Al2O3 catalyst

0 downloads 9 Views 4MB Size Report
such as removal of trace amounts of CO from H2 rich feed gas or purification of the reformate gas for NH3 synthesis and fuel cell application, and ... natural gas (SNG) via syngas methanation approach in China [4], as the basic pattern of China ...
Applied Catalysis A: General 488 (2014) 37–47

Contents lists available at ScienceDirect

Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata

Enhanced catalytic performances of Ni/Al2 O3 catalyst via addition of V2 O3 for CO methanation Qing Liu a , Fangna Gu a,∗ , Xiaopeng Lu a , Youjun Liu a , Huifang Li a , Ziyi Zhong b , Guangwen Xu a,∗ , Fabing Su a,∗ a b

State Key Laboratory of Multiphase Complex Systems, Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100190, China Institute of Chemical Engineering and Sciences, A*star, 1 Pesek Road, Jurong Island 627833, Singapore

a r t i c l e

i n f o

Article history: Received 4 July 2014 Received in revised form 19 September 2014 Accepted 21 September 2014 Keywords: Ni/Al2 O3 catalyst Vanadium oxide Ni3 V2 O8 CO methanation CO2 methanation

a b s t r a c t Highly active and coking resistant Ni-V2 O3 /Al2 O3 catalysts were prepared by co-impregnation method for CO and CO2 methanation. The influence of vanadium oxide addition on catalyst structure, distribution and reducibility of Ni species, morphology and surface characteristics, was investigated in detail. Compared to the catalyst without vanadium, the Ni-V2 O3 /Al2 O3 catalysts showed significant improvement in the activity, thermal stability, and resistance to coke formation in CO methanation. In addition, these catalysts also showed high activities for CO2 methanation at both atmospheric and 2.0 MPa pressures. It was found that Ni3 V2 O8 was formed during the calcination of the Ni-V2 O3 /Al2 O3 catalysts, which led to the formation of smaller Ni particle sizes (ca. 3.0 nm) as compared to the case without vanadium oxide addition. The higher catalytic activity over the Ni-V2 O3 /Al2 O3 catalysts for CO methanation was mainly due to the larger H2 uptake, the higher Ni dispersion as well as the smaller metallic Ni nanoparticles. The oxidation–reduction cycle of V2 O3 could increase the oxygen vacancies, which enhanced the dissociation of CO2 by-product and generated surface oxygen intermediates, thus preventing carbon deposition on the Ni particles in CO methanation. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Methanation of carbon oxides (CO and/or CO2 ) with hydrogen to produce methane was firstly reported by Sabatier and Senderens in 1902 [1], and has been widely used in various industrial processes such as removal of trace amounts of CO from H2 rich feed gas or purification of the reformate gas for NH3 synthesis and fuel cell application, and processes in relation to Fischer–Tropsch synthesis [2,3]. Recently, there is an arising interest to produce synthetic natural gas (SNG) via syngas methanation approach in China [4], as the basic pattern of China energy is characterized by “rich in coal, poor in oil and natural gas”. The strategy that relies on the rich coal resources in China while energetically developing “Coal to Natural Gas” not only meets the development direction of coal clean utilization, but also guarantees the effective supplement of natural gas [5]. In recent years great efforts have been made to develop highly efficient catalysts for syngas methanation, and the requirements for the excellent methanation catalysts include high activity at low

∗ Corresponding authors. Tel.: +86 10 82544850; fax: +86 10 82544851. E-mail addresses: [email protected] (F. Gu), [email protected] (G. Xu), [email protected] (F. Su). http://dx.doi.org/10.1016/j.apcata.2014.09.028 0926-860X/© 2014 Elsevier B.V. All rights reserved.

temperature (ca. 300 ◦ C) and high stability at high temperature (ca. 600 ◦ C) [6]. It is acknowledged that Ru-based catalyst is the most active for syngas methanation, however, limited resource and high cost of Ru restrict its large-scale commercialization [7]. Therefore, Ni-based catalysts are more preferred for syngas methanation due to their low cost, good availability and high activity. However, for Ni-based catalysts, a high reaction temperature (above 320 ◦ C) is generally required to achieve the maximum CO conversion at atmospheric pressure [8–10], and the coke formation is usually serious under the harsh conditions, which reduces the catalytic activity obviously [11–13]. To overcome these limits, various promoters [10,12,14], supports [5,13,15] and preparation methods [6,16] have been selected and tested. However, the above reported catalysts seldom show both high low-temperature activity and good hightemperature stability, as most of approaches are only effective on one of them. Therefore, developing catalysts that can work well at both low and high temperatures still remain as the main challenges. For Ni-based catalysts, it has identified that the ratedetermining step in the methanation reaction is the hydrogenation of CHx species [17]; for promotion of catalytic activity, it is crucial to realize a fast removal of these surface CHx species by surfacedissociated hydrogen. It has reported that the surface defects in metallic Ni serve as capture traps for surface hydrogen which

38

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

reduces the activation energy of hydrogen dissociation [18], and the largely increased surface defects can be achieved by decreasing the supported Ni particles size into nano-scale with a simultaneously increased Ni dispersion [18]. Moreover, the rate of carbon filament formation is proportional to the particle size of Ni, and below a critical Ni particle size (d < 2 nm), the formation of carbon becomes dramatically slow [19]. Therefore, it is proposed that a Ni-based catalyst with small Ni nanoparticles will show much improved low-temperature activity and anti-coking property. Vanadium oxides are used extensively in industrial catalysts as active component or catalyst supports for many reactions including oxidation of sulphur dioxide, partial oxidation of hydrocarbons and selective reduction of nitric oxide [20]. It was reported that V could obviously restrain the Rh particle size over SiO2 support through formation of RhVO4 [21–25], and the Rh-V/SiO2 catalysts could have improved activity and selectivity in CO hydrogenation for the production of C2 oxygenates such as ethanol and acetic acid [21,23,26,27]. In addition, some authors have reported that V was a good promoter to Ni/Al2 O3 catalyst in CO hydrogenation to produce C2 –C4 hydrocarbons [28], but the correlation between the catalyst structure and performance has not been well clarified yet. In addition, it is known that nickel vanadates such as Ni3 V2 O8 , Ni2 V2 O7 and NiV2 O6 can be synthesized by the reaction of V2 O5 with different amount of NiO [29], which may have the similar performance to RhVO4 in reducing the metallic particle size. In continuation of our previous work on CO methanation [8–12,30–32], in this work we have developed a Ni-V2 O3 /Al2 O3 catalyst by co-impregnation method. It is found that the addition of vanadium oxide is capable of improving both the low-temperature catalytic activity and the anti-coking property of the Ni catalyst in the CO methanation even under very harsh conditions. In addition, CO2 methanation is also studied over the Ni-V2 O3 /Al2 O3 catalyst.

