Partial oxidation of propane to syngas over nickel supported catalysts modified by alkali metal oxides and rare-earth metal oxides

Applied Catalysis A: General 211 (2001) 145–152

Partial oxidation of propane to syngas over nickel supported catalysts modified by alkali metal oxides and rare-earth metal oxides Shenglin Liu a,∗ , Longya Xu a,b , Sujuan Xie a , Qingxia Wang a , Guoxing Xiong b a

Group 804, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, P.O. Box 110, Dalian 116023, PR China State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, CAS, P.O. Box 110, Dalian 116023, PR China

b

Received 21 June 2000; received in revised form 8 August 2000; accepted 7 October 2000

Abstract The reaction performance, catalyst acidity and basicity property, and carbon-deposition of various nickel-supported catalysts for partial oxidation of propane (POP) to syngas were investigated with a flow-reactor, FTIR, TG and UVRRS analyses. The NiO/␥-Al2 O3 catalyst is the most suitable for the POP reaction among NiO/␥-Al2 O3 , NiO/MgO and NiO/SiO2 . And the reaction performance of the NiO/␥-Al2 O3 shows little difference from those of the nickel-supported catalysts modified by alkali metal oxides and rare-earth metal oxides. However, modification with alkali metal oxide Li2 O and rare-earth metal oxide La2 O3 can reduce the Lewis acidity intensity of the NiO/␥-Al2 O3 and enhance its ability to suppress carbon-deposition during the POP reaction. The carbon-deposition contains graphite-like species that were detected by UVRRS. The nickel-supported catalysts modified by alkali metal oxides and rare-earth metal oxides possess good reaction performance and carbon-deposition resistance. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Partial oxidation of propane; Singes; Performance; Nickel-supported catalysts; Modification; Carbon-deposition

1. Introduction There are abundant supplies of mixture gases containing CH4 , C2 H6 , C3 H8 and C4 H10 , etc. from FCC (fluidized catalytic cracking) tail gas and refinery gas. Commonly, the mixture gases are primarily combusted to carbon dioxide, because the complete separation of CH4 , C2 H6 , C3 H8 and/or C4 H10 , from the mixture gases may not be economical. Provided that syngas (CO + H2 ) could be produced from the mixture gases over nickel-supported catalysts with high selectivity and conversion, it can be directly obtained from the mixture gases from FCC tail gas and refinery gas. This ∗ Corresponding author. E-mail address: [email protected] (S. Liu).

may lead to better utilization of the light fractions from FCC tail gas and refineries. The partial oxidation of methane to syngas has received intensive attention [1,2], and partial oxidation of ethane to syngas over nickel-based catalysts has also been investigated [3]. Schmidt et al. [4,5] investigated the partial oxidation of alkanes to syngas ranging from 1073 to1473 K over noble metal-coated monoliths at residence times between 10−3 and 10−2 s, and reported that syngas can be produced from CH4 , C2 H6 , C3 H8 , and n-C4 H10 over a supported Rh catalyst with high selectivity and conversion, and the presence of C2 H6 in natural gas will not lead to catalyst deactivation by carbon-deposition. C3 H8 oxidation over Rh catalyst has led to carbon deposition, but only in severely fuel-rich regimes. Jovanovic and Stankovic

0926-860X/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 0 ) 0 0 8 6 5 - 6

146

S. Liu et al. / Applied Catalysis A: General 211 (2001) 145–152

[6,7] indicated the highest selectivity for the reaction (2C3 H8 + 3O2 = 6CO + 8H2 ) at temperatures 1123 and 1173 K, was obtained when the catalyst was prepared by sixfold immersion in the solution of 0.75 mol Ni dm−3 , and the smaller Ni crystallites were more selective to CO and H2 formation than the larger ones. Previously, we reported partial oxidation of methane and ethane to syngas over the nickel-based catalysts modified by alkali metal oxide and rare-earth metal oxide, and pointed out that ABNiO/␥-Al2 O3 (A = Li, Na, K; B = La, Sm, Ce, Y) were excellent POM and POE reaction catalysts [2,3,8,9]. On the basis of these results, the partial oxidations of propane to syngas over nickel-supported catalysts were investigated. One of the aims for this investigation on the POP reaction is to search for a catalyst that is not only suitable for the POM and POE reactions, but also for the POP reaction, hence, enabling mixture gases containing CH4 , C2 H6 , C3 H8 and C4 H10 from FCC tail gas to be converted to syngas with high conversion and selectivity. In the present work, the POP reaction performance, catalyst acidity and basicity property, and carbon-deposition over the nickel-supported catalysts modified by alkali metal oxides and rare-earth metal oxides are discussed in detail.

