- Email: [email protected]
Al2O3 catalysts for the dry reforming of methane
Applied Catalysis A: General 301 (2006) 9–15 www.elsevier.com/locate/apcata
Effect of potassium content in the activity of K-promoted Ni/Al2O3 catalysts for the dry reforming of methane J. Juan-Juan *, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez Department of Inorganic Chemistry, University of Alicante, P.O. Box 99, E-03080 Alicante, Spain Received 15 July 2005; received in revised form 21 October 2005; accepted 9 November 2005
Abstract In this paper the effect of the potassium content in the structure and properties of the Ni active phase and in the activity and selectivity of the NiK/Al2O3 catalysts for dry reforming of methane has been studied. The following characterization techniques were used: SEM, TEM, temperature programmed reduction (TPR-H2) and reaction (TPR-CH4), temperature programmed oxidation (TPO) and XAFS. The reforming of methane with CO2 was carried out at 973 K using 0.18 g of catalyst and a mixture CH4:CO2 (50:50, 60 ml/min, space velocity of 22,500 h1). Catalytic tests for 6 and 24 h have been developed. TPR-H2 and XAFS results reveal that potassium does not modify the arrangement of Ni atoms although facilitates the reduction of nickel species by H2 due to the modification of the interaction between metallic species and the alumina support. The activity data indicate that the addition of a low amount of potassium (0.2 wt.% K2O) allows to obtain a catalyst with an acceptably high activity (over 63% methane conversion, close to thermodynamic equilibrium) and very low coke deposition (below 30 mg C/g cat. during 6 h reaction). Independently of the amount of potassium, the catalytic activity remains almost constant during at least 24 h. # 2005 Elsevier B.V. All rights reserved. Keywords: Dry methane reforming; NiK/Al2O3 catalyst; K content; XAFS; TPR
1. Introduction The dry reforming of methane allows the conversion of two undesirable greenhouse gases into synthesis gas with a low H2/ CO ratio, adequate for hydroformylation and carbonylation reactions as well as for both methanol and Fischer–Tropsch synthesis [1,2]: CH4 þ CO2 ! 2H2 þ 2CO
(1)
That is, this reforming reaction has an economical and environmental interests because it allows the conversion of two greenhouse gases (CH4 and CO2) into a valuable feedstock. Furthermore, considering the reaction stoichiometry (CH4/CO2 = 1), it is specially convenient for gases coming from sources where similar amounts of CO2 and CH4 are present as, for example, biogas or some high CO2 content gas natural streams.
* Corresponding author. Tel.: +34 965903975; fax: +34 965903454. E-mail address: [email protected] (J. Juan-Juan). 0926-860X/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2005.11.006
The reforming of methane with CO2 is accompanied by some side reactions that determine the selectivity of the process: CO2 þ H2 ! H2 O þ CO
(2)
2CO ! CO2 þ C
(3)
CH4 ! C þ 2H2
(4)
These competitive reactions modify the CO2 and CH4 equilibrium conversion: reaction (2) increases the CO2 conversion and the yield of CO, reaction (3) decreases the CO2 conversion and the yield of CO and, finally, reaction (4) increases the methane conversion and the yield of H2. The last two reactions are, also, responsible for coke formation during reaction. It is well known that supported metals of groups 8, 9 and 10 are good catalysts for the dry reforming of methane (reaction (1)) [2–10]. Among those metals, Ni is specially interesting for industrial application due, mainly, to its lower cost and higher availability. Thus, Ni based catalysts are under investigation in order to improve both their activity and, particularly, their selectivity with the objective of avoiding
10
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
or, at least hindering, coke formation. As variables of such investigation, the following are attracting most of the attention: the nature of the support [11–17], the preparation method [13,18–21] and the addition of promoters [12,19,22– 32]. In relation with the last variable, it can be mentioned that interesting points are the nature of the promoter, being K one the most investigated [13,29–32], and the amount used, generally considered as promoter/Ni ratio. In a previous paper, we have reported the characterization and catalytic performance of Ni/Al2O3 and NiK/Al2O3 (K/ Ni = 0.7) catalysts in the dry reforming of methane [3]. A very positive effect of potassium in decreasing coke deposition was observed, and the XAFS analysis of the catalysts allowed to conclude that potassium species behave, mainly, as an independent catalyst for the gasification of the deposited coke. Further research on this topic, subject of the present work, deals with the analysis of the effect of the potassium content both in the structure and properties of the Ni active phase and in the activity and selectivity of the catalyst for dry reforming of methane. It must be also mentioned that compared to the previous work [3], and in order to use reaction conditions closer to those in real systems (biogas), an undiluted CO2:CH4 (1:1) gaseous mixture has been used. 2. Experimental 2.1. Catalysts preparation A g-Al2O3 (supplied by Across, SBET = 90 m2/g) was used as catalysts support. It was first impregnated, by using excesssolution impregnation method, with a KNO3 aqueous solution of the required concentration to obtain different potassium contents, from 0.2 to 5 wt.% K2O. After solvent removal (by an air flow at 298 K), the catalysts were dried overnight at 373 K and calcined in air at 773 K for 2 h. Ni was also introduced by excess-solution impregnation, using an aqueous solution of Ni(NO3)26H2O of the appropriate concentration to obtain a 10 wt.% (nominal) nickel content, followed by drying overnight at 373 K. 2.2. Catalysts characterization Nickel and potassium content was determined by atomic absorption spectroscopy (AAS). The metals were extracted from the catalyst by a reflux treatment with diluted HCl for 8 h. The morphology of the catalysts and of the coke deposited during reforming reaction were analyzed by scanning electron microscopy using a Hitachi S-3000N equipment. Transmission electron microscopy (JEOL JEM-2010) was used to estimate the size of Ni particles in used catalysts. Temperature programmed reduction (TPR-H2) was performed in an automatic equipment (PulseChemisorb 2705, Micromeritics) with a thermal conductivity detector. The fresh catalyst (50mg) was submitted to a heat treatment (10 K/min up to 1173 K) in a gas flow (60 ml/min) of the mixture H2:N2 (5:95). Previous to the TPR experiments, the samples were heat treated under inert atmosphere, at 673 K for 2 h, to produce the
