Low-temperature catalytic performance of Ni-Cu/Al2O3 catalysts for gasoline reforming to produce hydrogen applied in spark ignition engines

: The performance of Ni-Cu/Al 2 O 3 catalysts for steam reforming (SR) of gasoline to produce a hydrogen-rich gas mixture applied in a spark ignition (SI) engine was investigated at relatively low temperature. The structural and morphological features and catalysis activity were observed by X-ray diffractometry (XRD), scanning electron microscopy (SEM), and temperature programmed reduction (TPR). The results showed that the addition of copper improved the dispersion of nickel and therefore facilitated the reduction of Ni at low temperature. The highest hydrogen selectivity of 70.6% is observed over the Ni-Cu/Al 2 O 3 catalysts at a steam/carbon ratio of 0.9. With Cu promotion, a gasoline conversion of 42.6% can be achieved at 550 ˝ C, while with both Mo and Ce promotion, the gasoline conversions were 31.7% and 28.3%, respectively, higher than with the conventional Ni catalyst. On the other hand, initial durability testing showed that the conversion of gasoline over Ni-Cu/Al 2 O 3 catalysts slightly decreased after 30 h reaction time.


Introduction
Hydrogen is an emerging alternative used increasingly to replace the depletion of conventional fossil fuels-it has long been recognized as a fuel having some highly desirable properties for application in engines [1]. These features make hydrogen an excellent fuel to potentially meet the increasingly stringent targets for exhaust emissions from combustion devices, including the reduction of greenhouse gas emissions. Hydrogen can be used in internal combustion engines (ICE) on its own or as a partial substitute for traditional ICE fuels to improve performance and emissions, as has been reported by many researchers [2][3][4][5][6][7][8][9][10][11].
The combustion speed of the mixture of hydrogen and air is much higher than other petroleum mixtures. Burning its mixture with air requires ignition by a spark. However, the spark needs only a minuscule amount of energy to ignite this mixture [12]. A hydrogen engine is easy to start in a cold winter because hydrogen remains in a gaseous state until it reaches the low temperature of 253˝C [13]. Its combustion products are clean, consisting of water and a tiny amount of NO x [14]. Hydrogen-rich gaseous fuel was produced by exhaust gas assisted reforming of gasoline or diesel fuel with steam (on-board production) [8,15]. This technique can potentially provide a feasible and practical engine system. It does not require a secondary fueling system, which is undesirable for the driver/operator.
Hydrogen-rich gaseous fuel involves hydrogen generation by direct catalytic interaction of liquid fuels with steam (steam reforming). Krumplet et al. reported the activity of different transition metals Table 1 shows the surface and pore properties of different catalysts. The specific surface area, pore volume, and average pore diameter of (Ni 0.5 -Cu 0.5 ) x /(Al 2 O 3 ) 1´x catalysts (x = 6 wt. %, 18 wt. %, and 36 wt. %) decreased with the increase of Ni-Cu loading; the reason could be the depositions of the active components in the pores and the micropores of γ-Al 2 O 3 . The surface area (152.8 m 2 /g) and total pore volume (0.22 cm 3 /g) of 18 wt. % Ni/γ-Al 2 O 3 catalyst demonstrated progressive substitution of Ni by Cu of 18 wt. % Ni 1´x -Cu x /Al 2 O 3 catalysts; this results in a slight increase of the BET area and total pore volume, which may be caused by the different CuO and NiO crystallite sizes and formation of mixed oxides. In addition, the surface area and pore structure of different catalysts are also shown in Table 1.  Figure 1 shows the X-ray diffractometry (XRD patterns of Ni 0.5 -Cu 0.5 /Al 2 O 3 catalysts (with varying Ni 0.5 -Cu 0.5 /Al 2 O 3 ratios). The γ-Al 2 O 3 peaks were observed with other ratios; the intensity of the γ-Al 2 O 3 peaks slightly decreases, indicating that CuO and NiO particles could be highly dispersed in γ-Al 2 O 3 [29,33]. The reflection peak intensities of the CuO and NiO phase occurred and became sharper with the increase of Ni-Cu loading, while no appearance of other phases was detected (Figure 1c,d).  Figure 1 shows the X-ray diffractometry (XRD patterns of Ni0.5-Cu0.5/Al2O3 catalysts (with varying Ni0.5-Cu0.5/Al2O3 ratios). The γ-Al2O3 peaks were observed with other ratios; the intensity of the γ-Al2O3 peaks slightly decreases, indicating that CuO and NiO particles could be highly dispersed in γ-Al2O3 [29,33]. The reflection peak intensities of the CuO and NiO phase occurred and became sharper with the increase of Ni-Cu loading, while no appearance of other phases was detected (Figure 1c,d).  XRD patterns of 18 wt. % different catalysts are shown in Figure 3. The reflection of CuO and cubic CeO2 are evidently observed (Figure 3a,b), while no separate reflections from Ni are detected for all samples. This provided further evidence that nickel promoted the dispersion of Cu and Ce [28]. It is   (Figure 2a), only the phases of CuO were observed. With the decrease of CuO loading, the reflection peak intensities of the CuO phase were largely reduced with peak broadening (Figure 2a-c). At 30 mol % CuO ratio, only the cubic NiO phase was observed (Figure 2d). On further reducing the CuO contents (Figure 2e,f), the emerged intensity of the cubic NiO phase was gradually increased, while no reflection peaks of CuO were observed.