2. Experimental 2.1. Preparation of catalysts and test of reaction performance NiO/␥-Al2 O3 , NiO/SiO2 and NiO/MgO catalysts were prepared by impregnating with an appropriate amount of Ni(NO3 )2 on ␥-Al2 O3 , SiO2 and MgO, respectively, for 24 h. Catalysts were dried at 393 K and then calcined in air at 823–1073 K for 4 h. Preparation of the ABCO/␥-Al2 O3 (A = Li, Na, K; B = La, Sm, Ce, Y; C = Fe, Co, Ni) catalysts has been described previously [2]. Catalysts were investigated in an atmospheric pressure fixed-bed flow microreactor. Reaction performance was tested using a microreactor with an internal diameter of 6 mm, and 100 mg of catalyst with an average particle size of 0.37–0.25 mm was employed, giving a catalyst bed length of 7 mm. An EU-2 type thermocouple fixed with the quartz reactor was placed at the entry of the catalyst bed to control

the electric furnace temperature, which was taken as the reaction temperature. After the catalyst was in situ reduced in H2 (20 ml min−1 ) at 1123 K for 1 h, it was cooled to 873 K in a flowing of Ar. The catalyst was then used the POP reaction. Products of the reaction were analyzed by gas chromatography using a TCD detector. The conversion and the selectivity were calculated on the basis of carbon numbers of the propane reacted. The amount of H2 was corrected by the external standard method. 2.2. Characterizations of catalysts IR spectra were recorded on a Perkin-Elmer 580 double-beam FT-IR spectrometer equipped with a liquid nitrogen-cooled mercury-calcium-telluride (MCT) detector. All IR spectra of adsorbed species are difference spectra obtained by subtracting the background spectra at the corresponding temperatures. The procedure for the measurement of the pyridine is as follows. The sample was outgased under a vacuum of 3 × 10−5 Torr for 2 h at 723 K. Then the temperature was slowly decreased to 298 K followed by the admission of pyridine. TG tests were recorded and treated by a PerkinElmer 3600 work-station at a programmed temperature velocity of 10 K min−1 in air, with the flow rate of 25 ml min−1 . UV resonance Raman spectra (UVRRS) characterization was performed in air at room temperature, using an UV resonance Raman spectroscope. The ultraviolet laser beam for exciting UV Raman spectra was generated by frequency doubling of the 488 nm output of an Ar+ ion laser to 244 nm using a BBO crystal. The Raman scattering from the sample surface was collected by an AlMgF2 coated ellipsoidal reflector using a back-scattering geometry, and then focused into a 0.32 nm single grating spectrograph through a notch filter. 3. Results and discussion 3.1. Performance of the LiLaNiO/γ -Al2 O3 catalyst for the POP reaction The POM and POE reactions over the nickelsupported catalysts were investigated under the high temperatures ranging from 973 to 1123 K [2,3], so

S. Liu et al. / Applied Catalysis A: General 211 (2001) 145–152

147

Fig. 1. Performance of LiLaNiO/␥-Al2 O3 catalyst at different temperatures (O2 /C3 H8 /Ar = 1.65/1/5.3, SV = 5.7 × 104 h−1 ).

we tried to carry out the POP reaction at the same high temperatures. The results indicate that, as the temperature ranges from 923 to 1123 K, the gas phase takes place obviously above the catalyst and on the reactor, and there exists a large amount of coke from the thermal splitting decomposition of propane. The coke will block the reactor and cause the reaction

not to continue in progress through a suitable time. While the reaction temperature is below 873 K, little gas phase occurs. Thus, the performance of the POP reaction was carried out below 873 K. The effect of reaction temperature on the performance of the LiLaNiO/␥-Al2 O3 is shown in Fig. 1. Propane and oxygen are almost converted completely

Fig. 2. Influence of O2 /C3 H8 ratio on the performance of LiLaNiO/␥-Al2 O3 catalyst (at 873 K).

