- Email: [email protected]
Steam reforming of methane over nickel catalysts at low reaction temperature
Applied Catalysis A: General 258 (2004) 107–114
Steam reforming of methane over nickel catalysts at low reaction temperature Yasuyuki Matsumura∗ , Toshie Nakamori Research Institute of Innovative Technology for the Earth, Kizu-cho, Soraku-gun, Kyoto 619-0292, Japan Received 23 April 2003; received in revised form 10 August 2003; accepted 21 August 2003
Abstract Effects of supports such as silica, ␥-alumina, and zirconia for nickel catalysts have been studied in steam reforming of methane at 500 ◦ C. The activity of nickel supported on silica reduced with hydrogen at 500 ◦ C decreases with oxidation of nickel particles by steam during the reaction. Nickel supported on ␥-alumina is not much reduced with hydrogen at 500 ◦ C and is inactive in the reforming at 500 ◦ C. However, the catalyst reduced at 700 ◦ C is fairly active while nickel is partially oxidized during the reaction. Nickel supported on zirconia is the most effective in the stream reforming at 500 ◦ C. In the initial stage of the reaction solely with methane at 500 ◦ C, surface hydroxyl groups on these catalysts react readily with methane to produce hydrogen and carbon dioxide; suggesting that the hydroxyl groups play an important role in the mechanism of the steam reforming to carbon dioxide. A significant quantity of water can be accumulated on the surface of zirconia-supported nickel in the reaction only with steam at 500 ◦ C, and this results in formation of significant quantities of hydrogen and carbon dioxide in the following reaction with methane. © 2003 Elsevier B.V. All rights reserved. Keywords: Steam reforming; Methane; Hydrogen production; Nickel catalyst; Surface hydroxyl group
1. Introduction Steam reforming of methane is employed for a large scale production of hydrogen and/or carbon monoxide. The reaction is industrially operated at a high temperature around 800 ◦ C over nickel-alumina based catalysts, because a reasonable conversion of methane is required in this endothermic process [1]. Such a high temperature causes some demerits, including the expensive tubular reformer made of high alloy nickel-chromium steel, irreversible carbon formation in the reactor, and large energy consumption. In order to reduce the reaction temperature, some membrane reactors have been proposed [2–8]. In the reactor, hydrogen is usually separated from the reaction mixture by a palladium-based membrane and the separation of hydrogen significantly increases the equilibrium conversion of methane. Uemiya et al. [4] reported methane conversion close to 90% at 500 ◦ C using this reactor. Since a high partial pressure of hydrogen in the reactor increases the efficiency of hydrogen separation, the reaction composition ∗
Corresponding author. Fax: +81-774-75-2318. E-mail address: [email protected] (Y. Matsumura).
0926-860X/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2003.08.009
should be close to the equilibrium all over the part of the membrane. In the above case, a commercial nickel catalyst was employed with the F/W of 0.5 dm3 h−1 g−1 [4], while much higher space velocity is required in an industrial process. Hence, high catalytic activity at a low reaction temperature around 500 ◦ C is required in a realistic membrane system. We have preliminarily studied effects of the supports on steam reforming over nickel catalysts at 500 ◦ C. Decomposition of methane on nickel surface is believed as a first step of the steam reforming of methane; then the carbon species formed on the surface react subsequently with steam or surface oxygen species [1]. To follow the reaction steps simply, we separately fed methane and steam to the catalysts and have analyzed the mechanism especially at the initial stage of the reaction in which surface active species participate highly. In this report, it will be shown that zirconium oxide, which can absorb water at 500 ◦ C, is an effective support and surface hydroxyl groups contribute significantly to the steam reforming at 500 ◦ C. Recently, some researches have been reported the activity of nickel supported on zirconium oxide to methane reforming above 700 ◦ C [9–11].
108
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
2. Experimental
3. Results
Nickel catalysts were prepared by an impregnation technique. Metal oxide supports such as ␥-alumina (ALO-4 supplied by Catalysis Society of Japan), silica (Fuji-Silicia, G-10), and zirconia (Daiichi Kigenso Kagaku Kogyo, RC-100) were impregnated with nickel nitrate (Wako, GR grade) by evaporation of the aqueous solution at 80 ◦ C. After drying in air at 110 ◦ C overnight, they were heated at 700 ◦ C for 3 h. The samples (Ni/Al2 O3 , Ni/SiO2 , and Ni/ZrO2 , respectively) contained 5–20 wt.% of nickel. Catalytic tests were performed in a fixed bed continuous flow reactor operated at atmospheric pressure. A catalyst (0.30 g) was placed in a tube reactor made of quartz glass (i.d.: 6 mm). After the pre-reduction in a stream of 10 vol.% hydrogen diluted with argon (6.0 dm3 h−1 ) at 500 ◦ C for 1 h, a mixture of methane (33 vol.%) and steam (67 vol.%) was introduced at the same temperature with a flow rate of 4.5 dm3 h−1 . The effluent gas, the water in which was trapped at ice temperature, was monitored with a quadrupole mass spectrometer (ULVAC MASSMATE-200) and analyzed with an on-stream Yanaco G2800 gas chromatograph (activated carbon, 2 m; Ar carrier) equipped with a thermal conductivity detector. The intensities of these instruments were corrected using standard gases. The reaction data was reproducible when the amount of a catalyst was 0.20 g with the same F/W. Methane conversion and selectivity were calculated on the basis of carbon numbers. The mass balance was within the error of ±5%. In other reactions, methane (33 vol.%; diluent, argon) and steam (67 vol.%) were separately fed on a catalyst (0.30 g) with a flow rate of 1.8 dm3 h−1 . In the interval of the gas flows, an argon stream was fed with the same flow rate. The conversion of methane and the selectivities were calculated on the basis of the material balance of hydrogen, methane, and carbon oxides in the effluent gas. The conversion of steam was calculated from the material balance, assuming that the quantity of steam fed is double that of argon. The selectivities of hydrogen and oxygen products were based on hydrogen and oxygen numbers, respectively. Temperature-programmed reduction of the catalyst (0.30 g) was carried out in a stream of 4 vol.% hydrogen diluted with argon at a flow rate of 6.0 dm3 h−1 . Temperature of the sample bed was risen linearly at a rate of 600 ◦ C h−1 . The hydrogen consumption was monitored by a mass spectrometer. The BET surface areas of the catalysts were determined from the isotherms of nitrogen physisorption. Powder X-ray diffraction (XRD) patterns of the catalysts were recorded with a Rigaku RINT 2000 diffractometer using nickel-filtered Cu K␣ radiation. Carbon contents in used catalysts were determined by high frequency induction combustion using a LECO CSS-244 testmeter.
3.1. Steam reforming over nickel catalysts Hydrogen and carbon oxides were formed at 500 ◦ C in the steam reforming over the nickel catalysts. No catalytic activity was found with 20 wt.% Ni/Al2 O3 pretreated with hydrogen at 500 ◦ C for 1 h, but the catalyst reduced at 700 ◦ C was quite active, with a methane conversion of 17.4% at 4 h-on-stream with 91.8% of a selectivity to carbon dioxide (Table 1). The equilibrium conversion is calculated to be 34% (selectivity to carbon dioxide, 84%) under the reaction conditions. At the initial stage of reaction, the activity steeply increased, then it stabilized (Fig. 1). The activity of 20 wt.% Ni/SiO2 was high at the beginning, but it decreased gradually and the catalyst was completely deactivated at 4 h-on-stream. The activity of nickel supported on zirconium oxide was high and stable. Even the catalyst containing
Table 1 Steam reforming of methane over nickel catalysts at 500 ◦ C Catalyst
Time-onstream (h)
CH4 conversion (%)
Selectivity (%) CO
CO2
20 wt.% Ni/SiO2 20 wt.% Ni/SiO2
0.5 2.0
21.8 14.8
9.7 9.6
90.3 90.4
20 wt.% Ni/Al2 O3 20 wt.% Ni/Al2 O3 a 20 wt.% Ni/Al2 O3 a
0.5 0.5 4.0
0.0 15.0 17.4
– 9.4 8.2
– 90.6 91.8
5 wt.% 5 wt.% 20 wt.% 20 wt.%
0.5 4.0 0.5 4.0
15.6 21.3 14.1 25.5
6.9 4.7 13.1 7.2
93.1 95.3 86.9 92.8
Ni/ZrO2 Ni/ZrO2 Ni/ZrO2 Ni/ZrO2
Reduction temperature of catalyst: 500 ◦ C unless otherwise mentioned; F/W: 15.0 dm3 h−1 g−1 (H2 O/CH4 , 2.0). a Reduction temperature: 700 ◦ C.
