Influence of calcination temperature on textural and structural properties, reducibility, and catalytic behavior of mesoporous γ-alumina-supported Ni–Mg oxides by one-pot template-free route

Journal of Catalysis 329 (2015) 151–166

Contents lists available at ScienceDirect

Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat

Influence of calcination temperature on textural and structural properties, reducibility, and catalytic behavior of mesoporous c-alumina-supported Ni–Mg oxides by one-pot template-free route Mingwu Tan, Xueguang Wang ⇑, Xinxing Wang, Xiujing Zou, Weizhong Ding, Xionggang Lu ⇑ State Key Laboratory of Advanced Special Steel, Shanghai University, Shanghai 200072, China

a r t i c l e

i n f o

Article history: Received 15 February 2015 Revised 12 May 2015 Accepted 14 May 2015

Keywords: Mesoporous alumina Nickel catalyst Nickel–magnesium oxides Calcination temperature Liquefied petroleum gas Steam reforming

a b s t r a c t c-Alumina-supported Ni–Mg oxides by one-pot route were calcined in the range of 300–700 °C. Influences of calcination temperature on catalyst structure, surface properties, interaction between metal oxides and support, reducibility of Ni2+ ions, and Ni particle dispersion were systematically investigated. The catalysts showed wormhole-like mesoporous structures with narrow pore size distributions. NiO species were homogeneously distributed with the strengthened interaction by support with calcination temperature, producing uniform Ni nanoparticles throughout c-alumina frameworks after reduction. The Ni particle sizes increased with calcination temperature due to Ni crystallite growth by Ostwald ripening rather than migration of Ni nanoparticles. The investigation for steam reforming of liquefied petroleum gas revealed that smaller Ni nanoparticles facilitated hydrocarbon reforming, methanation of carbon oxides, and water–gas shift and lowered the rate of coke deposition; mesopores with uniform sizes were favorable for diffusion of reactants and products to active sites, improving catalytic activity and stability. Ó 2015 Elsevier Inc. All rights reserved.

1. Introduction Alumina-supported nickel catalysts have widely been investigated for various reforming processes of hydrocarbons, including carbon dioxide reforming [1,2], partial oxidation reforming [3], steam reforming [4], and auto-thermal reforming [5], because of their wide availability, high intrinsic catalytic activity, and low cost. However, conventional alumina-supported nickel catalysts suffer from severe catalyst deactivation, which can be caused by carbon deposition of the active metal surface, nickel particle sintering, reaction with support, and phase transformation during the reforming processes [6–9]. It has been shown that the chemical composition and the preparation procedure exhibit significant effects on the structure and surface properties of the catalysts, thus resulting in substantial change in the catalytic performance. Therefore, many effects have been made to improve resistance to coke deposition and to prevent sintering of nickel catalysts supported by alumina through adding structural and electronic promoters, modifying catalyst supports, improving catalyst preparation routes, etc. [10–20]. A majority of alumina-supported nickel catalysts for the reforming process are prepared by incipient wetness impregnation ⇑ Corresponding authors. Fax: +86 21 56338244. E-mail addresses: [email protected] (X. Wang), [email protected] (X. Lu). http://dx.doi.org/10.1016/j.jcat.2015.05.011 0021-9517/Ó 2015 Elsevier Inc. All rights reserved.

technique, followed by a suitable thermal activation process. Numerous investigations have been directed toward the characteristics and distribution of surface nickel compounds on the support, the reducibility of nickel oxide species, the shape and size of metallic nickel particles, the catalyst composition, and the interplay of metal particles and the support properties, which strongly influence the catalytic activity, stability, and resistance to carbon deposition [21–27]. Although most of the obtained results are very interesting, the correlation among the active metal species, the catalyst structures, and the catalytic properties in the reforming reactions is still ambiguous, even though it is pivotal to understand and to abate its irreversible deactivation, due to the inherently inhomogeneous metal particle dispersion and porous structure in the catalysts obtained by the impregnation method. Recently, we have successfully synthesized for the first time high-surface area c-alumina (c-MA) with a homogeneous wormhole-like mesoporous network structure at a low temperature of 300 °C, which showed high thermal and hydrothermal stability, through a simple partial hydrolysis of Al(NO3)3 aqueous solution with (NH4)2CO3 without organic surfactants [28]. On this basis, we further extended this approach to prepare mesoporous c-alumina-supported Ni and Ni–Mg mixed oxides by one-pot hydrolysis method [29]. The obtained NiO–MgO/c-MA materials also possessed wormhole-like mesoporous structures with large surface areas, pore volumes, and narrow pore size distributions,

152

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

similar to pure c-MA. Ni–Mg oxide species were homogeneously distributed in c-MA frameworks and produced uniformly dispersed Ni nanoparticles with relatively narrow particle size distributions after the reduction with H2. The Ni–MgO/c-MA catalysts reduced were used for the steam reforming of liquefied petroleum gas (LPG) and showed much higher stability and coke resistance ability than either the ordered mesoporous alumina-supported Ni–MgO catalyst with amorphous framework walls obtained via one-pot surfactant-assisted strategy [18,30] or c-MA-supported Ni–MgO catalyst via the traditional impregnation route mainly, due to forming smaller metal Ni nanoparticles and stable support structures [29]. This information will provide an ideal opportunity for us to better investigate the nature of the active metal species, the catalytic roles of the promoter, and the interaction among the metal Ni, the promoter, and the alumina support in the alumina-supported nickel-based catalysts, which are of fundamental importance for optimally designing a high-efficient nickel catalyst for the reforming of hydrocarbons. The previous investigations [31–35] have shown that for the supported metal catalysts, the preparation conditions have direct influences on the surface properties of the catalyst, the dispersion of surface metal oxides on the support, and the interaction between metal oxides and the support. This implies that the particle sizes of metal Ni, the extents of metal–support interaction, the surface basicity of the catalyst, and the promotion efficiency of Mg promoter on the c-MA-supported Ni–Mg catalysts, which control the catalyst activity and stability and the resistance to coke deposition in the steam reforming of hydrocarbons, can be affected through varying the thermal treatment conditions such as calcination temperature in the preparation process. Liquefied petroleum gas, which consists mainly of propane and butane without sulfur or other electronegative atoms, is considered to be a promising fuel for distributed hydrogen production for fuel cell applications due to their higher energy density than natural gas, easy storage and transportation, and existing infrastructures available for its use [36–39]. Low-temperature steam reforming (or pre-reforming) of LPG is usually operated in the temperature range of 300–700 °C [29,40,41], where c-alumina phase is stable. Thus, LPG can be used as an eligible model hydrocarbon compound to investigate the catalytic behaviors for the steam reforming of hydrocarbons over c-alumina-supported Ni-based catalyst. In this work, mesoporous c-alumina-supported Ni–Mg oxide material, NiO–MgO/c-MA, prepared through one-pot hydrolysis technique without organic surfactants, was calcined at different temperatures in the range of 300–700 °C for use in the pre-reforming of LPG, where Ni species existed in the c-MA framework mainly in the form of highly dispersed NiOx (or surface nickel aluminate), avoiding the formation of nickel aluminate spinel. Influences of calcination temperature on the textural and structural properties, the interaction between metal oxide species and c-alumina support, the reducibility of Ni2+ ions, and the size of Ni particles were discussed in detail. A comprehensive insight into the interrelation among the active metal, the promoter, the catalyst structure, and the catalytic properties over c-alumina-supported Ni-based catalyst for the reforming reactions of hydrocarbons is provided, in combination with the catalytic results for the steam reforming of LPG as a model compound.

2. Experimental 2.1. Preparation of c-MA-supported Ni–Mg catalysts The c-MA-supported Ni–Mg oxide catalysts were prepared through one-pot hydrolysis and co-condensation of an aqueous

solution of inorganic metal salts with (NH4)2CO3 (Sinopharm Chemical Reagent Co., Ltd., China) as a hydrolytic agent in the similar way, as described in our previous study [29]. A total of 200 mL of a mixed aqueous solution of metal nitrates (Sinopharm Chemical Reagent Co., Ltd., China) with a total metal [Mg + Al + Ni] ion concentration of 2.0 mol/L was heated to 70 °C, and 1 mol/L (NH4)2CO3 aqueous solution was added slowly (3.6 mL/min) into the above aqueous solution of metal salts with vigorous magnetic stirring using a peristaltic pump until a sudden formation of monolithic green and transparent gel took place, and the stirring was stopped. At this time, the pH value of the gel was determined at 5.5. The gel beaker was covered with plastic film and aged at 30 °C for 48 h. After this, the crude gel was dispersed in an open glass dish at 100 °C in an oven in a laboratory fume hood for 24 h and followed by treating in two steps at 150 °C and 200 °C in air in a muffle furnace in the hood for 10 h, respectively, to remove ammonium nitrate. Caution! This process must be operated with care to avoid the explosion of the ammonium nitrate. The treated solid was divided into five equal portions, which were calcined in air for 10 h at a heating rate of 1 °C/min at 300 °C, 400 °C, 500 °C, 600 °C, and 700 °C, respectively. In the obtained catalysts, the nominal Mg/Al and Ni/Al molar ratios were fixed at 0.25 and 0.24 (or 18 wt% Ni), respectively. This metal composition was considered as optimal for the low-temperature steam reforming of LPG in our previous work [29]. The resulting c-MA-supported Ni–Mg oxide materials were designated as NiO–MgO/c-MA–T and read as Ni–MgO/c-MA–T after H2 reduction, where T represents the calcination temperature of the catalyst in °C. 2.2. Catalyst characterization The powder X-ray diffraction (XRD) analysis of the catalysts was carried out on a Rigaku D/MAX-2200 diffractometer with a Cu Ka radiation (k = 0.15418 nm) at a voltage of 40 kV and a current of 40 mA. The NiO–MgO/c-MA samples reduced and spent for the XRD analysis were immediately transferred in N2 atmosphere in a sealed container. The diffraction patterns were identified via comparing with those in the JCPDS database. The crystallite sizes of c-Al2O3 and metal Ni were estimated using the full widths at half maximum (FWHM) of the Al2O3 (4 4 0) and Ni (2 0 0) peaks, respectively, through the Scherrer equation: d = 0.94k/(b cos h), where k is the wavelength of the radiation, b is FWHM, and h is the half of the diffraction angle [29]. Scanning electron microscopy (SEM) images were obtained on a Hitachi S-4800 electron microscope. TEM micrographs were performed with a JEOL JEM-2010F field emission microscope operating at 200 kV. The sample was prepared by placing a drop of the ethanol solution of a well-ground catalyst powder on a carbon-coated copper grid followed by ethanol evaporation. Size distributions of the Ni particles were obtained by counting ca. 200–250 particles in TEM images. N2 adsorption and desorption isotherms were measured using a Micromeritics ASAP 2020 sorptometer at liquid nitrogen temperature (196 °C). Before the measurements, the samples were degassed at 300 °C for at least 6 h. The Brunauer–Emmett–Teller (BET) method was used to calculate the specific surface area (SBET) using adsorption data in the P/P0 range from 0.05 to 0.25. The pore size distribution (PSD) was calculated using the desorption branch of the isotherms with the Barrett–Joyner–Halenda (BJH) method. The pore size (Dp) was obtained from the peak position of the distribution curve, and the average pore size (Da) was calculated by the BJH method. The pore volume (Vp) was estimated from the adsorbed amount at the P/P0 = 0.990. X-ray photoelectron spectra (XPS) of the samples were measured on an ESCALAB 250Xi spectrometer. This spectrometer used an aluminum anode (Al Ka = 1486.6 eV) operated at 12 kV and