148

S. Liu et al. / Applied Catalysis A: General 211 (2001) 145–152

Fig. 3. Reaction performance of LiLaNiO/␥-Al2 O3 catalyst at different space velocities (at 873 K).

(not shown). When the reaction temperature is subsequently increased from 773 to 873 K, the CO and H2 selectivities increase, while the CH4 and CO2 selectivities. And the H2 /CO ratio decrease. Under the constant space velocity by keeping the flow rate of C3 H8 (12 ml min−1 ) and the total flow rate (95 ml min−1 ) of O2 and Ar constant, the respective flow rates of O2 and Ar are changed to obtain different O2 /C3 H8 ratios. The effect of the O2 /C3 H8 ratio on the performance of the LiLaNiO/␥-Al2 O3 is investigated (Fig. 2). As the O2 /C3 H8 ratio changes from 1.26 to 3.57, the CO, CH4 and H2 selectivities decreased, and the H2 /CO ratio and CO2 selectivity increase. Under 873 K and O2 /C3 H8 /Ar ratio of 1.65/1/5.3, the influence of space velocity on the performance of the LiLaNiO/␥-Al2 O3 is also studied (Fig. 3). The results indicate that the influence of space velocity is not appreciable, e.g. the LiLaNiO/␥-Al2 O3 keeps good reaction performance with a wide range of space velocity.

3.2. Performance of catalysts with different supports and components for the POP reaction The performances of partial oxidation of propane to syngas over NiO/␥-Al2 O3 , NiO/MgO and NiO/SiO2 were compared. The results are shown in Table 1. There are strikingly different performances when MgO or SiO2 displaces the ␥-Al2 O3 support in the NiO/␥-Al2 O3 catalyst. C3 H8 and O2 conversions, and H2 selectivity of the MgO or SiO2 supported nickel catalysts are much lower than that of the NiO/␥-Al2 O3 catalyst, and no CH4 appears over the NiO/MgO or NiO/SiO2 catalyst. Previously, we carried out the partial oxidation of ethane (POE) reaction over the same catalysts [3]. The results are shown in Table 2. Ethane and oxygen are almost converted completely (not shown), and CO selectivities under the same reaction conditions are as follows: NiO/␥-Al2 O3 > NiO/MgO > NiO/SiO2 . Some C2 H4 is produced

Table 1 Comparison of performances of different catalysts for POP reactiona Catalyst

C3 H8 conversion (%)

O2 conversion (%)

CO selectivity (%)

CH4 selectivity (%)

CO2 selectivity (%)

H2 selectivity (%)

NiO/␥-Al2 O3 NiO/MgO NiO/SiO2

100 41.3 5.8

100 79.2 16.5

42.6 45.4 13.4

27.1 0 0

30.3 54.6 86.6

80.6 57.8 7.0

a

O2 /C3 H8 /Ar = 1.65/1/5.3; SV = 5.7 × 104 h−1 ; T = 873 K.

S. Liu et al. / Applied Catalysis A: General 211 (2001) 145–152

149

Table 2 Comparison of performances of different catalysts for POE reactiona Catalyst

CO selectivity (%)

CH4 selectivity (%)

CO2 selectivity (%)

C2 H2 selectivity (%)

NiO/␥-Al2 O3 NiO/MgO NiO/SiO2

89.0 87.3 72.6

0.2 2.0 11.8

10.8 10.7 11.5

0 0 4.1

a

O2 /C2 H6 /He = 1.15/1/4; SV = 3 × 105 h−1 ; T = 1123 K.

over the NiO/SiO2 catalyst. Xiong et al. [9,10] investigated the performance of the partial oxidation of methane (POM) reaction over the same catalysts (Table 3). The results show that the sequence of CH4 conversions is as follows: NiO/␥-Al2 O3 > NiO/SiO2  NiO/MgO, while that of the CO selectivities as follows: NiO/SiO2 > NiO/␥-Al2 O3  NiO/MgO. Some C2 H4 is produced over NiO/MgO (for POM), but not over NiO/SiO2 (for POE). Xiong et al. claimed that the acidic property of catalyst favors keeping the reduced nickel and that the reduced nickel is necessary for the POM reaction, whereas the basic property favors keeping the oxidized nickel and the oxidized nickel is necessary for the oxidative coupling of methane to C2 hydrocarbons (OCM). According to the above results, the reaction performances of POE and POP are not directly related to the catalyst acid–base property. Even for the POE and POP reactions, there exist some differences, for example, some C2 H4 is produced over the NiO/SiO2 catalyst for the POE reaction (Table 2), whereas no C3 H6 appears over the same catalyst for the POP reaction (Table 1), etc. It can be concluded that the behavior of the POP reaction is different from those of the POE and POM reactions over the three different nickel-supported catalysts, and the activated behavior of propane is not the same as those of methane and ethane. So it is necessary to study the reaction of partial oxidation of propane over the nickel supported catalysts.