Fig. 1. Hydrogen production monitored by mass spectrometer (m/e = 2) in steam reforming of methane over nickel catalysts at 500 ◦ C. Reduction temperatures of the catalysts are given in the parentheses.
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
109
5 wt.% of nickel (5 wt.% Ni/ZrO2 ) produced a higher activity than that with 20 wt.% Ni/Al2 O3 . The activity increased with an increase in the content of nickel. The activity of 20 wt.% Ni/ZrO2 steeply increased at the initial stage, as observed with 20 wt.% Ni/Al2 O3 , and then stabilized. 3.2. Hydrogen production by separate feed of methane and steam In order to understand this reaction process, we separately fed methane on the nickel catalysts at 500 ◦ C; then steam was supplied instead of methane. At the initial stage of the reaction with methane over 20 wt.% Ni/SiO2 , hydrogen was produced with carbon oxides and the formation of hydrogen steeply decreased, then it increased gradually (Fig. 2). Since the amount of coke formed during the reaction was unable to be directly detected, the value was estimated from the material balance among the gas-phase products, assuming the following reactions (see Section 4): CH4 + 2OH (surface) → CO2 + 3H2 ,
(1)
CH4 + O (surface) → CO + 2H2 ,
(2)
and CH4 → C (surface) + 2H2 .
(3)
The conversions and selectivities in Table 2 are given on the basis of this assumption. After the initial increase, no formation of carbon dioxide was observed (see Fig. 2) and the selectivity to carbon increased to 96% at 1.1 h-on-stream.
Fig. 2. Composition of the effluent gas in the reaction with a separate feed of methane and steam at 500 ◦ C over 20 wt.% Ni/SiO2 reduced at 500 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
After the reaction with methane for 1.4 h, steam was introduced to the solid. Hydrogen formation increased gradually. At the initial stage hydrogen was mainly detected with small quantities of methane and carbon monoxide, showing that oxygen in steam was accumulated on 20 wt.% Ni/SiO2 (Table 3). The conversion and selectivities were calculated on the basis of the stoichiometric equations: H2 O + C (surface) → H2 + CO,
(4)
2H2 O + C (surface) → 2H2 + CO2 ,
(5)
2H2 O → H2 + 2OH (surface),
(6)
Table 2 Decomposition of methane over nickel catalysts at 500 ◦ C Catalyst
Time-on-stream (h)
CH4 conversion (%)
Selectivity (%) CO
CO2
C (surface)
20 wt.% Ni/SiO2 20 wt.% Ni/SiO2
0.1 1.1
23 17
16 4
52 0
32 96
20 wt.% Ni/Al2 O3 20 wt.% Ni/Al2 O3 a 20 wt.% Ni/Al2 O3 a
0.1 0.1 1.1
2 24 12
13 13 8
0 57 0
87 30 92
5 wt.% 5 wt.% 5 wt.% 5 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.%
0.1 1.1 0.1 1.1 0.1 1.1 0.1 1.1 0.1 1.1 0.1 1.1
10 (10) 11 21 (17) 10 13 (13) 15 25 (19) 15 44 (32) 14 0.1 17
11 0 11 2 10 2 11 5 6 2 – 3
Ni/ZrO2 Ni/ZrO2 Ni/ZrO2 b Ni/ZrO2 b Ni/ZrO2 Ni/ZrO2 Ni/ZrO2 b Ni/ZrO2 b Ni/ZrO2 c Ni/ZrO2 c Ni/ZrO2 d Ni/ZrO2 d
(11) (14) (10) (16) (8)
Reduction temperature of catalyst: 500 ◦ C unless otherwise mentioned; F/W: 6.0 dm3 h−1 g−1 (CH4 , 33 vol.%). a Reduction temperature: 700 ◦ C. b After reaction with steam for 1.4 h following methane decomposition for 1.4 h. c After contact with steam for 1.4 h. d Without reduction.
0 (0) 0 69 (86) 0 7 (8) 0 57 (80) 0 67(92) 0 – 0
89 100 20 98 83 98 32 95 27 98 – 97
(89) (0) (82) (4) (0)
110
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
Table 3 Reaction between water and nickel catalyst at 500 ◦ C after decomposition of methane for 1.4 h Catalyst
Time-on-stream (h)
H2 O conversion (%)
Selectivity (%) Hydrogen
Oxygen
H2
CH4
OH
CO
CO2
OH
20 wt.% Ni/SiO2 20 wt.% Ni/SiO2
0.1 1.1
4 12
42 95
11 1
47 4
7 0
0 91
93 9
20 wt.% Ni/Al2 O3 a 20 wt.% Ni/Al2 O3 a
0.1 1.1
2 12
41 94
15 0
44 6
11 0
0 88
89 12
5 wt.% Ni/ZrO2 5 wt.% Ni/ZrO2 20 wt.% Ni/ZrO2 20 wt.% Ni/ZrO2 20 wt.% Ni/ZrO2 b 20 wt.% Ni/ZrO2 b
0.1 1.1 0.1 1.1 0.1 1.1
4 14 4 13 3 2
26 89 19 90 50 50
24 1 31 2 0 0
50 10 50 8 50 50
0 79 0 83 0 0
100 21 100 15 100 100
0 0.4 0 2 0 0
Reduction temperature of catalyst: 500 ◦ C unless otherwise mentioned; F/W: 6.0 dm3 h−1 g−1 (H2 O, 67 vol.%). a Reduction temperature: 700 ◦ C. b Without methane decomposition before the feed of steam.
and 4H2 O + C (surface) → CH4 + 4OH (surface).
(7)
Formation of carbon dioxide accompanied the gradual increase in hydrogen formation and the accumulation of oxygen on the surface was quite small in the stabilized state. When methane was again fed on the sample after the reaction with steam, the phenomenon observed was similar to that in the first feed of methane (see Fig. 2). The reaction processes observed with 20 wt.% Ni/Al2 O3 were almost the same as those with 20 wt.% Ni/SiO2 except for the reaction with the secondary feed of methane, where the initial steep formation of hydrogen was absent (see Table 2and cf. Figs. 2 and 3). No steep hydrogen formation over Ni/ZrO2 was observed at the initial stage of methane feed (Fig. 4 for 5 wt.% Ni/ZrO2 and Fig. 5 for 20 wt.% Ni/ZrO2 ). The selectivity to
Fig. 3. Composition of the effluent gas in the reaction with a separate feed of methane and steam at 500 ◦ C over 20 wt.% Ni/Al2 O3 reduced at 700 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
Fig. 4. Composition of the effluent gas in the reaction with a separate feed of methane and steam at 500 ◦ C over 5 wt.% Ni/ZrO2 reduced at 500 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
Fig. 5. Composition of the effluent gas in the reaction with a separate feed of methane and steam at 500 ◦ C over 20 wt.% Ni/ZrO2 reduced at 500 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
Fig. 6. Composition of the effluent gas in the reaction with methane following steam treatment at 500 ◦ C over 20 wt.% Ni/ZrO2 reduced at 500 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
surface carbon was high even at the initial stage and formation of carbon dioxide was low (see Table 2). When water was fed, an induction period was present (see Figs. 4 and 5). In this period, hydrogen and methane were mainly detected in the gas-phase products as seen in the initial stage of the reaction over 20 wt.% Ni/SiO2 and Ni/Al2 O3 . The formation of hydrogen once decreased, then gradually increased with production of carbon dioxide. After the contact with steam for 1.4 h, introduction of methane produced abrupt formation of hydrogen, mainly with carbon dioxide (see Table 3). This phenomenon was repeated by the experiment without the first feed of methane. That is, steam was supplied on 20 wt.% Ni/ZrO2 reduced at 500 ◦ C for 1.4 h, then methane was fed (Fig. 6). In the reaction with steam, a small quantity of hydrogen was detected without formation of carbon oxides (see Table 3). Hydrogen and carbon dioxide were mainly formed in the initial stage of the reaction with methane (see Table 2). Methane was also fed on 20 wt.% Ni/ZrO2 preheated in argon at 500 ◦ C for 1 h and without reduction. After the induction period of 0.15 h, production of hydrogen started with small quantities of carbon oxides (Fig. 7).
