Synergy Effects of Cobalt Oxides on Ni/Co-Embedded Al2O3 for Hydrogen-Rich Syngas Production by Steam Reforming of Propane

The synergetic effects of Co oxides on the Ni/CoAl (NCA) catalysts were observed at an optimal molar ratio of Al/Co = 2 (NCA(2)) due to the partial formations of thermally stable spinel CoAl2O4 phases for the steam reforming of propane (SRP). The optimal content of the spinel CoAl2O4 phases on the NCA(2) was responsible for the formation of the relatively active oxophilic metallic Co nanoparticles with a smaller amount of less active NiAl2O4 on the surfaces by preserving the relative amount of metallic Co of 68% and 52% in the reduced and used catalysts, which enhanced the catalytic activity and stability with the largest specific rate of 1.37 C3H8/(Ni + Co)h−1 among the tested NCA catalysts. The larger or smaller amounts of Co metal on the less active NCA mainly caused the preferential formation of larger aggregated Ni nanoparticles ~16 nm in size due to their weaker interactions, or induced the smaller formations of active metal phases by selectively forming the spinel NiAl2O4 phases with ~60% in the NCA(4), resulting in a fast deactivation.


Introduction
Hydrogen has been remarkably spotlighted recently as clean energy due to its high efficiency with low emissions of greenhouse gases through a fuel cell application compared to conventional fossil fuel-based energies [1]. Moreover, since the emission regulations are getting stricter in regard to not only power generation but also vehicles, the needs of clean fueled vehicles, such as electric or fuel cell vehicles, are more attractive compared to the conventional fossil fuel combusting vehicles. For the fuel cell applications, hydrogen can be generally produced from a fuel processing technology composed of the following units; desulfurization, steam reforming, water-gas shift and preferential oxidation of carbon monoxide, which could convert hydrocarbons to hydrogen-rich gas with sparse CO concentration. Among those hydrocarbons, propane can be one of the promising feedstocks for the fuel processor since it can be easily liquified at room temperature, resulting in easy shipping and transportation compared to other gaseous feedstocks [2]. Therefore, the existing infrastructures can be directly utilized to supply the feedstock for the fuel processor applications. The steam reforming of light alkanes, which is the most economical way to produce the industrially applicable hydrogen, typically takes about 60% market share among overall hydrogen production plants due to its relatively stable syngas production as well as high H 2 /CO ratio compared to other various reforming processes [3]. Steam reforming of propane (SRP) is mainly composed of the reactions as listed below.
Steam reforming of propane (SRP): C 3 H 8 + 3H 2 O → 3CO + 7H 2 (∆H • = 497.7 kJ/mol) (1)  The increased amounts of cobalt metal on the NCA catalysts caused the decreased surface area due to the increased crystallite sizes and small contributions of mesoporous Al 2 O 3 , which were calculated by using the main diffraction peaks detected in the range of 44.2 • for Co (111)-44.5 • for Ni (111) with the help of the Scherrer equation as shown in Figure 1 and Figure S2. The metallic Ni crystallite sizes were found to be 6.5 nm on NCA(4) and 14.6 nm on NCA(0.5), which were decreased with an increase of the Al/Co ratio. It is worth noting that NCA(3) and NCA(4) had the smallest crystallite sizes of 6.3-6.5 nm (dispersion of 3.0-3.6%) among the reduced NCA catalysts. In addition, the characteristic diffraction peaks assigned to the spinel-type NiAl 2 O 4 , Co 3 O 4 and CoAl 2 O 4 phases were observed at 31 • , 37 • , 45 • , 56 • , 59 • and 65 • , which are scarcely classified due to their overlapped positions with the same crystal structures of AB 2 O 4 [7,22,[29][30][31][32]. Diffraction peaks assigned to the γ-Al 2 O 3 phases were also observed, with their smaller peaks at 38 • , 46 • and 67 • , which also suggests that the Al 2 O 3 seems to exist mainly in the spinel-types such as NiAl 2 O 4 and CoAl 2 O 4 in the NCA catalysts [12,33]. Interestingly, the peaks of the main diffraction at 44.2 • -44.5 • were shifted to the higher two theta values and broadened with a decrease of cobalt contents, especially on the NCA(3) and NCA(4), due to the smaller crystallite formations corresponding to larger surface areas. Even though those peaks were hard to be assigned to the different crystalline phases, the transformations of those major crystallite structures were observed from the metallic Co and Ni to the spinel-type NiAl 2 O 4 and CoAl 2 O 4 phases at higher Al/Co ratios on the reduced NCA catalysts ( Figure 1A). After the SRP reaction at 650 • C, the used NCA catalysts showed increased metallic crystallite sizes ( Figure 1B) due to the aggregations of those nanoparticles with their sizes of 7.3-16.2 nm from 6.3-14.6 nm, as summarized in Table 1. The extents of the aggregations were not significant on the NCA catalysts with a Al/Co ratio ≥2, such as the NCA(3) and NCA(4), while the much bigger crystallites were formed on the used NCA(1) and NCA(0.5) to 12.9 and 16.2 nm in size, respectively. The moderate aggregations of Ni nanoparticles on the NCA(2) (from 6.9 to 8.7 nm with dispersion of 2.5%) were mainly attributed to the formations of strongly interacted Co 3 O 4 -Al 2 O 3 particles with a moderate surface area (64 m 2 /g), which can be more beneficial for increasing the catalytic activity and stability.
