Ru – Pd Bimetallic Catalysts Supported on CeO 2-MnO X Oxides as Efficient Systems for H 2 Purification through CO Preferential Oxidation

The catalytic performances of Ru/ceria-based catalysts in the CO preferential oxidation (CO-PROX) reaction are discussed here. Specifically, the effect of the addition of different oxides to Ru/CeO2 has been assessed. The Ru/CeO2-MnOx system showed the best performance in the 80–120 ◦C temperature range, advantageous for polymer-electrolyte membrane fuel cell (PEMFC) applications. Furthermore, the influence of the addition of different metals to this mixed oxide system has been evaluated. The bimetallic Ru–Pd/CeO2-MnOx catalyst exhibited the highest yield to CO2 (75%) at 120 ◦C whereas the monometallic Ru/CeO2-MnOx sample was that one with the highest CO2 yield (60%) at 100 ◦C. The characterization data (H2-temperature programmed reduction (H2-TPR), X-ray diffraction (XRD), N2 adsorption-desorption, diffuse reflectance infrared Fourier transform spectroscopy (DRIFT), X-ray photoelectron spectroscopy (XPS)) pointed out that the co-presence of manganese oxide and ruthenium enhances the mobility/reactivity of surface ceria oxygens accounting for the good CO-PROX performance of this system. Reducible oxides as CeO2 and MnOx, in fact, play two important functions, namely weakening the CO adsorption on the metal active sites and providing additional sites for adsorption/activation of O2, thus changing the mechanism from competitive Langmuir–Hinshelwood into non-competitive one-step dual site Langmuir–Hinshelwood/Mars–van Krevelen. As confirmed by H2-TPR and XPS measurements, these features are boosted by the simultaneous presence of ruthenium and palladium. The strong reciprocal interaction of these metals between them and with the CeO2-MnOx support was assumed to be responsible of the promoted reducibility/reactivity of CeO2 oxygens, thus resulting in the best CO-PROX efficiency at low temperature of the Ru-Pd/CeO2-MnOx catalyst. The higher selectivity to CO2 found on the Ru–Pd system, which reduces the undesired H2 consumption, represents a promising result of this research, being one of the key aims of the design of CO-PROX catalysts.


Introduction
The modern energy crisis and urgent development of green technologies and energies have driven in the last decade intensive research in fuel cell technologies both for stationary and mobile uses.In particular, polymer-electrolyte membrane fuel cells (PEMFC), which work in the 80-100 • C temperature range, are among the most promising devices for automotive purposes [1].
The hydrogen used as fuel in PEMFC is currently produced by steam reforming of hydrocarbons or alcohols and a subsequent water-gas shift reaction.The anode of the PEMFC is, however, easily poisoned even at very low concentrations of CO (less than 10 ppm) [2].Different technologies have been investigated to purify H 2 from traces of CO including: catalytic methanation [3], membrane separation [4] and preferential oxidation of CO in excess of hydrogen (CO-PROX reaction) [5][6][7].The last one, considered highly valuable of investigation, involves two competitive reactions, namely the desired CO oxidation to CO 2 and the undesired H 2 oxidation to water.Therefore, the catalyst of this process must be highly selective towards the first reaction.Noble-metal catalysts, such as Pd [8], Rh [9], Ru [10,11], Au [12][13][14] and Pt [15,16], have revealed good CO-PROX performance.The selection of the proper support has been also found crucial in addressing the catalytic activity.Recently, the use of a reducible/active support, such as cerium oxide or iron oxide, resulted in remarkably low-temperature CO-PROX activities [12,17].The peculiar red-ox properties of CeO 2 involving the facile exchange between Ce 3+ and Ce 4+ states and the high mobility of O 2-ions in the lattice, are the key factors to explain the high relevance of this oxide as a metal support in the PROX reaction [6,12,13,18].
In order to optimize and increase the efficacy of ceria through the rise of the CeO 2 redox properties, one of the most used strategies has been the addition of chemical promoters or the mixing with another oxide such as, for example, CuO [19,20].In particular, the addition of Mn into a CeO 2 crystal lattice was found to increase the thermal stability and to improve the redox properties of ceria by generating a more efficient Ce 3+ /Ce 4+ redox couple [21].For these reasons, the CeO 2 -MnO x mixed oxides have reported good catalytic activity for various oxidation reactions [22][23][24].To the best of our knowledge, few papers deal with the PROX performance of Ru-based catalysts supported on CeO 2 or CeO 2 -based materials, although it has been reported that the Ru-based catalysts were more active and selective than Pt-based ones below 150 • C [10,25], an optimal temperature range for a possible use of these catalysts for the hydrogen purification in the PEMFC or for the alkaline fuel cells (AFC).
On the basis of the above considerations, here we report a study of the catalytic performance in the PROX reaction of Ru supported on ceria and ceria-mixed oxide catalysts.In particular, the effect of the addition of another oxide or another metal to Ru/CeO 2 has been explored both in terms of catalytic activity and chemico-physical properties.