decomposition of nickel nitrate, which would lead a strong interference in the TCD analysis of H2. Temperature programmed reaction with methane (TPRCH4) was performed in quartz reactor (0.4 cm of diameter) using a mass spectrometer (Balzers Thermostar) for the product analysis. The fresh catalyst was submitted to a heat treatment (10 K/min up to 1173 K) in a gas flow (60 ml/min) of the mixture CH4:He (5000 ppm of CH4). XAFS measurements were performed at the BL-7C, BL-9 and BL1B101 stations of the Photon Factory in the National Laboratory for High Energy Physics (KEK-PF) in Tsukuba, Japan. A Si (1 1 1) double crystal was used to monochromatize the X-ray beam form the 2.5 GeV electron storage ring. The NiK-edge absorption spectra were recorded in the transmission mode at room temperature, in a range of photon energy extending from 8080 to 9375 eV. A Fourier transform was done on the k3-weighted EXAFS oscillations over the range of 3– ˚ 1 (FT-EXAFS). XAFS measurements were done on the 12 A catalysts after a reduction treatment (pure H2, 973 K, 2 h) and after reaction (973 K, 6 h). After these treatments, the catalysts were transferred without exposure to air to a glovebox where they were vacuum packed. The used catalysts were submitted to temperature programmed oxidation (TPO) experiments to determine the amount of carbon deposited on the catalysts. These experiments were carried out in a thermobalance (TA SDT-2960) coupled to a mass spectrometer (Balzers Thermostar) under the following conditions: 20 mg of catalyst were submitted to a heat treatment (20 K/min up to 1223 K) with a gas mixture of 16 vol.% O2 in He (100 ml/min). To determine the amount of carbon deposited on the catalysts, m/z = 44 and m/z = 28 signals and the weight loss were simultaneously recorded. 2.3. Catalytic activity tests The reforming of methane with CO2 was carried out at 973 K using 0.18 g of catalyst and a gas mixture CH4:CO2 (50:50, 60 ml/min, space velocity of 22,500 h1). Catalytic tests were carried out for 6 and 24 h. The equipment used consisted basically of a quartz tube fixed-bed reactor coupled to gas chromatograph (HP5890 series II) with a thermal conductivity detector. Note that the catalysts have not been reduced before reaction because previous results [33] showed that this treatment does not modify the catalytic activity. 3. Results and discussion 3.1. Characterization of catalysts Table 1 shows the K and Ni content of the different catalysts prepared. Ni loading ranges from around 5 to 9 wt.%, approximately, and K loading (as K2O) from 0.18 to 5 wt.%. To describe the composition of the different catalysts, the K/Ni atomic ratio has been used. Fig. 1 shows the TPR-H2 curves of some catalysts: Ni/ Al2O3, NiK/Al2O3-3, NiK/Al2O3-6 y NiK/Al2O3-8 (nomen-
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
11
Table 1 Nickel and potassium content of catalysts Catalysts
Ni (wt.%)
K2O (wt.%)
K/Ni ratio
Ni/Al2O3 NiK/Al2O3-1 NiK/Al2O3-2 NiK/Al2O3-3 NiK/Al2O3-4 NiK/Al2O3-5 NiK/Al2O3-6 NiK/Al2O3-7 NiK/Al2O3-8
6.63 6.13 4.87 6.00 6.22 6.36 7.00 8.91 9.00
0.00 0.22 0.18 0.37 0.49 0.85 1.26 1.83 5.00
0.00 0.04 0.05 0.08 0.10 0.17 0.22 0.26 0.69
clature as indicated in Table 1). For the free-potassium catalyst (Ni/Al2O3), the maximum in the hydrogen consumption is located at about 873 K, which corresponds to the expected reduction temperature for NiO in intimate contact with alumina supports [13]. In the case of the potassium containing catalysts very different profiles are found. For the lowest potassium content catalyst analyzed (NiK/Al2O3-3) apart of the maximum at 873 K, a shoulder at around 673 K is detected. However, as the potassium content increases, the TPR-H2 profiles are more complex, that is, for the catalyst with the intermediate potassium content (NiK/Al2O3-6) several maxima are found at about 623, 673, 873 and 1000 K; and, for the highest potassium content catalyst (NiK/Al2O3-8), two clearly resolved maxima are observed at 673 and 1000 K, approximately. In this case, it is remarkable that the maximum at 873 K, assigned to the reduction of NiO in intimate contact with the support, is not observed. The results shown in Fig. 1 mean that in the potassium containing catalysts, nickel is present in several states with different reducibility [13,31]. Recently, Hou et al. [13], using calcium as promoter, observed a similar effect in the reducibility of nickel species. These authors assigned the maximum at 673 K to the reduction of NiO particles without interaction with the support [13,34] and, the maximum at 1100 K, approximately, to the reduction of NiAl2O4 [13,35– 37]. In our case, it seems that potassium also modifies the
Fig. 2. FT-EXAFS of reduced catalysts: Ni/Al2O3, NiK/Al2O3-1, NiK/Al2O3-3 and NiK/Al2O3-6.
interaction of nickel with the support, and, consequently, the reducibility of nickel species. Thus, for catalyst NiK/Al2O3-8, the most intense peak is located at about 673 K, so, it can be due to the reduction of NiO with very low interaction with support. The second peak, with a maximum at 1000 K, seems to be due to the reduction of NiO particles with a very strong contact with the support, probably NiAl2O4. Note that the intermediate potassium content catalyst (NiK/Al2O3-6) shows maxima close to the two above extreme peaks, indicating that NiO particles with different extent of contact with support have been formed. In conclusion, the TPR-H2 data clearly reveal that potassium modifies the contact between nickel oxide particles and the support and, because of that, favours the reduction, by H2, of a fraction of nickel. Fig. 2 shows the FT-EXAFS data obtained for some reduced catalysts: Ni/Al2O3, NiK/Al2O3-1, NiK/Al2O3-3 and NiK/ Al2O3-6. The signals corresponding to NiO and Ni foil have been also included as reference. Considering that Ni–Ni bonds ˚ , it can in metallic nickel appear at a distance slightly over 2 A be deduced, in agreement with Baley et al. [38] and with our previous results [3,39] that in these catalysts, Ni is only partially reduced. Note that NiK/Al2O3-6 shows a well defined ˚ . This fact indicates that the addition of a maximum at about 2 A certain amount of potassium favours the reduction of the Ni phase. It must be pointed out that for the catalysts with lower K/ Ni ratio (NiK/Al2O3-1 and NiK/Al2O3-3), the mentioned effect of potassium is not observed. The XANES data (not presented here) confirm these results because the intensity of the white line is the lowest for sample NiK/Al2O3-6. 3.2. Catalytic activity, coke deposition and characterization of used catalysts
Fig. 1. TPR-H2 of Ni/Al2O3, NiK/Al2O3-3, NiK/Al2O3-6 and NiK/Al2O3-8 catalysts.
Fig. 3 shows the data of catalytic activity for all catalysts, expressed as methane and carbon dioxide conversion (Fig. 3a), and the coke deposition (in mg of carbon per gram of catalyst) (Fig. 3b) after 6 h of time-on-stream versus the K/Ni ratio. In Fig. 3a, and in agreement with literature and previous results [3,12,22,23], a decrease in the conversion of both CH4
12
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
Fig. 3. Effect of K/Ni ratio on: (a) CO2 and CH4 conversion at 973 K and (b) amount of coke deposited during 6 h reaction at 973 K.
