Synthesis of CuNi/C and CuNi/γ-Al2O3 Catalysts for the Reverse Water Gas Shift Reaction

A new polyol synthesis method is described in which CuNi nanoparticles of different Cu/Ni atomic ratios were supported on both carbon and gamma-alumina and compared with Pt catalysts using the reverse water gas shift, RWGS, reaction. All catalysts were highly selective for CO formation. The concentration of CH 4 was less than the detection limit. Cu was the most abundant metal on the CuNi alloy surfaces, as determined by X-ray photoelectron spectroscopy, XPS, measurements. Only one CuNi alloy catalyst, Cu 50 Ni 50 /C, appeared to be as thermally stable as the Pt/C catalysts. After three temperature cycles, from 400 to 700C, the CO yield at 700C obtained using the Cu 50 Ni 50 /C catalyst was comparable to that obtained using a Pt/C catalyst.


Introduction
The carbon dioxide hydrogenation reaction has been proposed for use with carbon capture technologies for the production of industrially viable chemicals, such as long chain hydrocarbons, methanol, formic acid, and carbon monoxide [1,2]. When cofeeding CO 2 and H 2 over a hydrogenation catalyst, there are two main hydrogenation processes that can take place, the reverse water gas shift reaction and the subsequent hydrogenation of CO to either hydrocarbons or alcohols, depending on the values of , , and in One of the reactions in (2), the hydrogenation of CO to methane, is of particular interest in this study: When CO is selectively formed via (1) and mixed with H 2 , the resulting syngas can be a feed-stock for the Fischer Tropsch process [2] that produces liquid fuels. In contrast CH 4 formed via (3) is an undesirable by-product that is not convertible to liquid fuels in a Fischer Tropsch process. Wang et al. [1] have recently reviewed catalysts for the RWGS reaction. They reported that noble metals have been studied and shown to be among the best catalysts for the RWGS reaction because they generally promote H 2 dissociation. Among noble metals used for the RWGS reaction, platinum (Pt) has received considerable attention. It was found to produce high CO yields [3][4][5]. In addition, Pekridis et al. [6] tested the electrokinetics of the RWGS reaction in solid oxide fuel cells containing a Pt/YSZ catalyst. The Pt/YSZ catalyst was found to be stable at high temperatures and the cell achieved a maximum power density of 9 mW/cm 2 [6]. In spite of the recognized performance of noble metals as catalysts for the RWGS reaction, their main drawback is their high cost that limits their commercial application.
Transition metals such as copper (Cu), nickel (Ni), and iron (Fe) are promising alternatives to noble metal catalysts for the RWGS reaction. Both copper and nickel based catalysts have shown good conversion for the WGS reaction as well as the RWGS reaction [7][8][9][10][11][12]. Chen et al. [7] investigated Cu nanoparticles ranging in size from 2.4 to 3.4 nm and found that the catalyst becomes unstable at higher temperatures [13]. These researchers [14] also added Fe to Cu in an attempt to stabilize the catalyst. Although Fe alone had poor conversions, Fe stabilized the Cu catalyst for 120 h and caused an increase in conversion of approximately 7% at 600 ∘ C. On the other hand, the Cu catalyst without the iron stabilizer was deactivated rapidly and reached zero conversion after 120 hours.
Similar research was performed by Chen et al. [15] using Ni catalysts. Ni alone showed high selectivity towards methane. When they added potassium to a Ni/ -Al 2 O 3 catalyst they reported higher selectivity towards CO even though they did not notice an increase in CO 2 conversion. However with the potassium promoter they noticed the formation of coke.
Y. Liu and D. Liu [16] studied a Ni-Cu catalyst that was prepared by immersing gamma-alumina ( -Al 2 O 3 ) in an aqua ammonia solution of nickel nitrate and copper nitrate. Their catalysts were not selective in that they reported large yields of both CH 4 and CO. They interpreted their results as CO 2 being adsorbed on Cu and H 2 being adsorbed on Ni.