Catalytic Activity
A preliminary screening of the PROX activity of monometallic ruthenium catalysts supported on different ceria-based mixed oxides was carried out to determine the best-performing system.The results of investigated monometallic Ru catalysts in terms of CO and O 2 conversions and selectivity and yield towards the CO oxidation as a function of reaction temperature are depicted in Figure 1.
Figure 1a shows the CO conversion against the reaction temperature.Generally, CO conversion on all samples displays a volcano shape behavior as a function of the reaction temperature, i.e., the CO conversion reaches a maximum and then decreases.Reasonably at high temperature, the reverse water-gas shift (RWGS) reaction (CO 2 + H 2 ↔ CO + H 2 O) becomes significant (Ea = 120-160 kJ/mol, higher than that found for PROX, 70-100 kJ/mol [5,25]).In fact, on Ru-based catalysts RWGS has been reported to occur at T > 150 • C [10,25], therefore justifying the drop of activity at higher temperatures.Specifically, it is worth noting that the CO conversion upper value and its temperature rely on the mixed-oxide support used.This happens also for selectivity to CO 2 (Figure 1c).To comprehend this behavior, the fact that cerium oxide has a strong tendency to provide active oxygens with strong oxidation power [6,12,25,26], which are however reactive towards both CO and H 2 , must be taken into account.This causes a drop of selectivity at higher temperature due to the higher activation energy of H 2 oxidation compared to that of CO [27][28][29][30].
Interestingly, the Ru/CeO 2 -MnO x sample exhibited the highest value of CO conversion at lower temperature (70.6% at 100 • C), which is higher than those found for Ru/CeO 2 -CuO (62.5% at 170 • C), Ru/CeO 2 (54% at 140 • C) and Ru/CeO 2 -Fe 2 O 3 (45.6% at 140 • C) samples.In the case of the Ru/CeO 2 -CuO catalyst, CO conversion shifted to a higher temperature, probably due to the fact that Cu catalysts require higher temperatures to promote the CO oxidation [12,13,25].However, it cannot be fully ruled out that the low activity of the Ru/CeO 2 -CuO catalyst can be related to an unsuitable pretreatment of the CuO-CeO 2 system, for which calcination at 450 • C (adopted here for all samples) could not be ideal for the formation of the active Cu + sites [19,20].
The Ru/CeO 2 catalyst is, instead, the most active as to the O 2 conversion (Figure 1b), which is due to both CO and H 2 oxidation with 100% conversion reached at about 120 • C, i.e., 50  1c, the above behavior results in the much higher selectivity of the Ru/CeO 2 -MnO x sample towards the preferential CO oxidation in the 70-140 • C temperature range, with a maximum at 100 • C.This indicates that the contemporaneous presence of reactive oxygens of cerium and manganese oxides leads to a very good selectivity at low temperature.Accordingly, CO 2 yields as a function of temperature (Figure 1d) clearly show that the Ru/CeO 2 -MnO x catalyst is the most efficient among all investigated systems for PROX in the 80-120 pretreatment of the CuO-CeO2 system, for which calcination at 450 °C (adopted here for all samples) could not be ideal for the formation of the active Cu + sites [19,20].The Ru/CeO2 catalyst is, instead, the most active as to the O2 conversion (Figure 1b), which is due to both CO and H2 oxidation with 100% conversion reached at about 120 °C, i.e., 50 °C less than on Ru/CeO2-Fe2O3 and Ru/CeO2-CuO, whereas the Ru/CeO2-MnOx sample reached 80.9% conversion at 170 °C.As shown in Figure 1c, the above behavior results in the much higher selectivity of the Ru/CeO2-MnOx sample towards the preferential CO oxidation in the 70-140 °C temperature range, with a maximum at 100 °C.This indicates that the contemporaneous presence of reactive oxygens of cerium and manganese oxides leads to a very good selectivity at low temperature.Accordingly, CO2 yields as a function of temperature (Figure 1d) clearly show that the Ru/CeO2-MnOx catalyst is the most efficient among all investigated systems for PROX in the 80-120 °C temperature range.The other monometallic samples were far less active, with only the Ru/CeO2-CuO sample reaching 20% of CO2 yield in the 140-170 °C temperature range.Therefore, the CeO2-MnOx mixed oxide was chosen as a support to investigate other mono (Pt or Pd) and bimetallic (Ru-Pd) catalysts.The other monometallic samples were far less active, with only the Ru/CeO 2 -CuO sample reaching 20% of CO 2 yield in the 140-170 • C temperature range.Therefore, the CeO 2 -MnO x mixed oxide was chosen as a support to investigate other mono (Pt or Pd) and bimetallic (Ru-Pd) catalysts.
Figure 2 displays the CO conversion (Figure 2a), the O 2 conversion (Figure 2b), the selectivity towards the CO oxidation (Figure 2c) and the yield to CO 2 (Figure 2d) of Ru/CeO 2 -MnO x , Pt/CeO 2 -MnO x , Pd/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x .The Pd/CeO 2 -MnO x catalyst exhibited an increased CO conversion with the temperature reaching the maximum of 62% at 140 • C whereas the Pt/CeO 2 -MnO x showed the lowest activity in the inspected temperature range (Figure 2a).Pt/CeO 2 -MnO x reached 100% O 2 conversion (Figure 2b) at lower temperature (100-120 • C) compared to Ru/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x catalysts; therefore, CO selectivity (Figure 2c) and CO 2 yield (Figure 2d) did not exceed 30%.It is noteworthy that the Ru-Pd system leads to higher CO conversion at high temperature (T > 100 • C) compared to the corresponding monometallic samples (Figure 2a), providing at 120 • C the highest values of CO conversion (85%) (Figure 2a), CO selectivity (88.5%) (Figure 2c), and CO 2 yield (75.3%) (Figure 2d) among all catalysts.It must be stressed that the increased activity of the Ru-Pd system cannot be related to the concentration of the metal species, which on both mono and bimetallic samples remains 1 wt % (0.5-0.5% on bimetallics and 1% on monometallics).It has been seen that bimetallics can be often better performing than the related monometallics due to a synergy between the two metals and/or the support [6,12,31].
Catalysts 2018, 8, x FOR PEER REVIEW 4 of 18 Figure 2 displays the CO conversion (Figure 2a), the O2 conversion (Figure 2b), the selectivity towards the CO oxidation (Figure 2c) and the yield to CO2 (Figure 2d) of Ru/CeO2-MnOx, Pt/CeO2-MnOx, Pd/CeO2-MnOx and Ru-Pd/CeO2-MnOx.The Pd/CeO2-MnOx catalyst exhibited an increased CO conversion with the temperature reaching the maximum of 62% at 140 °C whereas the Pt/CeO2-MnOx showed the lowest activity in the inspected temperature range (Figure 2a).Pt/CeO2-MnOx reached 100% O2 conversion (Figure 2b) at lower temperature (100-120 °C) compared to Ru/CeO2-MnOx and Ru-Pd/CeO2-MnOx catalysts; therefore, CO selectivity (Figure 2c) and CO2 yield (Figure 2d) did not exceed 30%.It is noteworthy that the Ru-Pd system leads to higher CO conversion at high temperature (T > 100 °C) compared to the corresponding monometallic samples (Figure 2a), providing at 120 °C the highest values of CO conversion (85%) (Figure 2a), CO selectivity (88.5%) (Figure 2c), and CO2 yield (75.3%) (Figure 2d) among all catalysts.It must be stressed that the increased activity of the Ru-Pd system cannot be related to the concentration of the metal species, which on both mono and bimetallic samples remains 1 wt % (0.5-0.5% on bimetallics and 1% on monometallics).It has been seen that bimetallics can be often better performing than the related monometallics due to a synergy between the two metals and/or the support [6,12,31].