2. Experimental 2.1. Catalyst preparation All the chemicals with analytical grade including nickel (II) nitrate hexahydrate (Ni(NO3 )2 ·6H2 O) and ethanol were purchased from Sinopharm Chemical Reagent Co. Ltd., China, and used without further treatment. Vanadyl (IV) acetylacetonate was purchased from Acros. The commercial porous -Al2 O3 (Zibo Honghe Chemical Co. Ltd., China) with a surface area of 164 m2 g−1 was calcined at 400 ◦ C in air for 2 h before use. The NiO/Al2 O3 catalyst was prepared by the wet impregnation method. The stoichiometric quantity of Ni(NO3 )2 ·6H2 O was dissolved in ethanol, followed with addition of the -Al2 O3 support to form a slurry. The slurry was vigorous stirred at room temperature overnight, and then heated to 70 ◦ C to evaporate the liquid, followed with drying at 100 ◦ C for 24 h and further calcination at 500 ◦ C for 2 h in air with the heating rate of 2 ◦ C min−1 . The collected catalyst with a NiO loading of 20 wt% was denoted as 20NA. The NiO-V2 O5 /Al2 O3 catalysts with various V2 O5 contents were prepared by the co-impregnation method following with the above procedures by adding stoichiometric quantities of Ni(NO3 )2 ·6H2 O and vanadyl (IV) acetylacetonate in ethanol simultaneously. The finally obtained samples were denoted as 20NAxV (x = 2, 5, 10 and 16), where x represented the mass percent of V2 O5 . In addition, Ni3 V2 O8 was also prepared with the same procedure mentioned above only without addition of -Al2 O3 [29]. V2 O5 /Al2 O3 was prepared by the same method as 20NA10V only without addition of NiO, and the pure V2 O5 was obtained by calcining vanadyl (IV) acetylacetonate at 500 ◦ C for 2 h in air with the heating rate of 2 ◦ C min−1 .

2.2. Catalyst characterization N2 adsorption at −196 ◦ C was measured using a Quantachrome surface area and pore size analyzer NOVA 3200e. Prior to the measurement, the sample was degassed at 200 ◦ C for 4 h under vacuum. The specific surface area was determined according to the Brunauer–Emmett–Teller (BET) method in the relative pressure range of 0.05–0.3. The pore size distribution (PSD) was calculated with the Barett–Joyner–Halenda (BJH) method using the adsorption isotherm branch. X-ray diffraction (XRD) patterns were recorded on a PANalytical X‘Pert PRO MPD with a step of 0.02◦ using the Cu K˛ ˚ at 40 kV and 40 mA, and checked with the radiation ( = 1.5418 A) card number of Joint Committee on Powder Diffraction Standards (JCPDS). The crystal size of the sample was calculated using the Debye–Scherrer equation. H2 temperature-programmed reduction (H2 -TPR), H2 temperature-programmed desorption (H2 -TPD) and CO2 temperature-programmed desorption (CO2 -TPD) were carried out on Quantachrome Automated chemisorption analyzer (chemBET pulsar TPR/TPD). For H2 -TPR, 0.05 g sample was loaded in a quartz U-tube and heated from room temperature to 200 ◦ C at 10 ◦ C min−1 and maintain for 1 h under He flow. Then, the sample was cooled to room temperature and followed by heating to 1000 ◦ C at 10 ◦ C min−1 under a binary gas (10 vol% H2 /Ar) with a gas flow of 30 mL min−1 . For H2 -TPD, 0.2 g catalyst was used and reduced in situ by H2 /Ar flow previously. Then, the sample was cooled to room temperature and saturated with H2 . After removing the physically adsorbed H2 by flush with Ar for 2 h, the sample was heated to 1000 ◦ C ramping at 10 ◦ C min−1 in Ar flow (30 mL min−1 ). The consumed or desorbed H2 was detected continuously as a function of increasing temperature using a thermal conductivity detector (TCD). The number of surface Ni sites per unit mass of catalyst was determined by means of H2 -TPD assuming the adsorption stoichiometry of H/Ni = 1:1. The peak area of H2 -TPD profile was normalized by that of H2 -TPR of a standard CuO sample. The dispersion of Ni was calculated based on the volume of chemisorbed H2 using the following simplified equation [12]: D(%) =

2 × Vad × M × SF × 100 m × P × Vm × dr

where Vad (mL) represents the volume of chemisorbed H2 at the standard temperature and pressure (STP) conditions measured in the TPD procedure; m is the sample weight (g); M is the molecular weight of Ni (58.69 g mol−1 ); P is the weight fraction of Ni in the sample as determined by ICP; SF is the stoichiometric factor (the Ni:H molar ratio in the chemisorption) which is taken as 1 and Vm is molar volume of H2 (22414 mL mol−1 ) at STP; dr is the reduction degree of nickel calculated based on H2 -TPR. For CO2 -TPD, 0.2 g catalyst was used and reduced in situ by H2 /Ar flow previously. Then, the sample was cooled to room temperature in H2 flow and saturated with CO2 (30 mL min−1 ) for 1 h. After removing the physically adsorbed CO2 by flush with He for 2 h, the sample was heated to 500 ◦ C ramping at 10 ◦ C min−1 in He flow (30 mL min−1 ). The turnover frequency (TOF) was calculated according to the formula described in our previous work [9]. The surface chemical composition was analyzed by an Xray photoelectron spectroscopy (XPS) test conducted on a VG ESCALAB 250 spectrometer (Thermo Electron, U.K.) with a nonmonochromatized Al K˛ X-ray source (1486 eV). The microscopic feature of the samples was observed by field emission scanning electron microscope (SEM) (JSM-6700F, JEOL, Japan) and transmission electron microscopy (TEM) (JEM-2010F, JEOL, Japan). Before the TEM measurement, the H2 -reduced catalysts were cooled to room temperature in H2 flow and then

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

39

passivated in 1 vol% O2 /Ar gas mixture for 30 min to prevent bulk oxidation of the Ni nanoparticles. Thermogravimetric (TG) analysis was conducted on a Seiko Instruments EXSTAR TG/DTA 6300. 10 mg of the sample was used and heated under air (200 mL min−1 ) from room temperature up to 1000 ◦ C (10 ◦ C min−1 ). 2.3. Catalytic performance measurement The CO and CO2 methanation reactions were carried out in a fixed bed reactor at 0.1 MPa with a quartz tube, and at 3.0 MPa and 2.0 MPa for CO and CO2 methanation respectively a stainless steel tube respectively. The influence of mass transfer had been examined and eliminated before the catalytic tests. First, 0.1 g catalyst sample (20–40 mesh) diluted with 5.0 g (2.5 g at high pressures) quartz sands (20–40 mesh) was uploaded in quartz tube with an inner diameter of 8 mm. The addition of the quartz sands was to avoid the generation of hotspot in the catalyst bed. For test at 3.0 or 2.0 MPa, the above quartz tube was then packed in the stainless steel tube. A thermocouple was inserted into the furnace chamber and bonded to the outside of the reactor tube near the middle position of the catalyst bed to control the reaction temperature. The catalysts were reduced at 600 ◦ C in pure H2 (100 mL min−1 ) for 2 h and then cooled to the starting reaction temperature in H2 . The mixed H2 and CO or CO2 as well as N2 (as an internal standard) were introduced into the reactor at a molar ratio of H2 /CO/N2 = 3/1/1 or H2 /CO2 /N2 = 12/3/5. The outlet gas stream from the reactor was cooled using a cold trap. Inlet and outlet gases were analyzed on line by Micro GC 3000A (Agilent Technologies). The concentrations of H2 , N2 , CH4 and CO in gas products were detected by a TCD detector with a Molecular Sieve column while the concentrations of CO2 , C2 H4 , C2 H6 , C3 H6 and C3 H8 were analyzed by another TCD with a Plot Q column. A lifetime test of CO methanation was performed at 500 and 550 ◦ C, 3.0 MPa. The CO and CO2 conversion, CH4 selectivity and yield are defined as follows: Fi,in − Fi,out × 100 Fi,in