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

153

20 mA. The instrument was performed at a pressure of ca. 1  109 Torr. The catalyst powder was placed on a sample holder and pressed into self-supported wafer. Each spectrum was run at least two times. Binding energies were measured with a precision of ±0.2 eV or better. The spectra for the samples were calibrated using the binding energy of Al 2p peak at 73.8 eV with respect to c-Al2O3 support. CO2 temperature-programmed desorption (CO2-TPD) was conducted to analyze the surface basic properties of the catalyst on a homemade apparatus. Prior to the measurement, 100 mg of sample was pretreated in a helium flow (30 mL/min) at 300 °C for 0.5 h and then cooled to 50 °C and exposed to 10 vol% CO2/He atmosphere (30 mL/min) for 1 h. After this, the sample was purged with a helium flow (30 mL/min) for 1 h. Finally, CO2-TPD was carried out with a temperature ramp of 15 °C/min to 700 °C. The amount of desorbed CO2 was measured with a gas chromatograph equipped with a thermal conductivity detector (TCD). H2 temperature-programmed reduction (H2-TPR) was performed using the same setup as CO2-TPD in a conventional flow system with a moisture trap to observe the reducibility of the NiO–MgO/c-MA catalysts. Typically, 100 mg of sample placed in a quartz reactor was first pretreated in an Ar stream (30 mL/min) at 300 °C for 0.5 h to remove moisture and other absorbed impurities and then cooled to room temperature. After this pretreatment, H2-TPR was conducted with a gas mixture of 5 vol% H2 in Ar at 30 mL/min. The temperature was raised to 1000 °C at a heating rate of 10 °C/min. The amount of H2 uptake was measured with the TCD. The surface area of metal nickel was determined using the Micromeritics ASAP 2020 sorptometer by H2 chemisorption. Approximately 1.5 g of NiO–MgO/c-MA catalyst (Note! More than 0.8 g of the catalyst was found to be necessary for the reliability of metal surface area) was placed in a quartz sample cell and treated in a He flow (30 mL/min) at 100 °C for 1 h. The sample was evacuated and then purged in flowing H2 for 5 min and finally reduced in a H2 flow (50 mL/min) at 600 °C for 2 h at a heating rate of 10 °C/min. After this, the sample was evacuated at 400 °C to 10 lmHg to remove the impurities for 30 min and cooled to the analysis temperature of 45 °C for 5 min for H2 chemisorption. The method of the double isotherm was used to determine the irreversibly bound chemisorbed H2, determined by extrapolating the linear part of the isotherm in the range of 95–450 mmHg to zero pressure which should correspond to H2 adsorbed on the metal surface, considering an irreversible adsorption on the metal phase and a reversible adsorption on the support. The reproducibility of the experiments was performed at least three times with a margin of error of less than 2%. The nickel surface area was calculated by assuming that one hydrogen atom occupies on one surface nickel atom and that the cross-sectional area of the nickel atom is 6.49  1020 m2. The amount of carbon deposition on the used samples was determined with a thermogravimetric analyzer (STA 4449 F3). The samples were first heat preserved at 50 °C for 0.5 h and then with a heating speed of 10 °C/min from 50 °C in an air flow of 30 mL/min to 800 °C.

controllers and a HPLC pump, respectively. The temperature of the catalyst was monitored using a thermocouple placed in the middle of the catalyst bed. Typically, 100 mg of catalyst (40– 60 mesh) diluted with 400 mg of quartz particles (40–60 mesh) was placed between two layers of quartz wool in the center of the reactor for each run. Prior to the reaction, the catalyst was reduced in situ in a mixed flow of 20 vol% H2/N2 (50 mL/min) at a heating rate of 10 °C/min to 600 °C, maintained at 600 °C for 2 h, and then cooled to 450 °C for 0.5 h. After this, the reductive gas was switched to a gaseous mixture of N2 (50 mL/min) and steam (the amount required in the reaction) for 5 min to avoid initial carbon deposition, followed by introducing LPG into the reaction system under atmospheric pressure and closing the N2 valve. After 1-h reaction, the reaction system became stable, and the effluent gas was cooled in a condenser at room temperature and passed a drierite bed to remove all water. Finally, the dried gas products were analyzed using an online GC-FID gas chromatograph with a CP–sil 5 CB column for hydrocarbons, followed by another GC-TCD gas chromatograph for CH4, CO and CO2, and H2. The reaction was terminated by closing the LPG and water valves and cooled quickly by keeping the 100 mL/min N2 flow and opening the furnace. The flow rate of outlet gas was measured by a soap flow meter. Other than H2, CO, CO2, and CH4, no other products were observed in the exit gas. The overall mass balance was more than 97% on the basis of carbon in the starting reactants. The conversion of LPG was defined as the percentage of the total molar flow rate of the carbon-containing products, that is, CH4, CO, and CO2 in the exit gas to the molar flow rate of carbon in the feed gas, on the basis of the carbon balance in the effluent gas under the assumption of no carbon deposition. The selectivity of the carbon-containing product was designated as the percentage of the molar flow rate of the carbon-containing product to the total molar flow rate of the carbon-containing products, and H2 selectivity was defined as the percentage of the molar flow rate of H2 in the effluent gas to the molar flow rates of all products containing hydrogen, multiplied with the proper factor, as described in the Ref. [29,41]. The turnover frequencies (TOFs) of the Ni–MgO/c-MA catalysts for the steam reforming of LPG are defined as the number of converted carbon atoms in LPG per surface metal Ni atom and hour. Considering that the reforming reaction of LPG for the TOF measurement should be conducted at a low LPG conversion (620%) to precisely determine the actual reaction rate on the metal active site, which was believed to be governed approximately by chemical kinetics, the reforming reaction had to be operated at a very high gas hourly space velocity (GHSV) without channeling in the catalyst bed. For this, a 25 mg catalyst (60–80 mesh) diluted with 200 mg of quartz particles (60–80 mesh) was used for the reaction, in which the GHSV was fixed at 3.3  105 mL/gcat h.

2.3. Catalyst test and analysis

3.1.1. XRD characterization XRD measurement was used to investigate the effect of calcination temperature on the crystalline structure of the prepared NiO– MgO/c-MA materials. Fig. 1 presents the XRD patterns of the NiO– MgO/c-MA samples calcined at different temperatures (300 °C, 400 °C, 500 °C, 600 °C, and 700 °C). All the NiO–MgO/c-MA–T materials showed three strong diffraction peaks around 2h = 37°, 45°, and 66°, corresponding to the (3 1 1), (4 0 0), and (4 4 0) planes for spinel, respectively. This result demonstrated that aluminum hydroxides and oxyhydroxides in the gel precursors for the NiO–

The steam pre-reforming of LPG was performed in the conventional fixed-bed reactor as described in our previous study [29]. LPG (purchased from Shanghai Auto Energy Co., Ltd.), consisting of C2H6 of 3.1 vol%, C3H8 of 84.0 vol%, and C4H10 of 12.9 vol%, was used as a model compound to study the catalytic behavior. A preheater filled with quartz balls with a set temperature of 300 °C was used to vaporize water and fully mix the reaction gases. The reaction gases and water were controlled using the mass flow

3. Results and discussion 3.1. Effect of calcination temperature on physicochemical properties of NiO–MgO/c-MA materials

154

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

Ni(Mg)O γ-Al2O3 or Ni(Mg)Al2O4

Intensity (a.u.)

T = 700 T = 600 T = 500 T = 400 T = 300 10

20

30

40

50

2θ/

60

70

80

90

ο

Fig. 1. XRD patterns of the NiO–MgO/c-MA–T materials.

MgO/c-MA materials were transformed into c-Al2O3, just as for pure c-MA [28,29], or into MgAl2O4 and/or NiAl2O4 spinels in the range of 300–700 °C. To the best of our knowledge, this is the first report for the synthesis of crystalline c-alumina-supported metal oxide materials derived from an aqueous solution of metal salts via a single-step method at such a low temperature of 300 °C, which is much lower than that (ca. 500 °C) for producing conventional c-alumina through the thermal transformation of aluminum hydroxides and oxyhydroxides [42]. In the case of the NiO–MgO/ c-MA–300 and NiO–MgO/c-MA–400 samples, no characteristic diffraction peaks associated with MgO and/or NiO phases were observed, which are difficult to be identified with the XRD patterns because of the high similarity of the XRD patterns [5,36,43]. This result implied that Ni and Mg oxide species were highly dispersed in the amorphous form or with very small particle sizes below the detection limit of XRD, or interacted with alumina support to form MgAl2O4 and/or NiAl2O4 spinels in the NiO–MgO/c-MA matrices [4,36]. With elevating the calcination temperature to 500 °C, two new diffraction peaks located at ca. 43.2° and 62.7° occurred, which could be attributed to MgO and/or NiO crystalline phases, and their diffraction intensities increased with the elevated temperature. These results indicated that free MgO and/or NiO phases or larger crystalline oxide crystallites were formed in the matrices [24,29]. As the XRD patterns of NiAl2O4 and MgAl2O4 spinels are similar to that of c-Al2O3, their diffraction peaks are indistinguishable from each other due to the broadening and overlapping [44]. However, the formation of NiAl2O4 and MgAl2O4 phases can be identified with the variation in the relative intensities and the slight shifts of the diffraction peaks for spinel around 2h = 37.0°, 45.5°, and 67.0° [44,45]. For example, for NiAl2O4 (PDF no. 10-0339) and MgAl2O4 (PDF no. 21-1152) spinels, the (3 1 1) reflection is the strongest and the relative intensity of the (4 0 0) and (3 1 1) peaks is 0.65, whereas for c-Al2O3 (PDF no. 10-0425), the (4 0 0) reflection is stronger and the intensity ratio is 1.25. On the other hand, since the ionic radii of Ni (0.069 nm) and Mg (0.072 nm) are larger than that (0.057 nm) of aluminum, the incorporation of Ni and/or Mg ions expands the lattice; in other words, the NiAl2O4 and MgAl2O4 spinels have larger lattice parameters than crystalline c-Al2O3, resulting in the shifts of the diffraction peaks to smaller angles [29,44]. As illustrated in Fig. 1, the diffraction peaks of the (4 0 0) and (4 4 0) planes for the spinel phase in the NiO–MgO/c-MA–T samples shifted to smaller 2h values, compared with 2h = 45.6° and 66.4° for c-MA shown in our previous study [29], and meanwhile, the relative intensities of the (4 0 0) and (3 1 1) peaks had a decrease. This result demonstrated that Mg2+