Catalysts Characterization
In order to find relations between catalytic performances and chemico-physical properties, we investigated the reducibility H 2 -temperature programmed reduction (H 2 -TPR), the structural and textural properties (diffuse reflectance infrared Fourier transform spectroscopy (DRIFT), X-ray diffraction (XRD) and Brunauer-Emmett-Teller (BET) surface area) and the surface properties (X-ray photoelectron spectroscopy (XPS)) of the most active Ru-based samples.
2.2.1.Temperature Programmed Reduction (H 2 -TPR) Figure 3 shows the H 2 -temperature programmed reduction profiles of the analyzed samples between 50 • C and 500 • C. On the bare cerium oxide support, the H 2 consumption starts at around 350 • C, due to the partial reduction of the CeO 2 surface oxygen [12,13].The reduction of bulk oxygen was reported to take place at temperatures higher than 700 • C [32] and, therefore, is not visible in the TPR curve of Figure 3.The profile of the Ru/CeO 2 sample displays two reduction peaks at 165 • C and 190 • C and a shoulder at ca. 325 • C. According to literature data, the peak at 165 • C is characteristic of the reduction of the ruthenium precursor salt (Ru(NO)(NO 3 ) 3 ), while the peak at 190 • C is assigned to the reduction of well dispersed ruthenium oxide particles on the surface of the support [10,33].The shoulder at 325 • C reasonably originates from the reduction of the ceria surface oxygens shifted to lower temperature as a result of the interaction with the ruthenium species.
Catalysts 2018, 8, x FOR PEER REVIEW 5 of 18 In order to find relations between catalytic performances and chemico-physical properties, we investigated the reducibility H2-temperature programmed reduction (H2-TPR), the structural and textural properties (diffuse reflectance infrared Fourier transform spectroscopy (DRIFT), X-ray diffraction (XRD) and Brunauer-Emmett-Teller (BET) surface area) and the surface properties (X-ray photoelectron spectroscopy (XPS)) of the most active Ru-based samples.

Temperature Programmed Reduction (H2-TPR)
Figure 3 shows the H2-temperature programmed reduction profiles of the analyzed samples between 50 °C and 500 °C.On the bare cerium oxide support, the H2 consumption starts at around 350 °C, due to the partial reduction of the CeO2 surface oxygen [12,13].The reduction of bulk oxygen was reported to take place at temperatures higher than 700 °C [32] and, therefore, is not visible in the TPR curve of Figure 3.The profile of the Ru/CeO2 sample displays two reduction peaks at 165 °C and 190 °C and a shoulder at ca. 325 °C.According to literature data, the peak at 165 °C is characteristic of the reduction of the ruthenium precursor salt (Ru(NO)(NO3)3), while the peak at 190 °C is assigned to the reduction of well dispersed ruthenium oxide particles on the surface of the support [10,33].The shoulder at 325 °C reasonably originates from the reduction of the ceria surface oxygens shifted to lower temperature as a result of the interaction with the ruthenium species.A quite sharp reduction peak at 435 °C characterizes the TPR profile of the CeO2-MnOx catalyst.In this case, the reduction of ceria surface oxygen is anticipated by the presence of another reducible oxide as MnOx and probably includes the reduction of MnOx species to MnO [34][35][36], as confirmed also by the TPR quantification data (Table S1), being the total H2 uptake higher (around 1.8 times) than the bare CeO2.The presence of ruthenium (sample Ru/CeO2-MnOx), instead, leads to a wide multi-component peak in the 150-350 °C range, in which it is quite hard to distinguish and attribute with certainty the different reduction signals to each single component (Ru precursor, CeO2 and MnOx).The deconvolution procedure (Figure S1A) led to the distinction of three different components at around 199 °C, 246 °C and 314 °C, respectively.Reasonably, considering also the TPR quantification reported in Table S1, the low temperature feature at 200 °C can be attributed to the reduction of the Ru precursor shifted at higher temperature with respect to Ru/CeO2 due to a stronger metal-support interaction [37]; the signal at around 315 °C can be assigned to the reduction of ceria surface oxygen interacting with ruthenium and manganese oxide; and the feature at 245 °C is probably due to the reduction of manganese oxide species (Mn2O3 → Mn3O4) [36] overlapping with the reduction of dispersed RuOX particles on the surface of CeO2-MnOx.The Pd/CeO2-MnOx catalyst A quite sharp reduction peak at 435 • C characterizes the TPR profile of the CeO 2 -MnO x catalyst.In this case, the reduction of ceria surface oxygen is anticipated by the presence of another reducible oxide as MnO x and probably includes the reduction of MnO x species to MnO [34][35][36], as confirmed also by the TPR quantification data (Table S1), being the total H 2 uptake higher (around 1.8 times) than the bare CeO 2 .The presence of ruthenium (sample Ru/CeO 2 -MnO x ), instead, leads to a wide multi-component peak in the 150-350 • C range, in which it is quite hard to distinguish and attribute with certainty the different reduction signals to each single component (Ru precursor, CeO 2 and MnO x ).The deconvolution procedure (Figure S1A) led to the distinction of three different components at around 199 • C, 246 • C and 314 • C, respectively.Reasonably, considering also the TPR quantification reported in Table S1, the low temperature feature at 200 • C can be attributed to the reduction of the Ru precursor shifted at higher temperature with respect to Ru/CeO 2 due to a stronger metal-support interaction [37]; the signal at around 315 • C can be assigned to the reduction of ceria surface oxygen interacting with ruthenium and manganese oxide; and the feature at 245 • C is probably due to the reduction of manganese oxide species (Mn 2 O 3 → Mn 3 O 4 ) [36] overlapping with the reduction of dispersed RuO X particles on the surface of CeO 2 -MnO x .The Pd/CeO 2 -MnO x catalyst shows a wide reduction peak with maximum at circa 275 • C, attributable to the reduction of the PdO-CeO 2 -MnO X species strongly interacting with each other [38,39].
The bimetallic Ru-Pd/CeO 2 -MnO x sample shows a broad peak (around 210 • C) which is at a lower temperature compared to the monometallic palladium catalyst and much more intense than both Ru and Pd monometallics.Furthermore, the total H 2 uptake of Ru-Pd/CeO 2 -MnO x is slightly higher (2106 µmol/g cat ) than the sum of the H 2 uptakes of Pd/CeO 2 -MnO x and Ru/CeO 2 -MnO x (2067 µmol/g cat ).It is rational to assume that the reduction both of Ru and Pd species, together with that of the superficial oxygens of the mixed CeO 2 -MnO x support, contribute to this peak.The deconvolution of the Ru-Pd/CeO 2 -MnO x reduction peak (Figure S1B) revealed two other contributions, one at low temperature (140 • C) and one at high temperature (275 • C), probably due to the reduction of part of the metal precursors.It is possible in this case, in the same way as other bimetallic systems supported on cerium oxide reported in the literature [6,12], that a mutual interaction between the two metals leads a single reduction peak, distinct from the individual monometallic samples.Furthermore, this reduction peak being shifted at lower temperature than the monometallic Ru sample also suggests an increased reducibility/mobility of the ceria-manganese reactive oxygens [6,12,40].No peaks associated with metallic species or manganese oxide were detected.The absence of signals for the Ru and Pd species is plausibly due to the combination of the small size (less than 5 nm) and the low amount (1 wt %) of metals.With regard to manganese oxide, the low amount (5 wt %), the homogeneous distribution of this host oxide on the ceria, and the co-presence of CeO 2 diffraction peaks overlapping with those of manganese oxide, having a similar structure [41], do not allow recognition of variations of the cerium oxide pattern upon addition of MnO x .