CO conversion :

Xi (%) =

CH4 selectivity :

SCH4 (%) =

CH4 yield :

Y CH4 (%) =

FCH4 ,out Fi,in − Fi,out

Xi × SCH4 100

=

× 100

FCH4 ,out Fi,in

× 100

Here, i represents CO or CO2 , X is the conversion of CO or CO2 , S is the selectivity of CH4 , Y is the CH4 yield, Fi ,in and Fi ,out are the molar flow rates of species i (i = CO or CO2 ) at the inlet and outlet, respectively. The rate and activation energy for CO methanation over the catalyst were determined using the reactor above at 0.1 MPa. 1 g catalyst sample (20–40 mesh) diluted with 3.0 g quartz sands (20–40 mesh) was used. The experiments were performed with different total gas flow of 50, 100 and 200 mL min−1 in the temperature range of 190–220 ◦ C. The rate was determined using the following equation [33]: Rate(r) =

FCO × XCO XCO = W W/FCO

where FCO represents the flow of the CO in mol s−1 , W is the weight of the catalyst in g, and XCO the CO conversion. The variations of XCO with W/FCO were plotted, and then the rates of reaction were calculated at various temperatures from the slope of linear portion. The activation energy was calculated using the Arrhenius equation.

Fig. 1. XRD patterns of the samples: (a) as-calcined and (b) reduced; H2 -TPR profiles of the samples (c) and H2 -TPD profiles of the catalysts (d).

3. Results and discussion 3.1. Characterization of the catalysts 3.1.1. Textural and structural properties of the catalysts The adsorption–desorption isotherms of all the samples are all in compliance with the type IV features of mesoporous materials (Fig. S1A), and the PSD curves of the catalysts are similar with the most probable pore diameters in the range of 8–10 nm (Fig. S1B), indicating that the addition of vanadium species has only little influence on PSD of 20NA. Table 1 compiles the surface areas and pore volumes of the catalysts. The specific surface areas and total pore volumes of the catalysts are almost the same, except a little increase in surface area after addition of proper amount of vanadium species, which may be due to the fact that some vanadium oxide nanoparticles make the surface of catalysts become coarser. However, comparing with the other catalysts, the 20NA16V catalyst has the smallest surface area and pore volume, probably because of partial blockage of pores by the excess vanadium species. 3.1.2. XRD characterization As seen from Fig. 1a, for the as-calcined 20NA catalyst, the observed XRD peaks at 37.4, 46.0 and 66.7◦ correspond to -Al2 O3 (JCPDS 00-034-0493), while those at 37.4, 43.1, 62.8, 75.4 and 79.4◦ to NiO (JCPDS 01-089-3080). After the addition of vanadium species, there is no appearance of any new peaks corresponding to V2 O5 (JCPDS 01-089-0612) in all the XRD patterns of the catalysts; however, there is a significant decrease in NiO peaks’ intensity. For example, for 20NAxV (x = 5, 10 and 16), no NiO diffraction peaks can be observed. As we all know, some promoters such as La2 O3 [34], CeO2 [12] and MgO [10] can effectively restrain the growth of NiO nanoparticles. It is anticipated that vanadium species may play similar role to the above promoters in this work. Interestingly, there are three new peaks at about 36.1, 44.3 and 64.1◦ in the XRD pattern of 20NA16V, which correspond to Ni3 V2 O8 (JCPDS 01-074-1484) that is formed via the solid phase reaction of NiO and V2 O5 during the calcination process. Similar phenomena were observed in Rh based

40

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

Table 1 Physical and chemical properties of the catalysts. Sample

SBET a (m2 g−1 )

Vp b (cm3 g−1 )

Ni particle size (nm)c

H2 uptake (␮mol g−1 )

D (%)d

TOFCO , 220 ◦ C (s−1 )e

20NA 20NA2V 20NA5V 20NA10V 20NA16V

138 152 144 143 131

0.35 0.35 0.33 0.33 0.31

10.7 3.8 3.2 2.8 11.2

98.5 187.5 240.1 175.6 133.3

7.4 14.0 17.9 13.1 9.9

0.90 × 10−3 1.11 × 10−3 1.38 × 10−3 1.22 × 10−3 0.91 × 10−3

a b c d e

SBET , surface area derived from BET equation. Vp , pore volume obtained from the volume of nitrogen adsorbed at the relative pressure of 0.97. Ni particle size, calculated by the XRD diffraction peak (2 = 44.6◦ ) using the Debye–Scherrer equation. D, Ni dispersion calculated based on the H2 -TPR and H2 -TPD results. Calculated based on the metal dispersion and the CO conversion at 220 ◦ C.

catalysts added with other oxides, in which complex oxides such as RhVO4 , RhNbO4 and Rh2 MnO4 could be formed by mutual interaction between Rh2 O3 and the oxide during calcination in air or O2 at high temperature (700–900 ◦ C) [21]. In this work, the observation of very weak Ni3 V2 O8 XRD peaks indicates its high dispersion over the catalyst surface. In Fig. 1b, it is seen that after treated in H2 flow at 600 ◦ C for 2 h, V2 O5 is reduced to V2 O3 (JCPDS 01-071-0344) [35], and Ni3 V2 O8 to V2 O3 and Ni (JCPDS 01-070-1849). For the reduced catalysts, the peaks at 44.6, 51.8 and 76.6◦ correspond to (1 1 1), (2 0 0) and (2 2 0) planes of metallic Ni. However, there are no any new peaks corresponding to V2 O3 , indicating a high dispersion of V2 O3 on the catalyst surface. The Ni nanoparticle sizes in these catalysts are calculated from the XRD patterns and listed in Table 1. The Ni particle size of 20NA is 10.7 nm, and those of 20NAxV (x = 2, 5 and 10) are only 2.8–3.8 nm, while that of 20NA16V is 11.2 nm, which is much larger than others. Moreover, the Ni particle size of Ni3 V2 O8 after reduction at 600 ◦ C could be as large as 25.6 nm. Hence, it is clear that the addition of V2 O5 and formation of proper amount of Ni3 V2 O8 can greatly reduce the Ni particle size and improve the Ni dispersion on the Al2 O3 support. 3.1.3. H2 -TPR Fig. 1c presents the H2 -TPR profiles of V2 O5 /Al2 O3 , Ni3 V2 O8 and all the catalysts. As seen in the reduction profile of V2 O5 /Al2 O3 , there is only one reduction peak at 517 ◦ C for V2 O5 , consistent with the results of Kip et al. [27] and Kijenski et al. [36]. At the same time, the reduction peak of the pure NiO is at 462 ◦ C. For Ni3 V2 O8 , there is only one reduction peak at about 507 ◦ C, which is higher than that of the pure NiO but a little lower than that of V2 O5 /Al2 O3 , consistent with the result of Bao et al. [37]. For 20NA, a broad reduction peak appears at about 567 ◦ C, which is assigned to the reduction of NiO species with middle strength of interaction with the support [9,31]. Compared with 20NA, the main reduction peaks of 20NAxV shift to lower temperatures with the increment of V2 O5 , and the lowest peak temperature is observed at 517 ◦ C for 20NA16V. This may be because that NiO and Ni3 V2 O8 are coexistent on the catalysts surface, and the interaction of NiO–Al2 O3 is stronger than that of NiO–V2 O5 , thus with the increase of Ni3 V2 O8 , the reduction temperature of the catalyst shifts to lower temperature range. In addition, there is a new peak at low temperature of 384, 409 and 397 ◦ C for 20NA5V, 20NA10V and 20NA16V respectively, which is attributed to the reduction of the free NiO [9,31]. With the increase of V loading, the integrated area of low temperature peak becomes larger, which may be because the occurrence of the competitive adsorption of Ni2+ and vanadyl (IV) acetylacetonate during the co-impregnation process [12]. The V2 O5 precursor adsorbs on the Al2 O3 support preferentially with high dispersion [26], which may increase the difficulty of Ni2+ ions adsorption due to the preoccupation of the adsorption sites by vanadyl (IV) acetylacetonate, resulting in poor dispersion of Ni2+ ions and formation of more and larger free NiO particles in 20NA10V and 20NA16V.