A series of ABCO/␥-Al2 O3 (A = Li, Na, K; B = La, Sm, Ce, Y; C = Fe, Co, Ni) catalysts were prepared with the same preparation method and condition in order to investigate the action of different components. The results are presented in Table 4. There are strikingly different performances when Fe or Co displaces the Ni component in the LiLaNiO/␥-Al2 O3 catalyst. C3 H8 conversion and H2 selectivity of the Fe or Co-containing catalyst are much lower than that of the Ni-containing catalyst, and no CH4 appeared over the LiLaFeO/␥-Al2 O3 or LiLaCoO/␥-Al2 O3 . Previously, Xiong et al. [2] investigated the POM reaction over the same catalysts. The results show that there are also strikingly different performances when Fe or Co displaces the Ni component in the LiLaNiO/␥-Al2 O3 catalyst for the POM reaction. Zhang et al. [11] have done comparative studies of the POM catalytic performances of the first series transition metals (Mn, Fe, Co, Ni, Cu). They claimed that the different performances of those transition metals containing catalysts were due to their different performance redox ability determined by their different electric potentials. As far as the LiCLaO/␥-Al2 O3 (C = Ni, Fe, CO) catalysts for the POM reaction, Xiong et al. [2] reported, except for their different standard inherent electric potentials, the interaction between transition metal and Al2 O3 support was one of the most important factors that determined the redox ability. Here, the different behavior of Fe or Co from that of Ni com-

Table 3 Comparison of performances of different catalysts for POM reactiona Catalyst

CH4 conversion (%)

O2 conversion (%)

C2 selectivity (%)

CO selectivity (%)

CO2 selectivity (%)

NiO/␥-Al2 O3 NiO/MgO NiO/SiO2

88.7 38.6 76.4

99.7 99.8 99.8

0 19.8 0

99.1 31.1 99.6

0.9 49.1 0.4

a

O2 /CH4 = 1/2; SV = 2.7 × 103 h−1 ; T = 1123 K.

150

S. Liu et al. / Applied Catalysis A: General 211 (2001) 145–152

Table 4 Comparison of performances of different catalysts for POP reactiona Catalyst

C3 H8 conversion (%)

O2 conversion CO selectivity (%) (%)

CH4 selectivity (%)

CO2 selectivity (%)

H2 selectivity (%)

H2 /CO (ration)

NiO/␥-Al2 O3 LiNiLaO/␥-Al2 O3 LiCoLaO/␥-Al2 O3 LiFeLaO/␥-Al2 O3 NaNiLaO/␥-Al2 O3 KNiLaO/␥-Al2 O3 LiNiCeO/␥-Al2 O3 LiNiYO/␥-Al2 O3 LiNiSmO/␥-Al2 O3

100 100 57.1 35.4 100 100 100 100 100

100 100 91.7 92.9 100 100 100 100 100

27.1 29.5 0 0 27.7 27.4 27.1 27.3 26.5

30.3 25.0 50.1 88.1 27.0 28.4 28.4 28.2 27.0

80.6 89.3 67.8 5.1 86.4 84.5 88.0 91.1 80.5

1.5 1.5 1.8 0.6 1.5 1.5 1.6 1.6 1.4

a

42.6 45.5 49.9 11.9 45.3 44.2 44.5 44.5 46.5

O2 /C3 H8 /Ar = 1.65/1/5.3; SV = 5.7 × 104 h−1 ; T = 873 K.

