Nanosized gold catalysts on Pr-modified ceria for pure hydrogen production via WGS reaction

Materials Chemistry and Physics 157 (2015) 138e146

Contents lists available at ScienceDirect

Materials Chemistry and Physics journal homepage: www.elsevier.com/locate/matchemphys

Nanosized gold catalysts on Pr-modified ceria for pure hydrogen production via WGS reaction L. Ilieva a, *, P. Petrova a, I. Ivanov a, G. Munteanu b, M. Boutonnet c, J.W. Sobczac d, W. Lisowski d, Z. Kaszkur d, P. Markov e, A.M. Venezia f, T. Tabakova a a

Institute of Catalysis, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria Institute of Physical Chemistry, Romanian Academy, Bucharest 77208, Romania KTH Royal Institute of Technology, 10044 Stockholm, Sweden d Institute of Physical Chemistry, PAS, Kasprzaka 44/52, 01-224 Warsaw, Poland e Institute of General and Inorganic Chemistry, BAS, 1113 Sofia, Bulgaria f Istituto per lo Studio di Materiali Nanostrutturati, CNR, I- 90146 Palermo, Italy b c

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


PreCe oxide supports were prepared by impregnation (IM) or microemulsion method (ME).
No substantial differences in the WGS activity of fresh gold catalysts were observed.
Used and reoxidized IM catalyst with 5 at% Pr exhibited significantly higher activity.
Substantially lower activity was shown by used and reoxidized ME sample (10 at% Pr).
Interpretation based on the gold dispersion and supports features was given.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 18 June 2014 Received in revised form 13 March 2015 Accepted 15 March 2015 Available online 20 March 2015

The WGS activity of gold catalysts on ceria doped with Pr (5 and 10 at% Pr) was studied. The supports were synthesized by impregnation (IM) and microemulsion (ME) method. Gold (3 wt%) was introduced by deposition-precipitation method. The samples were characterized by means of XRD, TEM, XPS and TPR measurements. The catalytic activity in WGSR of fresh catalysts did not differ significantly. After reoxidation following the first catalytic run, the activity of gold catalyst on IM support with 5 at% Pr increased, while a significant drop in activity of the gold catalyst on ME support containing 10 at% Pr was registered. In the latter case the presence of very small gold particles were responsible for the initial WGS activity, while their agglomeration after the catalytic reaction caused the observed activity decrease. In the former case the higher activity during the second catalytic run could be explained by: (i) the redox behavior, showing that only in this case the most mobile oxygen could be fully recovered; (ii) the morphology of the support, revealing better crystallization of the support after WGS catalytic test as compared to the presence of partially amorphous ceria in the fresh gold catalyst. © 2015 Elsevier B.V. All rights reserved.

Keywords: Oxides Nanostructures Chemical synthesis Surface properties Defects

* Corresponding author. Institute of Catalysis, Bulgarian Academy of Sciences, Acad. G. Bonchev St., Block 11, 1113 Sofia, Bulgaria. E-mail address: [email protected] (L. Ilieva). http://dx.doi.org/10.1016/j.matchemphys.2015.03.028 0254-0584/© 2015 Elsevier B.V. All rights reserved.