Catalysts 2020, 10, x FOR PEER REVIEW 4 of 13 with a Al/Co ratio ≥2, such as the NCA(3) and NCA(4), while the much bigger crystallites were formed on the used NCA(1) and NCA(0.5) to 12.9 and 16.2 nm in size, respectively. The moderate aggregations of Ni nanoparticles on the NCA(2) (from 6.9 to 8.7 nm with dispersion of 2.5%) were mainly attributed to the formations of strongly interacted Co3O4-Al2O3 particles with a moderate surface area (64 m 2 /g), which can be more beneficial for increasing the catalytic activity and stability.

Reduction Properties of NCA Catalysts and Catalytic Activities of CA Supports Themselves
As shown in Figure 2A, the reduction behaviors of the NCA catalysts were measured by temperature programmed reduction (TPR) analysis. With lower contents of cobalt species on the NCA(3) and NCA(4), one reduction peak at the respective temperatures of 842 and 854 °C was clearly observed. In general, cobalt-promoted Ni-based catalysts show the lower temperature shifts of those reduction peaks by facilitating an easy reduction of Ni/Al2O3 [12] as observed in Figure 2A from 854 °C on the NCA(4) to 833 °C on the NCA(0.5). In case of the Al/Co ≥ 3, the higher temperature reduction peaks above 800 °C were mainly attributed to the reduction of solid solutions, such as the spinel CoAl2O4 and NiAl2O4 (or strongly interacted NiO species), which were scarcely reduced below 700 °C to metallic Co and Ni due to their stronger metal-support interactions [3,12,25]. With the decreases of Al/Co ratios below two, many additional reduction peaks were observed with broader reduction patterns below 800 °C, which originated from the stepwise reductions of spinel-type oxides [12,13,19,31] such as CoAl2O4, Co3O4 and NiAl2O4. In detail, the apparent reduction peaks of cobalt oxides through stepwise reductions (Co3O4 → CoO → Co) were clearly observed below 700 °C, which suggests the preferential segregations of the active cobalt species were on the outer surfaces of the NCA(0.5) and NCA(1) compared to the supports themselves (CA(0.5) and CA(1)) which have lower activity for SRP reactions as well. Interestingly, on the NCA(2), the smaller segregated cobalt species (below ~700 °C) as well as spinel-type metal oxides (at ~839 °C) were observed, which were responsible for an enhanced catalytic activity and stability through the formation of thermally stable mixed metal interfaces.
The catalytic activity of the CA supports and NCA(1) with an impregnated Ni/Al2O3 and CoAl2O4 reference catalysts were preliminarily measured at a lower SV = 100,800 L/(kgcat•h), and the results are displayed in Figure 2B and summarized in Table S1. The CA supports themselves were active for the SRP reaction without any Ni species, which suggests that the cobalt nanoparticles were active for the reaction [3,25,34]. It was clear that the CA supports with the molar ratio of Al/Co ≤ 2 were also active for SRP reaction, and their higher and stable activities were characterized compared

Reduction Properties of NCA Catalysts and Catalytic Activities of CA Supports Themselves
As shown in Figure 2A, the reduction behaviors of the NCA catalysts were measured by temperature programmed reduction (TPR) analysis. With lower contents of cobalt species on the NCA(3) and NCA(4), one reduction peak at the respective temperatures of 842 and 854 • C was clearly observed. In general, cobalt-promoted Ni-based catalysts show the lower temperature shifts of those reduction peaks by facilitating an easy reduction of Ni/Al 2 O 3 [12] as observed in Figure 2A from 854 • C on the NCA(4) to 833 • C on the NCA(0.5). In case of the Al/Co ≥ 3, the higher temperature reduction peaks above 800 • C were mainly attributed to the reduction of solid solutions, such as the spinel CoAl 2 O 4 and NiAl 2 O 4 (or strongly interacted NiO species), which were scarcely reduced below 700 • C to metallic Co and Ni due to their stronger metal-support interactions [3,12,25]. With the decreases of Al/Co ratios below two, many additional reduction peaks were observed with broader reduction patterns below 800 • C, which originated from the stepwise reductions of spinel-type oxides [12,13,19,31] such as CoAl 2 O 4 , Co 3 O 4 and NiAl 2 O 4 . In detail, the apparent reduction peaks of cobalt oxides through stepwise reductions (Co 3 O 4 → CoO → Co) were clearly observed below 700 • C, which suggests the preferential segregations of the active cobalt species were on the outer surfaces of the NCA(0.5) and NCA(1) compared to the supports themselves (CA(0.5) and CA(1)) which have lower activity for SRP reactions as well. Interestingly, on the NCA(2), the smaller segregated cobalt species (below~700 • C) as well as spinel-type metal oxides (at~839 • C) were observed, which were responsible for an enhanced catalytic activity and stability through the formation of thermally stable mixed metal interfaces.