and CO2 with the amount of potassium in the catalysts is observed. Methane conversion decreases from 70% (near to the thermodynamic equilibrium at the reaction conditions) for the K-free catalyst, to 47% for the sample with K/Ni = 0.69 (NiK/Al2O3-8). Note that the rate of methane conversion decrease is higher at low K-contents. In literature [40], it is mentioned that potassium can migrate from the support to the nickel particles, and be placed, predominantly, on their surface. Recent studies [41] have demonstrated, by DFT calculations that a stepped nickel surface is more active than a surface with close-packed terraces for hydrocarbon reforming reaction and, also, for coke formation. Additionally, these studies conclude that potassium (and other promoters) is preferentially located on the step edges of the nickel surface. The authors observed that, in order to prevent the nucleation of coke, a small fraction of active sites have to be blocked on the stepped nickel surface. This blockage, however, does not stop the reforming reaction because it can still take place on the unblocked, less active, sites of the nickel surface (close-packed terraces). Snoeck and Froment [40], after their study about the effect of potassium addition to nickel catalyst for CH4 steam reforming, proposed that the most important effect of potassium is the decrease of the value of the lumped forward rate coefficient for methane craking. This is due to a reduction of the number of sites available for methane decomposition as a consequence of the presence of potassium on the metal surface. Also, Bengaard et al. [41,42] reported that the methane sticking probability is strongly diminished by potassium. On the basis of these conclusions, the results obtained in this work with the K-
containing catalysts can be explained. A fraction of potassium migrates from the support to the nickel surface and it is located, firstly, on the step sites that are the most active sites for reforming reaction. Consequently, a low potassium content produces a large decrease of methane conversion. When all the step sites are blocked, potassium occupies, gradually, other less active nickel sites and the decrease of methane conversion is lower. Fig. 3a shows that the CO2 conversion decreases in a lower extent than methane conversion. It is well known [3,43] that potassium is a good catalyst for carbon gasification and the increase of the K/Ni ratio could enhance the rate of coke gasification by CO2. This contributes to the increase of CO2 conversion, explaining the different behaviour of CH4 and CO2 conversion at higher K/Ni ratio. The analysis of the reaction products (H2 and CO) reveals that the presence of the promoter produces a slight decrease of the H2/CO ratio (from 0.86 with Ni/Al2O3 to 0.77 with Ni/Al2O3-7). This result, which is interesting from a practical point of view, is consequence of the two effects of potassium previously mentioned: the blockage of Ni active sites for methane decomposition (reaction (4)) and the enhancement of coke gasification (reaction (3)). In relation with carbon accumulation, Fig. 3b shows that the addition of a small amount of potassium has an important and clear effect in reducing the amount of coke deposited (from 280 mg C/g cat. for the K-free catalyst to 30 mg C/ g cat. for the sample with a K/Ni ratio of 0.04, NiK/Al2O3-1). A higher K/Ni ratio produces a further decrease in the amount of the deposited carbon, which becomes negligible for a K/Ni ratio of about 0.26 (less than 2 wt.% K2O). These results are consistent with the fact that the most active sites, responsible for the carbon formation, are firstly blocked by the alkali promoter [41]. The SEM images of two potassium-containing samples (NiK/Al2O3-2 and NiK/Al2O3-7), used in the reforming reaction, are shown in Fig. 4. A comparison between them confirms that a lower amount of coke is deposited on the sample with higher potassium content (Fig. 4B). In addition, it can be observed the filamentous nature of the deposited coke, as it was also observed for the potassium-free sample (not shown) [3]. This fact indicates that the promoter, in the conditions used in this work, decreases the amount of coke deposited but it does not modify its structural characteristics. Thus, it can be concluded that, by addition of a low amount of potassium, it is possible to obtain a catalyst with an acceptably high activity for methane reforming and very low carbon deposition. Among the analyzed catalysts, the one with about 0.2 wt.% K2O (NiK/ Al2O3-1) shows the optimal features: over 63% methane conversion (close to thermodynamic equilibrium) and a coke deposition lower than 30 mg C/g cat. To go more deeply into the effect of potassium in methane conversion, TPR-CH4 experiments were carried out (see experimental section for details) with three selected catalysts (Ni/Al2O3, NiK/Al2O3-4 and NiK/Al2O3-8). The results obtained, shown in Fig. 5, indicate that the presence of potassium, as expected, notably affects the performance of Ni/ Al2O3 catalysts for CH4 conversion [40,41]. In fact, a shift of
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
13
Fig. 6. FT-EXAFS of used catalysts (6 h): Ni/Al2O3, NiK/Al2O3-1, NiK/Al2O33 and NiK/Al2O3-6.
Fig. 4. SEM images of used catalysts (6 h): (A) NiK/Al2O3-2 and (B) NiK/ Al2O3-7.
the onset temperature for CH4 conversion to higher values (from around 773 K for Ni/Al2O3 to 900 K for NiK/Al2O3-4 and to 973 K for NiK/Al2O3-8) is observed. To explain this delay it must be taken into account that: (i) methane decomposition occurs on metallic Ni sites, (ii) as the samples
Fig. 5. TPR-CH4 of Ni/Al2O3, NiK/Al2O3-4 and NiK/Al2O3-8 catalysts.
have not been previously reduced, metallic Ni sites are being formed during the TPR-CH4 by reaction with methane and (iii) the presence of K hinders the formation of Ni active sites, probably because, due to its electropositive character, decreases the affinity of Ni for methane. The TPR-CH4 profiles found can be interpreted considering that, as soon as the active metallic sites are formed, methane decomposition takes place quickly. In the case of sample NiK/Al2O3-8, the shape of the TPR-CH4 profiles indicates that the Ni active sites are being formed slowly as the temperature increases. Fig. 6 shows the FT-EXAFS obtained for catalysts Ni/Al2O3, NiK/Al2O3-1, NiK/Al2O3-3 and NiK/Al2O3-6 after having been used in the dry reforming of methane for 6 h. The signal corresponding to Ni foil has been included as reference. The profiles presented in Fig. 6 indicate that the four catalysts show a similar Ni structure that is close to that of Ni foil, although probably, with a lower coordination number. This means, as previously observed [3,26,39] that the catalysts become further reduced under reaction conditions. For the K-containing samples, the intensity of the FT-EXAFS main signal is higher, this can be interpreted by considering that in these samples a higher fraction of nickel is in the metallic form (Ni0) or that the Ni particles are larger. The former hypothesis is in agreement with the conclusion extracted from TPR-H2 about the effect of potassium in the reducibility of nickel species. TEM analysis of two representative samples (Ni/Al2O3 and NiK/Al2O3-7) indicates that, in both cases, the mean size of Ni particles is between 10 and 15 nm. So, it can be concluded that the addition of potassium does not modify the Ni particle size. As the EXAFS results reveal a similar Ni structure in catalysts with and without potassium, it can be concluded that the promoter does not modify the arrangement of Ni atoms, although facilitates the reduction of nickel species by H2, probably due to the modification of the interaction between metallic species and the support. The catalytic activity data previously commented (Fig. 3) correspond to 6 h reaction time experiments. It is worth noting that in all cases, that is, independently of the K content, the catalytic activity remains almost constant during the whole
14
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
Fig. 7. CO2 and CH4 conversion at 973 K (24 h reaction) for Ni/Al2O3 and NiK/ Al2O3-7 catalysts.
Fig. 8. SEM image of used catalyst NiK/Al2O3-7 (24 h).
experiment, although a very slight decrease in the conversion of methane for the K-free catalyst is observed. The negative effect of the deposited coke on the catalysts performance should be patent, if so, after a longer reaction time. In order to check this point, some catalysts have been tested in a period of 24 h. Fig. 7 shows the results obtained with the K-free catalyst and with sample NiK/Al2O3-7 (K/Ni ratio of 0.26). The most remarkable about these data is that no deactivation has been observed during this long reaction period. It means that the deposited coke does not deactivate the Ni active phase, that is, Ni particles are not being blocked by the coke formed. SEM data of this work (Fig. 4) and of previous publications [3,14,30,44,45] show that the coke deposited on the catalysts surface has a fibrous structure. The growth of carbon filaments occurs on a metal surface where the active metal is carried on the top of the carbon filament [46] and, as consequence, the catalytic activity can be kept with reaction time because the active metal is still exposed to the reactants. Regarding the amount of carbon deposited during the 24 hexperiments on Ni/Al2O3 and NiK/Al2O3-7 catalysts, as expected, a larger amount of coke has been accumulated on the K-free catalyst. However, the rate of coke accumulation is not constant with time. Thus, the amount of carbon deposited on sample Ni/Al2O3 is lower (about 680 mg C/g cat.) than the expected value (1120 mg C/g cat., calculated considering a linear progression from the data obtained at 6 h). However, for the potassium containing catalyst, it is higher than the expected value (about 150 mg C/g cat. in front of the expected 20 mg C/g cat.). Fig. 8 shows the SEM image of NiK/Al2O3-7 catalyst after 24 h reaction. It can be observed that the amount of coke deposited is higher that in the 6 h reaction experiments (see Fig. 4B) and that the structure is also fibrous. The behaviour of the Ni/Al2O3 catalyst can be explained considering that the coke being accumulated produces the blockage of the most active sites for coke formation, hindering methane cracking and, as consequence, decreasing the amount of coke deposited. This explanation also justifies the slight decrease with time in methane
conversion (related to Eq. (4)) shown in Fig. 7 for the potassium-free catalyst. In order to explain the behaviour of the K-containing catalyst it has to be considered that, as the coke deposit grows (in form of carbon filaments, see Fig. 8), the contact between potassium species (mainly on the support), and the nickel particles and the carbon material formed on them, decreases. This implies that the formation of coke is less restricted and the gasification reaction takes place in a lower extent. In addition a possible loss of potassium from the catalyst (by volatilization) under reaction conditions could also contribute the observed behaviour. However, it has been proved (by XRF measurements) that the amount of promoter remains constant after reforming reaction.