The polyol synthesis method has been used extensively in the past for the synthesis of metal particles from a metal salt precursor [Cu(NO 3 ) 2 or Ni(NO 3 ) 2 represented here as Me(NO 3 ) 2 ]. Bonet et al. [17,18] indicated that the overall reactions at 180 ∘ C included the following reactions to form acetic acid, glycolaldehyde, and glycolic acid: Bock et al. [19] found that oxalic acid, HOOC-COOH, is also formed and suggested that the majority of the metal is formed by the oxidation of ethylene glycol to glycolic acid. At the boiling point of ethylene glycol, 196 ∘ C, Poul et al. [17] indicated that the overall reactions included the formation of diacetyl: where Me 0 represents either copper or nickel in the metallic state. Poul et al. [17] also commented on the reaction mechanism and indicated that intermediate solid phases (metal glycolates) precipitate before the metal powder is formed. Bonet et al. [18] indicated that oxidation products containing carboxylic acids act as stabilizers for metal colloid particles. Specifically they indicated that glycolate anions, the deprotonated form of glycolic acid, are good stabilizers for colloidal metal particles and that their concentration increases when the pH is greater than 6. In their work, they increased the pH by the addition of NaOH and reported that PtRu bimetallic nanoparticle sizes decreased when the pH was increased. Their explanation of glycolate anions on the exterior of the metal particles preventing metal colloid agglomeration seems to be consistent with their nanoparticle size results. The polyol synthesis method has become known for its simplicity and accurate control of particle size [17,18].
Other researchers [16,18,20] have also used the polyol synthesis method to obtain bimetallic particles. When alloying two metals together, the resulting reaction properties can often be enhanced compared to the pure metal. In the past both Cu and Ni have been alloyed with other metals to form alloys. For instance, Viau et al. [20] prepared Co-Ni and Fe-Ni particles using the polyol synthesis method.
Bonet et al. [18] used a polyol synthesis method to obtain CuNi particles. When nickel carbonate and copper carbonate were used at 140 ∘ C they obtained a CuNi powder composed of both a Ni rich CuNi solid solution and a Cu rich CuNi solid solution. When the carbonates were used at 196 ∘ C they obtained a CuNi powder composed of a Cu rich CuNi solid solution and a solid Ni metal phase. They noted that the reduction temperature for Cu is less than that for Ni. Their particles had a particle size of 140 nm. They did not report any reaction results.
In this work base metal CuNi nanoparticle catalysts were prepared by a new synthesis technique and used for the RWGS reaction. Cu was chosen because it is selective for the formation of CO [16], although it is unstable (sinters) at the higher temperatures where the equilibrium for the RWGS reaction is more favourable. Ni was chosen because it also produces CO [15] although it also can form unwanted byproducts, CH 4 , and coke.
The main objective of this investigation was to obtain physical characterization of the CuNi nanocatalyst and to use this catalyst in the RWGS reaction. The results obtained with the CuNi catalyst were then compared to Pt nanoparticle catalysts that were already synthesized and tested using a variety of reactions: ethylene oxidation [21], CO oxidation [22], and toluene oxidation [23]. These Pt catalysts are considered to be among the best catalysts for the RWGS reaction because they achieve reaction equilibrium at some conditions. Another purpose of the investigation was to determine if sintering of pure Cu could be prevented by the addition of Ni in the same way that it was prevented by the addition of Fe [14]. In what follows, we report CuNi catalyst compositions that promote CO formation and inhibit CH 4 formation at specific reaction conditions.

Catalyst Preparation.