The catalytic activity of the CA supports and NCA(1) with an impregnated Ni/Al 2 O 3 and CoAl 2 O 4 reference catalysts were preliminarily measured at a lower SV = 100,800 L/(kg cat ·h), and the results are displayed in Figure 2B and summarized in Table S1. The CA supports themselves were active for the SRP reaction without any Ni species, which suggests that the cobalt nanoparticles were active for the reaction [3,25,34]. It was clear that the CA supports with the molar ratio of Al/Co ≤ 2 were also active for SRP reaction, and their higher and stable activities were characterized compared to the reference catalysts of Ni/Al 2 O 3 and CoAl 2 O 4 . In addition, the CA supports were highly stable compared to the Ni/Al 2 O 3 , which was gradually deactivated for five hours at 550 • C from 51.4% to Catalysts 2020, 10, 461 5 of 13 42.8% due to possible coke depositions at a relatively low temperature [35,36]. With an increased Co content, the catalytic activity was still higher on the CA(1) and CA(0.5) with the C 3 H 8 conversion of 57.4 and 57.2%, while the CA(1) was more active than the CA(0.5) above 650 • C (respective values of 79.6% and 86.9% at 650 • C), which was attributed to the preferential aggregations of Co particles on the CA(0.5). Interestingly, an impregnation of Ni metal on the NCA(1) catalyst revealed a 100% conversion of C 3 H 8 at all temperatures from 550 to 750 • C without any deactivation phenomena. This observation revealed an extraordinary activity compared to the reference Ni/Al 2 O 3 , owing to the synergetic effects of Ni and Co [3,12,25]. However, the CA supports with Al/Co ≥ 3 and commercially available spinel CoAl 2 O 4 were not active below 650 • C since there could be little metallic Co nanoparticles even after reduction step due to their stronger metal-support interactions [37].
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 13 to the reference catalysts of Ni/Al2O3 and CoAl2O4. In addition, the CA supports were highly stable compared to the Ni/Al2O3, which was gradually deactivated for five hours at 550 °C from 51.4% to 42.8% due to possible coke depositions at a relatively low temperature [35,36]. With an increased Co content, the catalytic activity was still higher on the CA(1) and CA(0.5) with the C3H8 conversion of 57.4 and 57.2%, while the CA(1) was more active than the CA(0.5) above 650 °C (respective values of 79.6% and 86.9% at 650 °C), which was attributed to the preferential aggregations of Co particles on the CA(0.5). Interestingly, an impregnation of Ni metal on the NCA(1) catalyst revealed a 100% conversion of C3H8 at all temperatures from 550 to 750 °C without any deactivation phenomena. This observation revealed an extraordinary activity compared to the reference Ni/Al2O3, owing to the synergetic effects of Ni and Co [3,12,25]. However, the CA supports with Al/Co ≥ 3 and commercially available spinel CoAl2O4 were not active below 650 °C since there could be little metallic Co nanoparticles even after reduction step due to their stronger metal-support interactions [37].