Fig. 7. Composition of the effluent gas in the reaction with methane over 20 wt.% Ni/ZrO2 without reduction. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
peak at around 450 ◦ C was clearly present in the profile of 20 wt.% Ni/ZrO2 while the major peak was at ca. 610 ◦ C. 3.4. Physical properties of catalysts Two major XRD peaks, attributed to Ni(1 1 1) at 44.5◦ in 2θ and to Ni(2 0 0) at 51.8◦ , were recorded with 20 wt.% Ni/SiO2 just after reduction with hydrogen at 500 ◦ C (Fig. 9a), but after the steam reforming for 5 h (see Fig. 1) the peaks attributed to NiO(1 1 1) at 37.3◦ and NiO(2 0 0) at 43.3◦ appeared with the peaks for metallic nickel (Fig. 9b) [9]. The XRD pattern of 20 wt.% Ni/Al2 O3 treated with hydrogen at 500 ◦ C was very similar to that of the sample as prepared (not shown). In the pattern, the NiO peak at 43.4◦ was recorded with the peaks attributed to ␥-alumina [12],
3.3. Temperature-programmed reduction (TPR) of nickel catalysts Reduction of 20 wt.% Ni/SiO2 started at ca. 330 ◦ C in a TPR experiment and the maximum of the reduction peak was at 430 ◦ C (Fig. 8). Small hydrogen consumption from 500 ◦ C was recorded with 20 wt.% Ni/Al2 O3 and the consumption gradually increased up to 800 ◦ C, where the TPR experiment was terminated. In the case of 5 wt.% Ni/ZrO2 , hydrogen consumption was observed from 370 ◦ C and the peak maximum was at ca. 600 ◦ C. A small consumption
111
Fig. 8. Temperature-programmed reduction of nickel catalysts.
112
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
Discernible increases in the BET surface areas were observed after the reaction in comparison with those for the as-prepared samples except 20 wt.% Ni/Al2 O3 prereduced at 700 ◦ C (see Table 4). The contents of carbon in the catalysts after the reaction are also shown in Table 4. 4. Discussion 4.1. Catalytic activity of Ni/SiO2 and Ni/Al2 O3
Fig. 9. X-ray diffraction patterns of catalysts. (a) 20 wt.% Ni/SiO2 reduced at 500 ◦ C, (b) after steam reforming for 5 h, (c) 20 wt.% Ni/Al2 O3 reduced at 700 ◦ C, (d) after reforming for 5 h, (e) 20 wt.% Ni/ZrO2 reduced at 500 ◦ C, and (f) after reforming for 5 h.
but no peaks attributed to metallic nickel were recorded. On the other hand, peaks attributed to metallic nickel at 44.5◦ and 51.8◦ were recorded with the sample reduced at 700 ◦ C (Fig. 9c). After the reaction, the intensities of these peaks decreased appreciably, and a small peak at 43.4◦ was separated in the profile (Fig. 9d). In the pattern for 20 wt.% Ni/ZrO2 reduced at 500 ◦ C, the peaks at 44.6◦ and 51.8◦ were present with small peaks attributed to NiO at 37.2◦ and 43.4◦ (Fig. 9e), while the other peaks were attributed to ZrO2 (mainly baddeleyite phase) [12]. After the reaction, the peaks attributed to NiO were diminished and the Ni peaks were intensified (Fig. 9f), showing that the nickel particles were completely reduced during the reaction. The XRD pattens for 5 wt.% Ni/ZrO2 were similar to those for 20 wt.% Ni/ZrO2 while the peaks attributed to nickel oxide and metallic nickel were weaker. Mean crystallite sizes were determined from the line broadening of the XRD peaks of Ni(1 1 1) and NiO(2 0 0) [13]. The crystallite sizes of the catalysts after the reaction were almost the same as those observed just after the reduction (Table 4). Table 4 Properties of nickel catalysts after steam reforming for 5 h Catalyst
Crystallite sizea (nm) Ni
BET surface areab (m2 g−1 )
Carbon content (wt.%)
NiO
20 wt.% Ni/SiO2
17 (15) 17
244 (237)
0.20
20 wt.% Ni/Al2 O3 c
12 (11) 11
135 (129)
0.12
5 wt.% Ni/ZrO2 20 wt.% Ni/ZrO2
11 (11) 47 (42) 12 (11) 38 (33)
0.32 0.24
Reduction temperature of catalyst: 500 ◦ C unless otherwise mentioned. a The size just after reduction is given in parentheses. b The surface area of the samples as prepared is given in parentheses. c Reduction temperature: 700 ◦ C.
Metallic nickel particles whose mean crystallite size is 15 nm are oxidized to nickel oxide on 20 wt.% Ni/SiO2 during the steam reforming at 500 ◦ C, as evidenced in the XRD experiment (cf. Fig. 9a and b); this means that steam can gradually oxidize the surface of nickel particles on silica. The same phenomenon can be seen with 20 wt.% Ni/Al2 O3 reduced at 700 ◦ C (cf. Fig. 9c and d), but metallic nickel is dominant even after the reaction. The lower reducibility of Ni/Al2 O3 (see Fig. 8) is probably caused by formation of surface spinel (NiAl2 O4 ) during the reduction process [1]. The deactivation of 20 wt.% Ni/SiO2 is probably caused by the oxidation of nickel during the reaction because nickel oxide is inactive as can be seen in 20 wt.% Ni/Al2 O3 pretreated at 500 ◦ C in which nickel oxide is dominant. Since coke formation on the catalyst is slight in the reaction period (see Table 4), it is not responsible for the deactivation. 4.2. Mechanism of the reaction with separate feed of methane and steam The rate of the reaction with methane in the absence of steam is significantly lower than that of the steady steam reforming (cf. Tables 1 and 2), suggesting that the presence of steam, which can oxidize the surface of nickel, accelerates the decomposition of methane to hydrogen. In actual, the formation rate of hydrogen is high at the initial stage of the reaction with methane on Ni/SiO2 and Ni/Al2 O3 catalysts (see Figs. 2 and 3), and formation of carbon oxides, which evidences presence of surface oxygen species, is accompanied. Carbon monoxide is selectively formed in the reaction with methane over the catalysts containing NiO, such as 20 wt.% Ni/Al2 O3 pretreated at 500 ◦ C and 20 wt.% Ni/ZrO2 without reduction (see Table 2and Fig. 7). Hence, the reduction of nickel oxide results in formation of carbon monoxide (Eq. (2)) and a reaction such as CH4 + 2O (surface) → CO2 + 2H2 ,
(8)
is less probable. Formation of hydroxyl groups on the surface of nickel (Ni−OH) is estimated in the reduction process of NiO particles; thus, it is supposed that the hydroxyl group is rather more reactive to methane and/or carbon monoxide than the lattice oxygen of nickel oxide. Formation of carbon dioxide on 20 wt.% Ni/SiO2 and 20 wt.% Ni/Al2 O3 reduced at 700 ◦ C after the reaction with steam (see Figs. 2 and 3) supports this supposition, because the contact of steam and
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
nickel often results in formation of hydroxyl groups [14]. Accumulation of oxygen in water on the surface of these catalysts (Eqs. (6) and (7)) can be detected at the initial stage of the reaction with steam after methane decomposition. In this process, reactions of H2 O → H2 + O (surface),
(9)
and 2H2 O + C (surface) → CH4 + 2O (surface),
(10)
are also probable. However, the contribution will be small because formation of carbon monoxide, which is produced in presence of nickel oxide (Eq. (2)), is small in the reaction with methane after the contact with steam (see Figs. 2 and 3). Since the formation rates of hydrogen and carbon dioxide increases gradually in the reaction with steam, carbon dioxide cannot be formed in the direct reaction between steam and surface carbon, but is probably formed by the reaction with surface hydroxyl groups whose concentration should be saturated under the continuous feed of steam. That is, C (surface) + 2OH (surface) → H2 + CO2 ,
(11)
