Relationship Between the Pore Structure of Mesoporous Silica Supports and the Activity of Nickel Nanocatalysts in the CO 2 Reforming of Methane

: The question remains over the role of the pore structure of the support material on the catalytic behaviour of Ni catalysts during the CO 2 /dry reforming of methane (DRM). For this reason, a series of mesoporous materials with different pore structures, namely MCM-41, KIT-6, tri-modal porous silica (TMS), SBA-15 and mesostructured cellular foams (MCFs) were synthesised via hydrothermal synthesis methods and further impregnated with 15 wt.% NiO (11.8 wt.% Ni). It was observed that synthesised TMS is a promising catalyst support for DRM as Ni/TMS gave the highest activity and stability among these materials as well as the Ni catalysts supported on classic ordered mesoporous silicates support reported in the literature at the relatively low temperature (700 °C). On the other hand, Ni supported on CMC-41 exhibited the lowest activity among them. To understand the reason for this difference, the physicochemical properties of these materials were characterised in detail. The results show that the thickness of the silica wall and the pore size of the support material play a critical role in the catalytic activity of Ni catalysts in the CO 2 reforming of methane.


Introduction
The discovery of Mobil Crystalline Materials, such as MCM-41 in 1992 by Mobil Oil Corporation scientists [1], has initiated intensive attention in the development of mesoporous silica materials and their applications in research and industry [2][3][4][5][6][7][8][9][10][11][12][13][14]. Afterwards, a range of highly ordered mesoporous silica materials such as SBA-15 and KIT-6 have been created in 1998 and 2003, respectively [15,16]. The unique physicochemical properties of these materials (such as excellent thermostability and high specific surface areas) make them suitable supports for catalysts that are operable at high temperatures. Among the relevant high-temperature reactions, the catalytic CO2 (dry) reforming of methane (DRM) is an interesting process that has recently attracted substantial attention due to its environmental and economic benefits [17]. DRM converts two atmospheric pollutant greenhouse gases (CH4 and CO2) into the versatile intermediate product syngas (CO + H2), as is shown in Equation (1) [18][19][20][21][22][23][24]. Syngas can be used for the manufacturing of several substantial high-value-added materials such as synthetic fuels and methanol [25][26][27]: CO2(g) + CH4(g) ⇌ 2CO(g) + 2H2(g) (ΔH° = 59 kCal/mol) Nickel-based catalysts are typically used in the CO2 reforming of methane due to widespread availability, high activity and low price [28][29][30]. However, DRM is strongly endothermic and requires high temperatures (>700 °C) [31]. Therefore, sintering of the nickel would be likely to happen if the reaction takes place above its Tammann temperature (591 °C) [32]. Among the proposed solutions for this drawback, utilising the confinement consequence of the ordered meso pores in the support material to avoid the Ni particles from sintering is a very attractive option [19,[33][34][35][36]. Although the ordered mesoporous materials have many distinct advantages, they also have some disadvantages as support for Ni-based catalysts when employed in the CO2 reforming of methane. The greatest drawback of these materials is that they have small pores, which are not able to provide sufficient space for the reactants with an adequate number of Ni active centres to drive the reaction [27,[36][37][38][39]. As a result, the catalytic capability of Ni catalysts supported on these materials can be negatively influenced. A range of approaches has been developed to overcome this obstacle. Optimising the pore size and the development of hierarchical porous materials including different pore size systems have received special attention in recent years [9][10][11][12][13][14]37,38,[40][41][42][43][44][45][46][47][48][49][50][51][52][53]. This is because the hierarchical porous structure of the silica supports can not only increase the mass transport of the reactants to the Ni active sites, but also the mesoporous matrix hinders the thermal sintering of the Ni particles by the confinement effect [24,[54][55][56].
Although specific consideration has been paid to the development of hierarchical porous materials, the role of the pore size and the thickness of the pore walls of the catalyst support on the activity of nickel for the CO2 reforming of methane has yet to be addressed. For this purpose, this work focuses specifically on the correlation between the pore properties of a series of mesoporous supports (including tri-modal porous silica (TMS), mesostructured cellular foam (MCF), MCM-41, KIT-6 and SBA-15) and the catalytic activity of the nickel supported on these materials in the CO2 reforming of methane.