and/or Ni2+ ions might diffuse into the structure of c-MA to a certain extent, likely due to homogeneous mixing and strong interaction between alumina and metal ions during the calcination processes. From the XRD patterns in Fig. 1, with the increase in calcination temperature from 300 to 700 °C, the diffraction peaks positioned at 65.7° and 45.0° for NiO–MgO/c-MA–300 shifted to 2h angles at 65.2° and 44.6° for NiO–MgO/c-MA–700, and correspondingly, the relative intensity of the (4 0 0) and (3 1 1) peaks decreased from 1.23 to 0.79. These results implied that more Mg and/or Ni ions reacted with the c-Al2O3 support to form NiAl2O4 and/or MgAl2O4 spinels with the increased calcination temperature. It is noteworthy that the interaction of NiO with Al2O3 to form NiAl2O4 spinel on the surface of alumina support has been reported only at higher than 400 °C in the previous documents [46,47]. Thus, it could be anticipated that the variation in the diffraction peaks for NiO–MgO/c-MA–300, relative to pure c-Al2O3 [28], might be mainly due to the insertion of Mg ions in the lattice of c-Al2O3. Although the intensities of the diffraction peaks for the NiO– MgO/c-MA–T materials in Fig. 1, which revealed the degree of crystallinity of spinel phases, increased with elevating the calcination temperature, the FWHMs of the diffraction lines showed only smaller changes, for example, 4.29° of FWHM for the (4 4 0) plane of NiO–MgO/c-MA–300 and 2.46° for the (4 4 0) plane of NiO–Mg O/c-MA–700. Correspondingly, the mean c-alumina crystallite sizes from the Scherrer equation increased from 2.2 to 3.8 nm, which were approximate to those (2.0–3.1 nm) of pure c-MA calcined in the same temperature range [28]. This result suggested that all the NiO–MgO/c-MA–T materials might still have wormhole-like mesoporous network structures built up of small c-Al2O3 and/or c-Al2O3–MgAl2O4–NiAl2O4 nanoparticles, similar to pure c-MA shown in the previous study [28]. 3.1.2. SEM and TEM characterizations The influences of calcination temperature on the morphologies and mesoporous structures of the NiO–MgO/c-MA materials were characterized by SEM and TEM. It could be seen that the SEM image of the NiO–MgO/c-MA–300 sample in Fig. 2(a) consisted of various types of monoliths without mesoporous structure, which was similar to that of the as-prepared crude gel for c-MA dried at 100 °C [28]. The previous investigation showed that NH4NO3 was completely decomposed and c-alumina with the well-defined mesopores was formed from the dried gel for c-MA at 300 °C [28]. Therefore, it could be reasonably judged that Mg and/or Ni hydroxides and oxyhydroxides were not decomposed into the corresponding metal oxides to form c-aluminasupported Ni–Mg mixed oxides at 300 °C. The TEM image of the NiO–MgO/c-MA–300 sample in Fig. 2(c) also revealed that there were almost no mesopores in the c-alumina framework. All the other NiO–MgO/c-MA–T samples calcined in the temperature range of 400–700 °C showed similar frameworks with a homogenous wormhole-like mesoporous structure and a pore size of ca. 2.0–5.0 nm to pure c-MA [28], and the SEM and TEM images of the representative NiO–MgO/c-MA sample are illustrated in Fig. 2(b) and (d), respectively. The mesoporous c-alumina frameworks were fabricated through cross-linking nanoparticles together. It was believed that this kind of monolithic skeleton with relatively large pores via the non-surfactant sol–gel process possessed the better mass transfer properties, resulting in higher catalytic reaction rates and higher speed separations than mesoporous materials of distinct particles obtained using surfactants as templates owing to emulsion biphase chemistry [48–50]. However, there were no larger MgO and/or NiO crystallites found in the NiO–MgO/c-MA samples calcined at higher than 500 °C, as indicated by the XRD patterns in Fig. 1, which was due to a less striking contrast among MgO and/or NiO and c-alumina crystallites for TEM measurement.

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

155

Fig. 2. SEM images of (a) NiO–MgO/c-MA–300 and (b) NiO–MgO/c-MA–700, and TEM images of (c) NiO–MgO/c-MA–300 and (d) NiO–MgO/c-MA–700.

3.1.3. N2 adsorption–desorption analysis The textural properties of the NiO–MgO/c-MA–T materials were characterized by N2 adsorption–desorption measurements as displayed in Fig. 3. The N2 adsorption isotherm of the NiO–MgO/cMA–300 sample was close to the type I characteristic with a very low N2 adsorption volume according to the IUPAC classification, and no apparent hysteresis was observed. This result suggested that there existed only micropores or negligible mesopores in the sample with a low specific surface area, as shown in the SEM and TEM images of Fig. 2(a) and (c). All the other NiO–MgO/c-MA–T (T = 400, 500, 600, and 700) materials exhibited characteristic type IV isotherms with apparent hysteresis loops within a P/P0 range of 0.40–0.65, indicative of the presence of mesopores with relatively uniform pore sizes, which was in agreement with the SEM and TEM results in Fig. 2(b) and (d). BJH pore size distribution analysis showed that the NiO–MgO/c-MA–300 sample did not contain mesopores and the NiO–MgO/c-MA–T (T = 400, 500, 600, and 700) samples contained mesopores of narrow PSDs in the range of 2.5–4.5 nm with a FWHM of less than 1.0 nm. The basic physical and textural properties of the NiO– MgO/c-MA materials calcined at different temperatures are listed in Table 1. The NiO–MgO/c-MA–300 sample showed a small specific surface area of 21 m2/g and a very low pore volume of 0.01 cm3/g, due to the blockage of Mg and/or Ni hydroxides and oxyhydroxides thermally undecomposed in the c-MA framework. When the calcination temperature was elevated to 400 °C, the Ni O–MgO/c-MA–400 °C sample produced a large specific surface area (SBET) of 209 m2/g, a pore volume (Vp) of 0.17 cm3/g, and a BJH pore size (Dp) of 3.7 nm; however, the values of these parameters were much lower than those (SBET, 401 m2/g; Vp, 0.33 cm3/g; Dp, 3.9 nm) of pure c-MA [29], due to the variation in mass density of the materials, the coating of c-alumina surface with Mg and Ni

metal oxides, and the interaction between Al2O3 and Mg and Ni metal oxides [41,51]. With the further increase in the calcination temperature, the specific area, the pore volume, and the BJH pore size gradually decreased to 144 m2/g, 0.12 cm3/g, and 3.4 nm for NiO–MgO/c-MA–700, respectively. On the other hand, the average pore size (Da) of the NiO–MgO/c-MA samples increased progressively from 3.2 nm at 400 °C to 3.7 nm at 700 °C. These results were attributed to the structural shrinkage as well as to the blockage of micropores during the calcination processes. 3.1.4. CO2-TPD analysis The previous investigations have been shown that the calcination temperature may have an important influence on the surface basicity of alumina and alumina-supported metal oxide catalysts, resulting in the variation in the catalytic properties including the activity, the stability, and the resistance to coke deposition for the reforming of hydrocarbons [32,52]. Therefore, in accordance with the earlier studies [29,52], the influence of the calcination temperature on the strength and number of the basic sites on the surfaces of the NiO–MgO/c-MA materials was characterized by the TPD technique using CO2 as a probe molecule, and the CO2-TPD profiles are illustrated in Fig. 4. In the case of the NiO– MgO/c-MA–300 sample, only a very strong CO2 desorption with a high degree of symmetry centered around 345 °C was seen, which was indicative of a homogeneous distribution of medium strength basic sites. In combination with the CO2-TPD profile of pure c-MA calcined at 400 °C shown in our earlier study [29], this CO2 desorption peak could be assigned to CO2 chemisorbed uniformly on Mg and/or Ni hydroxides and oxyhydroxides thermally undecomposed in the c-MA framework. The CO2 desorption associated with c-alumina was not observed, implying that the surface basic sites of c-alumina crystallites were fully covered with Mg

156

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

a

Table 1 Textural properties of the prepared NiO–MgO/c-MA–T materials.

T = 700

Volume adsorbed a.u.

T = 600 T = 500

Sample

SBET (m2/g)

Vp (cm3/g)

Dp (nm)

Da (nm)

NiO–MgO/c-MA–300 NiO–MgO/c-MA–400 NiO–MgO/c-MA–500 NiO–MgO/c-MA–600 NiO–MgO/c-MA–700

21 209 188 165 144

0.01 0.17 0.15 0.14 0.12

– 3.7 3.6 3.5 3.4

2.3 3.2 3.3 3.5 3.7

T = 400

0.0

0.2

0.4

0.6

0.8

1.0

Relative pressure P/P0

Pore volume distribution a.u.

b

T = 700 T = 600 T = 500

T = 400

T = 300 2

3

4

5

6

7

8

9

10

Pore size (nm) Fig. 3. (a) Nitrogen adsorption–desorption isotherms and (b) BJH pore size distributions of the NiO–MgO/c-MA–T materials.

and/or Ni hydroxide species. For the other NiO–MgO/c-MA–T (T = 400, 500, 600, and 700) samples, the CO2-TPD profiles showed two distinct CO2 desorption zones in the range of ca. 50–120 °C and 200–500 °C, respectively. This result was an indication that there were different types of basic sites with separate site strengths distributed on the surfaces of the samples. The low-temperature CO2 desorption peaks centered at ca. 94 °C, the intensities of which gradually decreased with the increase in the calcination temperature, could be attributed to both the weakly chemisorbed CO2 of the framework and the physically adsorbed CO2 due to the large pore volumes and specific surface areas of the materials [18,29]. The high-temperature CO2 desorption peaks positioned at 267 °C for the stronger basic sites on the surfaces, which were always predominant for each sample, were associated with the c-MA support and the Ni and/or Mg oxides in the framework [29]. Both of them showed no obvious shifts, indicating that the calcination temperature did not have a significant effect on the basic site strength of the NiO–MgO/c-MA materials in the range of 400–700 °C. Table 2 lists the amounts of CO2 desorbed over the NiO–MgO/ c-MA–T materials in the range of 50–600 °C relative to that (100 as the reference) of the NiO–MgO/c-MA–400 sample, which were obtained by the integration of the CO2 desorption curves. It could

3.1.5. XPS analysis It has been proven that the preparation conditions strongly influence the characters of the nickel oxide species on the surface of the alumina support, resulting in the variation in the catalytic properties. Since XPS is a surface sensitive and quantifiable measurement technique, the differences in the binding energies (BEs), the band shapes, and the intensities of the XPS peaks were employed to analyze the oxidation state of Ni, the interaction of NiO species with MgO and c-MA support, and the chemical compositions of the surfaces of the prepared NiO–MgO/c-MA materials calcined at different temperatures. The XPS spectra of the Ni 2p regions and the corresponding BEs of the Ni 2p3/2 levels are shown in Fig. 5 and in Table 2, respectively. The NiO–MgO/c-MA–300 sample exhibited a primary Ni 2p3/2 BE of 856.3 eV with a FWHM of 2.6 eV, accompanied by a satellite peak at 862.0 eV, which corresponded to those for Ni(OH)2 [53–55]. In the case of the NiO–MgO/c-MA–400 sample, the Ni 2p3/2 peak maximum shifted to lower value at 855.2 eV and showed a wider FWHM of 3.1 eV as a result of the thermal decomposition of nickel hydroxides and/or oxyhydroxides to nickel oxide species in c-alumina framework. However, as the calcination temperature was elevated in the range of 400–700 °C, the Ni 2p3/2 peak started to shift to higher BE values and the band shapes of the peaks became more broadened. When the temperature was raised up to 700 °C, the

345

CO2 amount desorbed (a.u.)