will take place on the surface. Little methane is formed after the formation rate of carbon dioxide is saturated, implying that the surface is mostly oxidized, because formation of methane does not proceed in absence of metallic nickel which accepts the oxygen in steam (Eqs. (7) and/or (10)). After the feed of steam following the reaction with methane, formation of carbon dioxide (Eq. (1)) can be observed with re-feed of methane (see Figs. 2 and 3), suggesting formation of surface hydroxyl groups during the previous reaction with steam (Eq. (6)). When methane is decomposed on 5 and 20 wt.% Ni/ZrO2 reduced at 500 ◦ C (see Figs. 4 and 5, and Table 2), the yield of carbon dioxide in the initial stage of the reaction with methane (1% or less) is significantly smaller than that with 20 wt.% Ni/SiO2 and Ni/Al2 O3 (11 and 13%, respectively). This would mean that the quantity of the surface oxygen species such as hydroxyl groups is small on Ni/ZrO2 . Selective formation of methane in the initial stage of the reaction with steam over Ni/ZrO2 (see Table 3) suggests formation of surface oxygen species (e.g. Eqs. (7) and (10)). Although the oxygen species accumulate during the initial stage, carbon oxides are scarcely formed in the induction period. Thus, the species are inactive and different from the hydroxyl groups formed on Ni/SiO2 or Ni/Al2 O3 . The accumulation of the oxygen species is considered to take place on the surface of nickel, because the quantity of methane formed in the induction period is small on 5 wt.% Ni/ZrO2 whose nickel content is small in comparison with 20 wt.% Ni/ZrO2 (cf. Figs. 4 and 5). The reason for the formation of inactive oxygen species is not clear, but presence of zirconia should contribute to the nature of the species; thus, it is speculated that the species are rather present on the perimeter of nickel particles. The reduction temperature of
113
Ni/ZrO2 is significantly higher than that of 20 wt.% Ni/SiO2 (see Fig. 8), suggesting stronger interaction between zirconia support and nickel particles. After the feed of steam, a significantly large quantity of carbon dioxide and hydrogen is formed with the second feed of methane. The total quantity of carbon dioxide formed with 20 wt.% Ni/ZrO2 during the second feed can be evaluated as ca. 10 mmol g−1 , while the number of nickel atoms in the catalyst is only 3.4 mmol g−1 (see Fig. 5). Thus, the major source of oxygen in carbon dioxide is different from nickel hydroxide or nickel oxide and these results imply that accumulation of water takes place on the surface of the zirconia support. In the reaction between 20 wt.% Ni/ZrO2 and steam (see Fig. 6) the total quantity of oxygen atoms consumed in oxidation of nickel (Eq. (6)) is evaluated as ca. 1 mmol from the amount of hydrogen formed, and that of oxygen atoms in carbon oxides formed in the successive reaction with methane is ca. 20 mmol; these results show that methane is reacted with water adsorbed on the surface. Since water molecules are hard to be preserved at 500 ◦ C on the surface of Ni/ZrO2 , dissociation of Zr−O−Zr with H2 O to 2Zr−OH is supposed to take place. Hence, the values in the parentheses of Table 2 were calculated assuming a reaction of CH4 + 2H2 O (surface) → CO2 + 4H2 ,
(12)
instead of Eq. (1). This reaction is rather advantageous to methane decomposition to carbon (Eq. (3)) at the initial stage of the reaction with methane on Ni/ZrO2 after contact with steam (see Table 2). Since activation of methane probably takes place on the surface of nickel, the adsorption species such as Zr−OH cannot react directly with methane. Hence, it is supposed that the adsorbed water migrates to nickel surface and forms the active hydroxyl group which reacts with methane to carbon dioxide (Eq. (1)). The inactive oxygen species formed in the initial stage of the reaction with steam and surface carbon would be involved in the mechanism, but further investigation is necessary for the clarification. 4.3. Mechanism of methane steam reforming at low temperature Surface oxygen species involved in the reaction mechanism at high temperatures are shown in [1], that is, CH4 + nS∗ → CHx –S∗n +
4−x H2 , 2
(13)
CHx –S∗n + O–S∗ → CO + 21 xH2 + (n + 1)S∗ ,
(14)
H2 O + S∗ → O–S∗ + H2 ,
(15)
H2 + 2S∗ → 2H–S∗ ,
(16)
In the mechanism, where S∗ represents an active site, the presence of oxide causes formation of carbon monoxide and the process is similar to Eq. (2) in the reaction solely with methane. No reaction was observed at the initial stage of the
114
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
methane decomposition over 20 wt.% Ni/ZrO2 without reduction (see Fig. 7), suggesting that methane does not reduce nickel oxide. Just after the induction period, CHx is probably formed on the partially reduced surface and this results in formation of carbon monoxide and hydrogen (Eq. (14)), which also reduces the surface. Thus, formation of carbon monoxide in the steam reforming at 500 ◦ C will be caused from surface nickel oxide species. Since formation of carbon dioxide takes place in the presence of surface hydroxyl groups (Eqs. (1) and (11)), CHx –S∗n + 2HO–S∗ → CO2 + ( 21 x + 1)H2 + (n + 2)S∗ , (17) may proceed on the surface in the steam reforming. The rate of hydrogen production was 112 mmol h−1 g−1 at the initial stage of the reaction with methane after supplying steam over 20 wt.% Ni/ZrO2 (see Fig. 6 and Table 2), while the rate in the steam reforming over the same catalyst was 122 mmol h−1 g−1 at 0.5 h-on-stream (see Table 1). Hence, formation of hydroxyl groups on the surface should be an important step also in the steam reforming at 500 ◦ C. In the case of Ni/ZrO2 , accumulation of water on the support assists formation of the hydroxyl groups, and this is probably the reason why zirconia is an effective support of nickel in the steam reforming at 500 ◦ C. Formation of carbon dioxide from carbon monoxide by water–gas shift reaction is possible in the reaction mechanism, but no relationship between production of carbon monoxide and carbon dioxide can be found in the reactions solely with methane or steam.
Acknowledgements The financial support of the New Energy and Industrial Technology Development Organization of Japan is acknowledged. References [1] J.R. Rostrup-Nielsen, in: J.R. Anderson, M. Boudard, (Eds.), Catalysis, Science and Technology, vol. 5, Springer, New York, 1984, p. 3. [2] M. Oertel, J. Schmitz, W. Weirich, D. Jendryssek-Newmann, R. Schulten, Chem. Eng. Technol. 10 (1987) 248. [3] A.M. Adris, S.S.E.H. Elnashaie, R. Hughes, Can. J. Chem. Eng. 69 (1991) 1061. [4] S. Uemiya, N. Sato, H. Ando, T. Matsuda, E. Kikuchi, Appl. Catal. 67 (1991) 223. [5] US Patent 5,229,102 (1993). [6] M. Chai, M. Machida, K. Eguchi, H. Arai, Appl. Catal. A: Gen. 110 (1994) 239. [7] J. Galuszka, R.N. Pandey, S. Ahmed, Catal. Today 46 (1998) 83. [8] E. Kikuchi, Catal. Today 56 (2000) 97. [9] M.E.S. Hegarty, A.M. O’Connor, J.R.H. Ross, Catal. Today 42 (1998) 225. [10] J.-M. Wei, B.-Q. Xu, J.-L. Li, Z.-X. Cheng, Q.-M. Zhu, Appl. Catal. A 196 (2000) L167. [11] H.-S. Roh, K.-W. Jun, W.-S. Dong, S.-E. Park, Y.-S. Baek, Catal. Lett. 74 (2001) 31. [12] JCPDS Files 4-0850, 4-0835, 10-0427 and 37-1484. [13] H.P. Klug, L.E. Alexander, X-ray Diffraction Procedures, Wiley, New York, 1954. [14] J.R.H. Ross, M.C.F. Steel, J.C.S. Faraday Trans. I 69 (1973) 10.