However, it seems there is a contradiction that the Ni particles in the reduced 20NA10V are the smallest while those in 20NA16V are the largest (Table 1), which may be explained as follows: the proper amount of Ni3 V2 O8 can improve the dispersion of the Ni species and reduce the Ni particle size as observed in 20NA10V, but the excess vanadium species may cause an adverse effect due to the increased competitive adsorption by the nickel and vanadium precursors in the co-impregnation process which will obviously weaken the interaction between NiO species and the support; furthermore, the formation of excess Ni3 V2 O8 in the calcination process as well as the weakened interaction of Ni–V in Ni3 V2 O8 will also result in the formation of large Ni particles after reduction as observed in 20NA16V. 3.1.4. H2 -TPD Fig. 1d shows the H2 -TPD profiles of all the catalysts, and Table 1 lists the calculated hydrogen uptakes and the Ni dispersion of the catalysts based on the H2 -TPR and H2 -TPD results. For 20NA, two main H2 desorption peaks are located at around 231 and 833 ◦ C, respectively. The first peak at low temperature is attributed to the chemisorbed hydrogen on the highly dispersed Ni nanoparticles possessing a large density of surface defects, which often serve as traps for surface hydrogen diffusion that can reduce the activation energy of hydrogen dissociation [18]; while the second peak locates at 833 ◦ C can be assigned to the H2 adsorbed in the subsurface layers of Ni atoms and/or to the spillovered H2 [38]. For the peak at 231 ◦ C, there is an increment in intensity for 20NA2V and 20NA5V but a decline for 20NA10V and 20NA16V as compared to that in 20NA, indicating that the addition of proper amount of vanadium species can increase the dispersion of Ni nanoparticles, but the excess vanadium oxide may reduce the H2 chemisorption due to partial coverage of some Ni particles by vanadium species. For 20NAxV, a new strong peak appears at around 390 ◦ C that can be assigned to hydrogen adsorbed on bulk or lowlydispersed Ni nanoparticles [18]. At the same time, for the peaks at high temperatures (above 550 ◦ C), there is an obvious increment in the integrated areas and shift to low temperature after V addition compared to those of 20NA, and the changes follow the order of 20NA16V > 20NA10V > 20NA5V > 20NA2V in magnitude. The H2 uptake and the dispersion of Ni are calculated based on the H2 TPD (below 550 ◦ C) and H2 -TPR results and listed in Table 1. The results reveal that the addition of proper amount of V can increase the H2 uptakes and the dispersion of Ni. As the result, 20NA5V has the highest total H2 uptake of 240.1 ␮mol g−1 cat and Ni dispersion of 17.9%. However, the H2 uptakes and Ni dispersions of 20NA10V and 20NA16V become lower again compared to those of 20NA5V, which may be because the reduction of Ni3 V2 O8 will result in partial coverage of Ni atoms by V2 O3 particles. Luo et al. [39,40] once reported that the hydrogen adsorbed by Rh at high temperature over Rh-V/SiO2 catalyst could spill over to the lower valence V and be stored there. During desorption, the stored hydrogen on the lower valence vanadium could flow back to Rh, and desorb

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

41

Fig. 2. SEM image of 20NA10V (a), elemental mapping images of O (b), Al (c), V (d) and Ni (e).

from Rh. In this work, the V2 O3 over the reduced 20NAxV catalysts should be able to play similar role, causing the spillover of hydrogen from Ni to the nearby V species. The increment of peak area and the shift of the peak towards lower temperature range can be attributed to the enhanced hydrogen mobility due to larger amount of V2 O3 in the catalysts. These results reveal that the vanadium promoter seems to play important role in hydrogen storage and transfer which is conducive to the hydrogenation reaction. 3.1.5. CO2 -TPD As shown in Fig. S2, all catalysts have similar CO2 -TPD profiles with two peaks at about 137 and 273 ◦ C, respectively. The low-temperature desorption peak at 137 ◦ C can be associated to weak basic sites on the catalyst surface, which can form surface monodentate with CO2 on the catalysts [41]; while the relatively high-temperature desorption peak at 273 ◦ C can be associated to moderate basic sites, which lead to the formation of surface bidentate CO2 [41] or desorption of formate [42]. The latter is formed in the presence of surface OH groups [43]. In Fig.

S2, the peak areas gradually decrease following the sequence: 20NA > 20NA2V > 20NA5V > 20NA10V > 20NA16V, indicating that both the basicity and the number of basic active sites decline with the increase of V2 O3 , which is accordant with the result of the literature [44]. 3.1.6. SEM Fig. S3 presents the SEM images of the reduced catalysts. For 20NA (Fig. S3a), it can be seen that the Ni nanoparticles are highly dispersed on the Al2 O3 surface. After loading with vanadium oxide, no obvious change is seen in the images of 20NAxV (Fig. S3b–f), indicating that both the Ni and V2 O3 particles are well dispersed on the surface of the catalysts. The elemental mappings of the reduced 20NA10V are shown in Fig. 2, which confirm that Ni and vanadium promoter nanoparticles are homogeneously dispersed on the catalyst surface. 3.1.7. TEM The TEM images of the reduced catalysts are shown in Fig. 3. For 20NA, agglomeration of Ni particles is observed, and the Ni particle

42

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

Fig. 3. TEM images of reduced catalysts: (a) 20NA, (b) 20NA2V, (c) 20NA5V, (d and f) 20NA10V and (e) 20NA16V.

size is varied from 5 to 28 nm (Fig. 3a). After addition of vanadium promoter, the Ni particles are decreased obviously in size but there is no obvious Ni particle agglomeration in the reduced 20NAxV catalysts. The Ni particle sizes are about 2–8, 2–6, 2–7 and 5–16 nm in 20NA2V (Fig. 3b), 20NA5V (Fig. 3c), 20NA10V (Fig. 3d) and 20NA16V (Fig. 3e), respectively. Fig. 3f shows the HRTEM images of 20NA10V.

