Reforming Methane with CO 2 over Hierarchical Porous Silica-Supported Nickel Catalysts Modified with Lanthanum Oxide

,


Introduction
As the key greenhouse gas, carbon dioxide (CO 2 ) is primarily originated from the utilization of fossil fuels [1,2]. With the booming economic growth around the world, the global CO 2 emission is increasing year by year. This is bound to have a great negative impact on the living environment. At present, the conversion routes of using CO 2 as a valuable carbon source to manufacture useful chemicals have received extensive attention, such as hydrogen reduction of CO 2 to produce methanol [3], methane reforming to prepare synthesis gas (a mixture of H 2 +CO) [4,5], and CO 2 reaction with methanol to form dimethyl carbonate [6].
In DRM reaction, nickel-based catalysts have been extensively investigated because of its wide source, satisfactory activity, and competitive cost with most noble metal catalysts [11][12][13]. Yet, the rapid catalyst deactivation caused by carbon deposition and metal sintering is a tricky challenge in the large-scale industrial application in the future [14][15][16]. Carbon deposition originated from CH 4 decomposition (Equation (2)) and CO disproportionation (Equation (3)) is the main factor related to the deactivation [17][18][19]. Some research works have illustrated the feasibility of obtaining nickel-based catalysts with high stability and high activity by using some basic lanthanoid oxides [20][21][22][23][24]. Lanthanum oxide (La 2 O 3 ) [25] and cerium oxide (CeO 2 ) [26,27] as promoters or supports promoted carbon elimination by enhancing the adsorption ability of CO 2 molecules, thus achieving good catalytic performance. CeO 2 doped in the Al 2 O 3 framework could accelerate the oxygen mobility and play as the active center for CO 2 adsorption, subsequently suppressing the carbon deposition on the doped Ni/Al 2 O 3 catalyst [26,28]. In the ordered mesoporous La 2 O 3supported nickel catalyst [29], the increased interfaces between nickel and La 2 O 3 promoted the formation of bidentate carbonate, which participated in the carbon elimination and improved the resistance of carbon deposition.
In addition, the mesoporous materials have been extensively used as catalyst supports due to their superior textural properties such as large surface areas and controllable mesopore structures [30,31]. Immobilizing nickel nanoparticle within the channel of mesoporous supports can significantly control the particle size and preserve the initial crystallite sizes by the confinement effect of the mesoporous structure [32][33][34], which was beneficial to minimize the aggregation and carbon depositions on the nickel nanoparticles. Nickel particles incorporated into SBA-15 [35,36], SBA-16 [37,38], and MCM-41 [39,40] have displayed better carbon resistance in DRM reaction. Alternatively, the "one-pot" synthesis strategy can not only minimize crystalline size of nickel particles but also strengthen the interaction between nickel and support, thus notably weakening the motivation of carbon diffusion and further promoting long-term stability [41][42][43][44]. For instance, during the hydrolysis of urea, the precipitation of nickel salt precursor occurred concurrently with the formation of silica support. This process produced very fine metal particles with an average particle size of 3 nm [45]. Similarly, the nickel-based catalyst supported on the ordered mesoporous alumina synthesized via the evaporation-induced self-assembly method had intimate metal-support interaction and strong resistance to carbon deposition, accounting for no deactivation after a 100 h long-term stability test at 700°C [46]. The improved catalytic performance was suggested to be closely associated with both the amount of "accessible" active centers for the reactants on the mesopore wall surface and the stabilization of the active sites by the alumina matrix due to the "confinement effect" of the mesopores.
Motivated by the promotion effect of the abovementioned strategies, our previous work [47] reported nickel-based catalysts supported on porous silica modified with La 2 O 3 , CeO 2 , Sm 2 O 3 , and Gd 2 O 3 . Kinetic results indicated that the presence of lanthanum accelerated CO 2 dissociation, hence significantly suppressing the accumulation of carbon on the catalyst surface. However, the effect of lanthanum content on the textural properties of the catalysts and the relationship between La loading and carbon deposition or metal sintering were not clear. This was extraordinarily meaningful to thoroughly understand the influence of lanthanum on the deactivation of the supported nickel-based catalysts. To this end, the hierarchical porous silicasupported nickel-based catalysts modified with different lanthanum content were designed and synthesized via "onepot" method in this study. The main purpose of this study was to improve the carbon resistance of nickel-based catalyst by optimizing the lanthanum content, so as to deeply understand the effect of lanthanum loading on the physiochemical properties, carbon resistance, metal sintering, and catalytic performance in DRM reaction.