L. Ilieva et al. / Materials Chemistry and Physics 157 (2015) 138e146

1. Introduction With respect to clean energy sources, hydrogen is recognized as an efficient energy carrier without harmful emissions. Nowadays, H2 is mostly produced via reforming of hydrocarbons from fossil and renewable resources. For development of fuel cell technology with H2 utilization, the concentration of CO in the obtained “syn gas” has to be supplementary reduced. The preferred process for the first step of hydrogen purification is the wateregas shift (WGS) reaction as the simplest and the most cost effective method for increasing the H2 yield accompanied by CO removal (CO concentration to ~1%). Currently, a subject of appreciable investigations is the application of gold catalysts for low temperature (LT) WGS reaction. Many authors have shown the beneficial application of ceria as support of active WGS catalysts. The defect chemistry of ceria favors the exchange of lattice oxygen with the gas phase via Ce3$Ce4þ transition. Due to this effect, ceria is an oxide with unique catalytic properties, including also the promotion of the precious metal dispersion and the enhancement of the catalytic activity at the interfacial metal-ceria sites. Modification with different metal dopants could improve the oxygen mobility and facilitate the generation of active centers at the border between gold and support. In addition to the dopant nature and amount, the preparation method of ceria based supports is of great importance. Some researchers have reported the WGS results over gold catalysts on rare earth (RE) doped ceria carriers. Deng et al. [1] have studied gold catalysts on Gd-doped ceria. In the absence of gold the positive effect on the WGS activity was lost. Gold catalysts on La doped ceria have been investigated by Zhang et al. [2] and Wang et al. [3]. Andreeva et al. [4] compared the WGS activity over gold catalysts supported on ceria modified by various RE metals (RE ¼ La, Sm, Gd, Yb, Y) depending on the dopant nature and the method of mixed support preparation: co-precipitation and mechanochemical activation. The RE dopants used in these studies are with stable (3þ) oxidation state. Trovarelli et al. have established that the doping of aliovalent non-reducible cations (Gd3þ, La3þ) into the ceria lattice enhances the oxygen storage capacity (OSC) chiefly through the extrinsic oxygen vacancies [5]. The subject of the present work is the investigation of gold catalysts on Pr-doped ceria. The motivation for choosing Pr as dopant was related to its own redox couples (Pr4þ$Pr3þ). The preferential oxidation state of Pr in the most common form of praseodymium oxide (Pr6O11) is less than (4þ) and even in quite earlier investigations, it was expected that the CeePr mixed oxides would have relatively higher oxygen vacancy concentration due to the effect of ceria doping with lower-valence Pr3þ ions [6]. The conclusion that the incorporation of Pr ions into the ceria lattice induced both intrinsic and extrinsic oxygen vacancies thereby enhancing the oxygen mobility and OSC was in agreement to the ability of Pr to exist in both (3þ) and (4þ) oxidation states and the fact that oxygen could be easily removed from praseodymia compared to ceria [7e9]. OSC derives from the ability of the oxide to be easily but also reversibly reduced. Putna et al. [10] has argued that the activity of praseodymia-supported catalysts might be limited by slow re-oxidation kinetics of Pr2O3. However, Zhou and Gorte [7] have shown that the problem of slow re-oxidation could be solved by forming a solid solution with ceria. The influence of noble metals on the reduction and re-oxidation behavior of CeePr solid solutions has been investigated as well. Logan and Shelef [11] have reported that the addition of small amounts of noble metals to the mixed oxides shifts the accessibility of the “stored” oxygen to lower temperatures with the effect of Pd being more pronounced than that of Pt. Chun et al. [12] carrying out reduction experiments with Pd/PrxCe1xO2d, have shown that the addition of 2 wt% Pd shifts the reduction temperatures below the temperature observed