CuNi nanoparticles were synthesised using a modified polyol technique. Nickel nitrate (Ni (NO 3 ) 2 ) (hexahydrate 99.999% metal basis, Alfa Aesar) was dissolved in ethylene glycol (anhydrous 99.8%, Sigma Aldrich). Next, an increase in pH to 11 was achieved by sodium hydroxide (NaOH) pellets (EM Science, ACS grade) to obtain the first solution. Separately, copper nitrate (Cu (NO 3 ) 2 ) (hexahydrate 99.999% metal basis, Alfa Aesar) was also dissolved in ethylene glycol. Its pH was also increased to A wet impregnation technique was used to deposit the colloidal particles on supports. The powdered support was first placed into a beaker. Then, a specific amount of colloidal particles was injected in the powder. A nominal amount of 10 wt% of CuNi was chosen. The mixture was sonicated for 1 hour and stirred for 24 hours. The catalyst was then centrifuged and washed with deionized water several times in order to remove any salts remaining from the synthesis procedure. The supports used were carbon black (Vulcan XC-72R, Cabot Corp., specific surface area of 254 m 2 /g) and gamma-alumina (Alfa Aesar, specific surface area of 120 m 2 /g). A freeze dryer was used to dry the catalyst. The catalyst was finely crushed prior to all experiments.
CuNi particles of three different compositions were prepared. The combined solution for each composition was prepared with a different ratio of Solution 1 to Solution 2. The ratio was selected to obtain CuNi colloidal particles of 80 wt% Cu/20 wt% Ni (nominally Cu 80 Ni 20 ), 50 wt% Cu/50 wt% Ni (nominally Cu 50 Ni 50 ), and 20 wt% Cu/80 wt% Ni (nominally Cu 20 Ni 80 ).
Pt nanoparticles were synthesized using a different polyol synthesis method described here [22]. In short, PtCl 4 was dissolved in a 0.06 M NaOH solution of ethylene glycol and refluxed at 160 ∘ C for 3 hours. The nanoparticles were then deposited on C and -Al 2 O 3 using the same deposition technique described previously. A nominal value of 1 wt% of Pt was chosen.

Physical
Characterization. Supported Pt nanoparticles were analyzed using transmission electron microscopy (TEM) with a JEOL JEM 2100F FETEM operating at 200 kV. A particle size distribution was obtained using ImageJ software. More in-depth characterisation of the Pt nanoparticle is described elsewhere [20].
Scanning electron microscopy (SEM) was conducted on carbon supported CuNi nanoparticles using a JEOL model JSM-7500F field emission scanning electron microscope, FESEM, in both lower-secondary electron image, LEI, and compositional, COMPO, modes set at a distance of 8 mm with an acceleration voltage of 5 kV. In addition, an energydispersive X-ray spectroscope (EDS) operating with the SEM was used to obtain a quantifiable amount of Cu and Ni in the CuNi particles.
CuNi colloidal particles were analyzed using X-ray diffraction (XRD) with a Rigaku Ultima IV diffractometer which used a Cu K X-ray (40 ma, 44 kV) operating with focused beam geometry and a divergence slit of 2/3 degree, a scan speed of 0.17 deg min −1 , and a scan step of 0.06 degrees were used while operating between 35 ∘ and 55 ∘ . Table 1 demonstrates the crystal sizes obtained through XRD.

Reaction Experiments.
The performances of the supported CuNi catalysts were evaluated using the RWGS reaction. 50 mg of powdered catalyst was placed on a fritted quartz bed within a 35 mL quartz tube to act as a fixed bed reactor. A gas mixture of 1 kPa H 2 (Grade 4.0, Linde), 1 kPa CO 2 (Grade 3.0, Linde), and the balance He (Grade 4.7 Linde) flowed through the reactor at a total flow rate of 510 mL/min. The reaction was performed at atmospheric pressure using three consecutive temperature cycles. Each temperature cycle consisted of a series of experiments over the temperature range from 400 ∘ C to 700 ∘ C. Before each experiment, the temperature was held constant for 30 min. Although the same mass of catalyst was used in each experiment, the gas hourly space velocity (GHSV) was different because the supports had different bulk densities (288 g/L for the carbon support and 461 g/L for the alumina support). The GHSV values were 176000 h −1 and 282000 h −1 , respectively, for CuNi/C and CuNi/Al 2 O 3 catalysts. The effluent was dehumidified by flowing through an adsorbent and was analyzed by flowing through a mass spectrometer (Ametek Proline DM 100) and a nondispersive infrared CO gas analyzer (Horiba VIA-510). Each set of experiments was repeated three times (24 hrs total) in order to examine reproducibility and stability. The yield of CO was calculated using the following formula: The mass spectrometer identified any by-products that were formed via side reactions such as CO methanation. The mass spectrometer indicated the presence of gases with a molecular weight of up to 50 atomic units and had a detection limit of 50 ppm.