4. Conclusions The presence of potassium in Ni/Al2O3 catalysts hinders the accumulation of coke on the catalyst surface during the dry reforming of methane, but produces a decrease in the catalytic activity. There is, however, an optimum amount of potassium (0.2 wt.% K2O) that leads to a catalyst with a very low coke deposition and a high catalytic activity (more than 90% reduction in coke deposition and less than 10% decrease in the catalytic activity). Independently of the amount of potassium, the catalytic activity remains almost constant during at least 24 h. This indicates that the coke deposit does not deactivate the catalysts because Ni particles are on the top of carbon filaments formed during reaction. So, the effect of deposited coke would be a plug up of the reactor. The study about the effect of the amount of potassium allows to confirm the following ideas: (i) the addition of potassium increases the reducibility of nickel species by hydrogen because it modifies the interaction of nickel oxide with the Al2O3 support, (ii) potassium migrates from the support to the Ni surface and neutralizes a fraction of the active sites, (iii) potassium catalyzes the gasification of the coke formed during
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
reaction without modification of its structure and (iv) potassium does no modify the size nor the structure of the Ni particles. Acknowledgments Dr. H. Yamashita is gratefully acknowledged for performing the XAFS measurements. The authors thank the Spanish government (PPQ2002-01025) and the Photon Factory (Japan) project 2003G067, for financial support and J. Juan-Juan thanks his thesis grant to MCYT of Spain. References [1] M.C.B. Bradford, M.A. Vannice, Catal. Rev. Sci. Eng. 41 (1999) 1. [2] J.R. Rostrup-Nielsen, J. Catal. 144 (1993) 38. [3] J. Juan-Juan, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez, Appl. Catal. A: Gen. 264 (2004) 169. [4] M.C.B. Bradford, M.A. Vannice, Catal. Today 50 (1999) 87. [5] M.C.B. Bradford, M.A. Vannice, Appl. Catal. A: Gen. 142 (1996) 97. [6] A. Erdo¨helyi, J. Cserenyi, F. Solymosi, J. Catal. 141 (1993) 287. [7] Z. Zhang, U.A. Tsipouriari, A.M. Efstathiou, X.E. Verkyos, J. Catal. 158 (1996) 51. [8] U.L. Portugal, A.C.S.F. Santos, S. Damyonova, C.P.P. Marques, J.M.C. Bueno, J. Mol. Catal. A: Chem. 184 (2002) 311. [9] A. Erdo¨helyi, J. Cserenyi, E. Papp, F. Solymosi, Appl. Catal. A: Gen. 108 (1994) 205. [10] A. Erdo¨helyi, K. Forder, F. Solymosi, Stud. Surf. Sci. Catal. 107 (1997) 525. [11] K.S. Hwang, H.Y. Zhu, G.Q. Lu, Catal. Today 68 (2001) 183. [12] A. Lemonidu, M.A. Goula, I.A. Vasalos, Catal. Today 46 (1998) 175. [13] Z. Hou, O. Yokota, T. Tanaka, T. Yashima, Appl. Catal. A: Gen. 253 (2003) 381. [14] J. Juan-Juan, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez, in: Proceedings of the Sixth European Europacat-VI Congress, Austria, (2003. [15] K. Takehira, T. Shishido, P. Wang, T. Kosaka, K. Takaki, J. Catal. 221 (2004) 43. [16] K. Nagaoka, A. Jentys, J.A. Lercher, J. Catal. 230 (2005) 185. [17] A. Golebioswski, K. Stolecki, U. Prokop, A. Kusmieroska, T. Borowiecki, A. Denis, C. Sikorska, React. Kinet. Catal. Lett. 82 (2004) 179. [18] S. Tang, L. Ji, J. Lin, H.C. Zeng, K.L. Tan, K. Li, J. Catal. 194 (2000) 424. [19] J.A.C. Dias, J.M. Assaf, Catal. Today 85 (2003) 59.
15
[20] K. Takehira, T. Shishido, D. Shoro, K. Muramaki, M. Honda, T. Kaabata, K. Takaki, Catal. Commun. 5 (2004) 209. [21] K. Takanabe, K. Nagaoka, K. Nariair, K. Aika, J. Catal. 230 (2005) 75. [22] F. Frusteri, F. Arena, G. Galogero, T. Torre, A. Parmaliana, Catal. Commun. 2 (2001) 29. [23] J.C. Rodrı´guez, E. Romero, J.L.G. Fierro, J. Santamarı´a, A. Monzo´n, Catal. Today 37 (1997) 255. [24] J.J. Richardson, M. Lei, B. Turk, K. Foster, M.V. Twigg, Appl. Catal. A: Gen. 110 (1994) 217. [25] K. Tomishige, O. Yamazaki, Y. Chen, K. Yokoyama, X. Li, K. Fujimoto, Catal. Today 45 (1998) 35. [26] S. Takeneka, E. Kato, Y. Tomikubo, K. Otsuka, J. Catal. 219 (2003) 176. [27] H. Chen, C. Wang, C. Yu, L. Teng, P. Liao, Catal. Today 97 (2004) 173. [28] A. Valentini, N. Villareal, L. Dias, P. Lisboa, W. Schreiner, E. Leite, E. Longo, Appl. Catal. A: Gen. 255 (2003) 211. [29] Z. Hou, Z. Xiaoming, T. Yashima, React. Kin. Catal. Lett. 84 (2) (2005) 229. [30] F. Frusteri, L. Spadaro, F. Arena, A. Chuvilin, Carbon 40 (7) (2002) 1063– 1070. [31] T. Osaki, T. Mori, J. Catal. 204 (2001) 89. [32] J.S. Chang, S.E. Park, H. Chon, Appl. Catal. A: Gen. 145 (1996) 111. [33] J. Juan-Juan, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez, in: Proceedings of the SECAT 03 Congress, Spain, (2003), p. 83. [34] C.W. Hu, J. Yao, H.Q. Yang, Y. Chen, A.M. Tian, J. Catal. 166 (1997) 1. [35] A.A. Lemonidou, I.A. Vasalos, Appl. Catal. A: Gen. 228 (2002) 227. [36] A.M. Siskin, R.H. Cunningham, R.M. Ormerod, Catal. Today 46 (1998) 147. [37] R. Molina, G. Poncelet, J. Catal. 173 (1998) 257. [38] K.M. Baley, T.K. Campbell, J.L. Falconer, Appl. Catal. 54 (1989) 159. [39] J. Juan-Juan, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez, Photon Factory Activity Report, 21 (Part B) (2003) 82. [40] W. Snoeck, G.F. Froment, Ind. Eng. Chem. Res. 41 (2002) 3548. [41] H.S. Bengaard, J.K. Norskov, J. Sehested, B.S. Clausen, L.P. Nielsen, A.M. Molenbroek, J.R. Rostrup-Nielsen, J. Catal. 209 (2002) 365. [42] H.S. Bengaard, I. Alstrup, I. Chorkendorff, S. Ullmann, J.R. RostrupNielsen, J.K. Norskov, J. Catal. 187 (1999) 238. [43] R. Meijer, H.J. Mu¨hlen, F. Kapteijn, J.A. Moulijn, Fuel Process. Technol. 28 (1991) 5. [44] O.S. Joo, K.P. Jung, Bull. Korean Chem. Soc. 23 (2002) 1149. [45] F. Arena, F. Frusteri, A. Parmaliana, Appl. Catal. A: Gen. 187 (1999) 127. [46] M. Parvary, S.H. Jazayeri, A. Taeb, K. Petit, A. Kiennermann, Catal. Commun. 2 (2001) 357.