Structural and Textural Properties
Catalysts 2018, 8, x FOR PEER REVIEW 6 of 18 shows a wide reduction peak with maximum at circa 275 °C, attributable to the reduction of the PdO-CeO2-MnOX species strongly interacting with each other [38,39].The bimetallic Ru-Pd/CeO2-MnOx sample shows a broad peak (around 210 °C) which is at a lower temperature compared to the monometallic palladium catalyst and much more intense than both Ru and Pd monometallics.Furthermore, the total H2 uptake of Ru-Pd/CeO2-MnOx is slightly higher (2106 µmol/gcat) than the sum of the H2 uptakes of Pd/CeO2-MnOx and Ru/CeO2-MnOx (2067 µmol/gcat).It is rational to assume that the reduction both of Ru and Pd species, together with that of the superficial oxygens of the mixed CeO2-MnOx support, contribute to this peak.The deconvolution of the Ru-Pd/CeO2-MnOx reduction peak (Figure S1B) revealed two other contributions, one at low temperature (140 °C) and one at high temperature (275 °C), probably due to the reduction of part of the metal precursors.It is possible in this case, in the same way as other bimetallic systems supported on cerium oxide reported in the literature [6,12], that a mutual interaction between the two metals leads a single reduction peak, distinct from the individual monometallic samples.Furthermore, this reduction peak being shifted at lower temperature than the monometallic Ru sample also suggests an increased reducibility/mobility of the ceria-manganese reactive oxygens [6,12,40].The size of the crystalline grains of investigated catalysts, calculated by the Scherrer equation referring to the main peak of cerium oxide at 2θ = 28.6°(1 1 1), their BET surface area and mean pore diameters, calculated with the Dollimore-Heal method [42], are reported in Table 1.The size of the crystalline grains of investigated catalysts, calculated by the Scherrer equation referring to the main peak of cerium oxide at 2θ = 28.6 • (1 1 1), their BET surface area and mean pore diameters, calculated with the Dollimore-Heal method [42], are reported in Table 1.The precipitation method with KOH used to synthesize the bare CeO 2 leads to a catalyst with relatively high surface area (107 m 2 /g) and small pores (3 nm) in line with literature data [43,44].A small increase (around 10%) of surface area (119 m 2 /g) was noticed in the presence of manganese oxide, consistent with the decrease of the crystallite size (from 10 nm of the bare CeO 2 to 8.5 nm of the mixed CeO 2 -MnO X oxide) and the pore diameter.Interestingly, the bimetallic Ru-Pd/CeO 2 -MnO x sample showed the highest surface area (123 m 2 /g) among all investigated samples.It is probable that the presence of a bimetallic phase, well spread on the mixed ceria-manganese oxide, favours the replacement of cerium ions with manganese ones in the fluorite structure, which is reasonably due to the crystal structure similarity [41].Moreover, a part of the well dispersed bimetallic components can move into the interlayers of the ceria support.

Structural and Textural Properties
The DRIFT spectrum (4000-500 cm −1 ) of the cerium oxide support is shown in Figure S2.On the basis of the literature, the signal at 3470 cm −1 can be assigned to the O-H stretching of H-bonded water [45].The bands at 1585 and 1346 cm −1 are due to the O-C-O stretching vibrations whereas the features at 1057 and 852 cm −1 can be assigned to the C-O stretching, thus pointing to the presence of carbonate species on the oxide sample.Finally, the band at 600 cm −1 can be attributed to the O-Ce-O phononic vibration of the crystal lattice of the cerium oxide [46].Figure 5a compares the DRIFT spectrum of the support with those of the investigated samples.It is evident that all mixed oxides reveal a shift of the bands assigned to carbonates.This is better highlighted in Figure 5b, where the zoomed 1800-1000 cm −1 region is reported.The presence of different carbonate species is typical of basic oxide materials, such as cerium oxide, owing to the interaction of basic sites with the atmospheric CO 2 [43,44].The inspection of literature data [47][48][49] points to the presence of variously coordinated carbonate species (O-C-O asymmetrical and symmetrical stretchings at 1600-1470 cm −1 and 1370-1300 cm −1 , respectively) and un-coordinated H-bonded carbonate (ν as CO 3 2− at 1450-1420 cm −1 ).In the case of the CeO 2 samples (CeO 2 and Ru/CeO 2 ), the coordinated carbonate bands were predominant, whereas for the CeO 2 -MnO x supported samples (CeO 2 -MnO x , Ru/CeO 2 -MnO x , Pd/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x ) the un-coordinated H-bonded carbonate band was the main feature.This behaviour suggests that the presence of manganese oxide significantly modified the acid-base properties of the system, causing a decrease of the basicity of the oxide support, thus strongly affecting the nature of CO 2 adsorption sites.

X-ray Photoelectron Spectroscopy (XPS) Measurements
The electronic structure of the Ru/CeO2, Ru/CeO2-MnOx and Ru-Pd/CeO2-MnOx samples was investigated by XPS, which provides information on the surface elemental composition, the oxidation status and the chemical surroundings of the studied species [50,51].
The Ru 3d states (Figure S3) overlap the C 1s levels, so we prefer to mainly discuss the XPS of the Ru 3p binding-energy region.The two Ru 3p3/2,1/2 spin-orbit components for Ru/CeO2 (Figure 6a) lie at 462.7 and 485.1 eV, respectively.These energy values and also the Ru 3d5/2 state, at 281.3 eV (Figure S3), are indicative of the presence of RuO2 on the surface sample [52][53][54][55].RuO2 is also present on the surface sample of Ru/CeO2-MnOx since the Ru 3d5/2 state is at 281.5 eV and the two Ru 3p3/2,1/2 spin-orbit components are at 463.3 and 485.8 eV, respectively.It is important to stress that XPS is a surface-sensitive technique that probes a thickness of ~60 Å in our conditions.The above energy values are significantly lower in the case of Ru-Pd/CeO2-MnOx, being the Ru 3d5/2 state at 280.8 eV and the two Ru 3p3/2,1/2 spin-orbit components at 462.2 and 484.6 eV, respectively.All these latter values are consistent with the presence of metallic Ru in the Ru-Pd/CeO2-MnOx sample [56].