XRD patterns of 18 wt. % different catalysts are shown in Figure 3. The reflection of CuO and cubic CeO 2 are evidently observed (Figure 3a,b), while no separate reflections from Ni are detected for all samples. This provided further evidence that nickel promoted the dispersion of Cu and Ce [28]. It is surprising to note that the reflections of NiMo 4 were observed over 18 wt. % Ni 0.5 -Mo 0.5 /Al 2 O 3 catalysts. The presence of the NiMo 4 phase was possibly due to the preparations of Ni-Mo catalysts [34,35]. surprising to note that the reflections of NiMo4 were observed over 18 wt. % Ni0.5-Mo0.5/Al2O3 catalysts. The presence of the NiMo4 phase was possibly due to the preparations of Ni-Mo catalysts [34,35].   Figure 4 shows a scanning electron microscopy (SEM) micrograph of Ni0.5-Cu0.5/Al2O3 catalysts with various Ni-Cu content ratios; high distributions of CuO-NiO nanoparticles on large spongy clusters of γ-Al2O3 were observed for all the samples. Ni-Cu nanoparticles of 10 nm were observed on the 6 wt. % Ni0.5-Cu0.5/Al2O3 catalysts (Figure 4a). With the increase of Ni-Cu loading, an increase in the particle size of Ni-Cu was clearly observed (Figure 4b,c) where aggregations of packed particles from a few nanometers to few hundred nanometers in size exist, which may cause agglomeration of powders, indicating sintering of the catalysts. Diffraction intensity (a.u.) surprising to note that the reflections of NiMo4 were observed over 18 wt. % Ni0.5-Mo0.5/Al2O3 catalysts. The presence of the NiMo4 phase was possibly due to the preparations of Ni-Mo catalysts [34,35].   Figure 4 shows a scanning electron microscopy (SEM) micrograph of Ni0.5-Cu0.5/Al2O3 catalysts with various Ni-Cu content ratios; high distributions of CuO-NiO nanoparticles on large spongy clusters of γ-Al2O3 were observed for all the samples. Ni-Cu nanoparticles of 10 nm were observed on the 6 wt. % Ni0.5-Cu0.5/Al2O3 catalysts (Figure 4a). With the increase of Ni-Cu loading, an increase in the particle size of Ni-Cu was clearly observed (Figure 4b,c) where aggregations of packed particles from a few nanometers to few hundred nanometers in size exist, which may cause agglomeration of powders, indicating sintering of the catalysts.   Figure 5 shows the corresponding SEM micrograph and energy dispersive X-ray (EDX) mapping of Ni0.5-Cu0.5/Al2O3 catalysts; it clearly shows the presence of Ni, Cu, and Al elements. In addition, the element mapping revealed a uniform distribution for Ni, Cu, and Al.   Figure 5 shows the corresponding SEM micrograph and energy dispersive X-ray (EDX) mapping of Ni 0.5 -Cu 0.5 /Al 2 O 3 catalysts; it clearly shows the presence of Ni, Cu, and Al elements. In addition, the element mapping revealed a uniform distribution for Ni, Cu, and Al.   Figure 5 shows the corresponding SEM micrograph and energy dispersive X-ray (EDX) mapping of Ni0.5-Cu0.5/Al2O3 catalysts; it clearly shows the presence of Ni, Cu, and Al elements. In addition, the element mapping revealed a uniform distribution for Ni, Cu, and Al.    Figure 6 shows the TPR profiles of 18 wt. % Ni-Cu/γ-Al 2 O 3 catalysts with various Cu ratios. For the Ni/γ-Al 2 O 3 catalysts (Figure 6a), the TPR pattern shows that the broad reduction signal appears from 400˝C, which can be attributed to the reduction of NiO particles [36,37]; it recognizes that with the Ni/γ-Al 2 O 3 catalysts reduced at 400˝C, there might be a reduction of dispersed Ni 2+ and a more intense broad reduction corresponding to bulk NiO species [38]. At lower Cu ratios, only two apparent peaks were observed (Figure 6b,c): A low-temperature peak between 200 and 270˝C, and a high-temperature peak between 580 and 620˝C. With the increase of Cu ratios, the H 2 -TPR profile patterns showed three reduction peaks (Figure 6d,e); one more low temperature peak between 260 and 380˝C was detected. The intensity of the low-temperature peaks increased with increasing Cu content; while the high-temperature peak shifted to a lower temperature, the range of reduction temperatures was wider. For pure Cu/γ-Al 2 O 3 catalysts, the reduction is characterized by rather-combined two peaks in the range of 200-400˝C (Figure 6f).

Morphology Characterizations
the Ni/γ-Al2O3 catalysts (Figure 6a), the TPR pattern shows that the broad reduction signal appears from 400 °C, which can be attributed to the reduction of NiO particles [36,37]; it recognizes that with the Ni/γ-Al2O3 catalysts reduced at 400 °C, there might be a reduction of dispersed Ni 2+ and a more intense broad reduction corresponding to bulk NiO species [38]. At lower Cu ratios, only two apparent peaks were observed (Figure 6b,c): A low-temperature peak between 200 and 270 °C, and a high-temperature peak between 580 and 620 °C. With the increase of Cu ratios, the H2-TPR profile patterns showed three reduction peaks (Figure 6d,e); one more low temperature peak between 260 and 380 °C was detected. The intensity of the low-temperature peaks increased with increasing Cu content; while the high-temperature peak shifted to a lower temperature, the range of reduction temperatures was wider. For pure Cu/γ-Al2O3 catalysts, the reduction is characterized by rathercombined two peaks in the range of 200-400 °C (Figure 6f).