T = 300

be seen that except the NiO–MgO/c-MA–300, the amount of the desorbed CO2 declined with increasing the calcination temperature. This result demonstrated that the number of the basic sites on the surfaces of the NiO–MgO/c-MA materials decreased with the elevated calcination temperature. However, the relative densities of basic sites on the catalyst surfaces in Table 2 showed no observable change. Therefore, it could be speculated that the variation in the number of the basic sites on the surfaces of the NiO– MgO/c-MA–T materials was mainly due to the decrease in the specific surface area with the calcinations temperature, as shown in Table 1.

267 94

T = 700 T = 600 T = 500 T = 400 T = 300

100

200

300

400

500

600

o

Tempereture C Fig. 4. CO2-TPD profiles for the NiO–MgO/c-MA–T samples.

157

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166 Table 2 Relative amounts of CO2 desorbed, binding energies of Ni 2p3/2, and surface atomic compositions of Ni, Mg, and Al elements for NiO–MgO/c-MA–T samples. Sample

Amount of CO2 desorbed relative to NiO–MgO/c-MA–400a

Relative density of base sitesb (/m2)

BE of Ni 2p3/2c (eV)

NiO–MgO/c-MA–300 NiO–MgO/c-MA–400 NiO–MgO/c-MA–500 NiO–MgO/c-MA–600 NiO–MgO/c-MA–700

288.1 100 86.3 74.5 66.1

3.70 0.48 0.46 0.45 0.46

856.3 855.2 855.9 856.3 856.8

(2.6) (3.1) (3.1) (3.2) (3.4)

Surface composition by XPS (atomic ratio) Ni/Al

Mg/Al

0.11 0.10 0.14 0.15 0.16

1.15 0.98 0.48 0.26 0.21

a Defined as the total amount of CO2 desorbed per gram of each catalyst divided by the total amount of CO2 desorbed (100 as the reference) per gram of NiO–MgO/c-MA– 400 in the range 50–600 °C. b The relative amount of CO2 desorbed per gram catalyst to NiO–MgO/c-MA–400 divided by the specific surface area of the catalyst. c FWHM values are given in parentheses.

T = 700

Intensity (a.u.)

T = 600

T = 500

T = 400

T = 300

850

855

860

865

870

875

880

Binding energy (eV) Fig. 5. Ni 2p3/2 XPS spectra of the NiO–MgO/c-MA–T materials.

Ni 2p3/2 BE increased to 856.8 eV and the corresponding FWHM value was 3.4 eV. It was reported that the Ni 2p3/2 BE was 854.6 eV for pure NiO and 857.3 eV for pure NiAl2O4 [53,54]. This implied that in the NiO–MgO/c-MA–T (T = 400, 500, 600, and 700) materials, there existed dispersed NiO species with an interaction with c-MA support or a small amount of NiAl2O4 formed in the NiO–MgO/c-MA materials as exhibited in the XRD patterns of Fig. 1. The calcination temperature enhanced the interaction between the NiO species and the support and contributed to the formation of NiAl2O4 spinel. Table 2 gives the surface atomic compositions of Ni and Mg relative to Al atoms for the NiO–MgO/c-MA–T materials calcined in the range of 300–700 °C. In the case of the NiO–MgO/c-MA–300 sample, the Ni/Al and Mg/Al molar ratios on the surface were 0.11 and 1.15, respectively; namely, the total [Ni + Mg]/Al molar ratio was 1.26, which was much higher than 0.49 estimated from the nominal composition used for the gel precursor. This result meant that the Mg and Ni species were mainly enriched on the surface of c-MA support. However, the Ni/Al molar ratio was significantly lower than 0.24 of the theoretical value for the gel precursor. This result demonstrated that the Ni hydroxides preferred to be deposited on the surface sites of c-alumina support, while Mg hydroxides were covered on the Ni species, in agreement with the observation in our previous study [29]. When the calcination temperature was increased to 400 °C, the Mg and Ni hydroxides were decomposed to the corresponding isolated or microcrystalline metal oxides on the alumina surface, and the small amounts of Mg and Ni ions might diffuse into the lattices

of c-alumina as illustrated in the XRD patterns of Fig. 1. As a consequence, both Ni/Al and Mg/Al molar ratios on the support surface decreased and the basic sites of c-alumina support were exposed, in agreement with the results of the CO2-TPD profiles in Fig. 4. With further elevating the calcination temperature, the Ni/Al molar ratio on the surface started to increase, while the Mg/Al molar ratio rapidly decreased. For example, on the surface of the NiO–MgO/c-MA–500, the Ni/Al molar ratio increased from 0.10 for the NiO–MgO/c-MA–400 to 0.14, and the Mg/Al molar ratio decreased from 0.98 to 0.48, while for the NiO–MgO/c-MA–700, the former slightly increased to 0.16, but the latter further reduced to 0.21. Since the physical properties of the c-alumina support and especially the pore size distributions of the materials were confirmed to have no obvious change, the Ni/Al and Mg/Al molar ratios estimated from the intensities of XPS peaks could basically reflect the variations in the element concentrations on the support surface [22]. On one hand, MgO reacted with alumina to form spinel more easily, resulting in relatively more MgO for the tridimensional growth of MgAl2O4 spinel at higher temperatures. On the other hand, the microcrystalline MgO on Ni species was agglomerated into larger crystallites, leading to more Ni exposed on the surface with increasing calcination temperature. 3.2. Effect of calcination temperature on physicochemical properties of Ni–MgO/c-MA catalysts 3.2.1. H2-TPR measurements of NiO–MgO/c-MA–T materials It is known that before starting the reforming reaction of hydrocarbons, Ni-based catalysts need to be reduced to produce metallic Ni active sites. Thus, it is important to further investigate the influences of the calcination temperature on the reducibility of the NiO–MgO/c-MA materials, the Ni crystallite sizes formed, and the textural and structural properties of the Ni–MgO/c-MA catalysts after the reduction, which have close relationships with the activity, the stability, and the resistance to carbon deposition of the catalysts for the reforming of hydrocarbons. H2-TPR measurements were used to study the reducibility of the NiO–MgO/c-M A–T materials and the interaction between nickel species and support, and the H2-TPR profiles are illustrated in Fig. 6. It could be seen that the calcination temperature had a significant influence on types of Ni species, the reducibility of Ni2+ ions, and the interaction of Ni species with the support. In terms of the NiO–MgO/cMA–300 sample, the H2-TPR profile showed a very strong reduction peak centered at ca. 382 °C and a wide H2 consumption curve in the range of 500–800 °C, which were ascribed to the reduction of Ni hydroxides undecomposed and certain amounts of highly dispersed NiO species with a strong interaction with the support, respectively. When the calcination temperature was increased to 400 °C, the reduction peak corresponding to the reduction of Ni hydroxide disappeared, and concomitantly, the H2 consumption

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

H2 consumption (a.u.)

T = 700

T = 600 T = 500 T = 400 T = 300

200

300

400

500

600

700

800

900 1000

o

Temperature C Fig. 6. H2-TPR profiles for the NiO–MgO/c-MA–T materials.

curve in the range of 500–800 °C became dominant and a new weak reduction peak occurred corresponding to the reduction of bulk nickel aluminate-like species in a higher temperature range of 810–910 °C [12,41]. This result demonstrated that Ni hydroxide or oxyhydroxide was decomposed into isolated or microcrystalline NiO dispersed on the support surface and the small amounts of Ni2+ ions reacted with alumina to form NiAl2O4 spinel. With further elevating the calcination temperature, the reduction peak temperatures for both dispersed NiO species and aluminate-like species in TPR profiles increased, implying that the interaction between Ni species and support strengthened with the calcination temperature. On the other hand, the incorporation of divalent Mg ions in the tetrahedral or octahedral sites in c-Al2O3 structure resulted in a decrease in the cationic deficiency of c-Al2O3, thus stabilizing c-Al2O3–NiAl2O4 nanoparticles and elevating the reduction temperature of Ni2+ ions in spinel phase [56]. As a whole, the increase in the reduction peak intensity for aluminate-like species with the calcination temperature suggested that more Ni ions diffused into the lattices of c-alumina to form spinel. These results were in line with the XRD patterns in Fig. 1 and the XPS spectra in Fig. 5. From the H2-TPR profiles in Fig. 6, no reduction peaks were observed due to free and bulk-like NiO species for all NiO–MgO/c-MA–T materials. This result further demonstrated that in the XRD patterns of Fig. 1, the diffraction peaks of NiO and/or MgO phases for the NiO–MgO/c-MA materials calcined at more than 500 °C were due to crystalline MgO or Ni–Mg–O solid solution instead of bulk-like NiO. The previous investigations have found that not all Ni2+ ions were reducible in the c-alumina-supported nickel catalysts [22,29]. There were two forms of Ni2+ ions reported in the lattices of c-alumina, that is, the Ni2+ ions in the tetrahedral sites and the Ni2+ ions in the octahedral sites of c-alumina. The former in the tetrahedral sites corresponds to ‘‘hard to reduce’’ nickel, and the latter in the octahedral sites corresponds to ‘‘readily reduced’’ nickel [22,57,58]. Increasing the calcination temperature contributes to the formation of Ni2+ ions of the tetrahedral sites, thus

affecting the extent of reduction of Ni species in the c-alumina-supported Ni catalysts [22]. The reduction degree of Ni2+ ions, which is defined as the percentage of the total molar amount of H2 consumed for a sample to the molar amount of H2 calibrated by CuO having the same molar amount of NiO in the sample between 200 and 1000 °C, was used to show the reducibility of Ni2+ ions in the NiO–MgO/c-MA–T materials, and the results are summarized in Table 3. The NiO–MgO/c-MA–300 sample showed a reduction degree of Ni2+ ions at 99.8%, indicating that all Ni2+ ions in the sample could be reduced into metal Ni atoms. The reducibility of Ni2+ in the NiO–MgO/c-MA materials lowered with raising the calcination temperature. When the calcination temperature was raised to 700 °C, the reduction degree of Ni2+ ions declined to 82.2%. This result agreed well with the formation of NiAl2O4 spinel with the calcination temperature. Therefore, it was proposed that some of the Ni2+ ions could react with the tetrahedral Al3+ sites or diffused from the octahedral sites to tetrahedral sites of c-alumina to form the inverse spinel structure at the moment when nickel ions interacted with the octahedral Al3+ sites to produce the normal spinel structure [22,59]. In addition, the dissolution of a fraction of NiO into MgO to form Ni–Mg–O solid solution at higher calcination temperature, which is more difficult to be reduced than NiAl2O4 spinel [5,36], might also be responsible for the decrease in the reduction degree of Ni2+ ions.

3.2.2. XRD analysis of Ni–MgO/c-MA–T catalysts The NiO–MgO/c-MA materials reduced in a 20 vol% H2/N2 flow at 600 °C for 2 h were reported to show the optimum catalytic performance for the pre-reforming of LPG [29]. Thus, the Ni– MgO/c-MA–T samples obtained under this condition were characterized by XRD, and the patterns are illustrated in Fig. 7. It could be seen that for all the reduced catalysts, there were two new diffraction peaks at 51.9° and 76.4° corresponding to (2 0 0) and (2 2 0) reflections, respectively, and a diffraction peak at ca. 44.4° due to

Ni Ni(Mg)O γ-Al2O3 and Ni(Mg)Al2O4

Intensity (a.u.)