X-ray Photoelectron Spectroscopy (XPS) Measurements
The electronic structure of the Ru/CeO 2 , Ru/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x samples was investigated by XPS, which provides information on the surface elemental composition, the oxidation status and the chemical surroundings of the studied species [50,51].
The Ru 3d states (Figure S3) overlap the C 1s levels, so we prefer to mainly discuss the XPS of the Ru 3p binding-energy region.The two Ru 3p 3/2,1/2 spin-orbit components for Ru/CeO 2 (Figure 6a) lie at 462.7 and 485.1 eV, respectively.These energy values and also the Ru 3d 5/2 state, at 281.3 eV (Figure S3), are indicative of the presence of RuO 2 on the surface sample [52][53][54][55].RuO 2 is also present on the surface sample of Ru/CeO 2 -MnO x since the Ru 3d 5/2 state is at 281.5 eV and the two Ru 3p 3/2,1/2 spin-orbit components are at 463.3 and 485.8 eV, respectively.It is important to stress that XPS is a surface-sensitive technique that probes a thickness of ~60 Å in our conditions.The above energy values are significantly lower in the case of Ru-Pd/CeO 2 -MnO x , being the Ru 3d 5/2 state at 280.8 eV and the two Ru 3p 3/2,1/2 spin-orbit components at 462.2 and 484.6 eV, respectively.All these latter values are consistent with the presence of metallic Ru in the Ru-Pd/CeO 2 -MnO x sample [56].

X-ray Photoelectron Spectroscopy (XPS) Measurements
The electronic structure of the Ru/CeO2, Ru/CeO2-MnOx and Ru-Pd/CeO2-MnOx samples was investigated by XPS, which provides information on the surface elemental composition, the oxidation status and the chemical surroundings of the studied species [50,51].
The Ru 3d states (Figure S3) overlap the C 1s levels, so we prefer to mainly discuss the XPS of the Ru 3p binding-energy region.The two Ru 3p3/2,1/2 spin-orbit components for Ru/CeO2 (Figure 6a) lie at 462.7 and 485.1 eV, respectively.These energy values and also the Ru 3d5/2 state, at 281.3 eV (Figure S3), are indicative of the presence of RuO2 on the surface sample [52][53][54][55].RuO2 is also present on the surface sample of Ru/CeO2-MnOx since the Ru 3d5/2 state is at 281.5 eV and the two Ru 3p3/2,1/2 spin-orbit components are at 463.3 and 485.8 eV, respectively.It is important to stress that XPS is a surface-sensitive technique that probes a thickness of ~60 Å in our conditions.The above energy values are significantly lower in the case of Ru-Pd/CeO2-MnOx, being the Ru 3d5/2 state at 280.8 eV and the two Ru 3p3/2,1/2 spin-orbit components at 462.2 and 484.6 eV, respectively.All these latter values are consistent with the presence of metallic Ru in the Ru-Pd/CeO2-MnOx sample [56].The Mn 2p 3/2,1/2 spin-orbit levels are at 640.4 and 652.4 eV (Figure 6b), respectively, in both Ru/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x systems.Moreover, there is evidence of satellite features at ~7 eV higher binding energy from the main peaks that are a fingerprint of the presence of MnO, since these satellites are absent in Mn 2 O 3 and in MnO 2 species [57].
Finally, Figure 6c shows the XPS of the Pd 3d states for the Ru-Pd/CeO 2 -MnO x system.The binding energy values of 335.3 and 340.6 due to the Pd 3d 5/2,3/2 spin-orbit components can be unambiguously assigned to metallic palladium [58][59][60].It must be noted that the full width at half maximum (FWHM) of the Ru 3p 3/2 states for the Ru/CeO 2 (Figure 6a) is 1.8 eV, that of the Mn 2p 3/2 states for the Ru/CeO 2 -MnO x (Figure 6b) is 1.65 eV and that of the Pd 3d 5/2 states for the Ru-Pd/CeO 2 -MnO x (Figure 6c) is 1.1 eV.
In all cases, the CeO 2 system gave the expected XPS 3d signals as reported in Figure S4.Concerning the Ce(IV)/Ce(III) ratio, it must be noted that peaks at 882.0, 888.6 and 897.9 eV (Figure S4) are from the Ce 3d 5/2 states and peaks at 900.5, 906.9 and 916.2 are from the 3d 3/2 states.The first two peaks at 882.0 and 888.6 eV are due to a mixing configuration of 3d 9 4f 2 (O 2p 4 ) and 3d 9 4f 1 (O 2p 5 ) Ce 4+ states, and the peak at 897.9 eV belongs to the 3d 9 4f 0 (O 2p 6 ) Ce 4+ state.The series of peaks at 900.5, 906.9 and 916.2 are the related spin-orbit components of the 3d 3/2 states of the same configurations.Moreover, there is no structure between the peaks at 882.0 eV, 888.6 eV and 897.9 eV, and this rules out any 3d 9 4f 1 (O 2p 6 ) Ce 3+ final state [61].
The O1s peaks of the representative Ru/CeO 2 can be fitted into two peaks referred to as the lattice oxygen O 2− and the chemisorbed OH − groups.The O1s peak at about 528.7 eV is due to the lattice oxygen (O 2− ) for CeO 2 , and that at 530.6 eV belongs to the chemisorbed OH − species.In the case of the Ru/CeO 2 -MnO x the energy positions of these two peaks are 528.9 and 530.4 eV while in the case of Ru-Pd/CeO 2 -MnO x these two peaks lie at 529.2 and 530.8 eV.The intensity ratio between these two peaks (1.2-1.3) is almost the same in all the three different investigated systems, thus excluding different affluent oxygen species on the samples surface [61].The deconvolution features are reported in the supporting information (Figure S5).
The XPS atomic concentration analysis of the three investigated systems clearly show some surface segregation of ruthenium [62] since the Ru/Ce is 0.18, 0.12 and 0.06 for the Ru/CeO 2 , Ru/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x systems, respectively.The Mn/Ce is 0.07 and 0.03 for the Ru/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x systems, respectively and, finally, the Pd/Ce is 0.02 for the Ru-Pd/CeO 2 -MnO x system.XPS also indicates that there is neither nitrogen, potassium nor chlorine signals in all these samples, thus inferring the purity of the samples with respect to the starting reagents (see Figure S6 with the survey spectrum of the Ru/CeO 2 sample).