Catalyst Preparation.
Lanthanum-modified hierarchical porous silica-supported nickel-based catalysts were prepared using "one-pot" method. The nominal weight loading of nickel was 10 wt%. In a typical synthesis, 2.2 g of cetyltrimethylammonium bromide (CTAB) as a template was dispersed in 66 mL of deionized water and 10 mL of aqueous solution containing of 0.98 g of Ni(NO 3 ) 2 ·6H 2 O and required amount of La(NO 3 ) 3 ·6H 2 O was then added under stirring. Separately, another 10 mL of aqueous solution containing 4 g of urea and 6.25 g of tetraethyl orthosilicate (TEOS) was mixed slowly with the above metal saltcontaining CTAB solution. Urea was added to the mixture to induce the hydrolysis reaction of TEOS and to promote the precipitation of the nickel and lanthanum phases onto the silica surface. During the dropwise and stirring, the mixture gradually became light green and cloudy. After stirring at room temperature for 1 h, the mixture was heated at 100°C in an oil bath for 60 h to complete crystallization of the silica support and the incorporation of metal ions into the silica. The resultant green solid was then recovered by centrifugation and successively washed with deionized water and ethanol. The obtained material was dried in an oven at 100°C overnight, followed by calcination in air at 550°C for 6 h in a muffle furnace to remove the organic compounds. The eventual lanthanum-modified catalyst was designated as Ni1.5La/SiO 2 , Ni3.0La/SiO 2 , and Ni4.5La/SiO 2 , which indicated that the molar ratio of Ni/La in the catalyst was 5 : 1.5, 5 : 3.0, and 5 : 4.5, respectively. The catalyst without lanthanum nitrate named as Ni/SiO 2 was prepared by the similar method.

Catalyst
Characterization. The actual nickel and lanthanum contents in the catalyst powder were measured using Optima 8300 Inductively Coupled Plasma-Optical Emission Spectrometer (ICP-OES, Perkin Elmer).

2
International Journal of Energy Research The surface area and pore diameter distribution of the catalyst were analyzed by nitrogen physisorption at -196°C using a Tristar II 3020 Surface Area and Porosity Analyzer (Micromeritics). Before the adsorption, the catalyst was degassed at 150°C for 6 h under vacuum. The specific surface area was calculated using the Brunauer-Emmett-Teller (BET) equation in the relative pressure range of 0.07-0.30. The pore diameter distribution was analyzed with the adsorption branches of the physisorption isotherm using the Barrett-Joyner-Halenda (BJH) method.
The phase compositions were identified by powder Xray diffraction (XRD) using a D8 Advance X-ray diffractometer (Bruker) equipped with a Cu Kα radiation source. The scattering angle (2θ) was recorded from 10°to 70°with a step size of 0.01°and an acquisition time of 0.1 s/step. The morphological property and particle size distribution were detected by transmission electron microscopy (TEM) using a JEM-2100F electron microscope (JEOL). A high-angle annular dark field (HAADF) detector was used to acquire scanning TEM micrographs (HAADF-STEM). The distribution maps of nickel and lanthanum elements were obtained using an XFlash 6T-60 energy dispersive Xray (EDX) detector (Bruker). The particle size distribution was determined by counting 100 particles, and the average particle diameter (d) was calculated by means of the following formula: where n i is the number of particles with diameter d i .
The reduction behavior and reducibility of the calcined catalyst were investigated by hydrogen temperatureprogrammed reduction (H 2 -TPR) using an AutoChem II 2910 Chemisorption Analyzer (Micromeritics). About 100 mg of catalyst was degassed in argon flow at 300°C for 1 h and then reduced under 50 mL/min of a 10% H 2 /Ar flow. The temperature program was set to room temperature as the initial temperature and 800°C as the final temperature, and a heating rate was 5°C/min. The consumption of hydrogen was recorded online with a thermal conductivity detector (TCD).
Hydrogen pulse chemisorption experiment was carried out using the same instrument as described for H 2 -TPR analysis. After completing the H 2 -TPR operation, the catalyst was sufficiently purged in argon flow and cooled to 30°C. Subsequently, 10% H 2 /Ar was repeatedly injected until the monitored peak area reached saturated. The amount of nickel sites on the catalyst surface and the percentage of the exposed nickel (i.e., Ni dispersion) were calculated from the assumption that the stoichiometry ratio of surface H : surface Ni was 1 : 1.
The surface basicity of calcined catalyst was studied by carbon dioxide temperature-programmed desorption (CO 2 -TPD), using the same instrument in H 2 -TPR analysis. Prior to the adsorption, about 50 mg of calcined catalyst was heated at 300°C for 0.5 h with a pure argon flow and then cooled to room temperature. Subsequently, the adsorption was conducted with a 10% CO 2 /Ar flow at a rate of 30 mL/ min for 90 min and then purged by pure argon flow for 90 min to eliminate the physical adsorbed CO 2 . Thereafter, the catalyst was heated to 800°C at a ramp rate of 5°C/min, and the desorbed CO 2 was monitored by TCD detector.
The oxidation state of nickel and lanthanum on the catalyst surface was examined by X-ray photoelectron spectroscopy (XPS) using an ESCALAB 250Xi spectrometer (Thermo Fisher Scientific) equipped with an Al Kα excitation source. The binding energies were calibrated with respect to the C 1s peak at 284.6 eV of the adventitious carbon.
The amount of carbon deposited on the catalyst surface was quantitatively measured by thermogravimetric (TG) analysis using a STA 449 F1 Jupiter Simultaneous Thermal Analyzer (Netzsch). About 5 mg of the spent catalyst was oxidized in air flow from 30 to 800°C at a heating rate of 10°C/min.
The types of carbon formed on spent catalysts were analyzed by Raman's spectroscopy using a LabRAM Aramis Raman spectrometer (Horiba Jobin Yvon) and also studied by temperature-programmed hydrogenation (TPH) technique using the same instrument for H 2 -TPR analysis. Typically, about 25 mg of the spent catalyst was heated in argon flow at 200°C for 2 h. After cooling to 50°C, the catalyst was hydrogenated from 50 to 850°C with a heating rate of 5°C/ min under 50 mL/min of a 10% H 2 /Ar flow.