139

for only PrxCe1xO2d, which increased as the Pr concentration increases. Our very recent study was focused on gold catalysts on Pr-modified ceria for selective CO oxidation in H2 reach steam (PROX) [13]. In the literature there is only limited number of studies related to WGS reaction over CeePr based catalysts. The investigation of Pd catalysts supported on PrOx and CeePr solid solution with equal amounts of both cations was performed by Bakhmutsky et al. [14]. They have reported that differently from ceria, which is easily reoxidized by steam, the loosely bound oxygen associated with praseodymia or ceriaepraseodymia is lost under WGS conditions and cannot be restored by oxidation in steam. The authors concluded that there is an optimum reducibility that should lead to the most active catalysts because the oxides that hold oxygen too weakly could be as ineffective as oxides that hold oxygen too tightly. Very recently Poggio-Fraccari [15] have studied WGS reaction over Cu and Ni catalysts (the last systems showing also methanation) supported on Pr-doped ceria. It was shown that in agreement with OSC, Raman and TPR results, the samples on supports with low Pr content exhibited better catalytic performances. In the absence of literature about the use of gold catalysts supported on Pr-modified ceria for WGS application and aiming to use an oxide carrier with optimal redox properties, gold catalysts supported on ceria doped with small praseodymium content (5 or 10 at%) were considered in the present study. The purpose was to study the effect of various Pr amounts and different support preparation methods on the catalysts features (gold dispersion, morphology of the support), crucial for the WGS activity. Besides, the established important role of surface re-oxidation for the catalytic activity in PROX [13] was additionally clarified by detailed TPR investigation. The comparison of the reducibility of the studied fresh gold catalysts and the reducibility in the case of re-oxidation performed with the used samples allow to determine the relationship between redox and catalytic behavior in WGSR. 2. Experimental 2.1. Sample preparation Two series of Pr-modified ceria supports with 5 and 10 at% Pr were prepared by impregnation (IM) and by water in oil (W/O) microemulsion (ME) method. The impregnation of ceria with aqueous solution of Pr(NO3)3.6H2O was carried out by vigorous stirring at room temperature for 4 h. Then, the suspension was evaporated under vacuum at 70  C in a rotary evaporator until water was completely removed. The solids were dried at 80  C and calcined in air at 400  C for 2 h. Ceria was prepared by precipitation of aqueous solution of Ce(NO3)3.6H2O with K2CO3 at constant pH ¼ 9.0 and at a temperature of 60  C. The resulting precursor was aged at the same temperature for 1 h, filtered and carefully washed  until removal of NOe 3 ions; then it was dried in vacuum at 80 C and calcined in air at 400  C for 2 h. The mixed supports were denoted as xPrCeIM (x ¼ 5 and 10 at% of Pr). The W/O ME synthesis was carried out as follows: two microemulsion systems with different compositions were prepared separately. The first one, ME1, contained the adequate amounts of the metal nitrates in the aqueous solution and the second one, ME2, contained the adequate amount of the precipitating agent (NH3) in the aqueous solution. Both ME1 and ME2 contained also surfactant (CTAB) and oil phase (n-octane and 1-butanol). ME1 was added dropwise to the ME2 and precipitated particles were obtained. The mixture was kept under vigorous stirring during 24 h at room temperature and after that the particles were collected by centrifugation. The excesses of surfactant, co-surfactant and oil phase were carefully washed first with a 50/50 ethanol/chloroform mixture and then with ethanol.