Catalytic Tests
A series of the Ni-based nanocatalysts over various porous siliceous supports, namely SBA-15, TMS, KIT-6, MCF and MCM-41 were studied for the CO2 reforming of methane. The methane and CO2 conversions as a function of time on stream (TOS) obtained are compared in Figure 1. To discover the role of support on activity and stability of catalysts, the catalytic performance of Ni supported on different ordered mesoporous silicates in DRM which have been reported in the literature along with the catalysts tested in this study is summarised in Table 1. Due to variances in preparation method, Ni% and reaction conditions such as gas hour space velocity (GHSV), pressure and temperature, it is difficult to compare support performances precisely. However, in this Table, the initial and final CH4 conversion (%) of CO2 reforming of methane are compared for DRM reaction which reported at ambient pressure and 700 °C over catalysts those were prepared via impregnation method.  [63] Although all catalysts had the same composition (Ni/SiO2), they showed significant differences in catalytic activity, as illustrated in Figure 1 and Table 1. As seen in Figure 1 and Table 1, the shape and size of pores of support have an important effect on both, the activity and the stability of Ni catalysts. Ni supported on TMS, SBA-15 and MCF supports exhibited higher initial catalytic activity arising from the higher accessibility of the Ni active sites. On the other hand, Ni supported on the ordered mesoporous silicas (SBA-15, KIT-6 and MCM-41) showed higher stability, which can be attributed to the higher sintering and coking resistance of catalyst due to the confinement effect of Ni particles inside silica pores [33,34,36,38,64].
This observation indicates that the morphology and pore structure of the supports significantly affected the catalytic activity and stability. Among the catalysts tested in this work and reported in the literature, the Ni/TMS catalyst showed the highest activity and the most stable catalytic performance. The reason for the high activity of the Ni/TMS catalyst has been explained in an earlier report by the authors [24]. It was found that the existence of 2nd and 3rd mode pores in TMS support provided more available active Ni sites to CH4 and CO2 gases compare to other silicate supports. The high accessible active sites in TMS support can lead to the high activity observed for Ni supported on this support [24].
On the contrary, the Ni/MCM-41catalyst showed relatively poor performance among all the catalysts in the CO2 reforming of methane. This observation created a question as to why Ni supported on a well-ordered mesoporous material (MCM-41) showed a significantly lower activity compared to Ni supported on SBA-15, KIT-6, TMS and MCF. The low activity of the Ni/MCM-41 catalyst compared with the Ni catalysts supported on classic ordered mesoporous silica supports has also been widely reported in the literature [34,36,39,[63][64][65][66]. Therefore, the properties of the catalysts were carefully analysed to elucidate the reason for this interesting behaviour. Figure 2 shows pore size distribution curves and N2 adsorption-desorption isotherms of the different silica materials. It can be seen from Figure 2a that the ordered mesoporous materials SBA-15, KIT-6 and MCM-41 show narrow pore size distributions, centred at 7.3 nm, 6.3 nm and 2.6 nm, respectively. In contrast, the pore size distribution for MCF sample showed broad peaks centred around 18.9 nm and 111.0 nm. The pore size distribution curve of TMS showed three characteristic peaks at 8.1 nm, 24.6 nm and 89.3 nm. A summary of the results obtained from the N2 adsorptiondesorption measurements is reported in Table 2. The following order of BET surface areas was observed for the different materials: MCM-41 > KIT-6 > SBA-15> TMS> MCF. The average pore size decreases in the following order: TMS > MCF > SBA-15 > KIT-6 > MCM-41. Table 2 shows that the average pore sizes of all the siliceous materials were located in the size range of 3.75-22.66 nm, and are therefore classified as mesopores according to the IUPAC definition [67]. As can be seen from the N2 adsorption-desorption curves in Figure 2b, the materials demonstrate hysteresis loops with different shapes. The ordered mesoporous materials MCM-41, KIT-6 and SBA-15 exhibit H1 hysteresis loops, which is typical of materials with independent pores [67]. The existence of an H4 hysteresis loop for the MCF sample provides evidence of narrow slit-like pores [24,67]. On the other hand, TMS showed a non-uniform pore structure, with an H3 hysteresis loop at low relative pressures (P/Po) created by the capillary condensation of N2 molecules highlighting the presence of cage-like mesopores, while the H2 hysteresis loop at P/Po = 0.81 is a sign of secondary interstitial multicellular pores [24]. In spite of these differences, it is clear that all materials exhibited type IV isotherms, demonstrating that all of the samples are mesoporous in nature.