Catalytic Activities of NCA Catalysts with Coke Analysis
The SRP reaction was carried out again on the NCA catalysts at a slightly higher space velocity of 170,000 mL/gcat•h with the mixed gas composition of C3H8/N2/H2O = 1/4/9 at a fixed temperature of 650 °C, and the results of C3H8 conversion and its specific rate (C3H8 converted/(moles of (Ni + Co)·h)) are respectively displayed in Figure 3(A,B). As summarized in Table 2, the conversions of C3H8 were steadily increased with the decrease of Al/Co ratio, such as 49.5% on the NCA(4) and 78.1% on to NCA(1), except for the structurally unstable NCA(0.5). Based on the preliminary tests of the CA supports, the metallic Co species on the CA supports with a lower Al/Co ratio such as NCA(1) and NCA(2) seem to promote the reforming activity through the synergetic effects of Ni nanoparticles due to its own catalytic activity of the surface metallic Co nanoparticles. However, the NCA(0.5) with the largest amount of Co metal showed a continuous activity decrease from an initial conversion of 83.0% to 73.2% and larger deactivation rate of 0.89%/h due to the severe aggregations of the active Ni and Co nanoparticles with the variations of Ni size from 14.6 to 16.2 nm during SRP reaction measured by X-ray diffraction (XRD) analysis ( Table 1). The optimal NCA(1) and NCA(2) showed the moderate propane conversions of 72.6%-78.1% with smaller deactivation rates of 0.27%-0.31%/h while the NCA(4) showed the lowest conversion (49.5%) due to the small amount of active cobalt metal content with its less synergetic effects, which was confirmed by TPR analysis of the CA supports ( Figure 2). However, the specific rates, defined as [converted mols of C3H8 divided by the moles of Ni and Co metals per hour], revealed the lower rate of 0.96 on the NCA(0.5) (largest metal content) with larger metal aggregates with smaller active sites [20] and higher rate of 1.37 on the

Catalytic Activities of NCA Catalysts with Coke Analysis
The SRP reaction was carried out again on the NCA catalysts at a slightly higher space velocity of 170,000 mL/g cat ·h with the mixed gas composition of C 3 H 8 /N 2 /H 2 O = 1/4/9 at a fixed temperature of 650 • C, and the results of C 3 H 8 conversion and its specific rate (C 3 H 8 converted/(moles of (Ni + Co)·h)) are respectively displayed in Figure 3A,B. As summarized in Table 2, the conversions of C 3 H 8 were steadily increased with the decrease of Al/Co ratio, such as 49.5% on the NCA(4) and 78.1% on to NCA(1), except for the structurally unstable NCA(0.5). Based on the preliminary tests of the CA supports, the metallic Co species on the CA supports with a lower Al/Co ratio such as NCA(1) and NCA(2) seem to promote the reforming activity through the synergetic effects of Ni nanoparticles due to its own catalytic activity of the surface metallic Co nanoparticles. However, the NCA(0.5) with the largest amount of Co metal showed a continuous activity decrease from an initial conversion of 83.0% to 73.2% and larger deactivation rate of 0.89%/h due to the severe aggregations of the active Ni and Co nanoparticles with the variations of Ni size from 14.6 to 16.2 nm during SRP reaction measured by X-ray diffraction (XRD) analysis ( Table 1). The optimal NCA(1) and NCA(2) showed the moderate propane conversions of 72.6-78.1% with smaller deactivation rates of 0.27-0.31%/h while the NCA(4) showed the lowest conversion (49.5%) due to the small amount of active cobalt metal content with its less synergetic effects, which was confirmed by TPR analysis of the CA supports ( Figure 2). However, the specific rates, defined as [converted mols of C 3 H 8 divided by the moles of Ni and Co metals per hour], revealed the lower rate of 0.96 on the NCA(0.5) (largest metal content) with larger metal aggregates with smaller active sites [20] and higher rate of 1.37 on the NCA(2) even though the metal particles on the NCA(2) were aggregated during the SRP reaction. This suggests that the Ni and Co Catalysts 2020, 10, 461 6 of 13 metals have some additional synergy effects through the proper distributions of the active metallic Co nanoparticles on the NCA surfaces as well as optimal formations of the thermally stable spinel CoAl 2 O 4 and NiAl 2 O 3 phases on the bulk CA(2) support [12,32,33,37], as explained in the following section. In addition, the highly reduced NCA(0.5) and NCA(1) revealed a slightly lower H 2 selectivity of 70.1-71.2% and higher CH 4 selectivity of 4.4-5.2% compared to the less reduced NCA (2) with those values of 73.2% and 2.0%, which can be possibly attributed to the more preferential methanation activity as well as coke depositions by possibly inducing the smaller contributions of SRM activity on the metallic Co and Ni surfaces [4][5][6][7][8]33].   The surface coke types and their relative amounts on the used NCA catalysts were characterized by temperature programmed surface reaction (TPSR), and TPSR patterns monitored with a mass spectrometer (CH4 fragment, m/z = 15) are displayed in Figure 3C, including their relative intensity ratio based on that of NCA(0.5). There were insignificant amounts of soft coke precursors formed on the NCA catalysts, which can be hydrogenated to CH4 at low temperature of ~300 °C, however, hard coke precursors detected at the temperature regions of 550-800 °C [38] were mainly observed on all the used NCA catalysts. The hydrogenation peak positions at ~600 °C shifted to higher temperatures with an increase of the Al/Co ratio, which was possibly attributed to the formations of much stronger interacting, harder cokes on the Al2O3-rich surfaces [32,33]. In addition, the relative amounts of those hard coke precursors were found to be decreased with an increase of Al/Co ratios, being especially smaller on the NCA(2) and NCA(3), owing to the smaller crystallites, which can be responsible for  The surface coke types and their relative amounts on the used NCA catalysts were characterized by temperature programmed surface reaction (TPSR), and TPSR patterns monitored with a mass spectrometer (CH 4 fragment, m/z = 15) are displayed in Figure 3C, including their relative intensity ratio based on that of NCA(0.5). There were insignificant amounts of soft coke precursors formed on the NCA catalysts, which can be hydrogenated to CH 4 at low temperature of~300 • C, however, hard coke precursors detected at the temperature regions of 550-800 • C [38] were mainly observed on all the used NCA catalysts. The hydrogenation peak positions at~600 • C shifted to higher temperatures with an increase of the Al/Co ratio, which was possibly attributed to the formations of much stronger interacting, harder cokes on the Al 2 O 3 -rich surfaces [32,33]. In addition, the relative amounts of those hard coke precursors were found to be decreased with an increase of Al/Co ratios, being especially smaller on the NCA(2) and NCA(3), owing to the smaller crystallites, which can be responsible for less formation of cokes since the larger metallic Ni and Co nanoparticles act as active sites for coke depositions [39].