Steam reforming of methane over nickel catalysts at low reaction temperature Yasuyuki Matsumura∗ , Toshie Nakamori Research Institute of Innovative Technology for the Earth, Kizu-cho, Soraku-gun, Kyoto 619-0292, Japan Received 23 April 2003; received in revised form 10 August 2003; accepted 21 August 2003
Abstract Effects of supports such as silica, ␥-alumina, and zirconia for nickel catalysts have been studied in steam reforming of methane at 500 ◦ C. The activity of nickel supported on silica reduced with hydrogen at 500 ◦ C decreases with oxidation of nickel particles by steam during the reaction. Nickel supported on ␥-alumina is not much reduced with hydrogen at 500 ◦ C and is inactive in the reforming at 500 ◦ C. However, the catalyst reduced at 700 ◦ C is fairly active while nickel is partially oxidized during the reaction. Nickel supported on zirconia is the most effective in the stream reforming at 500 ◦ C. In the initial stage of the reaction solely with methane at 500 ◦ C, surface hydroxyl groups on these catalysts react readily with methane to produce hydrogen and carbon dioxide; suggesting that the hydroxyl groups play an important role in the mechanism of the steam reforming to carbon dioxide. A significant quantity of water can be accumulated on the surface of zirconia-supported nickel in the reaction only with steam at 500 ◦ C, and this results in formation of significant quantities of hydrogen and carbon dioxide in the following reaction with methane. © 2003 Elsevier B.V. All rights reserved. Keywords: Steam reforming; Methane; Hydrogen production; Nickel catalyst; Surface hydroxyl group
1. Introduction Steam reforming of methane is employed for a large scale production of hydrogen and/or carbon monoxide. The reaction is industrially operated at a high temperature around 800 ◦ C over nickel-alumina based catalysts, because a reasonable conversion of methane is required in this endothermic process [1]. Such a high temperature causes some demerits, including the expensive tubular reformer made of high alloy nickel-chromium steel, irreversible carbon formation in the reactor, and large energy consumption. In order to reduce the reaction temperature, some membrane reactors have been proposed [2–8]. In the reactor, hydrogen is usually separated from the reaction mixture by a palladium-based membrane and the separation of hydrogen significantly increases the equilibrium conversion of methane. Uemiya et al. [4] reported methane conversion close to 90% at 500 ◦ C using this reactor. Since a high partial pressure of hydrogen in the reactor increases the efficiency of hydrogen separation, the reaction composition ∗
Corresponding author. Fax: +81-774-75-2318. E-mail address: [email protected] (Y. Matsumura).
0926-860X/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2003.08.009
should be close to the equilibrium all over the part of the membrane. In the above case, a commercial nickel catalyst was employed with the F/W of 0.5 dm3 h−1 g−1 [4], while much higher space velocity is required in an industrial process. Hence, high catalytic activity at a low reaction temperature around 500 ◦ C is required in a realistic membrane system. We have preliminarily studied effects of the supports on steam reforming over nickel catalysts at 500 ◦ C. Decomposition of methane on nickel surface is believed as a first step of the steam reforming of methane; then the carbon species formed on the surface react subsequently with steam or surface oxygen species [1]. To follow the reaction steps simply, we separately fed methane and steam to the catalysts and have analyzed the mechanism especially at the initial stage of the reaction in which surface active species participate highly. In this report, it will be shown that zirconium oxide, which can absorb water at 500 ◦ C, is an effective support and surface hydroxyl groups contribute significantly to the steam reforming at 500 ◦ C. Recently, some researches have been reported the activity of nickel supported on zirconium oxide to methane reforming above 700 ◦ C [9–11].
108
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
2. Experimental
3. Results
Nickel catalysts were prepared by an impregnation technique. Metal oxide supports such as ␥-alumina (ALO-4 supplied by Catalysis Society of Japan), silica (Fuji-Silicia, G-10), and zirconia (Daiichi Kigenso Kagaku Kogyo, RC-100) were impregnated with nickel nitrate (Wako, GR grade) by evaporation of the aqueous solution at 80 ◦ C. After drying in air at 110 ◦ C overnight, they were heated at 700 ◦ C for 3 h. The samples (Ni/Al2 O3 , Ni/SiO2 , and Ni/ZrO2 , respectively) contained 5–20 wt.% of nickel. Catalytic tests were performed in a fixed bed continuous flow reactor operated at atmospheric pressure. A catalyst (0.30 g) was placed in a tube reactor made of quartz glass (i.d.: 6 mm). After the pre-reduction in a stream of 10 vol.% hydrogen diluted with argon (6.0 dm3 h−1 ) at 500 ◦ C for 1 h, a mixture of methane (33 vol.%) and steam (67 vol.%) was introduced at the same temperature with a flow rate of 4.5 dm3 h−1 . The effluent gas, the water in which was trapped at ice temperature, was monitored with a quadrupole mass spectrometer (ULVAC MASSMATE-200) and analyzed with an on-stream Yanaco G2800 gas chromatograph (activated carbon, 2 m; Ar carrier) equipped with a thermal conductivity detector. The intensities of these instruments were corrected using standard gases. The reaction data was reproducible when the amount of a catalyst was 0.20 g with the same F/W. Methane conversion and selectivity were calculated on the basis of carbon numbers. The mass balance was within the error of ±5%. In other reactions, methane (33 vol.%; diluent, argon) and steam (67 vol.%) were separately fed on a catalyst (0.30 g) with a flow rate of 1.8 dm3 h−1 . In the interval of the gas flows, an argon stream was fed with the same flow rate. The conversion of methane and the selectivities were calculated on the basis of the material balance of hydrogen, methane, and carbon oxides in the effluent gas. The conversion of steam was calculated from the material balance, assuming that the quantity of steam fed is double that of argon. The selectivities of hydrogen and oxygen products were based on hydrogen and oxygen numbers, respectively. Temperature-programmed reduction of the catalyst (0.30 g) was carried out in a stream of 4 vol.% hydrogen diluted with argon at a flow rate of 6.0 dm3 h−1 . Temperature of the sample bed was risen linearly at a rate of 600 ◦ C h−1 . The hydrogen consumption was monitored by a mass spectrometer. The BET surface areas of the catalysts were determined from the isotherms of nitrogen physisorption. Powder X-ray diffraction (XRD) patterns of the catalysts were recorded with a Rigaku RINT 2000 diffractometer using nickel-filtered Cu K␣ radiation. Carbon contents in used catalysts were determined by high frequency induction combustion using a LECO CSS-244 testmeter.
3.1. Steam reforming over nickel catalysts Hydrogen and carbon oxides were formed at 500 ◦ C in the steam reforming over the nickel catalysts. No catalytic activity was found with 20 wt.% Ni/Al2 O3 pretreated with hydrogen at 500 ◦ C for 1 h, but the catalyst reduced at 700 ◦ C was quite active, with a methane conversion of 17.4% at 4 h-on-stream with 91.8% of a selectivity to carbon dioxide (Table 1). The equilibrium conversion is calculated to be 34% (selectivity to carbon dioxide, 84%) under the reaction conditions. At the initial stage of reaction, the activity steeply increased, then it stabilized (Fig. 1). The activity of 20 wt.% Ni/SiO2 was high at the beginning, but it decreased gradually and the catalyst was completely deactivated at 4 h-on-stream. The activity of nickel supported on zirconium oxide was high and stable. Even the catalyst containing
Table 1 Steam reforming of methane over nickel catalysts at 500 ◦ C Catalyst
Time-onstream (h)
CH4 conversion (%)
Selectivity (%) CO
CO2
20 wt.% Ni/SiO2 20 wt.% Ni/SiO2
0.5 2.0
21.8 14.8
9.7 9.6
90.3 90.4
20 wt.% Ni/Al2 O3 20 wt.% Ni/Al2 O3 a 20 wt.% Ni/Al2 O3 a
0.5 0.5 4.0
0.0 15.0 17.4
– 9.4 8.2
– 90.6 91.8
5 wt.% 5 wt.% 20 wt.% 20 wt.%
0.5 4.0 0.5 4.0
15.6 21.3 14.1 25.5
6.9 4.7 13.1 7.2
93.1 95.3 86.9 92.8
Ni/ZrO2 Ni/ZrO2 Ni/ZrO2 Ni/ZrO2
Reduction temperature of catalyst: 500 ◦ C unless otherwise mentioned; F/W: 15.0 dm3 h−1 g−1 (H2 O/CH4 , 2.0). a Reduction temperature: 700 ◦ C.