Catalytic Activity
Measurements. The evaluation of catalytic performance for lanthanum-modified silica-supported nickel catalyst in DRM reaction was performed under atmospheric pressure. The reactor was a vertical quartz tube (6 mm inner diameter and 450 mm length) containing 100 mg of the calcined catalyst. A nickel-chromium/nickelsilicon thermocouple covered with a quartz thermocouple well (1.5 mm inner diameter) was placed at the outlet of the catalyst bed. The catalyst was in situ reduced and activated at 750°C under a 60 mL/min of pure hydrogen flow for 30 min. Then, the reactant gas mixture CH 4 /CO 2 with the molar ratio of 1 : 1 was introduced to the reduced catalyst bed at the total flow rate of 80 mL/min. The composition of the effluent gas was determined by online analysis using a gas chromatograph (GC9800, China) equipped with a TCD. The conversions for CH 4 and CO 2 and the ratio of H 2 /CO and H 2 selectivity were calculated according to where X, S, R, and F represented the conversion, selectivity, molar ratio, and gas flow rate, respectively.

Phase Composition and Metal Dispersion of Lanthanum-
Modified Silica-Supported Nickel Catalysts. The nickel content in four catalysts was 8.8-10.7 wt% (Table 1), which 3 International Journal of Energy Research was approximately to the stipulated Ni content. As expected, the actual content of lanthanum was in the range of 6-12 wt%, growing gradually with the increasing dosage of lanthanum salt.
The formation of 1 : 1 Ni-phyllosilicate species in the calcined Ni/SiO 2 catalyst was evidenced by two diffraction peaks at 34°and 60°(JCPDS #22-0754) (Figure 1(a)). The intensity of these peaks gradually decreased with the increase of lanthanum content. This result indicated that the addition of La hindered the formation of Ni-phyllosilicate species during "one-pot" process. Besides, this addition also attenuated the formation of silica support, as evidenced by the gradual weakening of the broad peak at 21.6°attributed to amorphous silica support. Furthermore, no characteristic diffraction peaks of La 2 O 3 were detected, which was likely due to the high dispersion. Intriguingly, Ni4.5La/SiO 2 exhibited several new diffraction peaks at 13.1°, 22.7°, 29.6°, 30.7°, and 44.3°, attributable to the characteristic of monoclinic La 2 O 3 CO 3 (JCPDS #48-1113). The formation of La 2 O 3 CO 3 was associated with the reaction of La 2 O 3 with carbon dioxide in air atmosphere during calcination process [48,49].
After the catalysts were reduced with H 2 reduction, all catalysts did not present 1 : 1 Ni-phyllosilicate phase ( Figure 1(b)) but exhibited characteristic diffraction peaks at 44.5°and 52°attributable to cubic metallic nickel (JCPDS #04-0850). This result indicated the Ni-phyllosilicate phase was reduced to metallic nickel with pure H 2 flow at 750°C. The diffraction peaks for La 2 O 3 CO 3 vanished on the reduced Ni4.5La/SiO 2 ( Figure 1(b)), which was attributed to the transformation of La 2 O 3 CO 3 to La 2 O 3 with H 2 reduction [50]. Alternatively, a new diffraction peak at 26.7°attributable to moganite (JCPDS #38-0360) was observed. The average nickel crystallite sizes estimated by the Scherrer equation were 4.5, 6.7, 7.4, and 8.2 nm for the respective catalysts (Table 2). Apparently, adding La and increasing its content hindered the formation of finer Ni nanoparticle during "one-pot" process. However, all catalysts exhibited similar nickel dispersion in the range of 5.2-6.5% (Table 2).