Effect of potassium content in the activity of K-promoted Ni/Al2O3 catalysts for the dry reforming of methane J. Juan-Juan *, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez Department of Inorganic Chemistry, University of Alicante, P.O. Box 99, E-03080 Alicante, Spain Received 15 July 2005; received in revised form 21 October 2005; accepted 9 November 2005
Abstract In this paper the effect of the potassium content in the structure and properties of the Ni active phase and in the activity and selectivity of the NiK/Al2O3 catalysts for dry reforming of methane has been studied. The following characterization techniques were used: SEM, TEM, temperature programmed reduction (TPR-H2) and reaction (TPR-CH4), temperature programmed oxidation (TPO) and XAFS. The reforming of methane with CO2 was carried out at 973 K using 0.18 g of catalyst and a mixture CH4:CO2 (50:50, 60 ml/min, space velocity of 22,500 h1). Catalytic tests for 6 and 24 h have been developed. TPR-H2 and XAFS results reveal that potassium does not modify the arrangement of Ni atoms although facilitates the reduction of nickel species by H2 due to the modification of the interaction between metallic species and the alumina support. The activity data indicate that the addition of a low amount of potassium (0.2 wt.% K2O) allows to obtain a catalyst with an acceptably high activity (over 63% methane conversion, close to thermodynamic equilibrium) and very low coke deposition (below 30 mg C/g cat. during 6 h reaction). Independently of the amount of potassium, the catalytic activity remains almost constant during at least 24 h. # 2005 Elsevier B.V. All rights reserved. Keywords: Dry methane reforming; NiK/Al2O3 catalyst; K content; XAFS; TPR
1. Introduction The dry reforming of methane allows the conversion of two undesirable greenhouse gases into synthesis gas with a low H2/ CO ratio, adequate for hydroformylation and carbonylation reactions as well as for both methanol and Fischer–Tropsch synthesis [1,2]: CH4 þ CO2 ! 2H2 þ 2CO
(1)
That is, this reforming reaction has an economical and environmental interests because it allows the conversion of two greenhouse gases (CH4 and CO2) into a valuable feedstock. Furthermore, considering the reaction stoichiometry (CH4/CO2 = 1), it is specially convenient for gases coming from sources where similar amounts of CO2 and CH4 are present as, for example, biogas or some high CO2 content gas natural streams.
* Corresponding author. Tel.: +34 965903975; fax: +34 965903454. E-mail address: [email protected] (J. Juan-Juan). 0926-860X/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2005.11.006
The reforming of methane with CO2 is accompanied by some side reactions that determine the selectivity of the process: CO2 þ H2 ! H2 O þ CO
(2)
2CO ! CO2 þ C
(3)
CH4 ! C þ 2H2
(4)
These competitive reactions modify the CO2 and CH4 equilibrium conversion: reaction (2) increases the CO2 conversion and the yield of CO, reaction (3) decreases the CO2 conversion and the yield of CO and, finally, reaction (4) increases the methane conversion and the yield of H2. The last two reactions are, also, responsible for coke formation during reaction. It is well known that supported metals of groups 8, 9 and 10 are good catalysts for the dry reforming of methane (reaction (1)) [2–10]. Among those metals, Ni is specially interesting for industrial application due, mainly, to its lower cost and higher availability. Thus, Ni based catalysts are under investigation in order to improve both their activity and, particularly, their selectivity with the objective of avoiding
10
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
or, at least hindering, coke formation. As variables of such investigation, the following are attracting most of the attention: the nature of the support [11–17], the preparation method [13,18–21] and the addition of promoters [12,19,22– 32]. In relation with the last variable, it can be mentioned that interesting points are the nature of the promoter, being K one the most investigated [13,29–32], and the amount used, generally considered as promoter/Ni ratio. In a previous paper, we have reported the characterization and catalytic performance of Ni/Al2O3 and NiK/Al2O3 (K/ Ni = 0.7) catalysts in the dry reforming of methane [3]. A very positive effect of potassium in decreasing coke deposition was observed, and the XAFS analysis of the catalysts allowed to conclude that potassium species behave, mainly, as an independent catalyst for the gasification of the deposited coke. Further research on this topic, subject of the present work, deals with the analysis of the effect of the potassium content both in the structure and properties of the Ni active phase and in the activity and selectivity of the catalyst for dry reforming of methane. It must be also mentioned that compared to the previous work [3], and in order to use reaction conditions closer to those in real systems (biogas), an undiluted CO2:CH4 (1:1) gaseous mixture has been used. 2. Experimental 2.1. Catalysts preparation A g-Al2O3 (supplied by Across, SBET = 90 m2/g) was used as catalysts support. It was first impregnated, by using excesssolution impregnation method, with a KNO3 aqueous solution of the required concentration to obtain different potassium contents, from 0.2 to 5 wt.% K2O. After solvent removal (by an air flow at 298 K), the catalysts were dried overnight at 373 K and calcined in air at 773 K for 2 h. Ni was also introduced by excess-solution impregnation, using an aqueous solution of Ni(NO3)26H2O of the appropriate concentration to obtain a 10 wt.% (nominal) nickel content, followed by drying overnight at 373 K. 2.2. Catalysts characterization Nickel and potassium content was determined by atomic absorption spectroscopy (AAS). The metals were extracted from the catalyst by a reflux treatment with diluted HCl for 8 h. The morphology of the catalysts and of the coke deposited during reforming reaction were analyzed by scanning electron microscopy using a Hitachi S-3000N equipment. Transmission electron microscopy (JEOL JEM-2010) was used to estimate the size of Ni particles in used catalysts. Temperature programmed reduction (TPR-H2) was performed in an automatic equipment (PulseChemisorb 2705, Micromeritics) with a thermal conductivity detector. The fresh catalyst (50mg) was submitted to a heat treatment (10 K/min up to 1173 K) in a gas flow (60 ml/min) of the mixture H2:N2 (5:95). Previous to the TPR experiments, the samples were heat treated under inert atmosphere, at 673 K for 2 h, to produce the