Fig. 1. Hydrogen production monitored by mass spectrometer (m/e = 2) in steam reforming of methane over nickel catalysts at 500 ◦ C. Reduction temperatures of the catalysts are given in the parentheses.
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
109
5 wt.% of nickel (5 wt.% Ni/ZrO2 ) produced a higher activity than that with 20 wt.% Ni/Al2 O3 . The activity increased with an increase in the content of nickel. The activity of 20 wt.% Ni/ZrO2 steeply increased at the initial stage, as observed with 20 wt.% Ni/Al2 O3 , and then stabilized. 3.2. Hydrogen production by separate feed of methane and steam In order to understand this reaction process, we separately fed methane on the nickel catalysts at 500 ◦ C; then steam was supplied instead of methane. At the initial stage of the reaction with methane over 20 wt.% Ni/SiO2 , hydrogen was produced with carbon oxides and the formation of hydrogen steeply decreased, then it increased gradually (Fig. 2). Since the amount of coke formed during the reaction was unable to be directly detected, the value was estimated from the material balance among the gas-phase products, assuming the following reactions (see Section 4): CH4 + 2OH (surface) → CO2 + 3H2 ,
(1)
CH4 + O (surface) → CO + 2H2 ,
(2)
and CH4 → C (surface) + 2H2 .
(3)
The conversions and selectivities in Table 2 are given on the basis of this assumption. After the initial increase, no formation of carbon dioxide was observed (see Fig. 2) and the selectivity to carbon increased to 96% at 1.1 h-on-stream.
Fig. 2. Composition of the effluent gas in the reaction with a separate feed of methane and steam at 500 ◦ C over 20 wt.% Ni/SiO2 reduced at 500 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
After the reaction with methane for 1.4 h, steam was introduced to the solid. Hydrogen formation increased gradually. At the initial stage hydrogen was mainly detected with small quantities of methane and carbon monoxide, showing that oxygen in steam was accumulated on 20 wt.% Ni/SiO2 (Table 3). The conversion and selectivities were calculated on the basis of the stoichiometric equations: H2 O + C (surface) → H2 + CO,
(4)
2H2 O + C (surface) → 2H2 + CO2 ,
(5)
2H2 O → H2 + 2OH (surface),
(6)
Table 2 Decomposition of methane over nickel catalysts at 500 ◦ C Catalyst
Time-on-stream (h)
CH4 conversion (%)
Selectivity (%) CO
CO2
C (surface)
20 wt.% Ni/SiO2 20 wt.% Ni/SiO2
0.1 1.1
23 17
16 4
52 0
32 96
20 wt.% Ni/Al2 O3 20 wt.% Ni/Al2 O3 a 20 wt.% Ni/Al2 O3 a
0.1 0.1 1.1
2 24 12
13 13 8
0 57 0
87 30 92
5 wt.% 5 wt.% 5 wt.% 5 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.% 20 wt.%
0.1 1.1 0.1 1.1 0.1 1.1 0.1 1.1 0.1 1.1 0.1 1.1
10 (10) 11 21 (17) 10 13 (13) 15 25 (19) 15 44 (32) 14 0.1 17
11 0 11 2 10 2 11 5 6 2 – 3
Ni/ZrO2 Ni/ZrO2 Ni/ZrO2 b Ni/ZrO2 b Ni/ZrO2 Ni/ZrO2 Ni/ZrO2 b Ni/ZrO2 b Ni/ZrO2 c Ni/ZrO2 c Ni/ZrO2 d Ni/ZrO2 d
(11) (14) (10) (16) (8)
Reduction temperature of catalyst: 500 ◦ C unless otherwise mentioned; F/W: 6.0 dm3 h−1 g−1 (CH4 , 33 vol.%). a Reduction temperature: 700 ◦ C. b After reaction with steam for 1.4 h following methane decomposition for 1.4 h. c After contact with steam for 1.4 h. d Without reduction.
0 (0) 0 69 (86) 0 7 (8) 0 57 (80) 0 67(92) 0 – 0
89 100 20 98 83 98 32 95 27 98 – 97
(89) (0) (82) (4) (0)
110
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
Table 3 Reaction between water and nickel catalyst at 500 ◦ C after decomposition of methane for 1.4 h Catalyst
Time-on-stream (h)
H2 O conversion (%)
Selectivity (%) Hydrogen
Oxygen
H2
CH4
OH
CO
CO2
OH
20 wt.% Ni/SiO2 20 wt.% Ni/SiO2
0.1 1.1
4 12
42 95
11 1
47 4
7 0
0 91
93 9
20 wt.% Ni/Al2 O3 a 20 wt.% Ni/Al2 O3 a
0.1 1.1
2 12
41 94
15 0
44 6
11 0
0 88
89 12
5 wt.% Ni/ZrO2 5 wt.% Ni/ZrO2 20 wt.% Ni/ZrO2 20 wt.% Ni/ZrO2 20 wt.% Ni/ZrO2 b 20 wt.% Ni/ZrO2 b
0.1 1.1 0.1 1.1 0.1 1.1
4 14 4 13 3 2
26 89 19 90 50 50
24 1 31 2 0 0
50 10 50 8 50 50
0 79 0 83 0 0
100 21 100 15 100 100
0 0.4 0 2 0 0
Reduction temperature of catalyst: 500 ◦ C unless otherwise mentioned; F/W: 6.0 dm3 h−1 g−1 (H2 O, 67 vol.%). a Reduction temperature: 700 ◦ C. b Without methane decomposition before the feed of steam.
and 4H2 O + C (surface) → CH4 + 4OH (surface).
(7)
Formation of carbon dioxide accompanied the gradual increase in hydrogen formation and the accumulation of oxygen on the surface was quite small in the stabilized state. When methane was again fed on the sample after the reaction with steam, the phenomenon observed was similar to that in the first feed of methane (see Fig. 2). The reaction processes observed with 20 wt.% Ni/Al2 O3 were almost the same as those with 20 wt.% Ni/SiO2 except for the reaction with the secondary feed of methane, where the initial steep formation of hydrogen was absent (see Table 2and cf. Figs. 2 and 3). No steep hydrogen formation over Ni/ZrO2 was observed at the initial stage of methane feed (Fig. 4 for 5 wt.% Ni/ZrO2 and Fig. 5 for 20 wt.% Ni/ZrO2 ). The selectivity to
Fig. 3. Composition of the effluent gas in the reaction with a separate feed of methane and steam at 500 ◦ C over 20 wt.% Ni/Al2 O3 reduced at 700 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
Fig. 4. Composition of the effluent gas in the reaction with a separate feed of methane and steam at 500 ◦ C over 5 wt.% Ni/ZrO2 reduced at 500 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
Fig. 5. Composition of the effluent gas in the reaction with a separate feed of methane and steam at 500 ◦ C over 20 wt.% Ni/ZrO2 reduced at 500 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
Fig. 6. Composition of the effluent gas in the reaction with methane following steam treatment at 500 ◦ C over 20 wt.% Ni/ZrO2 reduced at 500 ◦ C. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
surface carbon was high even at the initial stage and formation of carbon dioxide was low (see Table 2). When water was fed, an induction period was present (see Figs. 4 and 5). In this period, hydrogen and methane were mainly detected in the gas-phase products as seen in the initial stage of the reaction over 20 wt.% Ni/SiO2 and Ni/Al2 O3 . The formation of hydrogen once decreased, then gradually increased with production of carbon dioxide. After the contact with steam for 1.4 h, introduction of methane produced abrupt formation of hydrogen, mainly with carbon dioxide (see Table 3). This phenomenon was repeated by the experiment without the first feed of methane. That is, steam was supplied on 20 wt.% Ni/ZrO2 reduced at 500 ◦ C for 1.4 h, then methane was fed (Fig. 6). In the reaction with steam, a small quantity of hydrogen was detected without formation of carbon oxides (see Table 3). Hydrogen and carbon dioxide were mainly formed in the initial stage of the reaction with methane (see Table 2). Methane was also fed on 20 wt.% Ni/ZrO2 preheated in argon at 500 ◦ C for 1 h and without reduction. After the induction period of 0.15 h, production of hydrogen started with small quantities of carbon oxides (Fig. 7).