The observed lattice spacing of ca. 0.20 nm corresponds to the Ni (1 1 1) plane, as reported previously [12], which is closely related to the active sites for the methanation reaction. The lattice spacing at ca. 0.25 nm corresponds to the V2 O3 (1 1 0) plane, and the V2 O3 nanocrystallites are well dispersed and located closely to or even on the Ni particles.

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

80

100

0.1 MPa

Thermodynamics 20NA16V 20NA10V 20NA5V 20NA2V 20NA

40 20

250 100

300 350 400 o Temperature ( C)

(d)

98

40 20 250

100

99 3.0 MPa

90

20NA16V 20NA10V 20NA5V 20NA2V 20NA

97 96 95

300

400 500 o Temperature ( C)

0.1 MPa

Thermodynamics 20NA16V 20NA10V 20NA5V 20NA2V 20NA

60

80

300 350 o Temperature ( C)

(e)

0.1 MPa

20 0

400

250

90

20NA16V 20NA10V 20NA5V 20NA2V 20NA

400 500 o Temperature ( C)

Thermodynamics 20NA16V 20NA10V 20NA5V 20NA2V 20NA

40

100

3.0 MPa

300

60

CH4 Yield (%)

60

(c)

80

80

0

CO Conversion (%)

(b)

CH4 Selectivity (%)

100

CH4 Yield (%)

(a)

CH4 Selectivity (%)

CO Conversion (%)

100

43

80

300 350 400 o Temperature ( C)

(f)

3.0 MPa

20NA16V 20NA10V 20NA5V 20NA2V 20NA

300

400 500 o Temperature ( C)

Fig. 4. Catalytic properties of the catalysts using a WHSV of 120,000 mL g−1 h−1 : (a and d) CO conversion, (b and e) CH4 selectivity and (c and f) CH4 yield.

3.2. Catalytic properties of the catalysts 3.2.1. CO methanation The CO methanation reaction was carried out in the temperature range of 260–400 ◦ C at 0.1 MPa and weight hourly space velocity (WHSV) of 120,000 mL g−1 h−1 , and the results are shown in Fig. 4a–c. For 20NA, the catalytic activity is very poor below 300 ◦ C, and the CO conversion only reaches the thermodynamics equilibrium when the temperature is as high as 400 ◦ C. However, the addition of vanadium promoter can considerably improve the lowtemperature catalytic activities. For 20NAxV (x = 2, 5 and 10), the CO conversion can reach thermodynamics equilibrium in the temperature range of 300–400 ◦ C, and on 20NA5V, it even can reach 85% CO conversion at a reaction temperature as low as 280 ◦ C. On the other hand, the improvement in CH4 selectivity is not as obvious as that of CO conversion after V addition. The measured CH4 yield of 20NAxV (x = 2, 5 and 10) is very close to the thermodynamics equilibrium data above 300 ◦ C. However, the catalytic activity of 20NA16V is deceased obviously compared to those of other catalysts containing vanadium, and is even lower than that of 20NA above 350 ◦ C. Clearly, the addition of proper amount of V2 O3 in Ni/Al2 O3 can remarkably enhance the CO methanation at atmospheric pressure. Mori et al. [45,46] found that the main role of the vanadium promoter was to enhance the rate of CO dissociation via interaction between the oxygen atom in the chemisorbed CO and a positively charged promoter center (V3+ ) [47]. Happel et al. found that the slowest step in the methanation reaction was the hydrogenation of CHx species [17], and the concentration of adsorbed hydrogen atoms on the surface was the limiting factor. Combined with the results of aforementioned H2 -TPD, the much improved activity of 20NAxV (x = 2, 5 and 10) for CO methanation can be attributed to the enhancement of CO dissociation and H2 uptakes by V2 O3 . Since CO methanation is a volume-reduced reaction and normally conducted at high pressure (2.94–3.43 MPa) in industry [12], this reaction was further conducted in the temperature range of 300–550 ◦ C at 3.0 MPa and WHSV of 120,000 mL g−1 h−1 , and the results are shown in Fig. 4d–f. The CO conversions of all catalysts are almost the same in the total temperature range. On the other

hand, both the CH4 selectivity and the CH4 yield of 20NAxV are slightly higher than those of 20NA, and among the above 20NAxV catalysts, 20NA5V is the best (except the data at 400 and 550 ◦ C). Overall, 20NA5V shows the highest activity at both atmosphere and 3.0 MPa pressures for CO methanation. 3.2.2. Apparent activation energies of the catalysts in CO methanation The apparent activation energy for CO methanation over the different catalysts was determined and the Arrhenius plots are shown in Fig. 5. The results show that the activation energies decrease in the order of 20NA16V (109.27 kJ mol−1 ) > 20NA (109.25 kJ mol−1 ) > 20NA2V (103.31 kJ mol−1 ) > 20NA10V −1 (95.69 kJ mol ) > 20NA5V (86.38 kJ mol−1 ), in good agreement

Fig. 5. Arrhenius plots for CO methanation on different catalysts.

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

80 60 Thermodynamics 20NA16V 20NA10V 20NA5V 20NA2V 20NA

40 20

0.1 MPa, 90000 mL g-1 h-1

100

300 400 o 500 Temperature ( C)

(d)

CO2 Conversion (%)

60 20NA16V 20NA10V 20NA5V 20NA2V 20NA

40 20

300

400 500o 600 Temperature ( C)

60

Thermodynamics 20NA16V 20NA10V 20NA5V 20NA2V 20NA

Thermodynamics 20NA16V 20NA10V 20NA5V 20NA2V 20NA

40 20

0.1 MPa, 90000 mL g-1 h-1

80

100

(e)

0.1 MPa, 90000 mL g-1 h-1

0

300 400 o 500 Temperature ( C)

20NA16V 20NA10V 20NA5V 20NA2V 20NA

90

300 400 o 500 Temperature ( C)

(f)

50

20NA16V 20NA10V 20NA5V 20NA2V 20NA

2.0 MPa, 120000 mL g-1 h-1

2.0 MPa, 120000 mL g-1 h-1

0

(c)

80

90

100

80

100

CH4 Selectivity (%)

0

(b)

CH4 Selectivity (%)

100

CH4 Yield (%)

(a)

CO2 Conversion (%)

100

CH4 Yield (%)

44

80

300

400 500 o Temperature ( C)

600

0

2.0 MPa, 120000 mL g-1 h-1

300

400 500 o Temperature ( C)

600

Fig. 6. Catalytic properties of the catalysts: (a and d) CO2 conversion, (b and e) CH4 selectivity and (c and f) CH4 yield.

with the activity trend of CO methanation at atmospheric pressure. Furthermore, based on the results of CO conversion at 220 ◦ C and the H2 adsorption amount, TOFCO is calculated and listed in Table 1. It can be seen that overall, the 20NAxV catalysts have higher TOFCO values than 20NA, and among them, 20NA5V has the highest TOFCO value, which further confirms proper amount of V2 O3 can significantly enhance the performance of Ni/Al2 O3 for CO methanation.