SAXS Characterisation
The small-angle XRD patterns of the synthesised silica materials are displayed in Figure 3. The observed 2θ values centred at ~0.9°, ~1.5° and ~1.8° in the patterns of SBA-15 and TMS samples correspond to the (100), (110) and (200) reflections, respectively, associated with p6mm hexagonal structures [15,68,69]. MCM-41 also gives Bragg peaks at the 2θ value of 2.3°, 4.0° and 4.6° which can be indexed to (100), (110) and (200) reflections, respectively, associated with a p6mm hexagonal structure [70][71][72]. For KIT-6, three peaks are indexed as (211), (220) and (420) reflections [16]. On the contrary, no such order was found for the MCF material, which can be attributed to the absence of an ordered porous structure or the poor distribution of the porous structure. These observations are in line with the findings from pore size distribution ( Figure 2a) and N2 adsorption-desorption isotherm studies (Figure 2b).

XRD Characterisation
The wide-angle XRD patterns of samples before and after reduction are presented in

XPS Characterisation
XPS was used to analyse the surface composition and elemental valence states of the calcined catalysts before reduction. Ni 2p3/2 and Ni 2p1/2 XPS spectra of the catalysts are illustrated in Figure  5a. The Ni 2p3/2 peaks in all samples are centred at a binding energy of 854.0 eV with a shoulder at 856.0 eV and a satellite peak at 861.1 eV which corresponds to NiO [20,[73][74][75] rather than metallic nickel [76]. In addition, the Ni 2p1/2 peaks in all samples are present at a binding energy of 872.0 eV, providing further evidence of the presence of NiO [20,[73][74][75]. As shown in Figure 5b, the Si 2p binding energy of all catalysts show a characteristic peak at 103.3-103.6 eV attributed to SiO2 [75], suggesting no interaction of NiO particles with support. From the XRD patterns and XPS profiles of these catalysts, it can be concluded that there were no obvious chemical interactions between the nickel and its support, and any significant difference in electronic structure among all catalysts. The percentage of Ni on the surface of the reduced Ni/SiO2 materials were evaluated using XPS, and the results are shown in Table 3. The amount of Ni on the surface of different catalysts was found to be inversely proportional to the average pore size of the ordered mesoporous silica materials (MCM-41, KIT-6 and SBA-15). As later confirmed by TEM study, the nickel particles can be more easily accommodated within the larger pores of these materials compared to the smaller pores. The smaller pore size of MCM-41 may hinder the dispersion of Ni particles inside the pores. Ni particles can easily infiltrate into the larger mesoporous channels during the impregnation process [34]. However, in the case of Ni/MCF and Ni/TMS catalysts, a different trend was observed. It seems that there was no direct relationship between the % Ni in the surface of these materials and their pore size. This is most likely due to their wide range of pore diameters.