Fig. 7. Composition of the effluent gas in the reaction with methane over 20 wt.% Ni/ZrO2 without reduction. H2 , m/e = 2; CH4 , m/e = 16; CO, m/e = 28; CO2 , m/e = 44.
peak at around 450 ◦ C was clearly present in the profile of 20 wt.% Ni/ZrO2 while the major peak was at ca. 610 ◦ C. 3.4. Physical properties of catalysts Two major XRD peaks, attributed to Ni(1 1 1) at 44.5◦ in 2θ and to Ni(2 0 0) at 51.8◦ , were recorded with 20 wt.% Ni/SiO2 just after reduction with hydrogen at 500 ◦ C (Fig. 9a), but after the steam reforming for 5 h (see Fig. 1) the peaks attributed to NiO(1 1 1) at 37.3◦ and NiO(2 0 0) at 43.3◦ appeared with the peaks for metallic nickel (Fig. 9b) [9]. The XRD pattern of 20 wt.% Ni/Al2 O3 treated with hydrogen at 500 ◦ C was very similar to that of the sample as prepared (not shown). In the pattern, the NiO peak at 43.4◦ was recorded with the peaks attributed to ␥-alumina [12],
3.3. Temperature-programmed reduction (TPR) of nickel catalysts Reduction of 20 wt.% Ni/SiO2 started at ca. 330 ◦ C in a TPR experiment and the maximum of the reduction peak was at 430 ◦ C (Fig. 8). Small hydrogen consumption from 500 ◦ C was recorded with 20 wt.% Ni/Al2 O3 and the consumption gradually increased up to 800 ◦ C, where the TPR experiment was terminated. In the case of 5 wt.% Ni/ZrO2 , hydrogen consumption was observed from 370 ◦ C and the peak maximum was at ca. 600 ◦ C. A small consumption
111
Fig. 8. Temperature-programmed reduction of nickel catalysts.
112
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
Discernible increases in the BET surface areas were observed after the reaction in comparison with those for the as-prepared samples except 20 wt.% Ni/Al2 O3 prereduced at 700 ◦ C (see Table 4). The contents of carbon in the catalysts after the reaction are also shown in Table 4. 4. Discussion 4.1. Catalytic activity of Ni/SiO2 and Ni/Al2 O3
Fig. 9. X-ray diffraction patterns of catalysts. (a) 20 wt.% Ni/SiO2 reduced at 500 ◦ C, (b) after steam reforming for 5 h, (c) 20 wt.% Ni/Al2 O3 reduced at 700 ◦ C, (d) after reforming for 5 h, (e) 20 wt.% Ni/ZrO2 reduced at 500 ◦ C, and (f) after reforming for 5 h.
but no peaks attributed to metallic nickel were recorded. On the other hand, peaks attributed to metallic nickel at 44.5◦ and 51.8◦ were recorded with the sample reduced at 700 ◦ C (Fig. 9c). After the reaction, the intensities of these peaks decreased appreciably, and a small peak at 43.4◦ was separated in the profile (Fig. 9d). In the pattern for 20 wt.% Ni/ZrO2 reduced at 500 ◦ C, the peaks at 44.6◦ and 51.8◦ were present with small peaks attributed to NiO at 37.2◦ and 43.4◦ (Fig. 9e), while the other peaks were attributed to ZrO2 (mainly baddeleyite phase) [12]. After the reaction, the peaks attributed to NiO were diminished and the Ni peaks were intensified (Fig. 9f), showing that the nickel particles were completely reduced during the reaction. The XRD pattens for 5 wt.% Ni/ZrO2 were similar to those for 20 wt.% Ni/ZrO2 while the peaks attributed to nickel oxide and metallic nickel were weaker. Mean crystallite sizes were determined from the line broadening of the XRD peaks of Ni(1 1 1) and NiO(2 0 0) [13]. The crystallite sizes of the catalysts after the reaction were almost the same as those observed just after the reduction (Table 4). Table 4 Properties of nickel catalysts after steam reforming for 5 h Catalyst
Crystallite sizea (nm) Ni
BET surface areab (m2 g−1 )
Carbon content (wt.%)
NiO
20 wt.% Ni/SiO2
17 (15) 17
244 (237)
0.20
20 wt.% Ni/Al2 O3 c
12 (11) 11
135 (129)
0.12
5 wt.% Ni/ZrO2 20 wt.% Ni/ZrO2
11 (11) 47 (42) 12 (11) 38 (33)
0.32 0.24
Reduction temperature of catalyst: 500 ◦ C unless otherwise mentioned. a The size just after reduction is given in parentheses. b The surface area of the samples as prepared is given in parentheses. c Reduction temperature: 700 ◦ C.
Metallic nickel particles whose mean crystallite size is 15 nm are oxidized to nickel oxide on 20 wt.% Ni/SiO2 during the steam reforming at 500 ◦ C, as evidenced in the XRD experiment (cf. Fig. 9a and b); this means that steam can gradually oxidize the surface of nickel particles on silica. The same phenomenon can be seen with 20 wt.% Ni/Al2 O3 reduced at 700 ◦ C (cf. Fig. 9c and d), but metallic nickel is dominant even after the reaction. The lower reducibility of Ni/Al2 O3 (see Fig. 8) is probably caused by formation of surface spinel (NiAl2 O4 ) during the reduction process [1]. The deactivation of 20 wt.% Ni/SiO2 is probably caused by the oxidation of nickel during the reaction because nickel oxide is inactive as can be seen in 20 wt.% Ni/Al2 O3 pretreated at 500 ◦ C in which nickel oxide is dominant. Since coke formation on the catalyst is slight in the reaction period (see Table 4), it is not responsible for the deactivation. 4.2. Mechanism of the reaction with separate feed of methane and steam The rate of the reaction with methane in the absence of steam is significantly lower than that of the steady steam reforming (cf. Tables 1 and 2), suggesting that the presence of steam, which can oxidize the surface of nickel, accelerates the decomposition of methane to hydrogen. In actual, the formation rate of hydrogen is high at the initial stage of the reaction with methane on Ni/SiO2 and Ni/Al2 O3 catalysts (see Figs. 2 and 3), and formation of carbon oxides, which evidences presence of surface oxygen species, is accompanied. Carbon monoxide is selectively formed in the reaction with methane over the catalysts containing NiO, such as 20 wt.% Ni/Al2 O3 pretreated at 500 ◦ C and 20 wt.% Ni/ZrO2 without reduction (see Table 2and Fig. 7). Hence, the reduction of nickel oxide results in formation of carbon monoxide (Eq. (2)) and a reaction such as CH4 + 2O (surface) → CO2 + 2H2 ,
(8)
is less probable. Formation of hydroxyl groups on the surface of nickel (Ni−OH) is estimated in the reduction process of NiO particles; thus, it is supposed that the hydroxyl group is rather more reactive to methane and/or carbon monoxide than the lattice oxygen of nickel oxide. Formation of carbon dioxide on 20 wt.% Ni/SiO2 and 20 wt.% Ni/Al2 O3 reduced at 700 ◦ C after the reaction with steam (see Figs. 2 and 3) supports this supposition, because the contact of steam and
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
nickel often results in formation of hydroxyl groups [14]. Accumulation of oxygen in water on the surface of these catalysts (Eqs. (6) and (7)) can be detected at the initial stage of the reaction with steam after methane decomposition. In this process, reactions of H2 O → H2 + O (surface),
(9)
and 2H2 O + C (surface) → CH4 + 2O (surface),
(10)
are also probable. However, the contribution will be small because formation of carbon monoxide, which is produced in presence of nickel oxide (Eq. (2)), is small in the reaction with methane after the contact with steam (see Figs. 2 and 3). Since the formation rates of hydrogen and carbon dioxide increases gradually in the reaction with steam, carbon dioxide cannot be formed in the direct reaction between steam and surface carbon, but is probably formed by the reaction with surface hydroxyl groups whose concentration should be saturated under the continuous feed of steam. That is, C (surface) + 2OH (surface) → H2 + CO2 ,
(11)