ponent in the LiLaNiO/␥-Al2 O3 catalyst for the POP reaction may possibly be attributed to the differences in their structure of d-orbitals, ability to remove hydrogen, redox ability and the interaction between transition metal and Al2 O3 support. The work is still underway. Nickel is the most suitable for the POP reaction according to the results listed in Table 4. The reaction performance of the NiO/␥-Al2 O3 shows little difference from those of the catalysts with different alkali metal oxides ALaNiO/␥-Al2 O3 (A = Li, Na, K) and rare-earth metal oxides LiBNiO/␥-Al2 O3 (B = La, Sm, Y, Ce). Propane and oxygen are almost converted completely, the selectivities of CO and H2 , H2 /CO ratio remain at about 45, 85 and 1.5%, respectively. The results demonstrate that the modification with alkali and rare-earth metal oxides does not significantly influence the reaction performance of the POP reaction over the NiO/␥-Al2 O3 under these conditions. This is in agreement with those of POM and POE reactions [2,3]. However, the modification can obviously affect the thermal stability of ␥-Al2 O3 and the sintering of nickel in the NiO/␥-Al2 O3 [12–14]. 3.3. Comparative study of the catalyst acidity and basicity property, and the carbon deposition between NiO/γ -Al2 O3 and LiLaNiO/γ -Al2 O3 catalysts NiO/␥-Al2 O3 and LiLaNiO/␥-Al2 O3 catalysts were prepared with the same preparation methods and conditions in order to investigate the effects of alkali metal oxide Li2 O and rare-earth metal oxide La2 O3 on

the catalyst acidity and basicity property, and on the carbon-deposition of the POP reaction. Comparison of the catalyst acidity intensity between NiO/␥-Al2 O3 and LiLaNiO/␥-Al2 O3 is shown in Fig. 4. No evidence is found for a band at 1540 cm−1 on the three samples indicating that there are no Bronstred acidity sites on the surface strong enough to react with pyridine. Only the band at 1450 cm−1 appears on the three samples, indicating that there are Lewis acidity sites [15]. And the Lewis acidity intensity of the LiLaNiO/␥-Al2 O3 is much lower than that of the NiO/␥-Al2 O3 . The results show that the addition of Li and La can reduce the acidity intensity of the NiO/␥-Al2 O3 . The introduction of Li and La not only improves the thermal stability of ␥-Al2 O3 and the acidity intensity of the NiO/␥-Al2 O3 catalyst, but more importantly, the addition of Li and La enhances the ability of carbon-deposition resistance during the POP reaction. The deposition of surface carbon over the NiO/Al2 O3 catalyst during the POM reaction was undesirable and resulted in the deactivation of the NiO/Al2 O3 [16]. A hydrocarbon with higher C/H ratios favors carbon-deposition on the surface of a catalyst in comparison with the results for hydrocarbons with lower C/H ratios. Therefore, it is reasonable that the amount of carbon-deposition for the POP reaction is more than for the POM reaction over the NiO/Al2 O3 , and carbon-deposition also results in the deactivation of the NiO/Al2 O3 for the POP reaction. It is well known that the acidity of the catalyst surface favors carbon-deposition, while the basicity of the catalyst surface prevents carbon-deposition [17]. Because the acidity intensity of the LiLaNiO/␥-Al2 O3 is much

S. Liu et al. / Applied Catalysis A: General 211 (2001) 145–152

151

Fig. 4. FT-IR spectra of pyridine on samples: (a) ␥-Al2 O3 ; (b) NiO/␥-Al2 O3 ; (c) LiLaNiO/␥-Al2 O3 .

lower than that of the NiO/␥-Al2 O3 , the addition of Li and La should be beneficial to the prevention of carbon-deposition over the catalyst surface. The TG results of the samples after the POP reaction for 5 h indicate that the resistant ability for carbon-deposition of the LiLaNiO/␥-Al2 O3 is better than that of the NiO/␥-Al2 O3 . The amount of carbon-deposition over the LiLaNiO/␥-Al2 O3 is 29.1 wt.%, while that over the NiO/␥-Al2 O3 is 52.1 wt.%, below 873 K, with a C3 H8 /O2 /Ar ratio of 1/1.25/5.3 and a space velocity

of 5 × 104 h−1 . It is shown that the introduction of Li and La can obviously suppress carbon-deposition of the nickel-based catalysts. Xiong et al. [2,3] studied carbon-deposition of the nickel-based catalysts for the POM and POE reactions. They also reported that the incorporation of Li and La can improve the ability of carbon-deposition resistance of the NiO/␥Al2 O3 . In order to determine the carbon-deposition species formed on the catalysts, the samples which performed

Fig. 5. UV Raman spectra of carbon-depositions formed on catalysts: (a) LiLaNiO/␥-Al2 O3 ; (b) NiO/␥-Al2 O3 .