Porosity Structure of Lanthanum-Modified Silica-Supported Nickel
Catalysts. The evolution of the phase composition shown by XRD patterns in Figure 1(b) affected the textural properties of the catalysts. All reduced catalysts showed typical type IV isotherm ( Figure S1A), which was a characteristic for mesoporous structure [51]. There were two types of H3 hysteresis, one at medium relative pressure (P/P 0 ) (about 0.4-0.8) and the other at high P/P 0 (>0.8), which indicated the presence of small-sized mesopore [52] and large-sized mesopore larger than 30 nm [53], respectively. The pore size distribution curves also clearly verified that the catalysts possessed hierarchical porous structure ( Figure S1B). Specifically, the narrow peak at about 3 nm represented small-sized mesopore, the sharp peak at 40 nm was associated with large-sized mesopores, and the broader region beyond 50 nm was attributed to macropore structure [52,54]. Although all catalysts exhibited the similar pore size distribution, there was obvious discrepancy in S BET , D p , and V p ( Table 1). Compared with the Ni/SiO 2 catalyst, except for Ni3.0La/ SiO 2 , the S BET and V p increased with lanthanum addition, being the largest on Ni4.5La/SiO 2 (83 m 2 /g, 0.13 cm 3 /g). This increase could be attributed to the presence of more and more lanthanum nitrate with hexahydrates in the initial preparation mixture, which hydrolyzed and induced the different pH levels in the respective mixtures. As a result, there were different amounts of La 2 O 2 CO 3 in the calcined catalyst and thus more external particles of La 2 O 3 on the reduced Ni4.5La/SiO 2 catalyst, which provided extra pores to adsorb nitrogen during monolayer physisorption. The adsorption branch of isotherm in Ni4.5La/SiO 2 catalyst displayed abrupt increase in adsorption quantities at very high P/P 0 (>0.95), suggesting the presence of macropore with large pore volume. The increase in surface area after the addition of metal nitrate was a notable characteristic of these nickel-phyllosilicate materials, especially when they were synthesized by one-pot method [55][56][57]. In the case of Ni3.0La/SiO 2 , the amount of lanthanum composite compensated the hydrolysis of TEOS in the urea; however, the reason was not yet clarified. This alteration could contribute to the slight loss of surface area, as seen in analogous isotherm for Ni3.0La/SiO 2 and La-free Ni/SiO 2 ( Figure S1A and Table 1).

Morphology of the Reduced Lanthanum-Modified Silica-Supported Nickel
Catalysts. The morphology of the reduced catalysts and the nickel dispersion were directly detected via HAADF-STEM images. These images ( Figure S2-S5) indicated that nickel and lanthanum species were evenly distributed in each catalyst. Nickel nanoparticles with different diameter randomly dispersed on flake-like silica ( Figure S6A-D). Over the lanthanum-modified catalysts, the Ni nanoparticle was larger than those on Ni/SiO 2 . The particle size is mainly distributed at the range of 4-9 nm for lanthanum-modified catalysts (Fig. S6B1-D1), slightly broader than that of Ni/SiO 2 (3-6 nm) (Fig. S6A1). On the  International Journal of Energy Research whole, the particle size was 4:5 ± 0:8, 5:8 ± 0:9, 6:0 ± 0:9, and 6:2 ± 1:2 nm for Ni/SiO 2 and respective lanthanummodified catalysts (Table 2). There was a small discrepancy between XRD and TEM results, which was due to the measurement principle. The data from XRD diffraction pattern was calculated through half peak width and diffraction angle, representing the average crystallite size. TEM analysis was implemented through statistical observation, reflecting the particle size on certain local area. However, the changing trend in particle size obtained from TEM coincided well with that from XRD, supporting the aforementioned result that the increasing amount of lanthanum gradually increased the formation of larger nickel nanoparticles.

Reducibility of the Calcined Lanthanum-Modified Silica-Supported Nickel
Catalysts. The modification with La affected the formation of Ni 2+ species with different reduction behaviors. The TPR profiles of all calcined catalysts showed broad reduction peak at the range of 350 to 800°C ( Figure 2). Consistent with the previous reports, the reduction peaks appearing at 450°C (peak 1, violet line) and 500°C (peak 2, olive line) were attributed to the reduction of NiO with weak and moderate interaction with SiO 2 , respectively [58]. The peak located at 650°C (peak 3, wine line) was derived from the reduction of Ni 2+ in 1 : 1 phyllosilicate, which possessed strong interaction with SiO 2 [59][60][61]. More importantly, in each catalyst, the area for the last reduction peak was considerably higher than those of peak 1 and peak 2, indicating that 1 : 1 Ni-phyllosilicate was predominant species on the calcined catalysts. Moreover, it should be noted that the H 2 consumption area for peak 3 gradually declined and those for peak 1 and peak 2 instead increased with the content of lanthanum (Figure 2(b)). This result clearly illustrated that the addition of La was not conducive to the generation of 1 : 1 Niphyllosilicate yet promoted the transformation from Niphyllosilicate to NiO species which had medium strength interaction with support. This led to a slightly higher reduction degree of the La-containing catalysts (58-64%) than Ni/ SiO 2 (57%) ( Table 2). This phenomenon supported the XRD result that the intensity of the obvious diffraction peaks for Ni-phyllosilicate gradually attenuated with the increase of lanthanum loading (Figure 1(a)).