140

L. Ilieva et al. / Materials Chemistry and Physics 157 (2015) 138e146

The obtained solid was dried at 60  C during 24 h and finally calcined at 500  C for 2 h. The mixed supports were denoted as xPrCeME (x ¼ 5 and 10 at% of Pr). Gold (3 wt%) was introduced by the deposition-precipitation method. It was deposited onto mixed oxides thoroughly suspended in water via chemical interaction between HAuCl4.3H2O and K2CO3, under vigorous stirring, while keeping constant pH ¼ 7. The precipitation was carried out using the ‘Contalab’ system (Switzerland) under full control of all parameters of preparation (pH, temperature, stirring speed, reactant feed flow rates, etc.). After filtering and careful washing, the precursors were dried under vacuum and calcined in air at 400  C for 2 h. The gold-containing samples were denoted as AuxPrCeIM and AuxPrCeME (x ¼ 5 and 10 at% of Pr). 2.2. Sample characterization The BET surface area (SBET) of the samples was determined on a Carlo Erba Sorptomat 1900 instrument. The fully computerized analysis of the nitrogen adsorption isotherm at 196  C allowed to estimate the specific surface areas of the samples in the standard pressure range 0.05e0.3 p/po. Prior to the measurements, the samples were outgassed at 200  C for 30 min under vacuum. X-ray powder diffraction (XRD) measurements were performed using D5000 diffractometer (Bruker AXS), with Cu sealed tube operating at 40 kV and 40 mA. The setup employed Bragg-Brentano focusing geometry with 1 deg. beam divergence and LynxEye strip detector. Data was analyzed using PeakFit program (Jandel Scientific) fitting XRD profiles to Ka 1,2 doublets having PEARSON VII analytical form. For Au fcc phase in most cases only 2e3 weak reflexes were visible and no analysis of kind of Williamson-Hall plot [16,17] was possible nor conclusions regarding possible anisotropy. The crystal size was estimated from FWHM (full width at half maximum) of the strongest 111 reflexion. On the other hand CeO2 phase could be analyzed on the basis of 14 well measured reflections and Williamson-Hall plot could reveal the average crystal size (average column length over all available crystallographic directions) as well as strain parameter (denoted as eps). Transmission electron microscopy (TEM) investigations were performed on a TEM JEOL 2100 instrument at accelerating voltage of 200 kV. The specimens were prepared by grinding and dispersing them in ethanol by ultrasonic treatment for 6 min. The suspensions were dripped on standard carbon/Cu grids. The size distribution of ceria particles was determined by the Program Image J. The X-ray photoelectron (XP) spectra were registered on a PHI 5000 VersaProbe scanning ESCA Microprobe using monochromatic Al-Ka radiation (hn ¼ 1486.6 eV) from an X-ray source operating at 200 mm spotsize, 50 W and 15 kV. The analyzer pass energy was 23.5 eV, an electron take-off angle was 80 and the energy step size was 0.1 eV. The samples were pressed into thin wafers and degassed in a preparation chamber before analysis. The Shirley background subtraction and peak fitting with GaussianeLorentzian product peak was performed using a CasaXPS processing program. A standard way of energy scale calibration for monochromatic Al Ka source was performed (norm ISO 15472:2010). Following this procedure the Ag 3d5/2 peak (binding energy of 368.21 eV), was taken as a reference. In our XPS spectrometer, a PHI's patented dual beam charge neutralization method was used (US Patent 5,990,476). In this method a low energy ion beam is used to eliminate the samples static charge allowing the low energy electron beam to reach the sample and neutralize the localized positive charge created by the x-ray beam. The charging effects were corrected by adjusting the well separated Ce 3d3/2 peak (BE equal to 917.0 eV), usually described as the u''' peak [18,19].