Relative Compositions of Active Metal (Oxides) Measured by X-ray Absorption Fine Structure (XAFS) Analysis
The X-ray absorption fine structure was further carried out to quantify the compositions of crystalline phases with their oxidation states of Ni and Co metals on the reduced and used NCA catalysts by observing the absorption energy of Ni K-edge (8333 eV) and Co K-edge (7709 eV). The spectra of X-ray absorption near-edge structures (XANES) and their phase compositions are summarized in Table 3 and displayed in Figures 4 and 5. The XANES spectra and their first derivatives were fitted to quantify the crystalline phases through linear combination fitting (LCF) by comparing them with various reference samples. With a decrease of the Al/Co ratio, the amounts of metallic Ni were found to be dominant with the ratio of 0.78 and 0.55 on the reduced and used NCA(0.5) with the respective composition of the NiAl 2 O 4 phases of 0.11 and 0.28, since the smaller amount of Al 2 O 3 can suppress the formations of the spinel NiAl 2 O 4 phases with much bigger Ni nanoparticles due to their weak metal-support interaction. With an increase of Al content, the composition of NiAl 2 O 4 phases was significantly increased up to 0.57 and 0.60 on the reduced and used NCA(4). In addition, the observed larger compositions of the NiO and NiAl 2 O 4 phases on the used NCA catalysts compared to that of the reduced ones can be attributed to a high temperature SRP reaction condition with excess oxidative water vapor [33]. The relative composition of Co species also clearly showed the oxidation states as well as their distributions on the NCA catalysts according to Al/Co ratios. The distributions of Co phases were found to be similar with that of the Ni nanoparticles, where the metallic Co species was decreased with the increasing Al/Co ratio from 0.82 on the reduced NCA(0.5) to 0.40 on the reduced NCA(4) and the content of metallic Co was decreased to 0.69 on the used NCA(0.5) to 0.13 on the used NCA(4) due to an in-situ oxidation during the SRP reaction under an excess water environment. On the contrary, the content of spinel CoAl 2 O 4 was increased with the same trends as the NiAl 2 O 4 phases. Interestingly, the oxidation phenomena of the metallic Co to CoO and CoAl 2 O 4 phases were possibly attributed to its facile oxidation-reduction characteristics of the oxophilic Co species [12]. Furthermore, the stable preservation of the metallic Co phases with small changes of the CoAl 2 O 4 formation on the reduced and used NCA(2) was observed, with their changes from 0.68 to 0.52 and 0.22 and 0.37, respectively, which suggests the preservation of more stable metal phases caused the increased specific rate with less deactivation by maintaining active metallic Co and Ni phases with less coke depositions in the redox reaction condition during the SRP reaction. This was in line with the results of the TPR analysis and preliminary tests, as the amount of metallic Co was dramatically increased at lower ratios of Al/Co ≤ 2. The additional reduction peaks appearing at T ≤ 700 • C were responsible for the active surface metallic Co and Ni phases for an enhanced activity for SRP reaction, which are unlike the commercially available bulk CoAl 2 O 4 phases due to the contribution of the Co 3 O 4 -Al 2 O 3 porosity, especially for the NCA catalysts with higher ratios of Al/Co ≥ 3. Therefore, the higher specific rate on the NCA(2) with a small deactivation rate, even though its little lower conversion of C 3 H 8 compared to the NCA(1), can be attributed to the small aggregations of Ni nanoparticles on the porous Co 3 O 4 -Al 2 O 3 surfaces due to its stronger metal-support interactions as well as small formations of less active spinel Co 3 O 4 phases by robustly preserving the active metallic Co nanoparticles on the NCA(2) surfaces, even under an excess water environment.