decomposition of nickel nitrate, which would lead a strong interference in the TCD analysis of H2. Temperature programmed reaction with methane (TPRCH4) was performed in quartz reactor (0.4 cm of diameter) using a mass spectrometer (Balzers Thermostar) for the product analysis. The fresh catalyst was submitted to a heat treatment (10 K/min up to 1173 K) in a gas flow (60 ml/min) of the mixture CH4:He (5000 ppm of CH4). XAFS measurements were performed at the BL-7C, BL-9 and BL1B101 stations of the Photon Factory in the National Laboratory for High Energy Physics (KEK-PF) in Tsukuba, Japan. A Si (1 1 1) double crystal was used to monochromatize the X-ray beam form the 2.5 GeV electron storage ring. The NiK-edge absorption spectra were recorded in the transmission mode at room temperature, in a range of photon energy extending from 8080 to 9375 eV. A Fourier transform was done on the k3-weighted EXAFS oscillations over the range of 3– ˚ 1 (FT-EXAFS). XAFS measurements were done on the 12 A catalysts after a reduction treatment (pure H2, 973 K, 2 h) and after reaction (973 K, 6 h). After these treatments, the catalysts were transferred without exposure to air to a glovebox where they were vacuum packed. The used catalysts were submitted to temperature programmed oxidation (TPO) experiments to determine the amount of carbon deposited on the catalysts. These experiments were carried out in a thermobalance (TA SDT-2960) coupled to a mass spectrometer (Balzers Thermostar) under the following conditions: 20 mg of catalyst were submitted to a heat treatment (20 K/min up to 1223 K) with a gas mixture of 16 vol.% O2 in He (100 ml/min). To determine the amount of carbon deposited on the catalysts, m/z = 44 and m/z = 28 signals and the weight loss were simultaneously recorded. 2.3. Catalytic activity tests The reforming of methane with CO2 was carried out at 973 K using 0.18 g of catalyst and a gas mixture CH4:CO2 (50:50, 60 ml/min, space velocity of 22,500 h1). Catalytic tests were carried out for 6 and 24 h. The equipment used consisted basically of a quartz tube fixed-bed reactor coupled to gas chromatograph (HP5890 series II) with a thermal conductivity detector. Note that the catalysts have not been reduced before reaction because previous results [33] showed that this treatment does not modify the catalytic activity. 3. Results and discussion 3.1. Characterization of catalysts Table 1 shows the K and Ni content of the different catalysts prepared. Ni loading ranges from around 5 to 9 wt.%, approximately, and K loading (as K2O) from 0.18 to 5 wt.%. To describe the composition of the different catalysts, the K/Ni atomic ratio has been used. Fig. 1 shows the TPR-H2 curves of some catalysts: Ni/ Al2O3, NiK/Al2O3-3, NiK/Al2O3-6 y NiK/Al2O3-8 (nomen-
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
11
Table 1 Nickel and potassium content of catalysts Catalysts
Ni (wt.%)
K2O (wt.%)
K/Ni ratio
Ni/Al2O3 NiK/Al2O3-1 NiK/Al2O3-2 NiK/Al2O3-3 NiK/Al2O3-4 NiK/Al2O3-5 NiK/Al2O3-6 NiK/Al2O3-7 NiK/Al2O3-8
6.63 6.13 4.87 6.00 6.22 6.36 7.00 8.91 9.00
0.00 0.22 0.18 0.37 0.49 0.85 1.26 1.83 5.00
0.00 0.04 0.05 0.08 0.10 0.17 0.22 0.26 0.69
clature as indicated in Table 1). For the free-potassium catalyst (Ni/Al2O3), the maximum in the hydrogen consumption is located at about 873 K, which corresponds to the expected reduction temperature for NiO in intimate contact with alumina supports [13]. In the case of the potassium containing catalysts very different profiles are found. For the lowest potassium content catalyst analyzed (NiK/Al2O3-3) apart of the maximum at 873 K, a shoulder at around 673 K is detected. However, as the potassium content increases, the TPR-H2 profiles are more complex, that is, for the catalyst with the intermediate potassium content (NiK/Al2O3-6) several maxima are found at about 623, 673, 873 and 1000 K; and, for the highest potassium content catalyst (NiK/Al2O3-8), two clearly resolved maxima are observed at 673 and 1000 K, approximately. In this case, it is remarkable that the maximum at 873 K, assigned to the reduction of NiO in intimate contact with the support, is not observed. The results shown in Fig. 1 mean that in the potassium containing catalysts, nickel is present in several states with different reducibility [13,31]. Recently, Hou et al. [13], using calcium as promoter, observed a similar effect in the reducibility of nickel species. These authors assigned the maximum at 673 K to the reduction of NiO particles without interaction with the support [13,34] and, the maximum at 1100 K, approximately, to the reduction of NiAl2O4 [13,35– 37]. In our case, it seems that potassium also modifies the
Fig. 2. FT-EXAFS of reduced catalysts: Ni/Al2O3, NiK/Al2O3-1, NiK/Al2O3-3 and NiK/Al2O3-6.
interaction of nickel with the support, and, consequently, the reducibility of nickel species. Thus, for catalyst NiK/Al2O3-8, the most intense peak is located at about 673 K, so, it can be due to the reduction of NiO with very low interaction with support. The second peak, with a maximum at 1000 K, seems to be due to the reduction of NiO particles with a very strong contact with the support, probably NiAl2O4. Note that the intermediate potassium content catalyst (NiK/Al2O3-6) shows maxima close to the two above extreme peaks, indicating that NiO particles with different extent of contact with support have been formed. In conclusion, the TPR-H2 data clearly reveal that potassium modifies the contact between nickel oxide particles and the support and, because of that, favours the reduction, by H2, of a fraction of nickel. Fig. 2 shows the FT-EXAFS data obtained for some reduced catalysts: Ni/Al2O3, NiK/Al2O3-1, NiK/Al2O3-3 and NiK/ Al2O3-6. The signals corresponding to NiO and Ni foil have been also included as reference. Considering that Ni–Ni bonds ˚ , it can in metallic nickel appear at a distance slightly over 2 A be deduced, in agreement with Baley et al. [38] and with our previous results [3,39] that in these catalysts, Ni is only partially reduced. Note that NiK/Al2O3-6 shows a well defined ˚ . This fact indicates that the addition of a maximum at about 2 A certain amount of potassium favours the reduction of the Ni phase. It must be pointed out that for the catalysts with lower K/ Ni ratio (NiK/Al2O3-1 and NiK/Al2O3-3), the mentioned effect of potassium is not observed. The XANES data (not presented here) confirm these results because the intensity of the white line is the lowest for sample NiK/Al2O3-6. 3.2. Catalytic activity, coke deposition and characterization of used catalysts
Fig. 1. TPR-H2 of Ni/Al2O3, NiK/Al2O3-3, NiK/Al2O3-6 and NiK/Al2O3-8 catalysts.
Fig. 3 shows the data of catalytic activity for all catalysts, expressed as methane and carbon dioxide conversion (Fig. 3a), and the coke deposition (in mg of carbon per gram of catalyst) (Fig. 3b) after 6 h of time-on-stream versus the K/Ni ratio. In Fig. 3a, and in agreement with literature and previous results [3,12,22,23], a decrease in the conversion of both CH4
12
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
Fig. 3. Effect of K/Ni ratio on: (a) CO2 and CH4 conversion at 973 K and (b) amount of coke deposited during 6 h reaction at 973 K.