will take place on the surface. Little methane is formed after the formation rate of carbon dioxide is saturated, implying that the surface is mostly oxidized, because formation of methane does not proceed in absence of metallic nickel which accepts the oxygen in steam (Eqs. (7) and/or (10)). After the feed of steam following the reaction with methane, formation of carbon dioxide (Eq. (1)) can be observed with re-feed of methane (see Figs. 2 and 3), suggesting formation of surface hydroxyl groups during the previous reaction with steam (Eq. (6)). When methane is decomposed on 5 and 20 wt.% Ni/ZrO2 reduced at 500 ◦ C (see Figs. 4 and 5, and Table 2), the yield of carbon dioxide in the initial stage of the reaction with methane (1% or less) is significantly smaller than that with 20 wt.% Ni/SiO2 and Ni/Al2 O3 (11 and 13%, respectively). This would mean that the quantity of the surface oxygen species such as hydroxyl groups is small on Ni/ZrO2 . Selective formation of methane in the initial stage of the reaction with steam over Ni/ZrO2 (see Table 3) suggests formation of surface oxygen species (e.g. Eqs. (7) and (10)). Although the oxygen species accumulate during the initial stage, carbon oxides are scarcely formed in the induction period. Thus, the species are inactive and different from the hydroxyl groups formed on Ni/SiO2 or Ni/Al2 O3 . The accumulation of the oxygen species is considered to take place on the surface of nickel, because the quantity of methane formed in the induction period is small on 5 wt.% Ni/ZrO2 whose nickel content is small in comparison with 20 wt.% Ni/ZrO2 (cf. Figs. 4 and 5). The reason for the formation of inactive oxygen species is not clear, but presence of zirconia should contribute to the nature of the species; thus, it is speculated that the species are rather present on the perimeter of nickel particles. The reduction temperature of
113
Ni/ZrO2 is significantly higher than that of 20 wt.% Ni/SiO2 (see Fig. 8), suggesting stronger interaction between zirconia support and nickel particles. After the feed of steam, a significantly large quantity of carbon dioxide and hydrogen is formed with the second feed of methane. The total quantity of carbon dioxide formed with 20 wt.% Ni/ZrO2 during the second feed can be evaluated as ca. 10 mmol g−1 , while the number of nickel atoms in the catalyst is only 3.4 mmol g−1 (see Fig. 5). Thus, the major source of oxygen in carbon dioxide is different from nickel hydroxide or nickel oxide and these results imply that accumulation of water takes place on the surface of the zirconia support. In the reaction between 20 wt.% Ni/ZrO2 and steam (see Fig. 6) the total quantity of oxygen atoms consumed in oxidation of nickel (Eq. (6)) is evaluated as ca. 1 mmol from the amount of hydrogen formed, and that of oxygen atoms in carbon oxides formed in the successive reaction with methane is ca. 20 mmol; these results show that methane is reacted with water adsorbed on the surface. Since water molecules are hard to be preserved at 500 ◦ C on the surface of Ni/ZrO2 , dissociation of Zr−O−Zr with H2 O to 2Zr−OH is supposed to take place. Hence, the values in the parentheses of Table 2 were calculated assuming a reaction of CH4 + 2H2 O (surface) → CO2 + 4H2 ,
(12)
instead of Eq. (1). This reaction is rather advantageous to methane decomposition to carbon (Eq. (3)) at the initial stage of the reaction with methane on Ni/ZrO2 after contact with steam (see Table 2). Since activation of methane probably takes place on the surface of nickel, the adsorption species such as Zr−OH cannot react directly with methane. Hence, it is supposed that the adsorbed water migrates to nickel surface and forms the active hydroxyl group which reacts with methane to carbon dioxide (Eq. (1)). The inactive oxygen species formed in the initial stage of the reaction with steam and surface carbon would be involved in the mechanism, but further investigation is necessary for the clarification. 4.3. Mechanism of methane steam reforming at low temperature Surface oxygen species involved in the reaction mechanism at high temperatures are shown in [1], that is, CH4 + nS∗ → CHx –S∗n +
4−x H2 , 2
(13)
CHx –S∗n + O–S∗ → CO + 21 xH2 + (n + 1)S∗ ,
(14)
H2 O + S∗ → O–S∗ + H2 ,
(15)
H2 + 2S∗ → 2H–S∗ ,
(16)
In the mechanism, where S∗ represents an active site, the presence of oxide causes formation of carbon monoxide and the process is similar to Eq. (2) in the reaction solely with methane. No reaction was observed at the initial stage of the
114
Y. Matsumura, T. Nakamori / Applied Catalysis A: General 258 (2004) 107–114
methane decomposition over 20 wt.% Ni/ZrO2 without reduction (see Fig. 7), suggesting that methane does not reduce nickel oxide. Just after the induction period, CHx is probably formed on the partially reduced surface and this results in formation of carbon monoxide and hydrogen (Eq. (14)), which also reduces the surface. Thus, formation of carbon monoxide in the steam reforming at 500 ◦ C will be caused from surface nickel oxide species. Since formation of carbon dioxide takes place in the presence of surface hydroxyl groups (Eqs. (1) and (11)), CHx –S∗n + 2HO–S∗ → CO2 + ( 21 x + 1)H2 + (n + 2)S∗ , (17) may proceed on the surface in the steam reforming. The rate of hydrogen production was 112 mmol h−1 g−1 at the initial stage of the reaction with methane after supplying steam over 20 wt.% Ni/ZrO2 (see Fig. 6 and Table 2), while the rate in the steam reforming over the same catalyst was 122 mmol h−1 g−1 at 0.5 h-on-stream (see Table 1). Hence, formation of hydroxyl groups on the surface should be an important step also in the steam reforming at 500 ◦ C. In the case of Ni/ZrO2 , accumulation of water on the support assists formation of the hydroxyl groups, and this is probably the reason why zirconia is an effective support of nickel in the steam reforming at 500 ◦ C. Formation of carbon dioxide from carbon monoxide by water–gas shift reaction is possible in the reaction mechanism, but no relationship between production of carbon monoxide and carbon dioxide can be found in the reactions solely with methane or steam.
Acknowledgements The financial support of the New Energy and Industrial Technology Development Organization of Japan is acknowledged. References [1] J.R. Rostrup-Nielsen, in: J.R. Anderson, M. Boudard, (Eds.), Catalysis, Science and Technology, vol. 5, Springer, New York, 1984, p. 3. [2] M. Oertel, J. Schmitz, W. Weirich, D. Jendryssek-Newmann, R. Schulten, Chem. Eng. Technol. 10 (1987) 248. [3] A.M. Adris, S.S.E.H. Elnashaie, R. Hughes, Can. J. Chem. Eng. 69 (1991) 1061. [4] S. Uemiya, N. Sato, H. Ando, T. Matsuda, E. Kikuchi, Appl. Catal. 67 (1991) 223. [5] US Patent 5,229,102 (1993). [6] M. Chai, M. Machida, K. Eguchi, H. Arai, Appl. Catal. A: Gen. 110 (1994) 239. [7] J. Galuszka, R.N. Pandey, S. Ahmed, Catal. Today 46 (1998) 83. [8] E. Kikuchi, Catal. Today 56 (2000) 97. [9] M.E.S. Hegarty, A.M. O’Connor, J.R.H. Ross, Catal. Today 42 (1998) 225. [10] J.-M. Wei, B.-Q. Xu, J.-L. Li, Z.-X. Cheng, Q.-M. Zhu, Appl. Catal. A 196 (2000) L167. [11] H.-S. Roh, K.-W. Jun, W.-S. Dong, S.-E. Park, Y.-S. Baek, Catal. Lett. 74 (2001) 31. [12] JCPDS Files 4-0850, 4-0835, 10-0427 and 37-1484. [13] H.P. Klug, L.E. Alexander, X-ray Diffraction Procedures, Wiley, New York, 1954. [14] J.R.H. Ross, M.C.F. Steel, J.C.S. Faraday Trans. I 69 (1973) 10.