3.2.3. CO2 methanation As another essential methanation process, CO2 methanation was also carried out over the catalysts in the temperature range of 260–500 ◦ C at 90,000 mL g−1 h−1 , as shown in Fig. 6a–c. Overall, the CO2 conversion and CH4 yield of all the catalysts present volcano-shaped trends with increase of the reaction temperature, because the reaction is a strongly exothermic reaction and the high temperature will have an adverse effect on it. The maximum CO2 conversion and CH4 yield over 20NA can reach 73 and 71% respectively at 420 ◦ C, while obvious enhancement is obtained for the catalyst after vanadium oxide addition, and the maximum CO2 conversion and CH4 yield over 20NA2V are both 82% at 400 ◦ C. Among all the tested catalysts, 20NA5V shows the best activity again. However, unlike CO methanation, even on 20NA5V, the CO2 conversion and CH4 yield cannot reach thermodynamics equilibrium in the total temperature range, and in fact, the difference in activity with different catalysts is quite small. Probably, there are two reasons that can explain this phenomenon. Firstly, despite CO2 methanation is favorable in thermodynamics [11], the eight-electron reduction of CO2 to CH4 by hydrogen is difficult to achieve due to the larger kinetic barriers as compared to that of CO methanation [48]; secondly, as mentioned above, V2 O3 can facilitate the H2 adsorption and CO dissociation [45,46], which is conducive to CO methanation; however, addition of V2 O3 is disadvantageous for CO2 dissociation according to the results of CO2 -TPD (Fig. S2), resulting in the unapparent promotion of V2 O3 for CO2 methanation. Although the CO2 adsorption on 20NAxV is weaker than that on 20NA, the greater H2 uptakes on the former still lead to the enhanced activity of 20NAxV for CO2 methanation.

CO2 methanation was further carried out at high pressure (2.0 MPa) in the temperature range of 300–550 ◦ C at 120,000 mL g−1 h−1 , as shown in Fig. 6d–f. Similar to the literature results [49], the promotion effect of pressure for CO2 methanation is less obvious than that for CO methanation. In general, the activities of the catalysts follows the order of 20NA5V > 20NA10V > 20NA2V > 20NA16V > 20NA. Combining with the results of CO and CO2 methanation, we can conclude that 20NA5V is a promising candidate not only for CO but also for CO2 methanation to produce SNG. Also, it is applicable in preferential methanation of CO from a syngas involving CO2 at low temperature. 3.3. Stability of the catalysts The lifetime test is important in evaluation of a catalyst, and particularly, maintaining the activity and stability of the catalyst under harsh reaction conditions is crucial for the industrial process [12]. However, for the CO methanation reaction, most of the studies only evaluated the catalysts at low temperature and WHSV, although some authors [10,15] claimed their catalysts were stable more than 100 h. As stated above, these data are not persuasive and sufficient for an industrial process. Therefore, we carried out the 140 h lifetime tests for 20NA and 20NA5V at 500–550 ◦ C, 3.0 MPa and high WHSVs of 120,000 or 240,000 mL g−1 h−1 , and the results are shown in Fig. 7. For 20NA, the activity remains stable after 21 h at 500 ◦ C using 120,000 mL g−1 h−1 , but it deceases slightly after 52 h (total time 73 h) using 240,000 mL g−1 h−1 . The obvious decline is seen at 550 ◦ C using 240,000 mL g−1 h−1 after 57 h (total time 130 h). On the other hand, for 20NA5V, the activity remains stable at every process, and only the CH4 selectivity and CH4 yield are slight declined at the end of the 140 h lifetime test. Apparently, 20NA5V shows a better stability especially at high temperatures and WHSVs. 3.4. Characterization of the used catalysts The SEM images of the used 20NA and 20NA5V catalysts are shown in Fig. 8a–b. Some filamentous carbon can be seen on the surface of 20NA (Fig. 8a). In contrast, for 20NA5V, almost no carbon

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

45

Fig. 7. Lifetime test of 20NA5V and 20NA under 3.0 MPa: (a) CO conversion, (b) CH4 selectivity and (c) CH4 yield.

filament can be observed even after the 140 h lifetime test (Fig. 8b). The amount of carbon deposited on the used catalysts is further measured by TG analysis, and the result is presented in Fig. 8c. The carbon content over the used 20NA and 20NA5V is estimated to be 1.7 and 0.4 wt% respectively, suggesting a higher coke resistance of

20NA5V. In addition, there is no new diffraction peak corresponding to graphitic carbon in the XRD patterns of the used catalysts (Fig. 8d), indicating that the amount of graphitic carbon is below the detection limit of XRD. The TEM images of the used 20NA and 20NA5V catalysts are shown in Fig. 8e–f. For the used 20NA (Fig. 8e),

Fig. 8. SEM images of used 20NA (a) and used 20NA5V (b), TG curves of the reduced and used catalysts in air (c), XRD patterns of the catalysts after lifetime tests (d), TEM images of used 20NA (e) and used 20NA5V (f).

46

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

Fig. 9. Ni 2p (a), V 2p (b) and O 1s (c) XPS spectra of the catalysts.

[27]. The XPS results reveal that V3+ , V4+ and V5+ coexist in both the reduced and used 20NA5V. It seems contradictory that only V2 O3 crystalline phase is observed in the XRD patterns of reduced V2 O5 , Ni3 V2 O8 and 20NA16V (Fig. 1b). This may be explained by the fact that because the component on the surface is very different from the bulk, and the surface of the vanadium oxide is rich in V5+ and V4+ oxidation states while the inner of it is rich in V3+ species [20]. As mentioned above, there are Al2 O3 , vanadium oxides (VOx ) and NiO species on the catalyst surface and the binding energies of O 1s for VOx are very close [27], so the O 1s spectra in Fig. 9c are further fitted to three peaks around 531.5 (OI ), 530.6 (OII ) and 529.8 eV (OIII ), respectively. For the reduced 20NA catalyst, there are only OI and OIII . The relative contents of different oxygen calculated for catalysts are shown in Table 2. Compared with the percentage of OII in the reduced 20NA5V, the used 20NA5V has less OII , indicating that the reduction of VOx to low valence occurred during the reaction process. The percentage of OIII in reduced 20NA5V is higher than that in the used 20NA5V, probably because the oxidation of the former is more severe.