Catalysts 2020, 10, x FOR PEER REVIEW 8 of 13 0.22 and 0.37, respectively, which suggests the preservation of more stable metal phases caused the increased specific rate with less deactivation by maintaining active metallic Co and Ni phases with less coke depositions in the redox reaction condition during the SRP reaction. This was in line with the results of the TPR analysis and preliminary tests, as the amount of metallic Co was dramatically increased at lower ratios of Al/Co ≤ 2. The additional reduction peaks appearing at T ≤ 700 °C were responsible for the active surface metallic Co and Ni phases for an enhanced activity for SRP reaction, which are unlike the commercially available bulk CoAl2O4 phases due to the contribution of the Co3O4-Al2O3 porosity, especially for the NCA catalysts with higher ratios of Al/Co ≥ 3. Therefore, the higher specific rate on the NCA(2) with a small deactivation rate, even though its little lower conversion of C3H8 compared to the NCA(1), can be attributed to the small aggregations of Ni nanoparticles on the porous Co3O4-Al2O3 surfaces due to its stronger metal-support interactions as well as small formations of less active spinel Co3O4 phases by robustly preserving the active metallic Co nanoparticles on the NCA(2) surfaces, even under an excess water environment.   0.22 and 0.37, respectively, which suggests the preservation of more stable metal phases caused the increased specific rate with less deactivation by maintaining active metallic Co and Ni phases with less coke depositions in the redox reaction condition during the SRP reaction. This was in line with the results of the TPR analysis and preliminary tests, as the amount of metallic Co was dramatically increased at lower ratios of Al/Co ≤ 2. The additional reduction peaks appearing at T ≤ 700 °C were responsible for the active surface metallic Co and Ni phases for an enhanced activity for SRP reaction, which are unlike the commercially available bulk CoAl2O4 phases due to the contribution of the Co3O4-Al2O3 porosity, especially for the NCA catalysts with higher ratios of Al/Co ≥ 3. Therefore, the higher specific rate on the NCA(2) with a small deactivation rate, even though its little lower conversion of C3H8 compared to the NCA(1), can be attributed to the small aggregations of Ni nanoparticles on the porous Co3O4-Al2O3 surfaces due to its stronger metal-support interactions as well as small formations of less active spinel Co3O4 phases by robustly preserving the active metallic Co nanoparticles on the NCA(2) surfaces, even under an excess water environment.

Discussion and Conclusions
The synergetic effects of the oxophilic Co nanoparticles for the Ni-based reforming catalysts were largely investigated in terms of the metal-support interactions and coke resistance [6,25]. In the case of catalytic reforming reaction of hydrocarbons, it was well reported that those interactions not only enhanced the metal dispersion to promote the catalytic activity, but also lessened the coke deposition to enhance the catalytic durability. However, a lot of research was focused on simultaneously using the expensive noble metal promoters such as Rh, Ru and Pt with much cheaper Ni metal, which changed the dispersion of metal clusters, reducibility and heat of adsorption through hydrogen spillover. Those promoters can keep the Ni nanoparticles in more active metallic states by preventing the coke depositions compared to the easily deactivated monometallic Ni-based catalysts [9,23]. The bimetallic Ni-Co-based reforming catalysts were reported to possess similar promoting effects to enhance the metal dispersion and reducibility. The more interesting points of the Ni-Co-based catalysts are the oxophilic natures of the Co species, which can help increase the adsorption natures of the oxidants, such as H 2 O and CO 2 , by effectively preventing the coke depositions [12,26]. However, the synergetic effects of the spinel-type CoAl 2 O 4 and NiAl 2 O 4 formation on the Ni-impregnated Co 3 O 4 -Al 2 O 3 (CA) support prepared at different Al/Co ratios were not well considered till now, therefore, the surface compositions of the bimetal clusters as well as contributions of spinel-type CoAl 2 O 4 formations are briefly discussed to verify the activity variations of the home-made Ni/Co 3 O 4 -Al 2 O 3 (NCA) catalysts for the SRP reaction.
In the present study, the synergetic effects of the Ni (fixed at~10 wt%)and Co metal (oxides) on the CA supports, prepared by a sol-gel method with different Al/Co ratios, were verified by the characterization of the surface and bulk properties of NCA catalysts to explain the reasons for the enhanced activity and stability with their higher coke resistances on the optimal NCA(1) and NCA (2). The bulk compositions of the CA supports affected the catalytic performances largely by changing the active Ni and Co metal distributions on the porous Co 3 O 4 -Al 2 O 3 structures and by generating the different amounts of the thermally stable spinel CoAl 2 O 4 phases. The Co phases on the NCA catalysts were found to be mainly the metallic Co species at a lower Al/Co ratio ≤ 2 with small amounts of less active spinel CoAl 2 O 4 phases formed under SRP reaction conditions. Those metallic Co phases were also found to be active for the SRP reaction as confirmed by preliminary tests of the CA supports by displaying better catalytic activity than the reference Ni/Al 2 O 3 under the temperature of 650 • C, which was in line with the TPR results. It suggests that the cobalt oxides can be reduced to selectively form the more active metallic Co phases on the NCA(1) and NCA(2) surfaces. However, the less active CoAl 2 O 4 phases were mainly formed on the NCA(3) and NCA(4) under an excess water environment due to their higher Al/Co ratios. Based on the XANES analysis, the partially formed bulk spinel CoAl 2 O 4 and NiAl 2 O 4 structures on the NCA(1) and NCA (2) can effectively enhance the thermal stability of the strongly interacted metallic Ni and Co nanoparticles with their smaller aggregation natures compared to the NCA(0.5), which was also responsible for suppressing the depositions of coke precursors. The higher specific rate of the NCA(2), with proper amounts the surface metallic Ni and Co phases with the bulk spinel CoAl 2 O 4 phases, was attributed to the synergetic effects of the surface Ni and Co metals, as well as thermally stable bulk spinel structures.