Fierro et al. [39] reported that the TPR characteristics could be affected by mass transfer limitations and experimental operating variables such as the initial amount of reducible species, the initial H2 concentration, the total gas flow rate, the heating rate, and the activation energy of the reaction. They claimed that desorption of H2 attached on the reduced Cu metal surface could exhibit the apparent double-peak behavior, which could be affected by water vapor produced by the reduction process, and that the H2-TPR profile of CuO depends on the particle size and surface area, where the peak top is higher by 288 °C for particle sizes of 425-850 microns compared with <100 microns [39]. A similar tendency but with a much larger peak was also reported by Luo et al. [40], who claimed that the hydrogen spillover effect was the reason for the difference between CO-TPR and H2-TPR. Kim et al. [41] reported that there is an incubation period prior to reduction, which is longer at lower temperatures. The tendency was in agreement with the general theory of nucleation and growth, where the number of newly-formed nuclei was copious at lower temperatures but since the growth was very much limited at lower temperatures, the phase existence was not easily detected by X-ray diffractometry. They claimed that CuO reduction was generally easier than Cu2O reduction Fierro et al. [39] reported that the TPR characteristics could be affected by mass transfer limitations and experimental operating variables such as the initial amount of reducible species, the initial H 2 concentration, the total gas flow rate, the heating rate, and the activation energy of the reaction. They claimed that desorption of H 2 attached on the reduced Cu metal surface could exhibit the apparent double-peak behavior, which could be affected by water vapor produced by the reduction process, and that the H 2 -TPR profile of CuO depends on the particle size and surface area, where the peak top is higher by 288˝C for particle sizes of 425-850 microns compared with <100 microns [39]. A similar tendency but with a much larger peak was also reported by Luo et al. [40], who claimed that the hydrogen spillover effect was the reason for the difference between CO-TPR and H 2 -TPR.
Kim et al. [41] reported that there is an incubation period prior to reduction, which is longer at lower temperatures. The tendency was in agreement with the general theory of nucleation and growth, where the number of newly-formed nuclei was copious at lower temperatures but since the growth was very much limited at lower temperatures, the phase existence was not easily detected by X-ray diffractometry. They claimed that CuO reduction was generally easier than Cu 2 O reduction with H 2 -TPR and that the apparent activation energy for Cu 2 O was close to twice that of CuO, but when the H 2 flow rate was not high enough to avoid the rate-limiting step of the reduction process, a sequential reduction process such as CuO Ñ (Cu 4 O 3 Ñ) Cu 2 O Ñ Cu could occur.
Hierl et al. [42] reported that the lowering of the reduction temperature was partially caused by the effect of Ni on the segregation of Cu from the surface in preference of Ni to occupy subsurface or bulk coordination sites. The enrichment of Cu may be enhanced by an increase in the nickel content. On the other hand, the reduction of CuO species could accelerate the reduction of NiO species due to the formation of atomic hydrogen in the environment [43]. Xiaolei Wang et al. [28] reported the interaction between CuO and NiO as well as their occupancy on surface acidic sites of γ-Al 2 O 3 reducing the interaction between metals and support, leading to lowering of the reduction temperature of CuO and NiO.
Consequently, the peak for pure CuO reduction in Figure 6f can be mainly attributable to direct reduction of CuO particles to Cu, contributing a low-temperature peak. The extended TPR peak profile indicates the difficulty of the reduction due to a longer diffusion path for H atoms or H 2 molecules as well as the escape of the produced H 2 O molecules. TPR profile analysis of 18 wt. % Ni-Cu/γ-Al 2 O 3 catalysts at small Cu ratios shows a low-temperature peak (Figure 6b,c) that corresponds to the reduction of CuO particles to Cu. The high-temperature peak corresponding to the reduction of NiO interacted with alumina is also observed in the Ni-Al profile. With increase of Cu content, the first peak at low temperature between 200 and 270˝C could be due to the reduction of CuO particles while the second peak between 260 and 380˝C could correspond to the reduction of CuO bulk [44]. The range of reduction temperatures is wider, which could possibly be due to agglomeration of the CuO bulk to form larger crystallites [45]. Furthermore, the high-temperature peak shifted to a lower temperature with an increase in the Cu content, which is due to the added Cu that produces spillover hydrogen, which considerably accelerates the nucleation of the Ni metal under these reduction conditions, and enhances the reducibility of Ni [46]. This suggests that Cu could enhance the reducibility of the dispersed NiO species, resulting in the shift to lower temperatures. Figure 7 shows the TPR profiles of different catalysts. In the case of the 18 wt. % Ni 0.5 -Mo 0.5 /γ-Al 2 O 3 catalysts (Figure 7a), a reduction peak at about 532˝C was observed and another peak could be assigned at 700˝C. It was reported that the presence of a Mo species impeded the incorporation of the Ni species into the lattice of alumina, thus preventing the growth of Ni particles, leading to less lattice expansion of alumina and highly dispersed Ni-Mo catalysts. This resulted in a weak metal-support interaction and, consequently, improved the reducibility of the catalysts [34,35,47]. Thus we assigned the peak at 532˝C to the reduction of NiMoO 