Results and Discussion
The X-ray diffraction spectra of the CuNi nanoparticles are shown in Figure 1. The positions of the peaks for both pure Cu (2 = 43.2, ICSD Collection Code 53246) and pure Ni (2 = 44.6, ICSD Collection Code 43397) are shown as straight vertical lines in Figure 1. They are the X-ray reflections from the 111 crystal lattice planes. These peaks have corresponding lattice constants of 3.627Å and 3.519Å for pure Cu and pure Ni, respectively. The smaller peaks near 2 values of 50.4 and 51.5 are the reflections from the 200 crystal lattice planes of Cu and Ni, respectively. No other species or oxides were identified.
There is a slight difference between the peak positions for the pure metals and the metals in the catalysts. Deviations exist because the catalysts are bimetallic solid solutions rather than pure metals. Because of these shifts, the lattice constants of each alloy are shifted. The catalyst that is nominally Cu 50 Ni 50 in Figure 1(b) had a first 2 peak position of 43.385 and a lattice constant of 3.612Å which is slightly greater than the one for pure Cu. A second peak is observed near the first 2 peak in Figure 1(b) which has a peak position of 44.346 and a lattice constant of 3.537Å for Ni, which is slightly less than the one for pure Ni. These changes in lattice constants suggest that the two peaks represent a Cu rich alloy and a Ni rich alloy.
There is only one peak for the Cu 80 Ni 20 catalyst in Figure 1(c). That means that all of the Ni was soluble in the Cu lattice. The single peak in Figure 1(c) is experimental evidence for a copper-rich CuNi solid solution in which the spacing between planes of the catalyst lattice is close to that of pure copper.
There are two peaks for the Cu 20 Ni 80 catalyst in Figure 1(a). Because the first peak has a 2 value at 43.394 and a lattice constant of 3.611Å, close to that for pure Cu, the spacing between its planes will be similar to pure copper. Because the second peak has a 2 value at 44.412 and a lattice constant of 3.533Å, close to that for pure Ni, the spacing between its planes will be similar to pure nickel. Even though there is a large disparity between the bulk Cu content and the bulk Ni content of the catalyst, the two peaks appear to have similar areas. Therefore a substantial amount of Ni must be dissolved in the first Cu-like peak, and the first peak must represent a CuNi solid solution. Since the second peak has a lattice constant slightly different from pure Ni it will contain some Cu making it a Ni rich NiCu solid solution.
The observation of solid solutions is consistent with other works reported in the literature. Bonet et al. [18] synthesized CuNi particles using a similar technique. In their work they refluxed copper and nickel carbonates starting materials in ethylene glycol. After 39 hours at 140 ∘ C they observed the presence of a copper-rich solid solution, Cu 81 Ni 19 , and a Ni rich solid solution, Ni 86 Cu 14 .
The crystalline size of the synthesized nanoparticles increases with Cu content. A summary of their diameters (15-65 nm) can be found in Table 1. They were calculated from the XRD data using Scherrer's formula. The CuNi particles described by Poul et al. [17] had diameters of 250-400 nm. It is possible that the longer refluxing times and the absence of NaOH may have provided more opportunity for agglomeration.
Pt/C nanoparticles characterisation can be found in [21]. In her work, Isaifan et al. demonstrate that the particles are mainly spherical with a reasonably narrow size distribution.
The dispersion also appears to be relatively high. Numerical data derived from several TEM images indicated that a typical particle size for the Pt/C particles was 2.8 nm (Table 1).