Surface
Basicity of Lanthanum-Modified Silica-Supported Nickel Catalysts. The surface basicity of the catalysts was measured by CO 2 -TPD ( Figure S7). In general, 10    5 International Journal of Energy Research the CO 2 desorption peak at higher temperature represented the stronger alkalinity [62]. The CO 2 desorption peak was mainly in the range of 50-800°C. Specifically, the peak at about 100°C represented the weak alkaline sites such as bicarbonates, and that at around 200°C corresponded to medium sites, which was associated to the bidentate carbonates generated on the Lewis acid-base sites [63][64][65]. Compared to Ni/SiO 2 , the increasing amount of La induced weak and medium alkaline sites because of the gradually increased pore volumes and specific surface areas. The desorption peak in the range of 300-800°C was attributed to the strong basic sites, which was related to the unidentate carbonate [66]. This peak was more distinct in the Ni4.5La/SiO 2 catalyst, due to the decomposition of the large amount of La 2 O 2 CO 3 [67].
3.6. Surface Chemistry of Lanthanum-Modified Silica-Supported Nickel Catalysts. XPS analysis revealed the valence state for nickel element in the reduced catalysts ( Figure 3). In the Ni 2p XPS spectra, all reduced catalysts showed six peaks. The distinct peaks (yellow lines) with binding energy at about 856 and 874 eV corresponded to the characteristic peaks of the Ni 2p 3/2 and Ni 2p 1/2 spin-orbit splitting doublets, respectively. The generated two corresponding satellite peaks are located at 862 and 880 eV [68][69][70][71]. These results suggested that the nickel species in the catalysts were in the +2 oxidation state, which was derived from the unreduced Ni 2+ in 1 : 1 Ni-phyllosilicate. Meanwhile, the spinorbit splitting of Ni 2p 3/2 at 852 eV and Ni 2p 1/2 at 871 eV was the value characteristic of metallic Ni (purple lines). Due to Ni 2p 3/2 energy regions overlapping with La 3d 3/2 ones ( Figure S8) [72][73][74][75][76], the peaks from Ni 2p 1/2 core level were preferentially examined. The peak area at 871 eV gradually increased in the following order: Ni/ SiO 2 < Ni1.5La/SiO 2 < Ni3.0La/SiO 2 < Ni4.5La/SiO 2 . This phenomenon signified that the reduced Ni 0 on the catalyst surface gradually increased with the augment of lanthanum content, which coincided well with the result of reduction degree in TPR profiles. Figure S9 displays the catalytic activity investigated at increasing temperature from 550 to 850°C with 50°C temperature intervals. The conversions of reactant gas against temperature showed that at constant GHSV of 48,000 mL/g·h, all catalysts displayed the similar tendency, in which CH 4 and CO 2 conversions significantly increased with temperature. The CH 4 conversion steadily increased from~15 to~80% when the reaction temperature increased from 550 to 850°C, which was related to the strong endothermic property of DRM reaction. Over all tested catalysts, the higher CO 2 conversion was observed ( Figure S9B), suggesting that while the DRM reaction took place, the reverse water-gas shift reaction (RWGS) occurred simultaneously [77,78]. It might also suggest that the activation of methane was more difficult and energy consuming than CO 2 dissociation, which was very consistent with the DFT result [79,80] Figure 2: H 2 -TPR profile (a) of lanthanum-modified silica-supported nickel catalysts and peak area for peak 1, peak 2, and peak 3 as a function of La loading in the catalysts (b). 6 International Journal of Energy Research and CO 2 conversions became small, implying the greater contribution of CH 4 conversion [81]. Four catalysts exhibited similar conversion levels; for instance, the CH 4 conversion for Ni4.5La/SiO 2 was about 58.8% at 750°C, slightly higher than that of Ni1.5La/SiO 2 (~57.7%). The CO 2 conversion showed almost the same tendency. According to the characterization results (Table 2), although there was relatively remarkable discrepancy of surface area, the modified catalysts exhibited slight difference in average particle size (4.5-8.2 nm) and nickel dispersion (5.2-6.5%). These similar parameters indicated that approximately the same number of active sites contacted with reactant gas and participated in DRM reaction, resulting in the similar catalytic performance, especially under short reaction duration (e.g., 5 min in Figure S9). Based on the activity results of four catalysts at different reaction temperatures, the temperature of 750°C was chosen for further comparative studies, which was higher enough to investigate the aggregations of nickel nanoparticles and carbon depositions. In order to further comprehend the effect of lanthanum on catalytic performance, the stability test was conducted with the time on stream for 10 h. The modified Ni/SiO 2 catalysts with various lanthanum contents displayed different catalytic performance in terms of CH 4 and CO 2 conversions, H 2 /CO molar ratio, H 2 selectivity, and stability ( Figure 4). Due to the occurrence of RWGS, the H 2 / CO ratio was about 0.80, less than 1 (Figure 4(d)). In addition, all catalysts exhibited relatively high H 2 selectivity (~90%, Figure 4(c)).