Catalyst Preparation and Activity Measurement
The Co 3 O 4 -Al 2 O 3 (CA) supports were prepared by a well-known sol-gel method by using a gelation agent of propylene oxide according to the references [12,40]. For more details, desired amounts of Al(NO 3 ) 3 ·9H 2 O (Daejung, 98% purity) and Co(NO 3 ) 2 ·6H 2 O (Samchun, 97% purity) were dissolved in 100 mL of anhydrous ethanol (Daejung, 99.9% purity) at fixed molar ratios of Al/Co = 0.5, 1, 2, 3 and 4 under vigorous stirring. After all the metal precursors were dissolved and well mixed, 35 mL of propylene oxide (Samchun, 99.0%) was added to the above solution as a gelation agent. The mixture was transformed into gel after 10 min under stirring followed by aging it at room temperature for 6 h. After drying at 80 • C for 60 h, the xerogel was further calcined at T = 700 • C for 3 h in an air atmosphere. The collected powder was crushed and named as CA(x), where x represents the molar ratio of Al/Co. The Ni metal precursor was further introduced to the CA supports by a wet impregnation method at a fixed 10 wt% Ni based on the CA support. In further detail, 1.12 g of Ni(NO 3 ) 2 ·6H 2 O (Samchun, above 98% purity) was dissolved in 100 mL ethanol to prepare the precursor solution. An amount of 2 g of the CA support was added to the above solution then stirred vigorously for 1 h. The solvent was evaporated at 50 • C by using a vacuum evaporator (NE-2001, EYELA). The mixture was further dried at 80 • C overnight, followed by the calcination at 700 • C for 3 h in an air atmosphere. As-prepared catalysts were denoted as NCA(x), where the x represents the molar ratio of Al/Co. For the preparation of reference catalysts, a typical Ni/Al 2 O 3 catalyst was prepared by using an impregnation method with commercially available γ-Al 2 O 3 (Puralox SCCa-5/170, Sasol, S g = 165 m 2 /g) support, which was further calcined at 800 • C for 2 h. The spinel CoAl 2 O 4 was purchased from Sigma-Aldrich (cobalt aluminum oxide nano-powder with particle size < 50 nm) for the comparative activity measurement as well.
The activity measurements for steam reforming of propane (SRP) on the as-prepared NCA catalysts were performed using an Inconel fixed-bed tubular reactor with 3/8 inch of an outer diameter. An amount of 0.15 g of the NCA catalyst was placed and it was brought into contact with a thermocouple to measure the temperature gradients in the catalyst bed at a controlled furnace temperature. Prior to the SRP reaction, the catalyst was reduced at 700 • C under a flow of 50 mL/min of 5 vol% H 2 /N 2 for 3 h. As the reactor was cooled down to the reaction temperature of 550 and 650 • C and stabilized, steam was fed to the reactor prior to propane feed. The SRP reaction was carried out with feed gas compositions of C 3 H 8 /N 2 /H 2 O = 1/4/9 at a space velocity of 108,000-170,000 mL/g cat ·h in the temperature range of 550-650-750 • C. After the SRP reaction, high purity N 2 was fed to the reactor to prevent the unintended coke formation while it is cooled to room temperature. The effluent gases were passed over a cold trap for dehydration, and analyzed by using a gas chromatography (Young Lin Instrument Co, Anyang, Republic of Korea, YL6100 GC) equipped with a Carboxen 1000 packed column to quantify the gaseous components such as H 2 , N 2 , CO, CO 2 , and CH 4 using a thermal conductivity detector (TCD), and GS-GasPro capillary column used to quantify the light hydrocarbons containing C 1 -C 3 species. The catalytic activity was compared with conversion of C 3 H 8 and product distribution, which was calculated based on the changes in effluent gas concentrations with an internal standard of N 2 . Catalytic performances were calculated by the following equations based on the carbon balance in a dry mole basis as described below: Product distribution (mol%) = (mols of the product)/(Sum of product mols in a gas phase) × 100 (7)

Characterization of the CA Supports as Well as NCA Catalysts
Powder X-ray diffraction (XRD) patterns of the NCA catalysts were analyzed to validate the crystalline phases and crystallite sizes of the Ni and Co species by using a X'Pert3 Powder (PANalytical) instrument operated with radiation of Cu-Kα of 0.15406 nm at 40 kV and 100 mA in the 2θ range of 10 • -80 • at a scanning rate of 4 • /min. The full width at half maximum (FWHM) values of the metallic Ni-Co peaks at Bragg angles of 2θ = 44 • -44.5 • were used for the crystallite size calculations with the help of Scherrer's equation by using a Scherrer constant of 0.9 with an assumption of spherical crystals. N 2 adsorption-desorption analysis was performed on a constant volume sorption instrument (Tristar II, Micromeritics) at a temperature of −196 • C after the sample was degassed at the temperature of 350 • C for 4 h in a vacuum condition. The textural properties, such as surface area, pore volume and average pore size of the fresh NCA catalysts were calculated by Brunauer-Emmett-Teller (BET) equation and Barrett-Joyner-Halenda (BJH) model using the desorption branch of N 2 isotherms.