158

T = 700 T = 600 T = 500 T = 400 T = 300

10

20

30

40

50

2θ /

60

70

80

90

ο

Fig. 7. XRD patterns of the Ni–MgO/c-MA–T samples.

Table 3 Degree of reduction of Ni2+ ions and physicochemical properties of the prepared catalysts after H2 reduction. Sample

Ni–MgO/c-MA–300 Ni–MgO/c-MA–400 Ni–MgO/c-MA–500 Ni–MgO/c-MA–600 Ni–MgO/c-MA–700

Degree of reduction (%)

99.8 93.6 86.7 84.2 82.2

Ni particle size (nm) by XRD

TEM

5.5 5.6 7.8 10.1 11.2

5.5 ± 0.7 5.8 ± 0.8 7.4 ± 0.9 10.2 ± 1.2 13.1 ± 1.4

SBET (m2/g)

Vp (cm3/g)

Dp (nm)

Da (nm)

231 167 163 158 129

0.18 0.15 0.15 0.15 0.14

– 3.5 3.6 3.7 3.7

2.1 3.7 3.9 3.9 3.9

159

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

the Ni (1 1 1) plane, which was overlapped with the diffraction peak at ca. 45° corresponding to the (4 0 0) plane for spinel. This result indicated that the NiO species in the samples were reduced to metal Ni. The intensity of the Ni diffraction peaks increased, and the peak width decreased with the increase in the calcination temperature. This was an indication of the increase in the Ni crystallite size with the calcination temperature. By carefully comparing the XRD patterns of the NiO–MgO/c-MA materials before and after the reduction in Figs. 1 and 7, respectively, it could be found that the 2h values of the (4 4 0) diffraction peaks for the Ni– MgO/c-MA–T (T = 300, 400, 500, and 600) catalysts slightly shifted to lower values and became almost the same at 2h = 65.3°, while the intensity of the diffraction peaks corresponding to MgO showed an observable decrease after the reduction. This implied that some amounts of MgO reacted with c-alumina to form MgAl2O4 spinel, and the reconstruction of the MgO species on the surface might occur in the reduction processes. The intensities and shapes of the diffraction peaks corresponding to the spinel phases had no significant changes except the overlapped (4 0 0) diffraction peaks with the Ni (1 1 1) reflection. This result implicated that the mesoporous c-Al2O3 frameworks might still be maintained through the cross-linking of small c-Al2O3 and/or c-Al2O3–MgAl2O4–NiAl2O4 nanoparticles. Table 3 lists the apparent Ni crystallite sizes estimated from the broadening of the Ni (2 0 0) reflection using the Scherrer formula. The Ni–MgO/c-MA–300 and Ni–MgO/c-MA–400 catalysts had approximate metal Ni crystallite sizes, which were 5.5 nm and 5.6 nm, respectively. The Ni crystallite sizes gradually increased with elevating the calcination temperature. When the calcination temperature increased to 700 °C, the Ni crystallite sizes increased to 11.2 nm. It was worth noting that the Ni 2p XPS spectra in Fig. 5 and the H2-TPR profiles in Fig. 6 for the NiO–MgO/c-MA–T (T = 400, 500, 600, and 700) materials showed that the interaction between the Ni species and the support strengthened with the calcination temperature. Generally, the strong interaction between metal oxide species and support favors the attachment of reduced metal to form stable metal nanoparticles with relatively high dispersion in the reduction process. Such an abnormal phenomenon could be attributed to a combination of several factors, such as the textural properties of the catalysts, the reduction rate of NiO species, the interaction between Ni crystallites and support, and the reduction temperature. In the case of the NiO–MgO/c-MA–30 0 and NiO–MgO/c-MA–400 samples, Ni2+ ions could be rapidly reduced into metal Ni atoms or small Ni clusters supersaturated on the support surface, forming uniform metal Ni crystallites with medium sizes confined by pore structures during a short time due to the weak interactions with the support. The free energy released by the growth was insufficient for driving the medium-size Ni crystallites to migrate on the surface and form larger particles in the cross-linking framework at the present reduction temperature. For the NiO–MgO/c-MA samples calcined at higher temperatures (P500 °C), the reduction of Ni2+ ions might be the rate-determining step and Ni atoms were released very slowly for the growth of Ni crystallites because of the stronger interaction with the support. In this case, Ostwald ripening, namely monoatomic species or small clusters diffuse from small to larger particles [23], might be the dominant mechanism for Ni crystallite growth rather than migration and coalescence of Ni nanoparticles. In a word, the Ni atoms or small Ni clusters with high-surface energies could overcome the attachment on the support to move toward the small Ni crystallites initially formed in the framework and produce larger crystallites, breaking the pore wall. 3.2.3. TEM analysis of Ni–MgO/c-MA–T catalysts TEM measurements were used to determine the dispersion and size of the Ni nanoparticles and the porous structures in the NiO–

5.5 ± 0.6

3

4

5

6

7

Particle size (nm) 5.8 ± 0.8

3

4

5

6

7

Particle size (nm) 7.4 ± 0.9

5

6

7

8

9

Particle size (nm) 10.2 ± 1.2

8

9

10 11 12

Particle size (nm) 13.1 ± 1.6

11 12 13 14 15

Particle size (nm) Fig. 8. TEM images and corresponding Ni particle size distributions of the Ni–MgO/ c-MA–T catalysts. (a) T = 300, (b) T = 400, (c) T = 500, (d) T = 600, and (e) T = 700.

MgO/c-MA–T materials after the H2 reduction at 600 °C. The TEM images of the Ni–MgO/c-MA–T catalysts are illustrated in Fig. 8. In terms of the Ni–MgO/c-MA–300 catalyst, there were a large amount of pores dispersed in the framework due to the

160

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

temperature was elevated to 700 °C, the Ni crystallite sizes increased to 13.1 ± 1.4 nm. These results were in good agreement with the XRD results in Fig. 7 and Table 3.

T = 700

a

Volume adsorbed a.u.

T = 600 T = 500 T = 400

T = 300

0.0

0.2

0.4

0.6

0.8

1.0

Relative pressure P/P0

Pore volume distribution (a.u.)

b

T = 700 T = 600 T = 500 T = 400 T = 300 2

3

4

5

6

7

8

9

10

Pore size (nm) Fig. 9. (a) Nitrogen adsorption–desorption isotherms and (b) BJH pore size distributions of the Ni–MgO/c-MA–T catalysts.

decomposition and reduction of Ni and Mg hydroxides and oxyhydroxides in the H2 atmosphere at 600 °C. All the other Ni– MgO/c-MA–T (T = 400, 500, 600, and 700) catalysts showed homogeneous wormhole-like mesoporous structures with pore sizes of ca. 2.0–5.0 nm, which were very similar to those before the H2 reduction. In all the Ni–MgO/c-MA–T catalysts, the characteristic metal Ni nanoparticles were formed and homogeneously dispersed in the mesoporous frameworks. Table 3 showed the mean sizes and standard deviations of Ni particles, which were calculated by measuring the sizes of individual particles in TEM images. It could be seen that in the Ni–MgO/c-MA–300 and Ni–MgO/c-MA–400 catalysts, the Ni particle sizes were 5.5 ± 0.7 nm and 5.8 ± 0.8 nm, respectively. With further elevating the calcination temperature, the Ni crystallite sizes gradually increased. When the calcination

3.2.4. N2 adsorption–desorption analysis of Ni–MgO/c-MA–T catalysts The N2 adsorption–desorption isotherms and BJH pore size distributions were used to analyze the changes in the textural properties of the NiO–MgO/c-MA–T catalysts after the H2 reduction at 600 °C, and the curves are displayed in Fig. 9. The Ni–MgO/c-M A–300 catalyst presented the type I characteristic of N2 adsorption isotherm without apparent hysteresis loop, which was similar to the precursor before the reduction. However, the reduced sample had a N2 adsorption volume much larger than the precursor. This result suggested that in the reduction process, the porous structure in the c-alumina framework was formed, resulting in a high specific surface area, as listed in Table 3. BJH curve in Fig. 9 demonstrated that the small pores less than 3.5 nm were dominant in the Ni–MgO/c-MA–300 catalyst, in agreement with the TEM image in Fig. 8. For the other Ni–MgO/c-MA–T (T = 400, 500, 600, and 700) catalysts, the similar shape N2 sorption isotherms and narrow pore size distribution curves to those for the corresponding precursors before the reduction were observed in Fig. 9. These results revealed that the mesoporous c-alumina frameworks after the reduction at 600 °C were still maintained, as shown in the TEM images of Fig. 8. Interestingly, the NiO–MgO/c-MA–300 materials calcined in air at 600 °C showed almost the same textural properties and the reduction properties as the NiO–MgO/c-MA–600 materials. This meant that the formation of mesoporous structures in the NiO–MgO/c-MA materials was strongly affected by the thermal treating conditions. Compared with the counterparts before the reduction listed in Table 1, the Ni–MgO/c-MA–T (T = 400, 500, 600, and 700) in Table 3 showed slightly lower specific surface areas and larger average pore sizes, but the pore volume and BJH pore size of the Ni–MgO/c-MA–400 decreased and those for the NiO–MgO/c-MA–600 and NiO–MgO/c-MA–700 had slight increases. Such phenomena could be attributed to a combination of the reduction of the dispersed NiO species into metal Ni nanoparticles on the surfaces, structural shrinkage as well as to the blockage of micropores at higher temperature. 3.2.5. H2 chemisorption analysis of Ni–MgO/c-MA–T catalysts Hydrogen chemisorptions were carried out to analyze the influence of the calcination temperature on the active sites of metal Ni available on the surface of the Ni–MgO/c-MA catalysts, and the results estimated by the double isotherm method are listed in Table 4. It could be seen that compared with the Ni–MgO/c-MA– 400, although it took on the higher reduction degree of Ni2+ ions and very approximate sizes of metal Ni particles, the Ni–MgO/cMA–300 catalyst exhibited smaller amounts of hydrogen uptake. Correspondingly, the metal Ni surface area over the Ni–MgO/c-M A–300 catalyst was 16.6 m2/gcat, lower than that over the Ni–Mg O/c-MA–400. This result indicated that the availability of Ni atoms in the catalysts had a significant correlation with the catalyst pore structures. The larger mesopores with narrow size distribution more favored the access of hydrogen to the surface of metal Ni

Table 4 Hydrogen chemisorption results, LPG conversions, and corresponding TOFs over the MgO/c-MA–400 and Ni–MgO/c-MA–T catalysts for the steam reforming of LPG. Sample

H2 uptake (lmol/gcat)

Ni surface area (m2/gcat)

LPG conversion (%)

TOF (h1)

MgO/c-MA–400 Ni–MgO/c-MA–300 Ni–MgO/c-MA–400 Ni–MgO/c-MA–500 Ni–MgO/c-MA–600 Ni–MgO/c-MA–700

0 210 250 220 150 110

0 16.6 21.4 17.1 11.5 8.8

<0.1 15.2 20.3 18.1 13.5 10.3

– 240 272 274 269 268

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

particles to chemisorb on the Ni atoms. With increasing the calcination temperature in the range of 400–700 °C, the metal Ni surface area gradually declined from 21.4 to 8.8 m2/gcat. These results were consistent with the variations in the particle size of metal Ni with the calcination temperature shown in the XRD results of Fig. 7 and the TEM images of Fig. 8.