Characterization of the Spent Catalyst.
To understand the variation in textural properties of the catalysts after 10 h in DRM reaction, four spent catalysts were analyzed by nitrogen sorption. Compared with the reduced counterparts, the adsorption-desorption isotherms for Ni1.5La/ SiO 2 , Ni3.0La/SiO 2 , and Ni4.5La/SiO 2 catalysts remained unchanged ( Figure S10A), indicating that the mesoporous structures were well preserved after DRM. However, the content for small-sized mesopore decreased ( Figure S1B and Figure S10B), which was likely due to the collapse and/or blockage of these pore structures. As a result, the S BET of three catalysts decreased markedly (Table 1), among which Ni1.5La/SiO 2 dropped the most. In detail, the S BET for Ni1.5La/SiO 2 decreased by 33%, from 76 to 51 m 2 /g. The severe decline in surface area was part of the reasons for the deactivation. On the other hand, the isotherm of spent Ni/SiO 2 was different with that of the reduced one, suggesting the obvious change in pore size    International Journal of Energy Research distribution. Specifically, the contents of large-sized mesopore and macropore increased greatly ( Figure S1B and Figure S10B), while the amount of small-sized mesopore decreased slightly. For this reason, Ni/SiO 2 retained relatively high surface area (Table 1). It was worth noting that the average pore diameters for spent catalysts were all larger than those of the reduced catalysts. It was because the decrease of small-sized mesopore correspondingly increased the percentage of large-sized pore, thus resulting in the enlargement of the average pore diameter.
Not only pore structure, the phase composition of the catalysts after DRM test also changed as well ( Figure 5). The spent Ni/SiO 2 catalyst exhibited a sharp diffraction peak at 26°that could be attributed to the graphitic carbon, which clearly confirmed that a large amounts of carbon formed on the Ni/SiO 2 surface. In contrast, the other three catalysts showed no characteristic peak of graphitic carbon at the corresponding position, indicative of the low carbon content or no carbon deposition. Intriguingly, the obvious diffraction peak of La 2 O 3 CO 3 (JCPDS 48-1113) appeared again on the spent Ni4.5La/SiO 2 , which was due to the insertion of CO 2 into La 2 O 3 under DRM reaction condition. The similar result was also observed on perovskite LaNiO 3 catalysts [82].
Lanthanum-containing catalysts showed obvious enlargement in nickel particle size. The average particle size for the respective spent catalyst was 7.4, 9.3, and 10.6 nm ( Table 2), suggesting the metal sintering during DRM process. The sintering resistance decreased in the order Ni1.5La/SiO 2 > Ni3.0La/SiO 2 > Ni4.5La/SiO 2 . The dramatic sintering on Ni3.0La/SiO 2 (7.4 vs. 9.3 nm) and Ni4.5La/ SiO 2 (8.2 vs. 10.6 nm) was mainly responsible for the deactivation. In contrast, Ni/SiO 2 exhibited strong resistance towards sintering with no appreciable particle enlargement. Combining the H 2 -TPR result with the above XRD result, it could be assumed that the presence of La on SiO 2 hindered the formation of nickel-phyllosilicate and weakened the interaction between metal and support; hence, it could not effectively prevent the sintering in DRM reaction.