152

S. Liu et al. / Applied Catalysis A: General 211 (2001) 145–152

under the same reaction conditions were characterized by UVRRS, which has been demonstrated to be a powerful tool for catalysis and surface science studies, with the advantage of avoiding the surface fluorescence which frequently occurs in visible Raman spectra of many catalysts [18]. The samples used for the UVRRS were manipulated in air prior to the analysis. The results are shown in Fig. 5. Only one band at ∼1580 cm−1 is present clearly in the spectra, which is close to the characteristic band of graphite at 1575 cm−1 [19]. The intensity of the band for the NiO/␥-Al2 O3 was stronger than that for the LiLaNiO/␥-Al2 O3 . The results reveal that the carbon-deposition formed on the nickel-based catalysts contains graphite like species (amorphous forms of carbon may not be detected by UVRRS), and the unmodified NiO/␥-Al2 O3 catalyst favors the formation of carbon-deposition. Previously, we determined the carbon-deposition species which formed on the same catalysts for the POM and POE reactions. The UVRRS results indicate that the carbon-deposition formed on the same nickel-based catalysts also contains graphite like species at 1123 K [3,14].

Acknowledgements The authors would like to thank Prof. Can Li and Dr. Meijun Li for their help with UVRRS analysis and for beneficial discussions.

References [1] A.T. Ashcroft, A.K. Cheetham, J.S. Foord, M.L.H. Green, C.P. Grey, A.J. Murrell, Nature 344 (1990) 319. [2] Q. Miao, G.X. Xiong, S.S. Sheng, W. Cui, L. Xu, X.X. Guo, Appl. Catal. A 154 (1997) 17. [3] S.L. Liu, G.X. Xiong, W.S. Yang, L.Y. Xu, G. Xiong, C. Li, Catal. Lett. 63 (1999) 167. [4] M. Huff, P.M. Torniainen, D.A. Hichman, L.D. Schmidt, Stud. Surf. Sci. Catal. 81 (1994) 315. [5] M. Huff, P.M. Torniainen, L.D. Schmidt, Catal. Today 21 (1994) 113. [6] N.N. Jovanovic, M.V. Stankovic, Appl. Catal. A 30 (1987) 3. [7] N.N. Jovanovic, M.V. Stankovic, Stud. Surf. Sci. Catal. 110 (1997) 1145. [8] S.L. Liu, G.X. Xiong, S.S. Sheng, Q. Miao, W.S. Yang, Stud. Surf. Sci. Catal. 119 (1998) 747. [9] Q. Miao, G.X. Xiong, S.S. Sheng, W. Cui, X.X. Guo, Stud. Surf. Sci. Catal. 101 (1996) 453. [10] Q. Miao, Study of the oxidative transformation of methane over the nickel-based catalysts modified by alkali metal oxides and rare-earth oxides, Doctoral Dissertation, Dalian, 1995. [11] P. Chen, H. Zhang, G. Lin, K. Tsai, Chem. J. Chin. Univ. 114 (1995) 323. [12] S.L. Liu, G.X. Xiong, S.S. Sheng, React. Kinet. Catal. Lett. 68 (1999) 243. [13] S.L. Liu, G.X. Xiong, H. Dong, W.S. Yang, Appl. Catal. A 198 (2000) 261. [14] S.L. Liu, G.X. Xiong, H. Dong, W.S. Yang, Z.L. Yu, W. Chu, Stud. Surf. Sci. Catal. 130 (2000) 3567. [15] E.P. Parry, J. Catal. 2 (1963) 371. [16] D. Dissanayake, M.P. Rosynek, K.C.C. Kharas, J.H. Lunsford, J. Catal. 132 (1991) 117. [17] S.B. Tang, F.L. Qiu, S.J. Lu, Catal. Today 24 (1995) 253. [18] C. Li, P.C. Stair, Stud. Surf. Sci. Catal. 101 (1996) 881. [19] M. Nakamizo, Carbon 29 (1991) 257.