4 (the XRD data showed a NiMoO 4 peak); the latter peak could be assigned to the reduction of NiO interacting with MoO 3 . This result is consistent with a previous report [34]. Figure 7b shows the TPR of 18 wt. % Ni 0.5 -Ce 0.5 /γ-Al 2 O 3 catalysts. It shows that a peak at about 535˝C is observed and it has been reported that the presence of Ce improves Ni dispersion, thus promoting the reduction of nickel species [48][49][50]. Incidentally, the reduction peak could be assigned to the strong interaction between NiO and CeO 2 . For Ni 0.5 -Cu 0.5 /Al 2 O 3 catalysts with varying Ni 0.5 -Cu 0.5 /Al 2 O 3 ratios (Figure 7c-f), the three reduction peaks were clearly observed at 36 wt. % Ni 0.5 -Cu 0.5 /Al 2 O 3 catalysts; the intensity of the reduction peaks decreased with no change to the position of peaks due to decreasing the Ni-Cu content. A summary of the TPR results including the temperature peak at maximum H 2 consumption and estimated quantity H 2 consumption during the catalysts reduction is also presented in Table 2. The amount of H 2 consumption during the reduction of Ni-Cu/Al 2 O 3 catalysis is higher than that for Ni-Mo/Al 2 O 3 and Ni-Ce/Al 2 O 3 catalysis and with a lower temperature peak.   Figure 8 shows the effect of steam/carbon (S/C) molar ratios on product selectivity over 18 wt. % Ni0.5-Cu0.5/γ-Al2O3 catalysts at 550 °C. At a small S/C molar ratio of 0.3, less H2 was formed, with selectivity in favor of CH4 and low formation of CO2 and CO was observed. When the ratio of S/C increased from 0.3 to 0.9, H2 selectivity gradually increased and the best H2 selectivity was obtained with a 0.9 ratio of S/C, the formation of CH4 decreased while the formation of CO and CO2 did not change much. When the S/C ratio increased over 0.9, H2 production decreased and the formation of CO2 increased slightly. This suggests that the formation of products derived from thermal decomposition, WGS, and catalytic cracking reactions is favored at stoichiometric reaction conditions due to the limited amount of steam for reforming reactions [44][45][46]. The 0.9 ratio of S/C is used for the rest of the Ni-Cu/γ-Al2O3 catalysts.   Figure 8 shows the effect of steam/carbon (S/C) molar ratios on product selectivity over 18 wt. % Ni 0.5 -Cu 0.5 /γ-Al 2 O 3 catalysts at 550˝C. At a small S/C molar ratio of 0.3, less H 2 was formed, with selectivity in favor of CH 4 and low formation of CO 2 and CO was observed. When the ratio of S/C increased from 0.3 to 0.9, H 2 selectivity gradually increased and the best H 2 selectivity was obtained with a 0.9 ratio of S/C, the formation of CH 4 decreased while the formation of CO and CO 2 did not change much. When the S/C ratio increased over 0.9, H 2 production decreased and the formation of CO 2 increased slightly. This suggests that the formation of products derived from thermal decomposition, WGS, and catalytic cracking reactions is favored at stoichiometric reaction conditions due to the limited amount of steam for reforming reactions [44][45][46]. The 0.9 ratio of S/C is used for the rest of the Ni-Cu/γ-Al 2 O 3 catalysts.

Effect of Temperature on Reformation Composition
Steam reforming of iso-octane is a complex process including several gas phase and chemical reactions, such as pyrolysis, oxidation, reforming, methanation, and WGS reactions [46]. Savage et al. [51] reported that hydrogen selectivity occurs through multiple pyrolysis reactions, as hemolytic dissociation of the C-C bond, radical recombination, beta scission, isomerization, and hydrogen abstraction. The product selectivity is determined by the strength of the C-H and C-C bonds in the reactant molecules; an alternative approach for H2 selectivity through steam reforming (SR) of iso-octane was proposed by Kopasz et al. [16]. iso-Octane is initially broken down by thermolysis into lighter hydrocarbons, which are reformed by steam on the catalytic surface. The following simplified reaction scheme has been proposed and includes reforming/decomposition reactions and secondary chemical steps involving the different intermediate products [45].
C + 2H2 → CH4 ∆H 0 = -75 kJ/mol (8) C + H2O → CO + H2 ∆H 0 = 131 kJ/mol (10) Figure 9 shows the effects of temperature on conversion and product selectivity over 18 wt. % Ni0.5-Cu0.5/γ-Al2O3 catalyst at a 0.9 ratio of S/C. Steam reforming of iso-octane results mainly in H2, CO, CO2, and CH4. As the reforming temperature increases from 500 to 750 °C, the CO2 concentration decreases slightly from 15.1% to 9.4%; the CO concentration remained almost constant (7%), while a strong increase in the CH4 concentration from 6.9% to 21.2% was obtained. The concentration of H2 dropped from the highest concentration of 71% to 61.8% due to the reforming temperature increasing from 500 to 750 °C. This indicates that the methanation reactions between the carbonaceous

Effect of Temperature on Reformation Composition
Steam reforming of iso-octane is a complex process including several gas phase and chemical reactions, such as pyrolysis, oxidation, reforming, methanation, and WGS reactions [46]. Savage et al. [51] reported that hydrogen selectivity occurs through multiple pyrolysis reactions, as hemolytic dissociation of the C-C bond, radical recombination, beta scission, isomerization, and hydrogen abstraction. The product selectivity is determined by the strength of the C-H and C-C bonds in the reactant molecules; an alternative approach for H 2 selectivity through steam reforming (SR) of iso-octane was proposed by Kopasz et al. [16]. iso-Octane is initially broken down by thermolysis into lighter hydrocarbons, which are reformed by steam on the catalytic surface. The following simplified reaction scheme has been proposed and includes reforming/decomposition reactions and secondary chemical steps involving the different intermediate products [45].