The temperature programmed reduction (TPR) measurements were carried out by means of an apparatus described elsewhere [20]. A cooling trap (40  C) for removing water formed during reduction was mounted in the gas line prior to the thermal conductivity detector. A hydrogeneargon mixture (10% H2), dried over a molecular sieve 5A (40  C), was used to reduce the samples at a flow rate of 24 mL min1. The temperature was linearly raised at a rate of 15  C min1. The amount of used sample was about 0.05 g. Such amount was selected by the criterion proposed by Monti and Baiker [21]. Hydrogen consumption during the reduction process was calculated using preliminary calibration of the thermal conductivity detector, performed by reducing different amounts of NiO to Ni0 (NiO e ‘analytical grade’, calcined at 800  C for 2 h to avoid the presence of non-stoichiometric oxygen). The TPR measurements of fresh gold catalysts as well as TPR of samples tested in WGS reaction and then re-oxidized were performed. The best procedure for deconvolution of complicated TPR spectra is that based on simulation of the TPR processes. In the present study the assignment of the obtained complex TPR profiles to individual reduction process was carried out based on the kinetic parameters assessment, following the algorithm described in more details by Munteanu et al. [ [22] and references therein]. The analysis of the obtained kinetic parameters will be a subject of another study. 2.3. Catalytic activity measurements The catalytic activity measurements in the WGS reaction were carried out in a flow reactor at atmospheric pressure with increasing the temperature in the range 140e350  C. A gas mixture of initial composition 4.5 vol% CO in argon was used. The following experimental conditions were applied: catalyst bed volume ¼ 0.5 cm3 (0.63e0.80 mm pellets), space velocity (GHSV) ¼ 4000 h1, partial pressure of water vapor ¼ 31.1 kPa. Carbon monoxide content at the reactor outlet was determined using an Uras 3G (Hartmann&Braun AG) gas analyzer. After the test of fresh gold catalysts, the samples were pre-treated in a flowing air at 200  C for 60 min at space velocity 2000 h1, followed by cooling down to 140  C in air. Then a second catalytic run up to 350  C was performed. The stability test after the second run with the Au5PrCeIM catalyst was conducted at 275  C keeping the sample under reaction conditions for 40 h. 3. Results and discussion 3.1. Catalytic activity in WGS reaction The catalytic activity in WGS reaction expressed as degree of CO conversion over the studied initial gold catalysts as well as during the second catalytic run after catalysts re-oxidation is illustrated in Fig. 1. Generally, the catalytic activity of fresh catalysts on Pr-doped ceria by IM is higher than that on the supports synthesized by ME but the differences depending on the preparation method and dopant amount are not significant. However, after re-oxidation, the catalytic activity of Au5PrCeIM increased noticeably, while a significant drop in the activity of Au10PrCeME catalyst was registered. The highest activity of Au5PrCeIM was kept stable during the performed long term test at 275  C. The explanation of the observed catalytic behavior was searched in the dispersion of gold and ceria particles as well as the features of the supports depending on the preparation method and dopant amount. 3.2. Sample characterization The effect of the preparation method of CeePr supports on the specific surface area of the studied gold catalysts is shown in

L. Ilieva et al. / Materials Chemistry and Physics 157 (2015) 138e146

100 90

CO conversion, %

80 70 60 50 40 30 20 10 150

200

250

300

350

o

Temperature, C Fig. 1. Temperature dependence of CO conversion during the WGS reaction over the studied fresh gold catalysts: ( ) e Au5PrCeIM, (B) e Au10PrCeIM, (◊) e Au5PrCeME, ( ) e Au10PrCeME; and WGS reaction over re-oxidized samples: (-) e Au5PrCeIM, (C) e Au10PrCeIM, (A) e Au5PrCeME, (:) e Au10PrCeME.

▫

▵

Table 1. ME method has been shown as successful for obtaining ceria particles in the nanosize range, high specific surface area and homogeneous insertion of the dopant ions in ceria structure [ [23] and references therein]. As it was expected on the basis of these observations, the SBET of the catalysts on ME supports was higher than that of the supports synthesized by IM. Increasing the dopant amount leads to the decrease of SBET for both preparation methods. This is in agreement with the study of Somacescu et al. [24] reporting the highest surface area after incorporation in ceria of the lowest praseodymium content. The XRD patterns of all studied gold catalysts on Pr-doped ceria showed only reflections typical of the symmetry of face-centered cubic fluorite-type structure of CeO2 (Fig. S1). The formation of PreCe solid solutions could be assumed since diffraction lines of separate Pr6O11 phase were not detected. From the analysis of the diffraction patterns of the fresh and used catalysts the average (volume weighted) size of ceria crystallites (Dceria), lattice constant (a) and microstrain parameter (eps), as listed in Table 1 were obtained. Surprisingly, a slight decrease in the average size of ceria particles was observed in the used catalysts except in the Au5PrCeIM exhibiting substantial difference (12 nm of fresh sample compared to the 7.5 nm of the used one). Such results of “re-dispersion” after catalytic work were in contradiction to the usually found particles agglomeration. The explanation could be an increased number of small detectable by XRD nanocrystals of ceria built up from the previously amorphous part of the fresh material, thus shifting the