and CO2 with the amount of potassium in the catalysts is observed. Methane conversion decreases from 70% (near to the thermodynamic equilibrium at the reaction conditions) for the K-free catalyst, to 47% for the sample with K/Ni = 0.69 (NiK/Al2O3-8). Note that the rate of methane conversion decrease is higher at low K-contents. In literature [40], it is mentioned that potassium can migrate from the support to the nickel particles, and be placed, predominantly, on their surface. Recent studies [41] have demonstrated, by DFT calculations that a stepped nickel surface is more active than a surface with close-packed terraces for hydrocarbon reforming reaction and, also, for coke formation. Additionally, these studies conclude that potassium (and other promoters) is preferentially located on the step edges of the nickel surface. The authors observed that, in order to prevent the nucleation of coke, a small fraction of active sites have to be blocked on the stepped nickel surface. This blockage, however, does not stop the reforming reaction because it can still take place on the unblocked, less active, sites of the nickel surface (close-packed terraces). Snoeck and Froment [40], after their study about the effect of potassium addition to nickel catalyst for CH4 steam reforming, proposed that the most important effect of potassium is the decrease of the value of the lumped forward rate coefficient for methane craking. This is due to a reduction of the number of sites available for methane decomposition as a consequence of the presence of potassium on the metal surface. Also, Bengaard et al. [41,42] reported that the methane sticking probability is strongly diminished by potassium. On the basis of these conclusions, the results obtained in this work with the K-
containing catalysts can be explained. A fraction of potassium migrates from the support to the nickel surface and it is located, firstly, on the step sites that are the most active sites for reforming reaction. Consequently, a low potassium content produces a large decrease of methane conversion. When all the step sites are blocked, potassium occupies, gradually, other less active nickel sites and the decrease of methane conversion is lower. Fig. 3a shows that the CO2 conversion decreases in a lower extent than methane conversion. It is well known [3,43] that potassium is a good catalyst for carbon gasification and the increase of the K/Ni ratio could enhance the rate of coke gasification by CO2. This contributes to the increase of CO2 conversion, explaining the different behaviour of CH4 and CO2 conversion at higher K/Ni ratio. The analysis of the reaction products (H2 and CO) reveals that the presence of the promoter produces a slight decrease of the H2/CO ratio (from 0.86 with Ni/Al2O3 to 0.77 with Ni/Al2O3-7). This result, which is interesting from a practical point of view, is consequence of the two effects of potassium previously mentioned: the blockage of Ni active sites for methane decomposition (reaction (4)) and the enhancement of coke gasification (reaction (3)). In relation with carbon accumulation, Fig. 3b shows that the addition of a small amount of potassium has an important and clear effect in reducing the amount of coke deposited (from 280 mg C/g cat. for the K-free catalyst to 30 mg C/ g cat. for the sample with a K/Ni ratio of 0.04, NiK/Al2O3-1). A higher K/Ni ratio produces a further decrease in the amount of the deposited carbon, which becomes negligible for a K/Ni ratio of about 0.26 (less than 2 wt.% K2O). These results are consistent with the fact that the most active sites, responsible for the carbon formation, are firstly blocked by the alkali promoter [41]. The SEM images of two potassium-containing samples (NiK/Al2O3-2 and NiK/Al2O3-7), used in the reforming reaction, are shown in Fig. 4. A comparison between them confirms that a lower amount of coke is deposited on the sample with higher potassium content (Fig. 4B). In addition, it can be observed the filamentous nature of the deposited coke, as it was also observed for the potassium-free sample (not shown) [3]. This fact indicates that the promoter, in the conditions used in this work, decreases the amount of coke deposited but it does not modify its structural characteristics. Thus, it can be concluded that, by addition of a low amount of potassium, it is possible to obtain a catalyst with an acceptably high activity for methane reforming and very low carbon deposition. Among the analyzed catalysts, the one with about 0.2 wt.% K2O (NiK/ Al2O3-1) shows the optimal features: over 63% methane conversion (close to thermodynamic equilibrium) and a coke deposition lower than 30 mg C/g cat. To go more deeply into the effect of potassium in methane conversion, TPR-CH4 experiments were carried out (see experimental section for details) with three selected catalysts (Ni/Al2O3, NiK/Al2O3-4 and NiK/Al2O3-8). The results obtained, shown in Fig. 5, indicate that the presence of potassium, as expected, notably affects the performance of Ni/ Al2O3 catalysts for CH4 conversion [40,41]. In fact, a shift of
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
13
Fig. 6. FT-EXAFS of used catalysts (6 h): Ni/Al2O3, NiK/Al2O3-1, NiK/Al2O33 and NiK/Al2O3-6.
Fig. 4. SEM images of used catalysts (6 h): (A) NiK/Al2O3-2 and (B) NiK/ Al2O3-7.
the onset temperature for CH4 conversion to higher values (from around 773 K for Ni/Al2O3 to 900 K for NiK/Al2O3-4 and to 973 K for NiK/Al2O3-8) is observed. To explain this delay it must be taken into account that: (i) methane decomposition occurs on metallic Ni sites, (ii) as the samples
Fig. 5. TPR-CH4 of Ni/Al2O3, NiK/Al2O3-4 and NiK/Al2O3-8 catalysts.
have not been previously reduced, metallic Ni sites are being formed during the TPR-CH4 by reaction with methane and (iii) the presence of K hinders the formation of Ni active sites, probably because, due to its electropositive character, decreases the affinity of Ni for methane. The TPR-CH4 profiles found can be interpreted considering that, as soon as the active metallic sites are formed, methane decomposition takes place quickly. In the case of sample NiK/Al2O3-8, the shape of the TPR-CH4 profiles indicates that the Ni active sites are being formed slowly as the temperature increases. Fig. 6 shows the FT-EXAFS obtained for catalysts Ni/Al2O3, NiK/Al2O3-1, NiK/Al2O3-3 and NiK/Al2O3-6 after having been used in the dry reforming of methane for 6 h. The signal corresponding to Ni foil has been included as reference. The profiles presented in Fig. 6 indicate that the four catalysts show a similar Ni structure that is close to that of Ni foil, although probably, with a lower coordination number. This means, as previously observed [3,26,39] that the catalysts become further reduced under reaction conditions. For the K-containing samples, the intensity of the FT-EXAFS main signal is higher, this can be interpreted by considering that in these samples a higher fraction of nickel is in the metallic form (Ni0) or that the Ni particles are larger. The former hypothesis is in agreement with the conclusion extracted from TPR-H2 about the effect of potassium in the reducibility of nickel species. TEM analysis of two representative samples (Ni/Al2O3 and NiK/Al2O3-7) indicates that, in both cases, the mean size of Ni particles is between 10 and 15 nm. So, it can be concluded that the addition of potassium does not modify the Ni particle size. As the EXAFS results reveal a similar Ni structure in catalysts with and without potassium, it can be concluded that the promoter does not modify the arrangement of Ni atoms, although facilitates the reduction of nickel species by H2, probably due to the modification of the interaction between metallic species and the support. The catalytic activity data previously commented (Fig. 3) correspond to 6 h reaction time experiments. It is worth noting that in all cases, that is, independently of the K content, the catalytic activity remains almost constant during the whole
14
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
Fig. 7. CO2 and CH4 conversion at 973 K (24 h reaction) for Ni/Al2O3 and NiK/ Al2O3-7 catalysts.
Fig. 8. SEM image of used catalyst NiK/Al2O3-7 (24 h).
experiment, although a very slight decrease in the conversion of methane for the K-free catalyst is observed. The negative effect of the deposited coke on the catalysts performance should be patent, if so, after a longer reaction time. In order to check this point, some catalysts have been tested in a period of 24 h. Fig. 7 shows the results obtained with the K-free catalyst and with sample NiK/Al2O3-7 (K/Ni ratio of 0.26). The most remarkable about these data is that no deactivation has been observed during this long reaction period. It means that the deposited coke does not deactivate the Ni active phase, that is, Ni particles are not being blocked by the coke formed. SEM data of this work (Fig. 4) and of previous publications [3,14,30,44,45] show that the coke deposited on the catalysts surface has a fibrous structure. The growth of carbon filaments occurs on a metal surface where the active metal is carried on the top of the carbon filament [46] and, as consequence, the catalytic activity can be kept with reaction time because the active metal is still exposed to the reactants. Regarding the amount of carbon deposited during the 24 hexperiments on Ni/Al2O3 and NiK/Al2O3-7 catalysts, as expected, a larger amount of coke has been accumulated on the K-free catalyst. However, the rate of coke accumulation is not constant with time. Thus, the amount of carbon deposited on sample Ni/Al2O3 is lower (about 680 mg C/g cat.) than the expected value (1120 mg C/g cat., calculated considering a linear progression from the data obtained at 6 h). However, for the potassium containing catalyst, it is higher than the expected value (about 150 mg C/g cat. in front of the expected 20 mg C/g cat.). Fig. 8 shows the SEM image of NiK/Al2O3-7 catalyst after 24 h reaction. It can be observed that the amount of coke deposited is higher that in the 6 h reaction experiments (see Fig. 4B) and that the structure is also fibrous. The behaviour of the Ni/Al2O3 catalyst can be explained considering that the coke being accumulated produces the blockage of the most active sites for coke formation, hindering methane cracking and, as consequence, decreasing the amount of coke deposited. This explanation also justifies the slight decrease with time in methane
conversion (related to Eq. (4)) shown in Fig. 7 for the potassium-free catalyst. In order to explain the behaviour of the K-containing catalyst it has to be considered that, as the coke deposit grows (in form of carbon filaments, see Fig. 8), the contact between potassium species (mainly on the support), and the nickel particles and the carbon material formed on them, decreases. This implies that the formation of coke is less restricted and the gasification reaction takes place in a lower extent. In addition a possible loss of potassium from the catalyst (by volatilization) under reaction conditions could also contribute the observed behaviour. However, it has been proved (by XRF measurements) that the amount of promoter remains constant after reforming reaction.