C`H 2 O Ñ CO`H 2 ∆H 0 " 131 kJ{mol (10) Figure 9 shows the effects of temperature on conversion and product selectivity over 18 wt. % Ni 0.5 -Cu 0.5 /γ-Al 2 O 3 catalyst at a 0.9 ratio of S/C. Steam reforming of iso-octane results mainly in H 2 , CO, CO 2 , and CH 4 . As the reforming temperature increases from 500 to 750˝C, the CO 2 concentration decreases slightly from 15.1% to 9.4%; the CO concentration remained almost constant (7%), while a strong increase in the CH 4 concentration from 6.9% to 21.2% was obtained. The concentration of H 2 dropped from the highest concentration of 71% to 61.8% due to the reforming temperature increasing from 500 to 750˝C. This indicates that the methanation reactions between the carbonaceous compounds (C, CO, CO 2 ) and H 2 are favored at high temperatures (reaction 10, 11, and 12). Therefore, a decrease in H 2 was observed at high temperatures, accompanied by an increase in CH 4 concentration. It should also be noted that the CO concentration was almost stable at high temperatures, probably implying a balance between the CO production and CO consumption (reactions 5, 8-10, and [13][14]. A similar result on product selectivity was also observed by Ming [52]. The conversion of iso-octane is also shown in Figure 9. At 500˝C, only 38.1% of iso-octane was converted to gas compositions, while nearly 100% conversion was observed at 750˝C. As the reaction temperature increased, the energies for steam reforming (reactions 5 and 6), cracking (reaction 7), and dry reforming reactions (reaction 8) were promoted, therefore conversion performance was improved.
Catalysts 2016, 6, 45 10 of 17 compounds (C, CO, CO2) and H2 are favored at high temperatures (reaction 10, 11, and 12). Therefore, a decrease in H2 was observed at high temperatures, accompanied by an increase in CH4 concentration. It should also be noted that the CO concentration was almost stable at high temperatures, probably implying a balance between the CO production and CO consumption (reactions 5, 8-10, and [13][14]. A similar result on product selectivity was also observed by Ming [52]. The conversion of iso-octane is also shown in Figure 9. At 500 °C, only 38.1% of iso-octane was converted to gas compositions, while nearly 100% conversion was observed at 750 °C. As the reaction temperature increased, the energies for steam reforming (reactions 5 and 6), cracking (reaction 7), and dry reforming reactions (reaction 8) were promoted, therefore conversion performance was improved. Figure 9. Effect of temperature on conversion and product selectivity over 18 wt. % Ni0.5-Cu0.5/γ-Al2O3 catalyst, N2 = 15 cm 3 /min; iso-octane feed rate was 0.03 g/min; S/C molar ratio: 0.9.

Effect of Ni-Cu
Loading on γ-Al2O3 Support Figure 10 shows the effects of conversion and product selectivity when Ni0.5-Cu0.5 increased from 6 wt. % to 36 wt. %. The results show that the iso-octane conversion increased with increasing Ni-Cu content. It is a fact that with the increase of Ni-Cu loading, a higher interaction between CuO and NiO as well as their occupancy of surface sites of γ-Al2O3 were seen, therefore enhancing the isooctane conversion [28]. The product selectivity were not dependent upon the metal loading; the results showed that product selectivity was almost identical for 6 wt. %, 18 wt. %, and 36 wt. % Ni-Cu loading ( Figure 10).   Figure 10 shows the effects of conversion and product selectivity when Ni 0.5 -Cu 0.5 increased from 6 wt. % to 36 wt. %. The results show that the iso-octane conversion increased with increasing Ni-Cu content. It is a fact that with the increase of Ni-Cu loading, a higher interaction between CuO and NiO as well as their occupancy of surface sites of γ-Al 2 O 3 were seen, therefore enhancing the iso-octane conversion [28]. The product selectivity were not dependent upon the metal loading; the results showed that product selectivity was almost identical for 6 wt. %, 18 wt. %, and 36 wt. % Ni-Cu loading ( Figure 10).
Catalysts 2016, 6, 45 10 of 17 compounds (C, CO, CO2) and H2 are favored at high temperatures (reaction 10, 11, and 12). Therefore, a decrease in H2 was observed at high temperatures, accompanied by an increase in CH4 concentration. It should also be noted that the CO concentration was almost stable at high temperatures, probably implying a balance between the CO production and CO consumption (reactions 5, 8-10, and [13][14]. A similar result on product selectivity was also observed by Ming [52]. The conversion of iso-octane is also shown in Figure 9. At 500 °C, only 38.1% of iso-octane was converted to gas compositions, while nearly 100% conversion was observed at 750 °C. As the reaction temperature increased, the energies for steam reforming (reactions 5 and 6), cracking (reaction 7), and dry reforming reactions (reaction 8) were promoted, therefore conversion performance was improved. Figure 9. Effect of temperature on conversion and product selectivity over 18 wt. % Ni0.5-Cu0.5/γ-Al2O3 catalyst, N2 = 15 cm 3 /min; iso-octane feed rate was 0.03 g/min; S/C molar ratio: 0.9.