Table 1 BET surface area (SBET); lattice constant of ceria (a), microstrain parameter (eps) in ceria lattice; average size of ceria (Dceria) and gold (DAu) particles. Sample Au5PrCeIM fresh Au5PrCeIM used Au10PrCeIM fresh Au10PrCeIM used Au5PrCeME fresh Au5PrCeME used Au10PrCeME fresh Au10PrCeME used

SBET (m2 g1) 86 69 102 91

a (Å)

eps

Dceria (nm)

DAu (nm)

5.411(1) 5.423(1) 5.415(3) 5.430(3) 5.411(1) 5.426(1) 5.409(1) 5.425(1)

0.0076 0.0089 0.0070 0.0118 0.0033 0.0016 0.0033 0.0051

12.0 7.5 10.6 9.7 8.4 8.0 7.9 7.3

4.3 2.3e2.8 3.4 3.5 10.0 10.0 ~30.0 ~30.0

141

average size of ceria in the used sample towards lower value. The suggestion was supported by means of TEM measurements. Fig. 2A and B illustrates the selected bright field micrographs of fresh (A) and spent (B) Au5PrCeIM catalyst. A higher homogeneity of the sample after WGS reaction is seen. It is also shown the comparison of the diffraction peaks of Fast Fourie Trasform (FFT) images, obtained from morphological TEM (Fig. 2 A0 and B0 ). In good accordance with XRD measurements, it is clearly visible the amorphous halo in the case of fresh sample (A0 ) and the better crystallized sample after using in WGSR (B0 ). The size distribution histograms of ceria particles are also given in Fig. 2 (C and D, respectively). The fractions of very small particles (below 5 nm) and small particles (5e10 nm) were relatively larger in the used gold catalyst. The calculated average size of ceria particles estimated by TEM follows the tendency observed by XRD: it was 11 nm for the as prepared Au5PrCeIM catalyst and 8 nm for the spent sample. The obtained results both by means of XRD and TEM were consistent with the assumption for the formation of well crystallized support after the first catalytic run as compared to the fresh Au5PrCeIM catalyst, that contains partially amorphous modified ceria. It could explain the improvement of WGS activity over this gold catalyst during the second catalytic run. Such behavior in WGSR depending on the support morphology was already reported. The gold catalysts on well crystallized Fe2O3 and ZrO2 exhibited high catalytic activity, while the activity significantly decreased when gold was deposited on amorphous or not well crystallized ZrO2, ZnO, Fe2O3eZnO and Fe2O3eZrO2 [25]. It has been also established that gold catalysts on well crystallized titania manifested significantly higher WGS activity in comparison with the gold samples on amorphous titania [26]. Analyzing the data of cell parameter a, it is worth noticing that there is no expansion of ceria lattice for the fresh catalysts in comparison to Au/CeO2 (a ¼ 5.415 Å). However, after the catalytic reaction the ceria lattice constant for all samples visibly increased. This result could be interpreted in relation to the state of the Prdopant. In the fresh catalysts in addition to Pr4þ the presence of Pr3þ cations is also expected based on the literature data listed in the Introduction. The Pr-dopant in (4þ) state does not cause a change of the ceria lattice constant (the ionic radius of Pr4þ cation is 0.96 Å, very close to 0.97 Å of Ce4þ cation). The amount of Pr3þ (ionic radius of 1.13 Å) is not so high (low Pr content of 5 and 10 at%) to produce a visible expansion of the ceria lattice. However, the Pr3þ amount increased after the WGS reaction as Pr4þ was easier to be reduced whereas Pr3þ oxidation was difficult [10], especially by steam. This produced the expansion of the lattice. Relatively smallest expansion was registered for the most active Au5PrCeIM (the lowest a ¼ 5.423 Å) as compared to the other spent samples, which could be related to a larger fraction of Pr3þ being re-oxidized to Pr4þ by steam. In the fresh catalysts the lattice microstrain eps for ME supports with 5 and 10 at% was the same (0.0033) and it was about twice lower as compared to that of IM supports (0.0076 and 0.0070, respectively). The lattice microstrain increased after catalytic reaction except for Au5PrCeME, but again lower values were observed for ME supports as compared to IM. This finding was consistent with the supposition that the microemulsion method leads to mixed supports with less defective structure. Reflections at 2Q ¼ 38.2 , corresponding to metallic Au (1 1 1), were well visible in the X-ray difractograms of the catalysts. The average crystallite size of gold calculated from the line broadening using the equation of Scherrer is given in Table 1. Although the ME supports were characterized by larger surface area, bigger average size of gold particles was observed on these supports in comparison to the carriers prepared by IM. Entirely unexpected was the average size of 30 nm for Au10PrCeME sample. It is well known that the