An SEM image for the Cu 50 Ni 50 /C catalyst is shown in Figure 2. The particles appear to be generally spherical and to vary in size. The largest observed particle was approximately 100 nm. There are visible signs of agglomeration which is to be expected since no antiagglomerate such as polyvinylpyrrolidone (PVP) was used.
International Journal of Chemical Engineering 5  Figure 2: SEM image of a Cu 50 Ni 50 /C catalyst.
Typical particle sizes for all of the CuNi catalysts measured by SEM are listed in Table 1. In general the CuNi particle sizes are an order of magnitude larger than the Pt particle size. Since the CuNi metal loading, for example, 10 wt%, is an order of magnitude larger than the Pt loading, for example, 1 wt%, a greater extent of metal particle agglomeration might be expected for the CuNi particles.
An energy-dispersive X-ray spectroscopy, EDS, analysis was also performed on the catalyst. For example, the Cu 80 Ni 20 /C catalyst had a measured composition of 82.7 wt% Cu and 17.3 wt% Ni. The EDS measurement was repeated at two different sites on the catalyst's surface with reproducible results. That EDS result is essentially the same as the XRD composition of Cu 82 Ni 18 that was mentioned above. Both the EDS results and the XRD results were consistent with the Cu 80 Ni 20 nominal composition of the synthesized particles. This indicates that the synthesis method was successful in obtaining the nominal Cu : Ni ratio that was intended.
XPS measurements were performed on the CuNi/C catalysts to determine their surface compositions. The results in Table 2 show that the Cu surface concentration was greater than that of the Cu bulk concentrations for all three of the CuNi/C catalysts. Furthermore, the surface concentration of the Cu always exceeded the surface concentration of the Ni, even for the nominal Cu 20 Ni 80 /C catalyst. That result is consistent with the literature. An early report by van der Plank and Sachtler [24] indicated that Cu was the dominant species on the surface of CuNi alloys. Subsequently Watanabe et al. [25] provided definitive experimental data for the phenomenon. Later Sakurai et al. stated that the phenomenon had been conclusively shown [26].
In order to examine the physical changes of the catalyst, Cu 50 Ni 50 /C was examined by SEM and XPS before and after exposure to high temperatures and reactants. There were no visible signs of additional agglomeration or any other physical changes to the metal in comparison to the unreacted catalyst. This suggests that the catalyst is compositionally stable at temperatures of at least 700 ∘ C. The reverse water gas shift reaction was performed using eight different catalysts. These results can be seen in Figures 3-10. In each figure the results for the 3 consecutive temperature cycles, over the temperature range from 400 ∘ C to 700 ∘ C, are shown. Some of the CuNi catalysts showed slight deactivation between the first and second cycles and also between the second and third cycles. The catalysts supported on gamma-alumina, -Al 2 O 3 , are shown in Figures 7-10. In general the CO yields on the -Al 2 O 3 supported catalyst were slightly greater than those on the carbon supported catalysts in Figures 3-6. For the first temperature cycle the CuNi metal -Al 2 O 3 supported catalysts produced CO yields at 700 ∘ C that was 2-4% less than those produced by the Pt metal -Al 2 O 3 supported catalyst. In addition, only the Pt metal -Al 2 O 3 supported catalyst produced a nonzero CO yield at 400 ∘ C during the first cycle.
The only observable components in the gas stream entering the mass spectrometer were CO 2 , H 2 , CO, trace amounts of H 2 O, and the carrier gas, He. These results indicate that CO was the main product having a typical concentration of 2000 ppm. Other products including CH 4 had concentrations less than the detection limit of the spectrometer, 50 ppm.