3.3. Steam reforming of LPG over Ni–MgO/c-MA–T catalysts The influence of the calcination temperature on the catalytic behavior of the Ni–MgO/c-MA–T catalysts for the steam reforming of hydrocarbons was investigated using LPG as model compounds at 450 °C and GHSV of 35,000 mL/gcat h with a steam to carbon molar ratio (S/C) of 2.0. LPG conversions as a function of reaction time are presented in Fig. 10. It could be seen that the calcination temperatures had significant influences on the activity and stability of the Ni–MgO/c-MA–T catalysts. The Ni–MgO/c-MA–300 catalyst showed the poorest initial LPG conversion of 50.2%, and the Ni–MgO/c-MA–400 catalyst had the highest initial LPG conversion of 97.9%. The initial conversions of LPG decreased with the calcination temperature in the range of 400–700 °C. When the calcination temperature increased to 700 °C, the initial conversion of LPG declined to 67.3%. This result was consistent with the change in the metal Ni surface area by hydrogen chemisorption shown in Table 4; namely, the larger the metal surface areas were, the higher the initial LPG conversions. However, as shown in Table 4, the Ni– MgO/c-MA–300 catalyst took on the metal Ni surface area close to the Ni–MgO/c-MA–500 catalyst, but much larger than the Ni–Mg O/c-MA–600 and Ni–MgO/c-MA–700 catalysts. It was known from the results of the TEM images in Fig. 8 and the N2 sorption isotherms in Fig. 9, the Ni–MgO/c-MA–300 catalyst presented smaller average pore size and poorer pore size distribution, while the other Ni–MgO/c-MA–T (T = 400, 500, 600, and 700) catalysts had similar mesoporous structures with larger pore sizes and narrow pore size distributions. Thus, it was concluded that the reaction performance of the Ni–MgO/c-MA catalysts for the steam reforming of hydrocarbons had close relationships not only with the number of active sites but also with the porous structures. The large mesopores with uniform sizes could improve the accessibility of the reactants to metal active sites, leading to the improved catalytic activity. In order to further understand the correlationship among the Ni active sites, the pore structures and the activities of the Ni– MgO/c-MA–T catalysts for the steam reforming of LPG, turnover frequencies (TOFs), which reflect the intrinsic activity of the active sites in the catalyst, were measured at a low LPG conversion (620%) and a high GHSV without channeling, where the possibility

LPG conversion (%)

100 80

T = 300 T = 400 T = 500 T = 600 T = 700

60 40 20 0 0

5

10

15

20

25

30

35

40

Reaction time (h) Fig. 10. Activity of the Ni–MgO/c-MA–T catalysts for the steam reforming of LPG. Reaction conditions: S/C, 2; SV, 35,000 mL/gcat h; reaction temperature, 450 °C.

161

of the external and internal diffusion limitations was almost completely eliminated by the experiments of varying the catalyst particle sizes and GHSV. The steam reforming of LPG was established to be governed by chemical kinetics. Taking into account the catalyst stability, the TOFs were estimated by the initial LPG conversions at 1 h. Table 4 gives the conversions of LPG and the corresponding TOFs over the Ni–MgO/c-MA–T catalysts for the steam reforming of LPG at 450 °C and SV of 3.3  105 mL/gcat h with the S/C of 2.0 at 1 h. It could be seen that the LPG conversions generally showed the same trend as the metal Ni surface areas over the catalysts, that is, the LPG conversions were proportional to the number of metal active sites on the catalyst surface. However, the TOF of Ni active sites over the Ni–MgO/c-MA–300 catalyst was found to be 240 h1, clearly smaller than those over the other Ni–MgO/c-MA–T (T = 400, 500, 600, and 700) catalysts, which were almost the same at ca. 270 h1. This result demonstrated that some parts of the Ni active sites by hydrogen chemisorption might be inaccessible to larger hydrocarbon molecules for the steam reforming of LPG due to the poorer porous structure and smaller pore sizes of the Ni–MgO/c-MA–300 catalyst. The calcination temperatures influenced both the activity and the stability of the Ni–MgO/c-MA catalysts as shown in Fig. 10. The Ni–MgO/c-MA–300 catalyst exhibited poor stability for the steam reforming of LPG. The LPG conversion gradually declined from 50.2% to 35.1% in the initial reaction time of 20 h, followed by the almost constant conversion of LPG at ca. 33% during the tested reaction time of 40 h. In contrast, the Ni–MgO/c-MA–400 catalyst showed not only the highest catalytic activity but also the most excellent stability for the steam reforming of LPG. The LPG conversion kept unchanged at ca. 97% in the entire reaction process. The further increase in the calcination temperature caused the decrease in the catalyst stability. For example, the LPG conversion over the Ni–MgO/c-MA–500 catalyst declined slowly from the initial 95.1% to 87.9%, while that of the Ni–MgO/c-MA–700 lowered steeply from 67.3% to 22.3%. It has been indicated in the CO2-TPD profiles of Fig. 4 and Table 2 that the NiO–MgO/c-MA–T (T = 400, 500, 600, and 700) materials have almost the same strength and densities of base sites on the catalyst surfaces. Therefore, it could be speculated that in the case of the NiO–Mg O/c-MA–T catalyst with similar mesoporous structures, the decrease in the stability might have close relationship with the increase in Ni particle sizes with the calcination temperature as displayed in the TEM results of Fig. 8 and Table 3. Generally for Ni catalysts, the surface carbons are accepted as key reaction intermediates in the reforming of hydrocarbons [9,59]. Besides reacting with surface oxygen atoms to form CO, the surface carbon atoms can also combine with other surface carbon atoms, producing coke deposition. However, it was reported that the carbon formation is a structure-sensitive reaction, and the rate of coke deposition strongly depended on the sizes of Ni particles [60,61]. The formation of the carbidic carbon (*C), which is a precursor of the inert carbon resulting in the deactivation of Ni catalysts, required an ensemble of higher synergetic Ni active sites than the gasification of CHx (1 6 x 6 3) species [61,62]. Thus, the larger metal Ni particles are favorable to the coke deposition, whereas the smaller Ni particles prevent CHx species from occupying the proper active sites to form *C, which are even unable to proceed at all when the crystallites are below a critical size [8,60], leading to the reduction of carbon deposited on the catalyst surface. In terms of the Ni– MgO/c-MA–300 catalyst with smaller sizes of Ni particles, its deactivation behavior of the Ni–MgO/c-MA–300 catalyst was likely due to the poorer porous structure. In the initial reaction period, the coke deposition formed on the surface gradually clogged the micropores and prevented the reactant gases from approaching Ni active sites. As the micropores were clogged, the catalyst activity tended to become stable in the subsequent period, similar to

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

Product selectivity (%)

80 60 60 40 40

CH4 CO2

20

CO H2

LPG conversion (%)

100

80

20

LPG

0

0 300

400

500

600

700

o

Calcination temperature C Fig. 11. LPG conversions and product selectivities (solid line) of the Ni–MgO/c-MA– T catalysts and the estimated equilibrium values of product selectivities (dot line) for the steam reforming of LPG. Reaction conditions: S/C, 2; SV, 35,000 mL/gcat h; reaction temperature, 450 °C; reaction time, 40 h.

that over the Ni–MgO/c-MA–400 catalyst with comparable Ni particle sizes as shown in Fig. 10. Fig. 11 illustrates the LPG conversions and the influence of calcination temperature on the selectivities of the products (CH4, CO2, CO, and H2) for the steam reforming of LPG after the 40-h reaction time. The calcination temperature showed a significant influence on the selectivities of CH4, CO2, CO, and H2. The CH4 selectivity had a similar trend of variation with LPG conversion, while the opposite was true for those of CO2, CO, and H2; namely, the higher the LPG conversion was, the higher the CH4 selectivity was and the lower the selectivities of CO2, CO and H2 were. For instance, the Ni– MgO/c-MA–400 catalyst with the highest LPG conversion of 96.9% at the 40-h reaction time had the highest CH4 selectivity at 65.5% and the lowest selectivities of CO2, CO, and H2 at 32.2%, 2.3%, and 34.3%, respectively. This result demonstrated that the Ni– MgO/c-MA catalyst with a high activity for the steam reforming of LPG was highly active for the formation of methane in the steam reforming.

Cn Hm þ H2 O ! COx þ H2

DH > 0 ðn P 2 and x ¼ 1 or 2Þ

COx þ ð2 þ xÞH2 $ CH4 þ xH2 O DH < 0 CO þ H2 O $ CO2 þ H2

DH ¼ 41:2 kJ=mol

Cn Hm þ ð2n  m=2ÞH2 ! nCH4

DH  < 0

ð1Þ ð2Þ ð3Þ

responsible for the hydrocracking of LPG (Eq. (4)), rather than for the methanation of CO and CO2 with H2, and the water–gas shift (Eq. (3)) was controlled by reaction kinetics. This phenomenon was also observed over hydrotalcite-derived Mg–Al mixed oxide-supported Ni catalysts [36] and c-MA-supported Ni–MgO catalysts [29] with lower Mg/Al molar ratios for the steam reforming of LPG. However, the case is not true over the Ni–CeO2/Al2O3 catalysts for the steam reforming of LPG [41]; namely, the CH4 selectivity was below the theoretical equilibrium values. In terms of the Ni–MgO/c-MA–T (T = 400, 500, and 600) catalysts, the selectivities of CH4, CO2, CO, and H2 were in vicinity of the equilibrium values, implying that both the methanation of CO and CO2 with H2 and the water–gas shift reached the chemical equilibrium in the reaction system. Combined with the variations in porous structures and metal Ni particle sizes with the calcination temperature, it could be proposed that the larger mesopores with narrow pore size distribution and the smaller Ni nanoparticles facilitated not only the steam reforming of hydrocarbons but also the subsequent water–gas shift and methanation of CO and CO2 with H2. On the contrary, the poorer pore structure and the larger Ni particles more favored the hydrocracking of LPG to form CH4 and further deep cracking of CHx to produce coke deposition, which is in agreement with the results of the previous studies [8,29,60]. These results also demonstrated that the steam reforming of hydrocarbons over Ni-based catalysts was rather complicated, but the catalytic properties could be improved by modifying supports, selecting promoters, and improving preparation method and calcination temperature [10–14].