International Journal of Energy Research
The amounts of carbon deposition on the spent catalyst surfaces were quantitatively determined by TG analysis. TG curves included three mass change stages ( Figure 6). The mass loss below 250°C was related to the evaporation of physically adsorbed water on the catalyst surface, while that at the range 500-750°C was attributed to the oxidization of deposited carbon [59,83]. The slight mass gain at 280-400°C was due to the oxidization of metallic Ni [84,85]. The total carbon amounts decreased in the following the order: Ni/SiO 2 (23.0%) > Ni1.5La/SiO 2 (2.8%) > Ni3.0La/ SiO 2 (0.9%) ≈ Ni4.5La/SiO 2 (1.0%). This result suggested that the presence of lanthanum substantially reduced the deposited carbon; what is more, the increment of La loading improved the carbon resistance. Considering the less carbon deposition on La-containing catalysts, it was reasonably assumed that the relatively severe shrinkage in S BET was mainly attributed to the collapse of pore channel rather than the pore blockage by carbon. In particular, although Ni3.0La/SiO 2 and Ni4.5La/SiO 2 catalysts exhibited excellent carbon resistance with negligible carbon deposition, these catalysts yet suffered relatively rapid deactivation compared to Ni1.5La/SiO 2 , which illustrated that the carbon amounts were not the dominant factor leading to the activity decline for these two catalysts.
The carbon species deposited on the spent catalysts were characterized by Raman's spectroscopy (Figure 7). Two remarkable peaks located at 1350 cm −1 (D-band) and 1580 cm −1 (G-band) were assigned to the carbon-carbon disorder-induced vibration and in-plane stretching vibration mode of carbon atoms that were bonded with sp 2 hybridization [86,87]. The intensity of D-band relative to G-band (I D /I G ) was used to verify the graphitization degree, in which higher I D /I G represented lower graphitization degree [88,89]. The I D /I G value gradually increased with lanthanum content (data in Figure 7), implying that the defect carbon predominated the carbon type. The Raman result validated that increasing lanthanum content was beneficial to improve the resistance to carbon deposition for nickelbased catalysts.
The changes in the morphology and particle distribution of the spent catalysts were studied by TEM images ( Figure S11). Some large nickel nanoparticles are scattered on the spent lanthanum-modified catalysts, and the particle size histogram (Fig. S11B1-D1) differed from the reduced counterparts. More specifically, the particle size intervals for Ni1.5La/SiO 2 and Ni3.0La/SiO 2 widened from 4-9 (Fig. S6B1, C1) to 3-12 nm (Fig. S11B1, C1), and that for Ni4.5La/SiO 2 changed from 3-11 (Fig. S6D1) to 5-12 nm (Fig. S11D1). Statistically, the average particle size was 6:2 ± 1:2, 6:6 ± 1:3, and 7:2 ± 1:1 nm for the respective catalysts ( Table 2). The enlargement in particle size indicated that the nickel particles were prone to sintering under DRM conditions. In contrast, Ni/SiO 2 catalyst     9 International Journal of Energy Research exhibited excellent sintering resistance, as confirmed by the almost same particle size distribution and average particle diameter (4:5 ± 0:7 nm) (Fig. S11A1) as the reduced catalyst (4:5 ± 0:8 nm). However, a large number of carbon nanotubes accumulated on Ni/SiO 2 surface (red circle, Figure S11A), suggesting its inferior carbon resistance. It should be emphasized that due to the inherent porous structure, the carbon nanotubes on the catalyst surface favored increasing the specific surface area and pore volume. Compared with the reduced Ni/SiO 2 catalyst (0.08 cm 3 /g, Table 1), the V p on spent Ni/SiO 2 became larger (0.13 cm 3 /g, Table 1), and the S BET remained almost unchanged. Contrary to Ni/SiO 2 , there were negligible carbon nanotubes on the surfaces of Ni1.5La/SiO 2 , Ni3.0La/SiO 2 , and Ni4.5La/SiO 2 . The absence of carbon nanotubes could not effectively offset the reduction in surface area and shrinkage in pore volume caused by the collapse or blockage of the pore structure; hence, these catalysts displayed obvious decrease in S BET and V p after DRM reaction (Table 2).
3.9. Long-Term Catalytic Tests. Long-term catalytic tests were carried out at 750°C for 34 h with the total flow rate of 80 mL/min to compare the carbon resistance of Ni1.5La/ SiO 2 and Ni/SiO 2 catalysts (Figure 8). Two catalysts exhibited relatively high initial CH 4 (~60%) and CO 2 conversion

10
International Journal of Energy Research (~65%) as well as almost same long-term stability. TG analysis showed that the carbon deposition on the Ni/SiO 2 was 21.6% (Figure 8(c)). In contrast, Ni1.5La/SiO 2 exhibited better carbon resistance that 14.9% of carbon was detected. The above observation corroborated that the addition of small amount of lanthanum could significantly suppress the carbon deposition and improve the long-term catalytic stability. TPH experiment was used to identify the types of carbon species on the spent Ni/SiO 2 and Ni1.5La/SiO 2 catalysts ( Figure 9). The peak in the range of 200-300°C was correlated to the hydrogenation of extremely active carbon [90,91]; the next one centered at 400-500°C was related with the amorphous carbon [92], and the last peak at 500-700°C was ascribed to the graphitic carbon with inert activity [93,94]. The TPH curves of spent Ni/SiO 2 catalyst contained an intensive peak at 590°C and a small peak at 410°C, representing the hydrogenation of the two types of carbon, and the graphitic carbon predominated the carbon species. It was worth noticing that the Ni1.5La/SiO 2 , which was also graphitic carbon-dominated, had relatively weak graphitic in nature. This could be testified from its lower decomposition temperature (~570°C) than Ni/SiO 2 (~590°C). More recently, the kinetic study [47] clearly verified that the addition of La into Ni/SiO 2 catalyst via "one-pot" method noticeably promoted CO 2 dissociation process, which improved the carbon resistance for NiLa/SiO 2 catalyst. Lanthanum provided a large number of active oxygen species to involve in the carbon removal, thus accelerating the gasification rate. This process resulted in a benign balance of carbon generation and removal, which could suppress the progressive accumulation of carbon to eventually form inert carbon.