The absence of CH 4 in the products was a highly desirable result, since CH 4 is an undesirable by-product if syngas for a Fischer Tropsch process is the goal. Cu is known to favour CO production while CH 4 is known to form on Ni catalysts [1]. Since some of the catalysts used in this work contained 80 wt% Ni the absence of CH 4 might be considered to be inconsistent with the literature [1]. The advantage of the CuNi alloys made using this particular polyol synthesis method is that more than one-half of the surface was composed of Cu, even for catalysts consisting of 80% Ni in bulk metal, as was shown by our XPS results. Perhaps the presence of sufficient Cu on the surface may allow CO to desorb before additional hydrogenation occurs to form CH 4 . The catalysts supported on carbon are shown in Figures  3-6. Cu 80 Ni 20 /C showed the highest yield among all catalysts during the first cycle. In addition, it was the only CuNi metal carbon supported catalyst that produced a CO yield at 400 ∘ C during the first cycle. It is well known that Cu alone has better performance for the RWGS reaction than Ni [1,16,27] since Ni tends to further hydrogenate CO to catalyst seems to cause the increase in catalytic performance observed here. However, because of copper's instability at high temperatures [28], a reduction in performance was expected and is observed over time as shown in Figure 3. The yields of CO at 700 ∘ C using all the CuNi carbon supported catalysts differed from those using Pt by no more than 3%. Deactivation was observed between the first and second temperature cycles for all of the CuNi catalysts supported on carbon. Virtually no deactivation was observed when using the Pt metal carbon supported catalyst. This suggests that the deactivation observed with the CuNi carbon supported catalyst may have been related to the CuNi metal and not to the catalyst support. The Cu 50 Ni 50 /C catalyst in Figure 4 was different from the Cu 20 Ni 80 /C and Cu 80 Ni 20 /C catalysts in that no deactivation occurred between the second and third temperature cycles. This suggests that after sufficient timeon-stream the performance of the Cu 50 Ni 50 /C may become invariant with time and that it may become a thermally stable catalyst.
Furthermore, XPS experiments of the Cu 50 Ni 50 /C showed no differences in carbon surface composition for both of the Cu and Ni atoms. These experiments suggest that coking did not occur. SEM images also showed no change in particle size either before or after testing. The sample seems to be compositionally stable throughout the experiments.
Deactivation was observed between the first and third temperature cycles for all the -Al 2 O 3 supported catalysts, for both CuNi and Pt. The smallest extent of deactivation, 2%, was observed with the Cu 20 Ni 80 -Al 2 O 3 supported catalyst. It is the CuNi catalyst with the smallest Cu content. In contrast the CO yields at 700 ∘ C for Cu 80 Ni 20 and Pt -Al 2 O 3 supported catalysts decreased by over 8% between the first and third temperature cycles. Conversely, no deactivation was observed for Pt metal carbon supported catalysts. This suggests that the -Al 2 O 3 support may contribute to the deactivation. It is known [29] that as the temperature is increased above 500 ∘ C, gamma-alumina can be converted to other phases such as delta alumina, theta alumina, and alpha alumina. Alpha alumina has a much smaller surface area than gamma-alumina. The tendency of the -Al 2 O 3 support to deactivate at high temperatures is consistent with literature data [14,30].
Y. Liu and D. Liu [16] tested a similar catalyst having C 50 Ni 50 / -Al 2 O 3 at a maximum temperature of 600 ∘ C. In their research, Y. Liu and D. Liu use a typical coimpregnation method where -Al 2 O 3 is immersed in a solution of nickel nitrate and copper nitrate. They used twice the metal content (20 wt%) and much smaller GHSVs (much greater residence times in the reactor) and obtained CO 2 conversions that exceeded the ones being reported here. They used a catalyst preparation method that was different compared to the one used here. In fact, it did not seem to be possible to make a CuNi alloy using this technique when it was attempted in our laboratories. Perhaps that may explain why the CH 4 selectivity (e.g., 28.2%) in their work was much greater than in this work. Even though their CO 2 conversions were greater than the ones reported here, their finite selectivity to CH 4 caused their CO yields to be similar to the ones reported here. Y. Liu and D. Liu [16] noticed an increase in CH 4 when the Ni content of the CuNi catalysts was increased.