Effect of Ni-Cu
Loading on γ-Al2O3 Support Figure 10 shows the effects of conversion and product selectivity when Ni0.5-Cu0.5 increased from 6 wt. % to 36 wt. %. The results show that the iso-octane conversion increased with increasing Ni-Cu content. It is a fact that with the increase of Ni-Cu loading, a higher interaction between CuO and NiO as well as their occupancy of surface sites of γ-Al2O3 were seen, therefore enhancing the isooctane conversion [28]. The product selectivity were not dependent upon the metal loading; the results showed that product selectivity was almost identical for 6 wt. %, 18 wt. %, and 36 wt. % Ni-Cu loading ( Figure 10).   Figure 11 shows the effect of the Cu molar ratios on conversion and product selectivity over 18 wt. % Ni x -Cu 1´x /Al 2 O 3 (x = 1, 0.7, 0.5, 0.3, 0) catalysts at the low temperature reaction of 550˝C. For the 18 wt. % Ni/γ-Al 2 O 3 catalysts, it exhibited poor catalytic activity with a maximum iso-octane conversion of 20.6%. It was also observed that iso-octane conversion increased from 20.6% to 42.6% with increasing Cu loading up to 0.5 molar, while H 2 and CH 4 selectivity reached values of 70.5% and 6.7%, respectively; CO and CO 2 selectivity decreased slightly. This fact was probably caused by Cu, which enhanced the reducibility of the dispersed Ni species, and/or by Ni, which favored the segregation of Cu 2+ ions to the surface [28,53]. This is in agreement with the analysis of XRD and H 2 -TPR; in this way the iso-octane conversion was enhanced. It seems that a catalyst containing about 0.5 molar of Cu is suitable for steam reforming reactions (reactions 5-8); as a result, high hydrogen concentration is observed. At higher Cu loadings of 0.5 molar, iso-octane conversion and H 2 selectivity decreased while higher CH 4 formation was favored. The phenomenon could be caused by agglomeration of bulk Cu particles that were dispersed on the Ni surface, therefore inhibiting the reducibility of Ni. This indicated that the methanation reaction (reactions 10-12) should take preference at higher Cu loadings of 0.5 molar [54]. For 18 wt. % Cu/γ-Al 2 O 3 catalysts, it also showed poor catalytic activity, with a maximum iso-octane conversion of 32.4%; hydrogen selectivity reached a poor maximum value of 55.4% which is much lower than that obtained for the bimetallic Ni-Cu catalyst, and the highest CH 4 selectivity of 21.5% was observed.
Catalysts 2016, 6, 45 11 of 17 Figure 11 shows the effect of the Cu molar ratios on conversion and product selectivity over 18 wt. % Nix-Cu1−x/Al2O3 (x = 1, 0.7, 0.5, 0.3, 0) catalysts at the low temperature reaction of 550 °C. For the 18 wt. % Ni/γ-Al2O3 catalysts, it exhibited poor catalytic activity with a maximum iso-octane conversion of 20.6%. It was also observed that iso-octane conversion increased from 20.6% to 42.6% with increasing Cu loading up to 0.5 molar, while H2 and CH4 selectivity reached values of 70.5% and 6.7%, respectively; CO and CO2 selectivity decreased slightly. This fact was probably caused by Cu, which enhanced the reducibility of the dispersed Ni species, and/or by Ni, which favored the segregation of Cu 2+ ions to the surface [28,53]. This is in agreement with the analysis of XRD and H2-TPR; in this way the iso-octane conversion was enhanced. It seems that a catalyst containing about 0.5 molar of Cu is suitable for steam reforming reactions (reactions 5-8); as a result, high hydrogen concentration is observed. At higher Cu loadings of 0.5 molar, iso-octane conversion and H2 selectivity decreased while higher CH4 formation was favored. The phenomenon could be caused by agglomeration of bulk Cu particles that were dispersed on the Ni surface, therefore inhibiting the reducibility of Ni. This indicated that the methanation reaction (reactions 10-12) should take preference at higher Cu loadings of 0.5 molar [54]. For 18 wt. % Cu/γ-Al2O3 catalysts, it also showed poor catalytic activity, with a maximum iso-octane conversion of 32.4%; hydrogen selectivity reached a poor maximum value of 55.4% which is much lower than that obtained for the bimetallic Ni-Cu catalyst, and the highest CH4 selectivity of 21.5% was observed. Figure 11. Effect of Cu molar ratio on conversion and product selectivity over 18 wt. % Cu-Ni/Al2O3 catalyst; T = 550 °C, N2 = 15 cm 3 /min; iso-octane feed rate was 0.03 g/min; S/C molar ratio: 0.9.