the agglomeration of Ni particles is much more serious than that in the reduced 20NA (Fig. 3a) with 10–31 nm Ni particle size, indicating occurrence of serious sintering of Ni particle during the lifetime test. On the other hand, for the used 20NA5V, the Ni particle size is 5–12 nm, bigger than that of the reduced 20NA5V (Fig. 3c). We thus calculated the average Ni particle size of the used 20NA and 20NA5V catalysts from their XRD patterns (Fig. 8d), which is 16.6 and 9.0 nm respectively, both bigger than that of the reduced catalysts (Table 1). Overall, 20NA5V exhibits high stability under harsh condition and excellent resistance to coke formation, although its resistance to Ni sintering is still not high enough. 3.5. Surface analysis of the samples Fig. 9a shows the Ni 2p3/2 XPS spectra of the catalysts. For the fresh 20NA5V, Ni is in the form of Ni2+ . It can be seen that the Ni0 peak is at the same position (852.3 eV) in the reduced 20NA and 20NA5V catalysts, suggesting that the interaction of Ni and vanadium oxide does not alter the binding energy of metallic Ni. In addition, the spectrum of the used 20NA5V is almost the same as that of the reduced 20NA5V, indicating there is no obvious change in the Ni status after the lifetime test. The peak of Ni 2p at 855.8 and 861.5 eV belongs to the Ni oxide states (Ni2+ ), which may be because the surface Ni is easily oxidized after exposure to air during the sample transfer [30]. Fig. 9b presents the V 2p3/2 XPS spectra of the 20NA5V catalyst. For the fresh 20NA5V, the peak of V 2p3/2 at 517.1 eV is assigned to V5+ (VI ). After reduction, the V 2p3/2 peak is shifted to a lower binding energy (516.6 eV), and is further fitted into three peaks at about 517.1, 516.1 and 515.6 eV, assigned to V5+ (VI ), V4+ (VII ) and V3+ (VIII ) [50], respectively. The relative contents of vanadium at different oxidation states for the catalysts are calculated and listed in Table 2. It is found that the percentages of V3+ and V5+ in 20NA5V-used are higher than that in the 20NA5V-reduced, but the V4+ percentage exhibits the opposite trend, which may be because, on the one hand, V4+ in 20NA5V-reduced can be reduced to V3+ by H2 during the reaction process, and on the other hand, V4+ can also be oxidized to V5+ in the reaction or during the storage in air

3.6. Schematic diagram of the catalysts Combining all the XRD, H2 -TPR, H2 -TPD, TEM and XPS results, the possible formation process of 20NAxV catalysts is schematically illustrated in Fig. 10a. Ni3 V2 O8 coexists with the excess NiO on the surface of the as-calcined 20NAxV catalysts, and is further reduced to Ni and V2 O3 in H2 flow at 600 ◦ C with partial coverage of the surface Ni particles by V2 O3 . Compared to the case with pure NiO, the reduction of Ni3 V2 O8 facilitates the formation of smaller Ni particle size, resulting in the large H2 uptakes and higher Ni dispersion of 20NAxV catalysts. It is known that one of the effects of vanadium oxide promoter is to enhance the CO dissociation rate [45,46]. During this process, a CO molecule is first adsorbed on the metal atoms to form M–CHOH, and an adjacent V3+ -ion extracts the oxygen atom from this hydroxycarbene intermediate and promotes the dissociation of CO into (CHx )ad and (OH)ad [46]. Simultaneously, V3+ is oxidized to V4+ in

Table 2 Surface XPS compositions of the catalysts. Sample

Atomic ratio by XPS





5

V3+ /

5

(Vi+ )(%)

V4+ /

i=3

20NA-reduced 20NA5V-fresh 20NA5V-reduced 20NA5V-used

– – 31.4 33.8

 5

(Vi+ )(%)

V5+ /

i=3

– – 30.9 25.1

(Vi+ )(%)

III 

OI /

i=3

– 100 37.7 41.1

III 

(Oi )(%)

OII /

i=I

97.7 63.0 73.9 86.8

III 

(Oi )(%)

OIII /

i=I

– 27.2 21.8 10.9

(Oi )(%) i=I

2.3 9.8 4.3 2.3

Q. Liu et al. / Applied Catalysis A: General 488 (2014) 37–47

47

References

Fig. 10. Schematic diagram of the formation process of 20NAxV catalysts (a) and coke elimination over the 20NA5V catalyst (b).

this process. In a word, the vanadium oxide promotes the CO dissociation by making use of an oxidation–reduction cycle. At the same time, the (OHy )ad species on the V4+ ion are hydrogenated to H2 O, following with the reduction of V4+ to V3+ [46], which can increase the oxygen vacancies. The oxygen vacancies act as the active site for activating CO2 which is the byproduct of CO methanation, and the activated CO2 can react with the deposited carbon and facilitate the removal of carbon formed on the Ni particles (seen in Fig. 10b) [51–53]. 4. Conclusions In this work a series of Ni-V2 O3 /Al2 O3 catalysts are prepared by the co-impregnation method for CO/CO2 methanation. The 20NA5V catalyst shows the highest catalytic performance at both 0.1 MPa and 3.0 MPa, achieving nearly 100% CO conversion and 89% CH4 selectivity at 0.1 MPa and a high WHSV of 120,000 mL g−1 h−1 . In a 140 h life test conducted at high temperatures (500 and 550 ◦ C), 3.0 MPa and WHSVs of 120,000 and 240,000 mL g−1 h−1 , 20NA5V still displays a high stability and resistance to coke formation, on which only 0.4 wt% deposited carbon is accumulated after the life test, much lower than that of the catalyst without vanadium promoter. The much improved catalytic performance is attributed to the formed small Ni particle sizes (ca. 3.0 nm) from Ni3 V2 O8 obtained during the calcination. The oxidation–reduction cycle of V2 O3 , on the one hand, enhances CO dissociation in CO methanation thus improves the low-temperature activity of the catalysts, and on the other hand, increases the oxygen vacancies. The later enhance the dissociation of CO2 to generate surface oxygen intermediates, thus preventing carbon deposition on the Ni particles and enhancing high-temperature stability of the catalysts. In addition, Ni-V2 O3 /Al2 O3 catalyst is also an active catalyst for CO2 methanation at both atmospheric and 2.0 MPa pressures. This work demonstrates that Ni-V2 O3 /Al2 O3 catalyst is a promising candidate not only for CO but also for CO2 methanation to produce SNG. Acknowledgments The authors gratefully acknowledge the supports from the National Natural Science Foundation of China (No. 21476238), National Basic Research Program (No. 2014CB744306), National Key Technology R&D Program of China (No. 2010BAC66B01), and “Strategic Priority Research Program” of the Chinese Academy of Sciences (Nos. XDA07010100 and XDA07010200). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.apcata. 2014.09.028.