3.4. Characterization and discussion of spent Ni–MgO/c-MA–T catalysts It was known that the deactivation of alumina-supported Ni catalysts for the reforming of hydrocarbons was generally caused by carbon deposition, Ni sintering, and phase transformation of alumina [8,9]. In order to gain some insight into the relationship between the calcination temperatures and these factors, the spent Ni–MgO/c-MA–T catalysts (denoted as Ni–MgO/c-MA–T–S) were characterized by TG, XRD, TEM, and N2 adsorption–desorption measurements. The carbon deposition on the Ni–MgO/c-MA–T–S catalyst was first investigated by TG analysis in air, and the results are illustrated in Fig. 12. All the Ni–MgO/c-MA–T–S catalysts showed a first weight loss at temperature lower than 350 °C corresponding to the desorption of physically adsorbed water. This was followed by a

ð4Þ

The thermodynamic equilibrium analyses demonstrated that the steam reforming of all higher hydrocarbons (n P 2) (Eq. (1)) could be considered irreversible; namely, higher hydrocarbons were completely converted to H2 and COx provided sufficient catalyst activity existed. This reaction was followed by the establishment of the equilibriums of the exothermic methanation reaction (Eq. (2)) and water–gas shift reaction (Eq. (3)) [63]. This implicated that the steam reforming of LPG over the Ni–MgO/c-MA catalysts was governed by reaction kinetics under the present operating conditions. Fig. 11 also presents the selectivities of the products at the thermodynamic equilibrium, which were calculated on the basis of the total small gas molecules (CO, CH4, H2, CO2, and H2O) except LPG in the exit gas. It was found that compared with those at the corresponding equilibrium states, the selectivities of CH4 and CO were higher and the selectivities of CO2 and H2 were smaller, over the Ni–MgO/c-MA–300 and Ni–MgO/c-MA–700. These results implied that the CH4 formation might be mainly

100

Weight loss (%)

162

80

T = 400

T = 600

T = 300

T = 500

60

T = 700

40

20 100

200

300

400

500

600

700

800

o

Temperature C Fig. 12. TG profiles of the spent Ni–MgO/c-MA–T catalysts for the steam reforming of LPG. Reaction conditions: S/C, 2; SV, 35,000 mL/gcat h; reaction temperature, 450 °C; reaction time, 40 h.

163

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166 Table 5 Ni particle sizes and deposited carbon over the spent Ni–MgO/c-MA–T catalysts for the pre-reforming of LPG.a

a

Sample

Carbon amount by TG (mg/gcat)

Pcarbon (%)

Ni particle size (nm) by XRD

TEM

Ni–MgO/c-MA–300–S Ni–MgO/c-MA–400–S Ni–MgO/c-MA–500–S Ni–MgO/c-MA–600–S Ni–MgO/c-MA–700–S

60 55 93 158 625

0.098 0.035 0.063 0.113 1.022

5.0 5.1 5.8 6.9 8.0

4.9 ± 0.5 5.0 ± 0.5 6.2 ± 0.7 7.3 ± 0.8 8.8 ± 0.9

SBET (m2/g)

Vp (cm3/g)

Dp (nm)

Da (nm)

199 163 163 143 120

0.11 0.15 0.12 0.12 0.11

3.4 3.9 3.5 3.9 4.0

3.1 3.6 3.6 3.8 4.4

Reaction conditions: S/C, 2; SV, 35,000 mL/gcat h; reaction temperature, 450 °C; reaction time, 40 h.

small weight augment in the range of 350–500 °C, assigned to the Ni metal oxidation in the catalyst, and subsequent large weight loss was attributed to the removal of the deposited carbon. The TG curve of the Ni–MgO/c-MA–300–S catalysts showed a slight difference from the other spent Ni–MgO/c-MA–T catalysts due to smaller porous structure, which adsorbed more water in the matrix. The calcination temperature had a substantial influence on the amount of deposited carbon over the Ni–MgO/c-MA–T–S catalysts for the steam reforming of LPG. Table 5 gives the amount of carbon estimated from 500 to 800 °C in the TG profiles. It could be found that the Ni–MgO/c-MA–300–S and Ni–MgO/c-MA–400–S catalysts showed very small amounts of the deposited carbon, which was 60 mg/gcat and 55 mg/gcat, respectively. Correspondingly, the percentages of the deposited carbon to the converted carbon in LPG during the reaction period (designated as Pcarbon) were only 0.098% and 0.035%. With elevating the calcination temperature, the amounts of deposited carbon over the Ni–MgO/c-MA–T–S catalysts were dramatically augmented. When the calcination temperature was elevated to 700 °C, the amount of the deposited carbon was augmented to 625 mg/gcat, where corresponding Pcarbon was 1.022%, 29 times higher than that over the Ni–MgO/c-MA–400–S catalyst. These results were in good agreement with the sizes of Ni particles in the Ni–MgO/c-MA–T catalysts shown in Table 3. Fig. 13 presents the XRD patterns of the Ni–MgO/c-MA–T–S catalysts used for the pre-reforming of LPG at 450 °C after the 40-h reaction time. The Ni–MgO/c-MA–T–S catalysts showed almost the same XRD diffraction patterns corresponding to the c-Al2O3 and/or MgAl2O4 spinel, MgO, and metal Ni as before the reaction. However, the relative intensities of the (4 4 0) and (3 1 1) peaks for spinel showed a minor decrease, and meanwhile, the peak

∗

Intensity (a.u.)

∗

Ni Ni(Mg)O γ-Al2O3 and Ni(Mg)Al2O4 Graphite carbon

T = 700 T = 600 T = 500 T = 400 T = 300

10

20

30

40

50

2θ /

60

70

80

90

ο

Fig. 13. XRD patterns of the spent Ni–MgO/c-MA –T catalysts for the steam reforming of LPG. Reaction conditions: S/C, 2; SV, 35,000 mL/gcat h; reaction temperature, 450 °C; reaction time, 40 h.

intensity for the crystalline MgO at 43.2° also had a certain reduction. This result indicated that in the steam reforming process of LPG, parts of MgO reacted with c-Al2O3 to form MgAl2O4 spinel. Interestingly, compared with those before the reaction, the Ni peak intensities for the Ni–MgO/c-MA–T–S catalysts were diminished and more diffused. This result meant that the reaction atmosphere could influence the interaction between metal Ni nanoparticles and the support, resulting in the re-construction of Ni crystallites on the catalyst surface in the steam reforming of LPG as reported in the previous papers [5,41]. Table 5 lists the Ni crystallite sizes over the Ni–MgO/c-MA–T–S catalysts for the pre-reforming of LPG at 450 °C after the 40-h reaction time. The sizes of Ni crystallites became smaller relative to those before the reaction shown in Table 3. It was proposed that during the steam reforming of LPG, the reversible reduction–oxidation between Ni metal and Ni2+ ions on the catalyst surface took place. As a consequence, Ni species were redistributed, producing Ni metal particles dispersed more finely [5]. Although the relative intensity for spinel after the reaction made some changes due to the formation of MgAl2O4 spinel, the FWHMs for spinel had no obvious change: For example, the FWHMs for the (4 4 0) planes of Ni–MgO/c-MA–300–S and Ni–Mg O/c-MA–700–S were 4.12° and 2.23°, respectively, close to those before the reaction. This result suggested that the mesoporous c-Al2O3 frameworks might still be maintained through the cross-linking of small c-Al2O3 and/or c-Al2O3–MgAl2O4 nanoparticles. In the XRD patterns of the Ni–MgO/c-MA–T–S (T = 400, 500, and 600) samples in Fig. 13, there was a clear diffraction peak observed at 26.2°, corresponding to graphitic carbon (PDF 26-1080), which was believed to be mainly responsible for the deactivation of Ni-based catalysts in the reforming of hydrocarbons [62]. The peak intensity increased with the increase in the calcination temperature, that is, the amounts of graphite carbon deposited on the catalysts increased with the calcination temperature, which were consistent with the variation in the stability of the Ni– MgO/c-MA–T catalysts. However, the Ni–MgO/c-MA–T–S (T = 300 and 400) samples showed no diffraction peaks associated with graphite carbon. The TEM images of the Ni–MgO/c-MA–T–S catalysts are illustrated in Fig. 14. It could be found that the Ni–MgO/c-MA–T–S catalysts showed similar frameworks with a homogenous wormhole-like mesoporous structures to the counterparts before the reaction shown in Fig. 8. The Ni nanoparticles were highly dispersed in the support frameworks. However, as summarized in Table 5, the Ni particle sizes after the reaction became smaller than before the reaction, which were in agreement with the XRD results displayed in Fig. 13 and Table 5. On the other hand, no noticeable accumulation of carbon deposition was observed in the TEM images, implying that the deposited carbon might be highly dispersed in the catalyst frameworks. The convincing evidence for the maintenance of the mesoporous structures was also provided by N2 sorption measurements,

164

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

Fig. 14. TEM images of the spent Ni–MgO/c-MA–T catalysts after the steam reforming of LPG at 450 °C for 40 h. (a) T = 300, (b) T = 400, (c) T = 500, (d) T = 600, and (e) T = 700.

and the results are displayed in Fig. 15. With the exception that the Ni–MgO/c-MA–300–S sample showed characteristic of type IV isotherm with a hysteresis loop and a weak pore size distribution centered at 3.4 nm due to the blockage of carbon deposited in micropores, which were significantly different from the Ni–MgO/ c-MA–300 catalyst, all the other Ni–MgO/c-MA–T–S samples displayed the similar shape N2 sorption isotherms and pore size distribution curves to the counterparts before the reaction. However, the pore size distribution peaks obviously diminished. Table 5 gives the specific surface areas, pore volumes, and pore

sizes. It could be seen that in general, the specific surface areas and pore volumes of the Ni–MgO/c-MA–T–S samples lowered due to the coke deposition, compared with those before the reaction. In combination with the results of the previous characterizations, it could be concluded that the size of metal Ni particles over the Ni–MgO/c-MA catalysts played a decisive role for the rate of the coke deposition, affecting the activity and stability of the Ni catalyst. The pore structures were mainly responsible for the diffusion of the reactants and products to the active sites, causing the variation in the catalytic properties.

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

a

T = 700

Volume adsorbed a.u.

T = 600

T = 500 T = 400 T = 300

0.0

0.2

0.4

0.6

0.8

1.0

Relative pressure P/P0

b Pore volume distribution a.u.

T = 700

T = 400

This research was supported by Innovation Program of Shanghai Municipal Education Commission, the Major State Basic Research Development Program of China (No. 2014CB643403), National Science Fund for Distinguished Young Scholars (No. 51225401), and Basic Major Research Program of Science and Technology Commission Foundation of Shanghai (No. 14JC1491400).

T = 300

References

T = 500

3

4

5

6

size distributions. However, the Ni particle sizes increased with the calcination temperature. This phenomenon might be explained by the dominant mechanism for Ni crystallite growth by Ostwald ripening rather than by migration and coalescence of Ni nanoparticles. The catalytic properties of the Ni–MgO/c-MA–T catalysts for the steam reforming of LPG showed that the Ni particle sizes on the surfaces exhibited a strong relationship with the catalytic activity and stability and resistance to coke deposition. Smaller metal Ni nanoparticles could facilitate the steam reforming of LPG and the subsequent methanation of CO and CO2 with H2 and water–gas shift and lower the formation rate of coke deposition. The porous structures of the Ni–MgO/c-MA–T catalysts had a significant correlation with the availability of the reactants to metal Ni active sites. The large mesopores with uniform pore sizes were more favorable for the diffusion of the reactants and products to the active sites, improving the catalytic activity and stability. We believe that the present investigations will provide a comprehensive insight into the interrelation among the active metal, the promoter, the catalyst structure, and the catalytic properties over c-alumina-supported Ni-based catalysts for the reforming reactions of hydrocarbons. Acknowledgments

T = 600

2

165

7

8

Pore size (nm) Fig. 15. (a) Nitrogen adsorption–desorption isotherms and (b) BJH pore size distribution of the Ni–MgO/c-MA–T–S catalysts.