Effect on Catalytic Performance
Conversion and product selectivity for steam reforming of iso-octane on Al2O3-supported catalysts at the low temperature of 550 °C are summarized in Table 3. Here it is presented that isooctane conversion over 18 wt. % Ni0.5-Cu0.5/γ-Al2O3 catalyst was much higher than that of 18 wt. % Ni0.5-Mo0.5/γ-Al2O3 and of 18 wt. % Ni0.5-Ce0.5/γ-Al2O3 catalysts; similar product selectivity was observed for all catalysts. Wang et al. [55] reported steam reforming of gasoline over the Ni-Ce/γ-Al2O3 catalyst and found that the Ce addition improved oxygen mobility and therefore enhanced the activity maintenance of the catalyst. A better activity maintenance of Ni-Ce/γ-Al2O3 than Ni/γ-Al2O3 was exhibited at a temperature of 690 °C. It was also reported that the largest and most desirable changes of nickel catalytic properties in steam reforming of hydrocarbon were observed with small amounts of Mo at high temperatures [47]. The best catalytic performance of Ni0.5-Cu0.5/γ-Al2O3 catalyst demonstrated in Table 3 is consistent with TPR results (Figure 6), of which the superior reducibility of the Ni-Cu/γ-Al2O3 catalyst was reached at low temperatures. Figure 11. Effect of Cu molar ratio on conversion and product selectivity over 18 wt. % Cu-Ni/Al 2 O 3 catalyst; T = 550˝C, N 2 = 15 cm 3 /min; iso-octane feed rate was 0.03 g/min; S/C molar ratio: 0.9.

Effect on Catalytic Performance
Conversion and product selectivity for steam reforming of iso-octane on Al 2 O 3 -supported catalysts at the low temperature of 550˝C are summarized in Table 3. Here it is presented that iso-octane conversion over 18 wt. % Ni 0.5 -Cu 0.5 /γ-Al 2 O 3 catalyst was much higher than that of 18 wt. % Ni 0.5 -Mo 0.5 /γ-Al 2 O 3 and of 18 wt. % Ni 0.5 -Ce 0.5 /γ-Al 2 O 3 catalysts; similar product selectivity was observed for all catalysts. Wang et al. [55] reported steam reforming of gasoline over the Ni-Ce/γ-Al 2 O 3 catalyst and found that the Ce addition improved oxygen mobility and therefore enhanced the activity maintenance of the catalyst. A better activity maintenance of Ni-Ce/γ-Al 2 O 3 than Ni/γ-Al 2 O 3 was exhibited at a temperature of 690˝C. It was also reported that the largest and most desirable changes of nickel catalytic properties in steam reforming of hydrocarbon were observed with small amounts of Mo at high temperatures [47]. The best catalytic performance of Ni 0.5 -Cu 0.5 /γ-Al 2 O 3 catalyst demonstrated in Table 3 is consistent with TPR results (Figure 6), of which the superior reducibility of the Ni-Cu/γ-Al 2 O 3 catalyst was reached at low temperatures. Table 3. Conversion and product selectivity for steam reforming of iso-octane on Al 2 O 3 supported catalysts, T = 550˝C, N 2 = 15 cm 3 /min; iso-octane feed rate was 0.03 g/min; S/C molar ratio: 0.9.  Figure 12 shows hydrogen selectivity and iso-octane conversion over different catalysts with the time on steam of 30 h. For Ni 0.5 -Cu 0.5 /γ-Al 2 O 3 catalyst, the conversion of iso-octane and H 2 selectivity were stabilized for the first hours and thereafter slightly decreased when the time on steam was extended to 30 h. The same tendency of hydrogen selectivity and iso-octane conversion was observed over the Ni 0.5 -Ce 0.5 /γ-Al 2 O 3 and Ni 0.5 -Mo 0.5 /γ-Al 2 O 3 catalysts, this result is consistent with previous studies [47,48]. It is widely accepted that the deactivation of the Ni catalyst in the steam reforming of fossil fuels is mainly due to carbon deposition on the catalyst surface [20]. A large amount of coke was deposited on the surface of the Ni catalyst, resulting in a loss of catalytic activity for C-C bond cleavage. It is believed that Mo species incorporated with the Ni catalyst suppress coke formation. Mo species not only decrease the coking rate but also prolong the induction period of coke formation [56]. It was also reported that the highly dispersed Mo species served as a barrier for preventing the growth of Ni particles [56]. This is the reason for the low rate of coke deposition on the Ni-Mo catalyst. It was also reported that the Ce content in Ni 0.5 -Ce 0.5 /γ-Al 2 O 3 catalyst can make highly dispersed Ni particles with a strong metal-support interaction, resulting in high coke resistance [48]. The results in Figure 12 also show a higher decrease in the conversion of iso-octane over Ni 0.5 -Cu 0.5 /γ-Al 2 O 3 catalyst. This phenomenon could be caused by agglomeration of copper particles. It was reported that Cu-based catalysts could easily be deactivated, caused by agglomeration of copper particles when reaction temperatures are over 300˝C [57]. Figure 13 shows SEM micrograph and EDX mapping of Ni 0.5 -Cu 0.5 /Al 2 O 3 catalysts after 30 h on steam. The results show that the Cu particles become larger and the carbon particles deposited on the surface can also be clearly observed.   Figure 12 shows hydrogen selectivity and iso-octane conversion over different catalysts with the time on steam of 30 h. For Ni0.5-Cu0.5/γ-Al2O3 catalyst, the conversion of iso-octane and H2 selectivity were stabilized for the first hours and thereafter slightly decreased when the time on steam was extended to 30 h. The same tendency of hydrogen selectivity and iso-octane conversion was observed over the Ni0.5-Ce0.5/γ-Al2O3 and Ni0.5-Mo0.5/γ-Al2O3 catalysts, this result is consistent with previous studies [47,48]. It is widely accepted that the deactivation of the Ni catalyst in the steam reforming of fossil fuels is mainly due to carbon deposition on the catalyst surface [20]. A large amount of coke was deposited on the surface of the Ni catalyst, resulting in a loss of catalytic activity for C-C bond cleavage. It is believed that Mo species incorporated with the Ni catalyst suppress coke formation. Mo species not only decrease the coking rate but also prolong the induction period of coke formation [56]. It was also reported that the highly dispersed Mo species served as a barrier for preventing the growth of Ni particles [56]. This is the reason for the low rate of coke deposition on the Ni-Mo catalyst. It was also reported that the Ce content in Ni0.5-Ce0.5/γ-Al2O3 catalyst can make highly dispersed Ni particles with a strong metal-support interaction, resulting in high coke resistance [48]. The results in Figure 12 also show a higher decrease in the conversion of iso-octane over Ni0.5-Cu0.5/γ-Al2O3 catalyst. This phenomenon could be caused by agglomeration of copper particles. It was reported that Cubased catalysts could easily be deactivated, caused by agglomeration of copper particles when reaction temperatures are over 300 °C [57]. Figure 13 shows SEM micrograph and EDX mapping of Ni0.5-Cu0.5/Al2O3 catalysts after 30 h on steam. The results show that the Cu particles become larger and the carbon particles deposited on the surface can also be clearly observed.