[1] P. Sabatier, J.B. Senderens, Compt. Rend. Acad. Sci. 134 (1902) 514–516. [2] M.B.I. Choudhury, S. Ahmed, M.A. Shalabi, T. Inui, Appl. Catal. A 314 (2006) 47–53. [3] A.M. Abdel-Mageed, S. Eckle, H.G. Anfang, R.J. Behm, J. Catal. 298 (2013) 148–160. [4] R. Razzaq, C. Li, S. Zhang, Fuel 113 (2013) 287–299. [5] J. Zhang, Z. Xin, X. Meng, M. Tao, Fuel 109 (2013) 693–701. [6] A. Zhao, W. Ying, H. Zhang, H. Ma, D. Fang, Catal. Commun. 17 (2012) 34–38. [7] P. Panagiotopoulou, D.I. Kondarides, X.E. Verykios, Appl. Catal. A 344 (2008) 45–54. [8] J. Gao, C. Jia, J. Li, M. Zhang, F. Gu, G. Xu, Z. Zhong, F. Su, J. Energy Chem. 22 (2013) 919–927. [9] C. Jia, J. Gao, J. Li, F. Gu, G. Xu, Z. Zhong, F. Su, Catal. Sci. Technol. 3 (2013) 490–499. [10] D. Hu, J. Gao, Y. Ping, L. Jia, P. Gunawan, Z. Zhong, G. Xu, F. Gu, F. Su, Ind. Eng. Chem. Res. 51 (2012) 4875–4886. [11] J. Gao, Y. Wang, Y. Ping, D. Hu, G. Xu, F. Gu, F. Su, RSC Adv. 2 (2012) 2358–2368. [12] Q. Liu, J. Gao, M. Zhang, H. Li, F. Gu, G. Xu, Z. Zhong, F. Su, RSC Adv. 4 (2014) 16094–16103. [13] C. Guo, Y. Wu, H. Qin, J. Zhang, Fuel Process. Technol. 124 (2014) 61–69. [14] Y. Wang, R. Wu, Y. Zhao, Catal. Today 158 (2010) 470–474. [15] S. Ma, Y. Tan, Y. Han, J. Nat. Gas Chem. 20 (2011) 435–440. [16] X. Yan, Y. Liu, B. Zhao, Z. Wang, Y. Wang, C. Liu, Int. J. Hydrogen Energy 38 (2013) 2283–2291. [17] J. Happel, I. Suzuki, P. Kokayeff, V. Fthenakis, J. Catal. 65 (1980) 59–77. [18] J. Liu, C. Li, F. Wang, S. He, H. Chen, Y. Zhao, M. Wei, D.G. Evans, X. Duan, Catal. Sci. Technol. 3 (2013) 2627–2633. [19] J.A. Lercher, J.H. Bitter, W. Hally, W. Niessen, K. Seshan, Stud. Surf. Sci. Catal. 101 (1996) 463–472. [20] N. Alov, D. Kutsko, I. Spirovova, Z. Bastl, Surf. Sci. 600 (2006) 1628–1631. [21] S.I. Ito, C. Chibana, K. Nagashima, S. Kameoka, K. Tomishige, K. Kunimori, Appl. Catal. A 236 (2002) 113–120. [22] S.I. Ito, S. Ishiguro, K. Kunimori, Catal. Today 44 (1998) 145–149. [23] S.I. Ito, S. Ishiguro, K. Nagashima, K. Kunimori, Catal. Lett. 55 (1998) 197–199. [24] T. Yamagishi, S.I. Ito, K. Tomishige, K. Kunimori, Catal. Commun. 6 (2005) 421–425. [25] T. Yamagishi, I. Furikado, S.I. Ito, T. Miyao, S. Naito, K. Tomishige, K. Kunimori, J. Mol. Catal. A: Chem. 244 (2006) 201–212. [26] B.J. Kip, P.A.T. Smeets, J. Vangrondelle, R. Prins, Appl. Catal. 33 (1987) 181–208. [27] B.J. Kip, P.A.T. Smeets, J. Vanwolput, H.W. Zandbergen, J. Vangrondelle, R. Prins, Appl. Catal. 33 (1987) 157–180. [28] R. Kraselcuk, A.I. Isli, A.E. Aksoylu, Z.I. Onsan, Appl. Catal. A 192 (2000) 263–271. [29] B. Zhaorigetu, W. Li, R. Kieffer, H. Xu, React. Kinet. Catal. Lett. 75 (2002) 275–287. [30] J. Gao, C. Jia, J. Li, F. Gu, G. Xu, Z. Zhong, F. Su, Ind. Eng. Chem. Res. 51 (2012) 10345–10353. [31] J. Gao, C. Jia, M. Zhang, F. Gu, G. Xu, F. Su, Catal. Sci. Technol. 3 (2013) 2009–2015. [32] J. Gao, C. Jia, M. Zhang, F. Gu, G. Xu, Z. Zhong, F. Su, RSC Adv. 3 (2013) 18156–18163. [33] V.M. Shinde, G. Madras, AlChE J. 60 (2014) 1027–1035. [34] G. Zhi, X. Guo, Y. Wang, G. Jin, X. Guo, Catal. Commun. 16 (2011) 56–59. [35] W. Reichl, K. Hayek, J. Catal. 208 (2002) 422–434. [36] J. Kijenski, A. Baiker, M. Glinski, P. Dollenmeier, A. Wokaun, J. Catal. 101 (1986) 1–11. [37] Z.R. Bao, W.Z. Li, H.Y. Xu, R. Kieffer, Catal. Lett. 94 (2004) 125–129. [38] S. Velu, S. Gangwal, Solid State Ionics 177 (2006) 803–811. [39] H.Y. Luo, W. Zhang, H.W. Zhou, S.Y. Huang, P.Z. Lin, Y.J. Ding, L.W. Lin, Appl. Catal. A 214 (2001) 161–166. [40] H.Y. Luo, H.W. Zhou, L.W. Lin, D.B. Liang, C. Li, D. Fu, Q. Xin, J. Catal. 145 (1994) 232–234. [41] L. Li, L. Song, H. Wang, C. Chen, Y. She, Y. Zhan, X. Lin, Q. Zheng, Int. J. Hydrogen Energy 36 (2011) 8839–8849. [42] J. Liu, H. Peng, W. Liu, X. Xu, X. Wang, C. Li, W. Zhou, P. Yuan, X. Chen, W. Zhang, H. Zhan, ChemCatChem (2014), http://dx.doi.org/10.1002/cctc.201402091. [43] Y. Pan, C. Liu, Q. Ge, J. Catal. 272 (2010) 227–234. [44] K.V.R. Chary, G. Kishan, C.P. Kumar, G.V. Sagar, Appl. Catal. A 246 (2003) 335–350. [45] T. Mori, A. Miyamoto, N. Takahashi, M. Fukagaya, H. Niizuma, T. Hattori, Y. Murakami, J. Chem. Soc. Chem. Commun. (11) (1984) 678–679. [46] T. Mori, A. Miyamoto, N. Takahashi, M. Fukagaya, T. Hattori, Y. Murakami, J. Phys. Chem. 90 (1986) 5197–5201. [47] W.M.H. Sachtler, D.F. Shriver, W.B. Hollenberg, A.F. Lang, J. Catal. 92 (1985) 429–431. [48] K.R. Thampi, J. Kiwi, M. Gratzel, Nature 327 (1987) 506–508. [49] R. Razzaq, H. Zhu, L. Jiang, U. Muhammad, C. Li, S. Zhang, Ind. Eng. Chem. Res. 52 (2013) 2247–2256. [50] A.B. Boffa, A.T. Bell, G.A. Somorjai, J. Catal. 139 (1993) 602–610. [51] S. Therdthianwong, C. Siangchin, A. Therdthianwong, Fuel Process. Technol. 89 (2008) 160–168. [52] J. Wei, B. Xu, J. Li, Z. Cheng, Q. Zhu, Appl. Catal. A 196 (2000) L167–L172. [53] N. Wang, K. Shen, L. Huang, X. Yu, W. Qian, W. Chu, ACS Catal. 3 (2013) 1638–1651.