4. Conclusions The presence of potassium in Ni/Al2O3 catalysts hinders the accumulation of coke on the catalyst surface during the dry reforming of methane, but produces a decrease in the catalytic activity. There is, however, an optimum amount of potassium (0.2 wt.% K2O) that leads to a catalyst with a very low coke deposition and a high catalytic activity (more than 90% reduction in coke deposition and less than 10% decrease in the catalytic activity). Independently of the amount of potassium, the catalytic activity remains almost constant during at least 24 h. This indicates that the coke deposit does not deactivate the catalysts because Ni particles are on the top of carbon filaments formed during reaction. So, the effect of deposited coke would be a plug up of the reactor. The study about the effect of the amount of potassium allows to confirm the following ideas: (i) the addition of potassium increases the reducibility of nickel species by hydrogen because it modifies the interaction of nickel oxide with the Al2O3 support, (ii) potassium migrates from the support to the Ni surface and neutralizes a fraction of the active sites, (iii) potassium catalyzes the gasification of the coke formed during
J. Juan-Juan et al. / Applied Catalysis A: General 301 (2006) 9–15
reaction without modification of its structure and (iv) potassium does no modify the size nor the structure of the Ni particles. Acknowledgments Dr. H. Yamashita is gratefully acknowledged for performing the XAFS measurements. The authors thank the Spanish government (PPQ2002-01025) and the Photon Factory (Japan) project 2003G067, for financial support and J. Juan-Juan thanks his thesis grant to MCYT of Spain. References [1] M.C.B. Bradford, M.A. Vannice, Catal. Rev. Sci. Eng. 41 (1999) 1. [2] J.R. Rostrup-Nielsen, J. Catal. 144 (1993) 38. [3] J. Juan-Juan, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez, Appl. Catal. A: Gen. 264 (2004) 169. [4] M.C.B. Bradford, M.A. Vannice, Catal. Today 50 (1999) 87. [5] M.C.B. Bradford, M.A. Vannice, Appl. Catal. A: Gen. 142 (1996) 97. [6] A. Erdo¨helyi, J. Cserenyi, F. Solymosi, J. Catal. 141 (1993) 287. [7] Z. Zhang, U.A. Tsipouriari, A.M. Efstathiou, X.E. Verkyos, J. Catal. 158 (1996) 51. [8] U.L. Portugal, A.C.S.F. Santos, S. Damyonova, C.P.P. Marques, J.M.C. Bueno, J. Mol. Catal. A: Chem. 184 (2002) 311. [9] A. Erdo¨helyi, J. Cserenyi, E. Papp, F. Solymosi, Appl. Catal. A: Gen. 108 (1994) 205. [10] A. Erdo¨helyi, K. Forder, F. Solymosi, Stud. Surf. Sci. Catal. 107 (1997) 525. [11] K.S. Hwang, H.Y. Zhu, G.Q. Lu, Catal. Today 68 (2001) 183. [12] A. Lemonidu, M.A. Goula, I.A. Vasalos, Catal. Today 46 (1998) 175. [13] Z. Hou, O. Yokota, T. Tanaka, T. Yashima, Appl. Catal. A: Gen. 253 (2003) 381. [14] J. Juan-Juan, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez, in: Proceedings of the Sixth European Europacat-VI Congress, Austria, (2003. [15] K. Takehira, T. Shishido, P. Wang, T. Kosaka, K. Takaki, J. Catal. 221 (2004) 43. [16] K. Nagaoka, A. Jentys, J.A. Lercher, J. Catal. 230 (2005) 185. [17] A. Golebioswski, K. Stolecki, U. Prokop, A. Kusmieroska, T. Borowiecki, A. Denis, C. Sikorska, React. Kinet. Catal. Lett. 82 (2004) 179. [18] S. Tang, L. Ji, J. Lin, H.C. Zeng, K.L. Tan, K. Li, J. Catal. 194 (2000) 424. [19] J.A.C. Dias, J.M. Assaf, Catal. Today 85 (2003) 59.
15
[20] K. Takehira, T. Shishido, D. Shoro, K. Muramaki, M. Honda, T. Kaabata, K. Takaki, Catal. Commun. 5 (2004) 209. [21] K. Takanabe, K. Nagaoka, K. Nariair, K. Aika, J. Catal. 230 (2005) 75. [22] F. Frusteri, F. Arena, G. Galogero, T. Torre, A. Parmaliana, Catal. Commun. 2 (2001) 29. [23] J.C. Rodrı´guez, E. Romero, J.L.G. Fierro, J. Santamarı´a, A. Monzo´n, Catal. Today 37 (1997) 255. [24] J.J. Richardson, M. Lei, B. Turk, K. Foster, M.V. Twigg, Appl. Catal. A: Gen. 110 (1994) 217. [25] K. Tomishige, O. Yamazaki, Y. Chen, K. Yokoyama, X. Li, K. Fujimoto, Catal. Today 45 (1998) 35. [26] S. Takeneka, E. Kato, Y. Tomikubo, K. Otsuka, J. Catal. 219 (2003) 176. [27] H. Chen, C. Wang, C. Yu, L. Teng, P. Liao, Catal. Today 97 (2004) 173. [28] A. Valentini, N. Villareal, L. Dias, P. Lisboa, W. Schreiner, E. Leite, E. Longo, Appl. Catal. A: Gen. 255 (2003) 211. [29] Z. Hou, Z. Xiaoming, T. Yashima, React. Kin. Catal. Lett. 84 (2) (2005) 229. [30] F. Frusteri, L. Spadaro, F. Arena, A. Chuvilin, Carbon 40 (7) (2002) 1063– 1070. [31] T. Osaki, T. Mori, J. Catal. 204 (2001) 89. [32] J.S. Chang, S.E. Park, H. Chon, Appl. Catal. A: Gen. 145 (1996) 111. [33] J. Juan-Juan, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez, in: Proceedings of the SECAT 03 Congress, Spain, (2003), p. 83. [34] C.W. Hu, J. Yao, H.Q. Yang, Y. Chen, A.M. Tian, J. Catal. 166 (1997) 1. [35] A.A. Lemonidou, I.A. Vasalos, Appl. Catal. A: Gen. 228 (2002) 227. [36] A.M. Siskin, R.H. Cunningham, R.M. Ormerod, Catal. Today 46 (1998) 147. [37] R. Molina, G. Poncelet, J. Catal. 173 (1998) 257. [38] K.M. Baley, T.K. Campbell, J.L. Falconer, Appl. Catal. 54 (1989) 159. [39] J. Juan-Juan, M.C. Roma´n-Martı´nez, M.J. Illa´n-Go´mez, Photon Factory Activity Report, 21 (Part B) (2003) 82. [40] W. Snoeck, G.F. Froment, Ind. Eng. Chem. Res. 41 (2002) 3548. [41] H.S. Bengaard, J.K. Norskov, J. Sehested, B.S. Clausen, L.P. Nielsen, A.M. Molenbroek, J.R. Rostrup-Nielsen, J. Catal. 209 (2002) 365. [42] H.S. Bengaard, I. Alstrup, I. Chorkendorff, S. Ullmann, J.R. RostrupNielsen, J.K. Norskov, J. Catal. 187 (1999) 238. [43] R. Meijer, H.J. Mu¨hlen, F. Kapteijn, J.A. Moulijn, Fuel Process. Technol. 28 (1991) 5. [44] O.S. Joo, K.P. Jung, Bull. Korean Chem. Soc. 23 (2002) 1149. [45] F. Arena, F. Frusteri, A. Parmaliana, Appl. Catal. A: Gen. 187 (1999) 127. [46] M. Parvary, S.H. Jazayeri, A. Taeb, K. Petit, A. Kiennermann, Catal. Commun. 2 (2001) 357.