Discussion
Characterization and activity evaluation results clearly indicated that the additional amount of lanthanum exerted significant influence on physiochemical properties and catalytic performance. Irrespective of content of La, the cat-alysts exhibited meso-and macroporous hierarchical pore structure ( Figure S1). Adding La promoter and increasing its content exerted negative influence on the nickel dispersion and relatively larger nickel particle formed on the modified catalysts (Table 2). Simultaneously, increasing La content hindered the formation of nickel-phyllosilicate and hence weakened the metal-support interaction (Figures 1 and 2), which was conducive to improve reduction degree (Table 2). However, the presence of La could not effectively suppress sintering. The La-modified catalysts displayed obvious sintering degree (10-30%, red diamond in Figure 10). In spite of this negative effect on resistance to sintering, La promoter had better carbon resistance (orange hexagon in Figure 10) and inhibited the graphitization process (pink pentagon in Figure 10). The stability for La-modified catalysts expressed by deactivation rate (gray histogram in Figure 10) vigorously proved the promotional effect of La on catalytic stability. From the overall image of carbon deposition, it was possible to correlate the amount of carbon deposit with the catalytic activity of the catalysts. For Ni/SiO 2 , the carbon deposition was mainly responsible for the rapid activity decay, as confirmed by the highest carbon content (23.0%) and graphitization degree (gray shadow in Figure 10). Lamodified catalysts encountered relatively dramatic decrease in surface area, attributable to blockage and/or collapse of porous channels, which hindered the active sites contacting with reactant gas. Therefore, the decrease in surface area was one of the factors leading to deactivation for Lamodified catalysts (blue star in Figure 10). The Ni3.0La/ SiO 2 and Ni4.5La/SiO 2 catalysts displayed lower amounts of carbon and graphitization degree; however, the deactivation rate was higher than Ni1.5La/SiO 2 , indicating that the carbon deposition was not the most important factor affecting activity decay. Apparently, apart from obvious decrease in surface area, Ni3.0La/SiO 2 and Ni4.5La/SiO 2 encountered relatively serious sintering (blue and pink shadow in Figure 10). Hence, for Ni3.0La/SiO 2 and Ni4.5La/SiO 2 , sintering of nickel nanoparticle was another significant factor affecting deactivation. Concerning Ni1.5La/SiO 2 , besides the decrease in surface area, the carbon deposition with relatively higher graphitization degree was also responsible for the activity decay (red shadow in Figure 10). It was concluded that addition of La could improve the resistance towards carbon deposition; however, the higher La content did not imply better stability. Therefore, the suitable La content was the key point to significantly improve catalytic performance for nickel-based catalysts.
The comparison between the obtained results in the present work and similar previous works on La-modified catalysts was summarized ( Table 3). The different performance was probably related to the different reaction parameters (e.g., reaction temperature and feed gas composition). First, the catalysts with high nickel loading (>10 wt%) generally showed much higher deactivation rates. Second, the methane conversion obtained with the one (Ni@SiO 2 ) with core-shell structure [95] at same reaction temperature (750°C) and gas flow rate (48,000 mL/g•h) was similar to

11
International Journal of Energy Research our result. Third, the addition of lanthanum via the one-pot process in our work had superiority in conversion compared with the poor catalytic activity for Ni10%La 2 O 3 /KIT-6 [96], which was prepared with impregnation method. However, it should be important to note that when the ordered mesoporous SBA-15 was chosen as support and the diluted gas which favored lower carbon deposition was used as feeding gas, the reactivity and stability were significantly higher [97,98]. Most noteworthy, the best example, Ni/40La-SBA16 catalyst, exhibited the highest methane conversion at 700°C [38]; however, it had twice deactivation rate compared with our result, pointing to its higher carbon deposition (4.73%) than our case (2.8%).

Conclusion
Lanthanum-modified hierarchical porous silica-supported nickel catalysts were synthesized by "one-pot" method and evaluated for the CO 2 reforming of CH 4 at 750°C and GHSV = 48,000 mL/g•h. All catalysts possessed hierarchical porous structure, including small-sized mesopore at about 3 nm, large-sized mesopore at 40 nm, and macropore larger than 50 nm. The addition of lanthanum exerted negative influence on the formation of 1 : 1 Niphyllosilicate on calcined catalysts, and the content of Ni-phyllosilicate gradually deceased with increasing La loading. The presence of La did not effectively suppress the metal sintering. Better carbon resistance was obtained after the modification of La, which significantly improved with the increase of La loading. The main positive effect of lanthanum was to inhibit the formation of carbon deposits. The catalyst stability followed the order of Ni/ SiO 2 < Ni3.0La/SiO 2 < Ni4.5La/SiO 2 < Ni1.5La/SiO 2 , which indicated that the catalysts with the highest La content did not imply a better stability. Ni1.5La/SiO 2 exhibited the highest activity ðCH 4 conversion = 61:3%, CO 2 conversion = 68:9%, and H 2 /CO ratio = 0:80Þ and demonstrated the highest stability within 10 h of longterm stability test with the decay rate of 0.33%/h. The deactivation for Ni1.5La/SiO 2 was mainly caused by the severe decrease in surface area and carbon deposition. As for Ni3.0La/SiO 2 and Ni4.5La/SiO 2 catalysts, the sharp  Table 2). ΔSurface area was the decrease in surface area, which was determined as the difference in S BET for the reduced and spent catalysts (Table 1).

Data Availability
The physisorption, HAADF-TEM, TEM, XPS, and CO2-TPD used to support the findings of this study are included within the supplementary information files.

Conflicts of Interest
The authors declare that they have no conflicts of interest.