CO chemisorption was carried out to verify the dispersions of the active metals on the NCA catalysts by using an ASAP2020C (Micromeritics) instrument. Initially, 0.2 g of the sample was loaded in a u-shaped quartz reactor followed by reduction under a flow of 30 mL/min of H 2 at the temperature of 700 • C for 3 h. The reactor was evacuated while it was cooled from 700 • C to the 35 • C for 1 h and kept for an additional 1 h at 35 • C to maintain the vacuum level of 0.02 mmHg. After the leak test, the reduced sample was evacuated for 20 min then analyzed with CO as a probe molecule. CO uptakes were determined by the introduction of CO pulses until the pressure increments of the sample cell were measured. For the calculation of CO chemisorption, the adsorption stoichiometry of H 2 to metals was assumed as one CO molecule per one molecule of the surface metals (Ni and Co).
Temperature programmed reduction with H 2 (TPR) was performed to clarify the reduction behaviors of the fresh NCA catalysts by measuring the consumed amount of H 2 by using a BELCAT-M (MicrotracBEL) instrument. An amount of 30 mg of the sample was loaded in a quartz tube reactor and heated at 500 • C for 1 h under an inert flow of Ar to eliminate contaminants and to clean the catalyst surfaces. After being cooled to 100 • C, 30 mL/min of 10 vol% H 2 /Ar was fed to the reactor. After the TCD signal was stabilized after 30 m, the sample was heated up to 1000 • C at a ramping rate of 10 • C/min and H 2 uptakes were measured by TCD after the effluent gases were dehydrated by passing over a molecular sieve 3A trap. Temperature programmed surface reaction (TPSR) of the used NCA catalysts was performed to verify the relative amounts of coke formation on the catalyst surfaces using a BELCAT-M instrument connected with a mass spectrometer (QMS-200, Pfeiffer vacuum, Yong-in, Republic of Korea). 15 mg of the used NCA catalyst was loaded and pretreated under a flow of 30 mL/min of Ar at 500 • C for 1 h in a quartz reactor. After the above pretreatment, the sample tube was cooled down to 50 • C and 10 vol% H 2 /Ar was introduced at a flow rate of 30 mL/min. The temperature was increased up to 700 • C at a ramping rate of 10 • C/min to hydrogenate the deposited coke precursors to CH 4 , which was detected by collecting a MS signal by measuring the specific fragments such as m/z = 15, which are assigned to the major fragment of CH 4 .
To observe the crystalline phases and oxidation states of the Ni and Co species on the reduced and used NCA catalysts, X-ray absorption fine structure (XAFS) analysis with Ni K-edge (8333 eV) and Co K-edge (7709 eV) was performed at 7D beamline located at a Pohang Light Source (PLS). The beamline was equipped with a Si(111) monochromator and ionization chambers to measure the transmitted beam intensity. The powder sample was previously loaded in the reactor, which was positioned between two Ohken ionization chambers filled with N 2 gas for the measurement of incidents and transmitted X-ray intensity. The absorption spectra of the samples, including the NCA catalysts as well as references of NiO, NiAl 2 O 4 , CoO, Co 3 O 4 and CoAl 2 O 4 , were recorded with a spectrum of metallic foil. Normalization and XAFS analysis were carried out by using an Athena graphical interface program [41]. Linear combination fitting (LCF) was also performed in the regions of X-ray absorption near-edge structures (XANES) to quantify the crystalline structures of the reduced and fresh NCA catalysts as well.
Supplementary Materials: The followings are available online at http://www.mdpi.com/2073-4344/10/4/461/s1; Figure S1. Textural properties of the fresh NCA catalysts measured by N 2 adsorption-desorption analysis of the (A) adsorption-desorption isotherms and (B) pore size distributions; Figure S2. Wide-angle X-ray diffraction (XRD) patterns of the (A) reduced and (B) used NCA catalysts; Table S1. Catalytic activities of the CA supports with some reference materials.