FTIR Characterisation of the Siliceous Materials
The FTIR spectra of supports and catalysts in the region between 400 and 4000 cm −1 are presented in Figure 6. No differences were detected in the spectra of the supports (Figure 6a). The FTIR spectra of all supports show infrared absorption peaks at 830-870 cm −1 , 950-980 cm −1 , 1090-1120 cm −1 , 1600 cm −1 , 3400 cm −1 and 3750 cm −1 , which are assigned to the Si-O-Si bending vibrations, Si-OH stretching vibrations of non-bridging oxygen, asymmetric stretching, deformation vibrations of adsorbed water molecules, stretching of adsorbed water molecules and OH vibrations of free silalon groups [77,78], respectively. The FTIR spectra of the Ni/SiO2 catalysts are illustrated in Figure 6b. The addition of Ni into the supports generated a new peak at 482 cm −1 (marked by a circle), which is attributed to the stretching vibration of bulk NiO [78]. With the exception of the NiO peak, all FTIR spectra of the Ni/SiO2 catalysts were found to be very similar to that of the pure supports, indicating that there was no noticeable chemical interaction between the nickel and the support. This result is in line with the XRD and XPS results. However, it should also be noted that the narrow peak at 3750 cm −1 disappeared in the FTIR spectrum of Ni/MCM-41. Thus, it can be concluded that in the case of MCM-41 the addition of NiO led to the fixing of the free silanol groups in the Si-O network, impeding the OH vibrations of free silanol groups [77] due to the presence of disordered narrow pore networks. The microstructure of TMS was discussed in detail in an earlier report by the authors [24]. From the HRTEM images of the MCM-41, KIT-6 and SBA-15, it can be seen that the long ordered parallel pores have been retained after the high-temperature calcination. The pore sizes of the siliceous materials were estimated by TEM analysis using ImageJ software. The average pore size of MCM-41, KIT-6 and SBA-15 were 3.3 nm, 6.9 nm and 9.4 nm, respectively. These observations are in line with the results from the N2 adsorption-desorption study (Table 1). MCF demonstrated an entirely different texture compared with the above materials, having uniform cells with an average diameter of 7.3 nm. The average cell diameter of the MCF synthesised under the present conditions is smaller than what has been reported in the literature (MCFs with cell diameters of 22 nm and 32 nm) [79]. It can be seen from Figure 7 that there are interconnected pores in the MCF, which could be the reason for their broad distribution of pores in Figure 2a. The TEM images of the reduced Ni/SiO2 materials are presented in Figure 8. The average pore wall thickness and particle size distribution of metal in these materials were determined by evaluating ca. 100 particles using ImageJ software and the results are shown in Figure 9. The following orders were observed for catalysts with different supports:

TEM of the Siliceous Materials
The order of average Ni particle sizes: Considerable differences were observed in the Ni nanoparticle size distribution patterns for the different catalysts, owing to the pore structures of supports. In the case of Ni/SBA-15, Ni/TMS and Ni/MCF catalysts some large nanoparticles were formed outside of the pores of these supports. However the majority of Ni particles have sizes that are smaller than these pores. The existence of these small-sized Ni particles provides evidence that these particles were trapped in the pore channels of the support materials. Such confinement of nickel particles has previously been shown to increase the thermal stability of Ni nanoparticles [80][81][82][83][84][85][86][87][88][89]. In the present work, it is concluded that the confinement effect prevented the occurrence of any observable sintering, even though the reduction temperature (700 °C) was much higher than the Tammann temperature of nickel (591 °C) [32]. An interesting phenomenon was detected for the Ni particle size distributions for the Ni/MCF, Ni/MCM-41 and Ni/KIT-6. As can be observed in Figure 9, the Ni particles in these materials presented a bimodal size distribution. In these catalysts, a portion of Ni particles has a size smaller than the pore size of their support material, which can be attributed to the confinement effect of pore walls. Besides this, another portion of Ni particles is larger than the pore size of their support materials, indicating that their size was not controlled by the confinement effect of the pore walls. This phenomenon could be explained by the presence of the two distinguishable types of nickel sites on these supports, namely the nickel particles inside and outside of pores of the support materials. It is therefore apparent that the size and position of nickel sites strongly depend on the size and wall thickness of the support material's pores. The ratio of Ni species inside the pores, which can be affected by the confinement effect, significantly depended on the size of the pore of the support. The percentage of inner Ni species was found to increase as the pore size of support increased, as can be expected. Figure 9. Ni particle size distribution, average Ni particle size, average pore wall thickness and average pore size in the catalysts (average pore size was measured by N2 adsorption-desorption method).
In the case of Ni/MCM-41, the majority of Ni particles have sizes that are larger than the pores of support materials, which means that most of the Ni particles were located outside of these pores. This is most likely due to the narrow channels of MCM-41 (with an average size of 3.8 nm) which reduces the amount of nickel nitrate solution that was delivered into the pore channels during the impregnation process [81].
As a follow up from closer examination by HRTEM and EDS-mapping (Figure 10), another interesting phenomenon was observed for the Ni/MCM-41 material. It was found that the mechanical strength of the pore walls in MCM-41 is not high enough to effectively prevent the thermal sintering of the Ni particles entrapped inside the porous channel at temperatures higher than the Tammann temperature of nickel [32]. HRTEM investigation showed the majority of nickel particles were located on the outer surface of the MCM-41, which does not seem to be greatly affected by the confinement effect of the pore walls.  Figure 11 presents high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) images of Ni particles inside the pores of the ordered supports, namely SBA-15, KIT-6 and MCM-41. HAADF STEM shows atomic number contrast for high scattering angles of the electrons; therefore Ni particles are clearly visible in the Z-contrast image [90]. As proven by HAADF-STEM imaging, although MCM-41 has a smaller pore size compared to SBA-15 and KIT-6 supports, its silica walls were not strong enough to resist the agglomeration of Ni crystals inside the pores during sintering at temperatures higher than the Tammann temperature of nickel [32]. This finding is in line with previous literature data that MCM-41 has the relatively smooth pore wall surfaces as compared to the SBA-15 material [91,92]. The schematic drawing for this behaviour is presented in Figure 12, which compares the confinement effect on the Ni nanoparticles trapped inside the channels of MCM-41 with those trapped inside the channels of other supports. Another important observation from the TEM studies is that a large number of the pore mouths were partially filled by the growth of the Ni particles inside the pores of MCM-41. Therefore, the blockage of pores (especially the small pores) can cause a decrease in the amount of accessible Ni sites to gas molecules. It has been shown in the literature that blocking the active sites can be one of the major reasons why Ni/MCM-41 exhibited the lowest activity among the other tested catalysts [93][94][95][96].
Based on the obtained results of this work, it is possible to have a clearer view of the effect of the support morphology on the performance of Ni-based porous silica catalysts in the CO2 reforming of methane. The results of FTIR, XRD, SAXS and XPS analyses did not show any distinguishable chemical interaction between the nickel and the supports that were tested in this study. Nevertheless, it was clear that the pore size and the thickness of the silica walls greatly influence the amount of accessible Ni for the reactants and the role of the mesoporous supports in the prevention of metallic Ni particles from sintering through the confinement effect at temperatures higher than the Tammann temperature of nickel. With a particular focus on Ni/MCM-41, it seems that the nickel nitrate precursor solution was not able to transfer smoothly into the narrow channels of MCM-41 during impregnation. Consequently, a large number of Ni particles remained on the external surface of the mesoporous channels, instead of embedding inside the pores of the support. As a result, the bimodal size distribution of nickel particles in Ni/MCM-41 was created. Besides this, the slim walls of MCM-41 are not able to effectively confine the growth of the Ni particles which were able to grow inside the pores of MCM-41 during the thermal sintering at high temperatures. This led to the partial blocking of the pore channels, which limits the availability of Ni sites inside pores to the reactants gases through the pore channels of MCM-41.

Synthesis of the Silica Supports
A family of mesoporous silica structures was synthesised through the hydrothermal assisted sol-gel method. The SBA-15, TMS and MCF materials were prepared via the methods reported previously by the authors [24], while the siliceous KIT-6 [16] and MCM-41 [1,70] materials were prepared according to pioneering procedures reported in the literature.
A schematic graphic of the typical synthesis route for the materials is presented in Figure 13.