Discussion
In this paper the catalytic activity of ceria-based mixed oxides (CeO 2 -Fe 2 O 3 , CeO 2 -CuO and CeO 2 -MnO x ) supported mono and bimetallic Ru-Pd samples was evaluated in the CO-PROX reaction.A preliminary comparison of the catalytic results among investigated mixed oxides supported Ru monometallic catalysts pointed out that the Ru/CeO 2 -MnO x sample was the most selective towards the CO oxidation (Figure 1c), providing the highest CO 2 yield (Figure 1d) in the temperature range of PEM fuel cells (80-120 • C).Therefore, the CeO 2 -MnO x mixed oxide was chosen for further investigation focusing on Ru-Pd bimetallic systems.
As reported in the literature [63,64], the manganese oxide is a reducible oxide as ceria, and then it can provide, in addition to those of CeO 2 , beneficial reactive oxygens for oxidation reactions.Moreover, CeO 2 -MnO x mixed oxides can also show stronger redox properties than CeO 2 alone, due to the multi-valence surface state and the mutual contact between ceria and manganese oxide [65,66].It has been reported that the oxygen storage capacity of CeO 2 is boosted by the presence of manganese oxide, with the surface oxygen mobility of ceria also improved by forming MnO x -CeO 2 solid solution [67].In our case, the occurrence of an interaction between ruthenium-CeO 2 and MnO x itself, proved by TPR data, can be one of the reasons for the excellent performance of this catalyst in the CO-PROX reaction.Interestingly, the Ru/CeO 2 -MnO X exhibited better performance compared to the other investigated monometallic systems (Pt and Pd, Figure 2).Furthermore, the Ru-Pd/CeO 2 -MnO x system showed higher selectivity (Figure 2c) and yield to CO 2 (Figure 2d) compared to the Ru/CeO 2 -MnO x catalyst at T > 100 • C, namely in the operative range of alkaline fuel cells (100-140 • C).
To elucidate the catalytic behavior of the investigated catalysts it is important to refer to the CO-PROX reaction mechanisms reported in the literature for reducible oxide-supported metal catalysts [5] and for non-reducible oxide-supported Ru catalysts [10].
For reducible oxide-supported metal catalysts, it was stated that the reaction occurs between CO and oxygen co-adsorbed on the surface-active sites.The oxygen can be activated at the metal/support-oxide interface and/or the reducible surface oxygens of the oxide can participate in the reaction [5].However, it is necessary to clarify that a greater activation of oxygen does not directly mean a better yield for CO oxidation reactions: activated oxygen can in fact react with both CO and H 2 , which means that the two species are in strong competition between each other for the oxidation reaction.In order to obtain a good yield, it is then necessary to activate CO and not 2 .It is well known that cerium oxide has good catalytic properties for oxidation reactions, since it has mobile/reducible oxygens [6,12,13,43,68].The mobility of such oxygens can be increased by the presence of metals on the surface, which introduce crystalline defects that increase the mobility of these oxygens and increasing the oxidative power of the substrate [12,43,67].Specifically, for the preferential oxidation of CO the most widely reported mechanism is that of Mars-van Krevelen (MvK) [6,13,43,[68][69][70].The aforesaid mechanism involves different reaction steps: the first is the formation of a chemical bond between the reagents (CO, O 2 or H 2 ) and the catalyst surface; then the formation and the desorption of products (CO 2 and H 2 O) take place with the generation of oxygen vacancies in the bulk of the ceria, which can then be replaced by the subsequent oxygen adsorption from the gaseous phase.In this case, an important parameter is the reactivity of the support oxide (ceria) oxygens, which are directly involved in the reaction, being able to react at the metal/CeO 2 interface with CO to form CO 2 .The reducibility of the cerium oxide reflects directly the mobility and therefore the reactivity of surface oxygens with an increased reducibility (mobility/reactivity) beneficial for the catalytic activity in the CO-PROX reaction [5,6,12,13,43,[68][69][70].
For ruthenium or platinum-based catalysts supported on non-reducible (inert) oxides, as reported in the literature, the oxidation of CO occurs according to the Langmuir-Hinshelwood mechanism [71,72].Both reagents (CO and O 2 ) bind to the surface at the active sites of the metal species.In particular, the following mechanism was assumed on ruthenium-based catalysts: (a) Reactants' adsorption on the Ru surface: However, also in this case, it must be considered that the reaction mixture of CO-PROX contains considerable hydrogen excess (88 vol.%) and, therefore, H 2 can also be activated by the metal surface sites, thus competing with CO in reacting with oxygen and influencing the selectivity of the CO-PROX reaction [10,73].
The mobility of the ceria oxygens can be increased both by interaction with a metal and by the presence of a host oxide [44, 74,75].From the TPR characterization (Figure 3), it is possible to infer that a strong interaction is established between Ru-Pd and the mixed oxide CeO 2 -MnO x support, thus enhancing the reducibility/reactivity of the ceria oxygens (the reduction peak of the bimetallic sample is at a lower temperature and more intense than the corresponding monometallic ones).This can explain the higher activity of this catalyst.As evidenced by the XRD measurements in both the monometallic and bimetallic samples, the small metal particle size (below 5 nm of diameter) and the good distribution of manganese oxide on the ceria can be decisive factors in explaining the excellent catalytic performances of Ru/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x catalysts.The strong interaction between Ru-Pd and the mixed oxide CeO 2 -MnO x was, furthermore, confirmed by XPS measurements, with the Ru-Pd/CeO 2 -MnO x sample that showed the presence of metallic Ru and Pd species on the surface of the sample a key factor explaining the higher activity of this sample compared to the Ru/CeO 2 -MnO x and Ru/CeO 2 samples where Ru 4+ species were also detected.As reported in the literature, metallic Ru 0 is more active compared to oxidized Ru n+ species [76,77].In this case, the simultaneous presence of Pd 0 also facilitates the redox process of the ceria-manganese oxide, thus favouring an easier CO activation.
To interpret the results obtained, it is then reasonable to hypothesize a mixed Langmuir-Hinshelwood/Mars-van Krevelen (LH/MvK) reaction mechanism.In this case, in fact, the reaction can take place in one step: the adsorbed CO (CO ad ) reacts very quickly with the mobile/active oxygens of the mixed CeO 2 -MnO x oxide, also allowing the rapid desorption of the CO 2 produced, while the oxygen vacancy is instantly replaced by the O 2 adsorbed.The fast desorption of the produced CO 2 can be favoured by the modification of the surface sites of the CeO 2 -MnO x with respect to the cerium oxide as revealed by the DRIFT data (Figure 5), while the adsorption of oxygen and CO from the gas phase is allowed by the increased surface area of the mixed oxide (Table 1).
In conclusion, the use of the mixed cerium-manganese oxide system allows excellent performance to be obtained in the preferential oxidation reaction of CO due, essentially, to: (1) a higher reducibility/reactivity of the ceria surface oxygen; (2) a higher surface area which favours an easier and fast adsorption of the reacting species and a faster desorption of the CO 2 produced, due also to the modification of the surface adsorption sites.The strong interaction of the mixed oxide with Ru in the monometallic Ru/CeO 2 -MnO x and the effective synergism between Ru and Pd in the Ru-Pd bimetallic system allows the design of proper catalysts for the specific use of H 2 purification in fuel cells working at low temperature (PEM and AFC).In particular, the Ru/CeO 2 -MnO x catalyst appears the most suitable for PEM fuel cells (operating temperature range of 80-120 • C) while the Ru-Pd/CeO 2 -MnO x system can be considered best for the AFC fuel cells (operating temperature range of 100-140 • C).It is interesting to note that the catalytic performances of these catalysts were comparable or even better, in terms especially of CO 2 selectivity, with respect to ceria and ceria-manganese oxide-based catalysts reported in the literature for CO-PROX (see Table 2) [12,[78][79][80][81][82], even though it must stressed that a realistic comparison is very difficult due to the different operative conditions of the various works.As displayed in Figure 7, the Ru/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x catalysts were quite stable after 14 h of streaming both at 100 • C and 120 • C, whereas the monometallic Pd sample began to deactivate at 120 • C after 8 h.The higher selectivity to CO 2 found on the Ru-Pd system, which reduces the undesired H 2 consumption, represents a promising result of this research, being one of the key aims of the design of CO-PROX catalysts.
Catalysts 2018, 8, x FOR PEER REVIEW 11 of 18 catalytic performances of Ru/CeO2-MnOx and Ru-Pd/CeO2-MnOx catalysts.The strong interaction between Ru-Pd and the mixed oxide CeO2-MnOx was, furthermore, confirmed by XPS measurements, with the Ru-Pd/CeO2-MnOx sample that showed the presence of metallic Ru and Pd species on the surface of the sample a key factor explaining the higher activity of this sample compared to the Ru/CeO2-MnOx and Ru/CeO2 samples where Ru 4+ species were also detected.As reported in the literature, metallic Ru 0 is more active compared to oxidized Ru n+ species [76,77].In this case, the simultaneous presence of Pd 0 also facilitates the redox process of the ceria-manganese oxide, thus favouring an easier CO activation.
To interpret the results obtained, it is then reasonable to hypothesize a mixed Langmuir-Hinshelwood/Mars-van Krevelen (LH/MvK) reaction mechanism.In this case, in fact, the reaction can take place in one step: the adsorbed CO (COad) reacts very quickly with the mobile/active oxygens of the mixed CeO2-MnOx oxide, also allowing the rapid desorption of the CO2 produced, while the oxygen vacancy is instantly replaced by the O2 adsorbed.The fast desorption of the produced CO2 can be favoured by the modification of the surface sites of the CeO2-MnOx with respect to the cerium oxide as revealed by the DRIFT data (Figure 5), while the adsorption of oxygen and CO from the gas phase is allowed by the increased surface area of the mixed oxide (Table 1).
In conclusion, the use of the mixed cerium-manganese oxide system allows excellent performance to be obtained in the preferential oxidation reaction of CO due, essentially, to: (1) a higher reducibility/reactivity of the ceria surface oxygen; (2) a higher surface area which favours an easier and fast adsorption of the reacting species and a faster desorption of the CO2 produced, due also to the modification of the surface adsorption sites.The strong interaction of the mixed oxide with Ru in the monometallic Ru/CeO2-MnOx and the effective synergism between Ru and Pd in the Ru-Pd bimetallic system allows the design of proper catalysts for the specific use of H2 purification in fuel cells working at low temperature (PEM and AFC).In particular, the Ru/CeO2-MnOx catalyst appears the most suitable for PEM fuel cells (operating temperature range of 80-120 °C) while the Ru-Pd/CeO2-MnOx system can be considered best for the AFC fuel cells (operating temperature range of 100-140 °C).It is interesting to note that the catalytic performances of these catalysts were comparable or even better, in terms especially of CO2 selectivity, with respect to ceria and ceriamanganese oxide-based catalysts reported in the literature for CO-PROX (see Table 2) [12,[78][79][80][81][82], even though it must stressed that a realistic comparison is very difficult due to the different operative conditions of the various works.As displayed in Figure 7, the Ru/CeO2-MnOx and Ru-Pd/CeO2-MnOx catalysts were quite stable after 14 h of streaming both at 100 °C and 120 °C, whereas the monometallic Pd sample began to deactivate at 120 °C after 8 h.The higher selectivity to CO2 found on the Ru-Pd system, which reduces the undesired H2 consumption, represents a promising result of this research, being one of the key aims of the design of CO-PROX catalysts.
The preparation of Ru/CeO 2 and Ru/CeO 2 -mixed oxide catalysts (1 wt % Ru) was achieved by incipient wetness impregnation of the supports with an aqueous solution of the ruthenium precursor (Ru(NO)(NO 3 ) 3 ) (Alfa Aesar, Haverhill, MA, USA).In particular, the stoichiometric amount of Ru(NO)(NO 3 ) 3 was added dropwise to the support, to fill the pores.The slurry then obtained was placed in an oven at 110 • C to evaporate the excess of solvent.This process was repeated until addition of the required amount of Ru(NO)(NO 3 ) 3 .Finally, the sample was dried overnight at 110 • C.