4. Conclusions

c-Alumina-supported Ni–Mg oxides (NiO–MgO/c-MA–T), which were prepared through one-pot hydrolysis and co-condensation of inorganic salts without organic surfactants, were obtained at different calcination temperatures in the range of 300–700 °C. The calcination temperatures had significant influences on the porous structure and surface properties of the NiO– MgO/c-MA catalysts, the interaction between metal oxides and the support and the particle sizes of metal Ni formed. With the exception of NiO–MgO/c-MA–300 with undecomposed metal hydroxides, the other NiO–MgO/c-MA–T (T = 400, 500, 600, and 700) materials showed similar mesoporous structures with high specific surface areas and narrow pore size distributions, and the specific surface areas and the pore volumes lowered with the calcination temperature. CO2-TPD results showed that on the surface of the NiO–MgO/c-MA–T (T = 400, 500, 600, and 700) materials, the number of basic sites gradually decreased, but the strength and density of basic sites was almost unchanged with the calcination temperature. Ni species were homogeneously distributed in the mesoporous c-alumina framework mainly in the form of highly dispersed NiO. The interaction between Ni species and the support strengthened with the calcination temperature. After the H2 reduction at 600 °C, all the Ni–MgO/c-MA–T catalysts produced uniformly dispersed Ni nanoparticles with relatively narrow particle

[1] L.L. Xu, H.L. Song, L.J. Chou, ACS Catal. 7 (2012) 1331–1342. [2] L.B. Raberg, M.B. Jensena, U. Olsbye, C. Daniel, S. Haag, C. Mirodatos, A.O. Sjastad, J. Catal. 249 (2007) 250–260. [3] B.S. Caglayan, A.K. Avci, Z.I. Onsan, A.E. Aksylu, Appl. Catal. A 280 (2005) 181– 188. [4] D.L. Li, I. Atake, T. Shishido, Y. Oumi, T. Sano, K. Takehira, J. Catal. 250 (2007) 299–312. [5] K. Takehira, T. Shishido, P. Wang, T. Kosaka, K. Takaki, J. Catal. 221 (2004) 43– 54. [6] S. Souentie, M. Athanasiou, D.K. Niakolas, A. Katsaounis, S.G. Neophytides, C.G. Vayenas, J. Catal. 306 (2013) 116–128. [7] B. Valle, B. Aramburu, A. Remiro, J. Bilbao, A.G. Gayubo, Appl. Catal. B 147 (2014) 402–410. [8] K.O. Christensen, D. Chen, R. Lødeng, A. Holmen, Appl. Catal. A 314 (2006) 9– 22. [9] J. Xu, L.W. Chen, K.F. Tana, A. Borgna, M. Saeys, J. Catal. 261 (2009) 158–165. [10] S. Natesakhawat, R.B. Watson, X.Q. Wang, U.S. Ozkan, J. Catal. 234 (2005) 496– 508. [11] J.J. Strolhm, J. Zheng, C.S. Song, J. Catal. 238 (2006) 309–320. [12] J.G. Seo, M.H. Youn, I.K. Song, Int. J. Hydrogen Energy 34 (2009) 1809–1817. [13] E. Nikolla, J. Schwank, S. Linic, J. Catal. 250 (2007) 85–93. [14] J. Yang, X.G. Wang, L. Li, K. Shen, X.G. Lu, W.Z. Ding, Appl. Catal. B 96 (2010) 232–237. [15] A.H. Chen, T. Miyao, K. Higashiyama, H. Yamashita, M. Watanabe, Angew. Chem. Int. Ed. 49 (2010) 9895–9898. [16] N. Wang, K. Shen, L.H. Huang, X.P. Yu, W.Z. Qian, W. Chu, ACS Catal. 3 (2013) 1638–1651. [17] F. Bimbela, D. Chen, J. Ruiz, L. García, J. Arauzo, Appl. Catal. B 119 (2012) 1–12. [18] L.L. Xu, H.L. Song, L.J. Chou, Appl. Catal. B 108 (2011) 177–190. [19] M.H. Amin, K. Mantri, J. Newnham, J. Tardio, S.K. Bhargava, Appl. Catal. B 120 (2012) 217–226. [20] W. Chen, G.F. Zhao, Q.S. Xue, L. Chen, Y. Lu, Appl. Catal. B 137 (2013) 260–268. [21] J. Liu, J. Yu, F.B. Su, G.W. Xu, Catal. Sci. Technol. 4 (2014) 472–481. [22] M. Wu, D.M. Hercules, J. Phys. Chem. 83 (1979) 2003–2008. [23] P. Munnik, M.E.Z. Velthoen, P.E. de Jongh, K.P. de Jong, C.J. Gommes, Angew. Chem. Int. Ed. 53 (2014) 9493–9497. [24] G.H. Li, L.J. Hu, J.M. Hill, Appl. Catal. A 301 (2006) 16–24. [25] J. Mazumder, H.I. de Lasa, Appl. Catal. B 168–169 (2015) 250–265. [26] K. Zakamura, T. Miyazawa, T. Sakurai, T. Miyao, S. Naito, N. Begum, K. Kunimori, K. Tomishige, Appl. Catal. B 86 (2008) 36–44. [27] L.P.R. Profeti, E.A. Ticianelli, E.M. Assaf, Int. J. Hydrogen Energy 34 (2009) 5049–5060.

166

M. Tan et al. / Journal of Catalysis 329 (2015) 151–166

[28] X.F. Shang, X.G. Wang, W.X. Nie, X.F. Guo, X.J. Zou, W.Z. Ding, X.G. Lu, J. Mater. Chem. 22 (2012) 23806–23814. [29] M.W. Tan, X.G. Wang, X.F. Shang, X.J. Zou, X.G. Lu, W.Z. Ding, J. Catal. 314 (2014) 117–131. [30] L. L Xu, H.L. Song, L.J. Chou, Int. J. Hydrogen Energy 38 (2013) 7307–7325. [31] Z. Boukha, L. Fitian, M. Lopez-Haro, M. Mora, J.R. Ruiz, C. Jimenez-Sanchidrian, G. Blanco, J.J. Calvino, G.A. Cifredo, S. Trasobares, S. Bernal, J. Catal. 272 (2010) 121–130. [32] J.F. Yu, Q.J. Ge, W. Fang, H.Y. Xu, Appl. Catal. A 395 (2009) 114–119. [33] J.L. Chen, Q. Ma, T.E. Rufford, Y.D. Li, Z.H. Zhu, Appl. Catal. A 362 (2009) 1–7. [34] F.W. Chang, M.S. Kuo, M.T. Tsay, M.C. Hsieh, J. Chem. Technol. Biotechnol. 79 (2004) 691–699. [35] J.G. Seo, M.H. Youn, J.S. Chung, I.K. Song, J. Ind. Eng. Chem. 16 (2010) 795–799. [36] K. Shen, X.G. Wang, X.J. Zou, X.X. Wang, X.G. Lu, W.Z. Ding, Int. J. Hydrogen Energy 36 (2011) 4908–4916. [37] S. Takenaka, K. Kawashima, H. Matsune, M. Kishida, Appl. Catal. A 321 (2007) 165–174. [38] D.J. Moon, Catal. Surv. Asia 12 (2008) 188–202. [39] K. Nagaoka, K. Sato, Y. Takita, J. Catal. 287 (2012) 86–92. [40] H. Jeong, M. Kang, Appl. Catal. B 95 (2010) 446–455. [41] X.J. Zou, X.G. Wang, L. Li, K. Shen, X.G. Lu, W.Z. Ding, Int. J. Hydrogen Energy 35 (2010) 12191–12200. [42] M. Trueba, S.P. Trasatti, Eur. J. Inorg. Chem. 17 (2005) 3393–3403. [43] L.J.I. Coleman, W. Epling, R.R. Hudgins, E. Croiset, Appl. Catal. A 363 (2009) 52– 63. [44] F. Waldner, R.M. Laine, S. Dhumrongvaraporn, S. Tayaniphan, R. Narayanan, Chem. Mater. 8 (1996) 2850–2857.

[45] U. Olsbye, D. Akporiaye, E. Rytter, M. Ronnekleiv, E. Tangstad, Appl. Catal. A 224 (2002) 39–49. [46] J. Zielinski, J. Catal. 76 (1982) 157–163. [47] P. Salagre, J.L.G. Fierro, F. Medina, J.E. Sueiras, J. Mol. Catal. A 106 (1996) 125– 134. [48] A.M. Siouffi, J. Chromatogr. A 1000 (2003) 801–818. [49] J.B. Pang, K.Y. Qiu, Y. Wei, X.J. Lei, Z.F. Liu, Chem. Commun. (2000) 477–478. [50] Y. Wei, J.G. Xu, H. Dong, J.H. Dong, K. Qiu, S.A. Jansen-Varnum, Chem. Mater. 11 (1999) 2023–2029. [51] Y. Li, X.X. Wang, C. Xie, C.S. Song, Appl. Catal. A 357 (2009) 213–222. [52] Z. Boukha, L. Fitian, M. López-Haro, M. Mora, J.R. Ruiz, C. Jiménez-Sanchidrián, G. Blanco, J.J. Calvino, G.A. Cifredo, S. Trasobares, S. Bernal, J. Catal. 272 (2010) 121–130. [53] C.D. Wagner, J.F. Moulder, L.E. Davis, W.M. Riggs, Standard XPS spectra of the elements, in: G.E. Muilenberg (Ed.), Handbook of X-ray Photoelectron Spectroscopy, II, Perking-Elmer Corporation, Eden Prairie, MN, 1979. [54] T.L. Barr, J. Phys. Chem. 82 (1978) 1801–1810. [55] G.P. Li, A. Proctor, D.M. Hercules, Appl. Spectrosc. 38 (1984) 880–886. [56] J.A. Wang, A. Morales, X. Bokhimi, O. Novaro, Chem. Mater. 11 (1999) 308–313. [57] M. Lo Jacono, M. Schiavello, A. Cimino, J. Phys. Chem. 76 (1971) 1044–1049. [58] A. Cimino, M. Lo jacono, M. Schiavello, J. Phys. Chem. 79 (1975) 243–249. [59] Y. Kathiraser, W. Thitsartarn, K. Sutthiumporn, S. Kawi, J. Phys. Chem. C 117 (2013) 8120–8130. [60] H.S. Bengaard, J.K. Nørskov, J. Sehested, B.S. Clausen, L.P. Nielsen, A.M. Molenbroek, J.R. Rostrup-Nielsen, J. Catal. 209 (2002) 365–384. [61] L.M. Aparicio, J. Catal. 165 (1997) 262–274. [62] C. Zheng, Y. Apeloig, R. Hoffmann, J. Am. Chem. Soc. 110 (1988) 749–774. [63] T.S. Christensen, Appl. Catal. A 138 (1996) 285–309.