Synthesis of the Ni Catalysts on the Supports
The wet impregnation method was employed to prepare the Nickel catalysts supported on the synthesised mesoporous silicas. The required amount of nickel (II) nitrate hexahydrate was dissolved in Milli-Q water while stirring. The siliceous supports were impregnated with this aqueous solution at ambient conditions overnight. The solutions were then dried at 110 °C overnight and then calcined in air at 600 °C for 4 h (with a ramp rate of 5 °C/min). Based on previous studies by the author [18,19,[22][23][24], the composition of the catalysts was kept constant at 15 wt.% NiO (11.8 wt.% Ni) with respect to silica with the introduction of an appropriate amount of the nickel precursor. For example, 1.84 g Ni(NO3)2.H2O6 was added to 5.00 g the synthesised mesoporous silicas.

Characterisation Techniques
BET surface areas and N2 adsorption-desorption isotherms of the synthesised materials were determined using an ASAP 2020 instrument Micromeritics at 0.003 mmHg and −196 °C. Before analysis, the samples were degassed under 0.350 mmHg for 30 min at 25 °C followed by fast-mode degassing at 280 °C overnight.
The powder X-ray diffraction (XRD) analyses were performed using a Rigaku diffractometer with Cu Kα radiation (λ = 0.1540 nm) as the source.
Small-angle X-ray scattering (SAXS) patterns of the materials were recorded using a D8 Advance X-ray diffractometer (Bruker) equipped with a Nanostar Hi Star 2D area detector using Cu-Kα1 radiation (λ = 0.15406 nm) with a voltage of 40 kV and a current of 40 mA.
Transmission electron microscopy (TEM) studies were undertaken using a JEM-2100F operating at an accelerating voltage of 200 kV. TEM instrument equipped with a Gatan Orius SC1000 CCD camera.
A K-5 Alpha XPS instrument (Thermo) at a pressure less than 10 −8 Torr was used to study the Xray photoelectron spectroscopy (XPS) of the surface of the catalysts. The survey scan and Si 3d, Ni 2p, and O 1s core level spectra were recorded using Al Kα radiation. All binding energies were calibrated with reference to the C 1s peak at 284.6 eV arising from the hydrocarbons adsorbed from the atmosphere.
Fourier transform infrared spectroscopy (FTIR) was performed using a Frontier FT-IR/FIR spectrometer (Perkin Elmer) in the range between 400 and 4000 cm −1 .

Catalytic Activity Test
The catalysts were pre-reduced in situ in a mixed flow of hydrogen and helium (10:40 mL min −1 ) at 700 °C for 2 h, which is henceforth referred to as reduced catalysts. The performance of all catalysts was tested at the same conditions over the reduced catalysts under a gas hourly space velocity (GHSV) of 52,000 mL g −1 h −1 at 700 °C and atmospheric pressure in a fixed-bed continuous flow quartz reactor. The catalyst mass was 100 mg. The catalyst particle size was kept within the range of 500-710 μm. It had been found that under these conditions there were no any internal or external mass transfer limitations (based on the calculation of the Weisz criterion [96]) [19,24,97]. The product gas mixture was analysed by on-line gas chromatography (gas chromatography-PerkinElmer Clarus 580 GC) equipped with a silica packed column and a thermal conductivity detector.

Conclusions
From this study, it could be seen that Ni supported on TMS showed the highest activity among all the catalysts that were tested as well as the Ni catalysts supported on classic ordered mesoporous silicates support reported in the literature under similar conditions. While Ni supported on MCM-41 exhibited the least activity among all examined catalysts. The results presented in the present study showed that the morphology of mesoporous supports, in particular, the thickness and the pore size of the silica wall play a substantial role in determining the catalytic performance of impregnated Ni catalysts on the mesoporous silica supports towards the CO2 reforming of methane. It has been shown that the small pore diameters and thin silica walls in the support material were less favoured for the catalytic activity of impregnated nickel catalysts, due to their inability to prevent nickel sintering and through reducing the accessibility of the reactants to Ni particles embedded inside the pores and removal of products. These findings will help researchers to develop more active supported Ni catalysts which can find wide application in heterogeneous catalysis such as the catalytic CO2 reforming of methane.

Conflicts of Interest:
There are no conflicts to declare.