Catalytic Performance Test
Catalytic activity runs were executed in the gas phase at atmospheric pressure in a continuous flow mode, using a reactor loaded with the catalyst (0.05 g, 80-140 mesh) and 0.45 g of an inert glass powder.Catalysts were pre-reduced in H 2 at 350 • C. The gas mixture (80 mL/min) was CO/O 2 /He/H 2 (1/1/10/88 vol.%).A gas hour space velocity of 3.92 × 10 −2 mol CO h −1 g cat −1 was chosen.The effluent gases were analyzed online with gas chromatography (Trace GC, Thermo Finnigan, Waltham, MA, USA) with a Carboxen 1000 packed column and TCD detector.
CO and O 2 conversions were estimated from the ratio of the consumed to the initial amount of each gas, whereas selectivity was determined in terms of ratio of O 2 used for the CO oxidation and the whole O 2 consumed, according to the subsequent formulas: H 2 -TPR was carried out heating 0.05 g of the sample (dried in a oven at 110 • C) at 10 • C min −1 in a home-made flow apparatus with a TCD detector using 5 vol.% H 2 in Ar [10].The system employed an oxygen adsorbent to purify the reducing mixture and an acetone trap (−80 • C) to block the water and other gases released during TPR.
X-ray photoelectron spectra (XPS) of the powder samples, compressed on indium foil were measured with a PHI 5600 Multi Technique System (Chanhassen, MN, USA) (base pressure of the main chamber 3 × 10 −8 Pa) [83,84].Samples were excited with the Al-Kα X-ray radiation using a pass energy of 5.85 eV.The instrumental energy resolution was ≤0.5 eV.XPS peak intensities were obtained after a Shirley background removal.Spectra calibration was achieved by fixing the main C 1s peak at 285.0 eV.Structures due to the Al-Kα X-ray satellites were subtracted prior to data processing.The atomic concentration analysis was performed by taking into account the relevant atomic sensitivity factors.The fittings of the C 1s and O 1s XP spectra were carried out using Gaussian envelopes after subtraction of the background until there was the highest possible correlation between the experimental spectrum and the theoretical profile.The residual or agreement factor R, defined by R = [Σ(F obs − F calc ) 2 /Σ(F obs ) 2 ] 1/2 , after minimization of the function Σ(F obs − F calc ) 2 , converged to the value of 0.02.The catalysts for XPS measurements were introduced in the XPS antechamber under vacuum as soon as they were reduced in H 2 at 350 • C.
Diffuse reflectance infrared (IR) spectra were recorded in the range 4000-600 cm −1 with a Perkin-Elmer FT-IR System 2000 (Perkin-Elmer, Waltham, MA, USA) using the specific set-up and mirrors for the diffuse reflectance measurements, KBr being employed to obtain the background spectrum.
XRD analysis was performed with a Bruker AXSD5005 (Bruker, Karlsruhe, Germany) X-ray diffractometer using a Cu Kα radiation.Peaks of crystalline phases were compared with those of standard compounds reported in the JCPDS Data File (Bruker, Diffrac.Suite™ Software package, Karlsruhe, Germany).Before the measurements, the well-grounded samples were placed in the powder sample holder.The analysis was made in locked couple mode.
Surface area determination was achieved with a Sorptomatic series 1990 (Thermo Quest, Milano, Italy) adopting the BET N 2 adsorption method.All samples were previously outgassed (10 −3 Torr) at 120 • C.

