained over the rhodium-free activity and increases with increasing rhodium loading, which may result from the appearance of the more catalytically active surfaces or sites modified by the rhodium species. The high selectivity to the formation of C2+ alcohols obtained with rhodium-modified samples is most probably due to the co-existence of cationic and metallic rhodium species stabilized by the interaction of rhodium with the molybdenum species.

Acknowledgements This work was supported by National Natural Science Foundation of China (No. 29773042). The experimental facility of EXAFS was supplied by BSRF. We are also grateful to Drs. Tian-dou Hu, Tao Liu, Ming Meng and Guo-zhu Bian for help in the EXAFS measurements and to Professor Kun-quan Lu for supplying the EXAFS analysis program. References [1] C.B. Murchison, M.M. Conway, R.R. Stevens, G.J. Quarderer, in: M.J. Phillips, M. Ternan (Eds), Proceedings of 9th International Congress on Catalysis, vol. 2, Calgary 1988, Chemical Institute of Canada, Ottawa, p.626. [2] J.G. Santiesteban, C.E. Boogdan, R.G. Herman, K. Klier, K., in: M.J. Phillips, M. Ternan (Eds), Proceedings of 9th International Congress on Catalysis, vol. 2, Calgary 1988, Chemical Institute of Canada, Ottawa. [3] T. Tatsumi, A. Muramatsu, T. Fulcunaga, H. Tominga, in: M.J. Phillips, M. Ternan (Eds), Proceedings of 9th International Congress on Catalysis, vol. 2, Calgary 1988, Chemical Institute of Canada, Ottawa. [4] H.B. Zhang, Y.Q. Yang, H.P. Huang, C.D. Lin, K.R. Tsai, Preprint and Abstract Book of 10th International Congress on Catalysis, Budapest, Hungary, 19–24 July 1992, p.253.

[5] Y. Avila, C. Kappenstain, S. Pronier, J. Barrault, Appl. Catal. A. 132 (1995) 97. [6] J. Zhang, Y. Wang, L. Chang, Ibid 126 (1995) 205. [7] Y.L. Fu, K. Fujimoto, P.Y. Lin, K. Omata, Y. Yu, Ibid 126 (1995) 273. [8] K. Fujimoto, T. Oba, Appl. Catal. 13 (1985) 289. [9] G.Z. Bian, Y.L. Fu, Y.S. Ma, Acta Physico-Sinca 12 (1996) 406. [10] D.A. Storm, Topics in Catalysis 2 (1995) 91. [11] S.F. Wong, D.A. Storm, M.S. Patel, U.S. Patents 4,980,380 and 4,983,638 (1991). [12] D.A. Storm, F. P Mertens, M.C. Cataldo, E.P. DaCanio, J. Catal. 141 (1993) 478. [13] E.C. DeCanio, D.A. Storm, J. Catal. 132 (1991) 375. [14] F.G.A. Van Den Berg, J.H.E. Glezer, W.M.H. Sachtler, J. Catal. 93 (1985) 340. [15] E.E. Lowenthal, S. Schwarz, H.C. Foley, J. Catal. 156 (1995) 96. [16] C. Sudhakar, N.A. Bhore, K.B. Bischoff, W.H. Manogue, G.A. Mills, Proceedings of 10th Meeting of the Catalysis Society of North America, San Diego, CA., May 1987. [17] H.C. Foley, A.J. Hong, L.F. Allard, A.J. Garratt-Reed, Appl.Catal. 61 (1990) 351. [18] D.E. Sayers, B.A. Bunker, In D.C. Koningsberger, R. Prins (Eds), X-ray Absorption, Principles, Applications, Techniques of EXAFS, SEXAFS and XANES, Wiley, 1988, p.211. [19] G.L. Schrader, C.P. Cheng, J. Catal. 80 (1983) 369. [20] M. Jiang, G.Z. Bian, Y.L. Fu, J. Catal. 146 (1994) 144. [21] C. Wong, R.W. McCabe, J. Catal. 107 (1987) 535. [22] H.C. Yao, J. Catal. 70 (1981) 440. [23] H.C. Yao, S. Japer, M. Shelef, J.Catal. 50 (1977) 407. [24] C.V. Caceres, J.L.G. Fierro, A. Lopez, J. Catal. 95 (1985) 501. [25] C.D. Wangner, W.M. Riggs, L.E. Davis et al., Handbook of X-ray Photoelectron Spectroscopy, Perkin-Elmer, Norwalk, CT, 1979. [26] J.T. Beng, L.S. Steven, J. Catal. 140 (1993) 564. [27] P.A. Specvack, S. Mcintyre, Appl. Catal. 64 (1990) 191. [28] E. Stern, M. Newville, B. Ravel, Y. Yacoby, D. Haskel, Physica B. 208/209 (1995) 117. [29] G.Z. Bian, Y.L. Fu, M. Yamada, Appl. Catal. A 144 (1996) 79. [30] M. Araki, V. Ponec, J. Catal. 44 (1976) 439. [31] H. Arakawa, T. Fukushima, M. Ichikawa, Chem. Lett. 163 (1985) 881. [32] H. Topsøe, B.S. Clausen, Catal. Rev.-Sci. Eng. 26 (1984) 395. [33] T.F. Hayden, J.A. Dumesic, J. Catal. 105 (1987) 299. [34] R. Prins, V.H.J. De Beer, G.A. Somorjai, Catal. Rev.-Sci 31(1&2) (1989) 1. [35] J. Valyon, W.K. Hall, J. Catal. 84 (1983) 216. [36] M.I. Zaki, B. Vielhaber, H. Knoezinger, J. Phys. Chem. 90 (1986) 3176.