Catalyst Characterization
The structures, the morphological aspects and the compositions of catalysts were analyzed by X-ray diffractometry (XRD) using Cu Kα radiation (RIGAKU RINT-2100CMT, Tokyo, Japan), scanning electron microscopy (SEM, Hitachi, SU6600 EVACSEQ, Tokyo, Japan and JOEL JSM7600F, Tokyo, Japan) and EDX (energy dispersive X-ray, X-Max50, Tokyo, Japan). The surface area was estimated by the N 2 adsorption at´196˝C, using the multipoint BET analysis method (ASAP 2010, Norcross, GA, USA). The total pore volume was calculated based on the adsorbed nitrogen at the highest relative pressure, while the average pore size diameter was determined by the Barrett, Joyner, and Halenda (BJH) method [25].
Hydrogen temperature programmed reduction (H 2 -TPR) was conducted to determine the reductive properties of catalysts; 500 mg of catalyst was put in a quartz flow micro reactor and heated to 700˝C at a heating rate of 3˝C/min under 10% H 2 /He with a total flow of 25 mL/min. Prior to TPR profile measurement, all samples were pretreated at 600˝C for 1 hour under 20% O 2 /N 2 and cooled to room temperature. The consumption of H 2 was detected by a gas chromatography (Thermal Trace GC, San Jose, CA, USA).

Catalytic Activity
iso-Octane (Merck Chemicals, Darmstadt, Germany, pure 99.999%) was used as a gasoline surrogate for the simplicity of the experimental procedure. Catalytic activities were carried out with about 500 mg of catalyst placed in a fixed bed continuous-flow quartz reactor. Before the steam reforming reaction, all catalysts were preheated by flowing air at 600˝C for 1 hour then reduced under 10% H 2 /He at 400˝C for 2 h. Flows of liquid hydrocarbon (iso-octane) and water were controlled by liquid pumps and preheated in an evaporator before being fed into the reactor. Pure 99.99% N 2 was used as the carrier gas and internal standard; the flow rate of N 2 was controlled by a mass flow rate-controller at a value of 15 mL/min. The products were withdrawn periodically from the outlet of a condenser installed at the reactor exit to collect water. The gas mixture was analyzed by a gas chromatograp, (Thermal Trace GC RGA) equipped with a thermal conductivity detector. Conversion, selectivity, and formation rates of products were calculated by an internal standard analyzing method as reported earlier [10,18],whereby, the flow rate of nonreactive internal standard gas (F st ) was kept constant.
F st " F outlet total X outlet st " F inlet total X inlet st (11) Here, F inlet total and F outlet total were the total flow rates at inlet and outlet of the reactor, respectively. X st refers to the concentration of the internal standard gas.
The formation rate of i product (F i ) can be calculated as: Hydrocarbon conversion (C HC ) can be calculated on the basis of the carbon balance as F inlet gasoline and F inlet H 2 O are the flow rates of the gasoline and water feed, respectively, which were controlled by liquid pumps.

Conclusions
Steam reforming of gasoline with hydrogen selectivity over Ni-Cu/γ-Al 2 O 3 catalysts was investigated at low temperatures. TPR analysis of Ni-Cu/Al 2 O 3 catalysts indicated a shift in T max towards a lower temperature; the addition of copper improved the dispersion of nickel and therefore facilitated the reduction of the Ni at lower temperatures. The catalytic performance of Ni-Cu/γ-Al 2 O 3 catalysts was improved by the addition of Cu, which enhanced the reducibility of the dispersed Ni species. The conversion of gasoline with Ni-Cu/Al 2 O 3 catalysts was significantly higher than that with nickel catalysts at a low temperature of 550˝C close to the normal SI engine exhaust gas temperature. On the other hand, initial durability testing showed that the conversion of gasoline slightly decreased after 30 h with steam, which was ascribed mainly to carbon deposition and agglomeration of copper particles. Therefore, Ni-Cu/Al 2 O 3 could be effective in hydrogen selectivity when applied in SI engines.