Conclusions
Different CeO 2 -mixed oxides were synthesized by the co-precipitation method and ruthenium was subsequently added by incipient wetness impregnation.Among monometallic samples, the Ru/CeO 2 -MnO x catalyst showed the highest activity in the CO-PROX reaction at 100 • C. The performance of the bimetallic Ru-Pd system on the ceria-manganese support was successively studied.The bimetallic sample exhibited the highest CO 2 yield at 120 • C. On the basis of characterization data, a one-step Langmuir-Hinshelwood/Mars-van Krevelen mechanism was proposed.The high mobility of ceria surface oxygens was enhanced by the simultaneous presence of manganese oxide and ruthenium.The increased reducibility/reactivity of CeO 2 oxygen was further promoted by the occurrence of a strong interaction between ruthenium and palladium, resulting in an increased CO-PROX activity of the Ru-Pd/CeO 2 -MnO x catalyst.

Figure 4
Figure 4 exhibits the XRD patterns of the bare CeO 2 and CeO 2 -MnO x supports and of the Ru/CeO 2 , Ru/CeO 2 -MnO x , Pd/CeO 2 -MnO x and Ru-Pd/CeO 2 -MnO x mono and bimetallic samples.The signals at 2θ = 28.6 • (1 1 1), 33.1 • (2 0 0), 47.5 • (2 2 0) and 56.4 • (3 1 1) are assigned, according to the Joint Committee on Powder Diffraction Standards (JCPDS) data, to the typical fluorite structure of the CeO 2 .No peaks associated with metallic species or manganese oxide were detected.The absence of signals for the Ru and Pd species is plausibly due to the combination of the small size (less than 5 nm) and the low amount (1 wt %) of metals.With regard to manganese oxide, the low amount (5 wt %), the homogeneous distribution of this host oxide on the ceria, and the co-presence of CeO 2 diffraction peaks overlapping with those of manganese oxide, having a similar structure[41], do not allow recognition of variations of the cerium oxide pattern upon addition of MnO x .

Figure 4
Figure 4 exhibits the XRD patterns of the bare CeO2 and CeO2-MnOx supports and of the Ru/CeO2, Ru/CeO2-MnOx, Pd/CeO2-MnOx and Ru-Pd/CeO2-MnOx mono and bimetallic samples.The signals at 2θ = 28.6°(1 1 1), 33.1° (2 0 0), 47.5° (2 2 0) and 56.4° (3 1 1) are assigned, according to the Joint Committee on Powder Diffraction Standards (JCPDS) data, to the typical fluorite structure of the CeO2.No peaks associated with metallic species or manganese oxide were detected.The absence of signals for the Ru and Pd species is plausibly due to the combination of the small size (less than 5 nm) and the low amount (1 wt %) of metals.With regard to manganese oxide, the low amount (5 wt %), the homogeneous distribution of this host oxide on the ceria, and the co-presence of CeO2 diffraction peaks overlapping with those of manganese oxide, having a similar structure[41], do not allow recognition of variations of the cerium oxide pattern upon addition of MnOx.

Figure 7 .
Figure 7. Conversion of CO as a function of time on stream at 100 • C (a) and 120 • C (b) over Ru/CeO 2 -MnO x , Ru/CeO 2 -MnO x and Ru/CeO 2 -MnO x catalysts.

Table 1 .
Structural and textural properties of investigated samples.

Table 2 .
Comparison of PROX performance of Ru and Ru-Pd supported CeO 2 -MnO x catalysts in this work and other ceria and ceria-manganese oxide-based catalysts reported in the literature.