A comparison of the average CO yields for the carbon supported catalysts is shown in Figure 11. The CO yields obtained with Cu 50 Ni 50 /C catalyst are smaller than those obtained with the other two CuNi catalysts. The Cu 50 Ni 50 /C catalyst was the one that appeared to be compositionally stable according to the XPS results discussed previously. It was also the one that appeared to be the most thermally stable in Figure 4. Perhaps there is an association between minimum CO yield and thermal stability. In other words, perhaps the reaction sites on CuNi catalysts with the largest turnover frequencies are the ones that are the most thermally unstable. Thermal stability was further investigated and is discussed in another work [27]. Here, testing using Cu 50 Ni 50 deposited on samarium-doped ceria showed no sign of deactivation when the catalyst is exposed to temperatures of 600-700 ∘ C for 48 consecutive hours. The high CO yields obtained using Cu 80 Ni 20 /C are believed to be associated with a high Cu concentration on the surface of the catalyst as shown in the XPS results. As mentioned previously, Cu is known to obtain higher CO yields than Ni for the RWGS reaction [16]. A comparison of the average CO yields for the alumina supported catalysts shown in Figure 12 is almost identical regardless of catalyst composition. Nevertheless substantial catalyst deactivation was evident in Figures 7-10. That suggests that the deactivation caused by thermal transitions in alumina (loss of surface area, spinel formation) may have had more influence on catalyst performance than variations in CuNi catalyst composition. In spite of the similarity of the results in Figure 12, the CO yields for the Cu 50 Ni 50 / -Al 2 O 3 catalyst appear to be slightly less than the other two catalyst compositions. That is consistent with the observation in Figure 11 that the Cu 50 Ni 50 /C catalyst also produced smaller yields than the other two CuNi catalysts.
As discussed above, some deactivation occurs when CuNi metal is supported on either carbon or -Al 2 O 3 . As a result, most catalysts composed of CuNi metal supported on either carbon or -Al 2 O 3 will not satisfy one of the main objectives of this research, namely, obtaining a thermally stable catalyst capable of operating under high temperatures. The one exception discussed here is the Cu 50 Ni 50 /C catalyst. Unless deactivation can be mitigated CuNi catalysts will not meet the requirements for an efficient, industrially viable catalyst.

Conclusion
The results of this investigation can be summarized by the following statements: The new polyol synthesis method permits both Cu and Ni metal salts to be reduced at the same time and at the same temperature, by first heating the Ni salt solution to 196 ∘ C and then adding the Cu salt solution that was at room temperature. In the past [18] a Ni rich surface was obtained because Cu(NO 3 ) 2 was reduced first at temperatures as low as 140 ∘ C followed by Ni(NO 3 ) 2 reduction as the solution continued to be heated to 196 ∘ C. Instead, a Cu rich surface is obtained which is ideal for the RWGS reaction because Cu has a higher selectivity towards CO than Ni. The CuNi alloy catalysts investigated in this work are similar to pure Cu catalysts in that they show selectivity for CO formation and the absence of CH 4 formation. The selectivity to CO was attributed to Cu being the most abundant metallic species on the surface of the catalyst, as determined by XPS measurements. Although some of the CuNi alloys show some deactivation, they are not nearly as thermally unstable as pure Cu (sintering) at the higher temperatures that are necessary for the equilibrium of the RWGS reaction to be thermodynamically favorable. Deactivation was observed in each case that an alumina catalyst support was used which was attributed to the instability of alumina at high temperatures. With one carbon supported catalyst, Cu 50 Ni 50 /C, there was no deactivation between the second and third temperature cycles and it appeared to be compositionally stable according to XPS and SEM results. Finally, CO yields at 700 ∘ C during the third temperature cycle of each CuNi catalyst were comparable to those with the Pt/C catalyst.
In conclusion, considering the difference in cost between CuNi alloys and Pt metal, these results suggest that more studies are warranted on the use of CuNi alloy catalysts for the RWGS reaction using the synthesis method described in this work.