142

L. Ilieva et al. / Materials Chemistry and Physics 157 (2015) 138e146

B

A

A'

B'

70

70

C

60

50 Frequency, %

Frequency, %

50 40 30 20 10

0

D

60

40 30 20 10

0-5

5-10

10-15

15-20

0

0-5

Particle size, nm

5-10

10-15

15-20

Particle size, nm

Fig. 2. Selected bright field TEM micrographs, FFT of the TEM images, and size distribution histograms of ceria particles in fresh (A, A0 and C, correspondingly) and used (B, B0 and D, correspondingly) Au5PrCeIM catalyst.

presence of finely dispersed gold is a key factor for the WGS activity. The simultaneously presence of big but also very small gold particles in Au10PrCeME can be assumed because the degree of CO conversion over this fresh catalyst was comparable to that of other studied gold catalysts. However the very small gold particles were not evidenced by XRD as the symmetric shape of gold reflections was not in accord with a bimodal distribution. It is worth noting that no differences in gold particle sizes were observed between the fresh and the used catalysts. X-ray photoelectron spectroscopy (XPS) allowed to study the surface composition and chemical state of gold, Ce and Pr cations. The XPS data of the examined fresh samples are summarized in Table 2. The Au 4f XP spectra were fitted successfully with 2e3 components, which could be assigned to differently charged Au particles. For all fresh catalysts, in addition to metallic gold (Au0) with binding energy (BE) of Au 4f7/2 peaks centered at ~84 eV, Au species with electropositive (Audþ) oxidation state were also detected. In the case of ME preparation method both partially

Table 2 The ratios Pr/Ce, Au/Ce and Au/Pr, calculated for the fresh and used in WGS gold catalysts. Catalyst

Ratios

Fresh sample

Used sample

Au5PrCeIM

Pr/Ce Au/Ce Au/Pr Pr/Ce Au/Ce Au/Pr Pr/Ce Au/Ce Au/Pr Pr/Ce Au/Ce Au/Pr

0.121 0.020 0.168 0.267 0.045 0.170 0.114 0.029 0.253 0.217 0.068 0.316

0.113 0.024 0.214 0.235 0.043 0.182 0.120 0.026 0.218 0.174 0.029 0.164

Au10PrCeIM

Au5PrCeME

Au10PrCeME

Intensity, a.u.

L. Ilieva et al. / Materials Chemistry and Physics 157 (2015) 138e146

Au5PrCeIM fresh

Au10PrCeME fresh

Au5PrCeIM spent

Au10PrCeME spent

92

90

88

86

84

82

80

92

90

88

86

84

82

80

Binding Energy, eV Fig. 3. Comparison of experimental and fitted Au 4f XPS spectra of Au5FeCeIM and Au10PrCeME catalysts: fresh and used in WGSR.

in the case of gold supported on Mn- and Sn-doped ceria [28,29]. As it was discussed above, according to XRD measurements the average size of gold particles in ME catalysts was bigger than in the case of IM preparation and relatively very big gold crystallites (average size of about 30 nm) in Au10PrCeME were detected. However, the XPS data of charged gold species reveal that also finely dispersed gold co-exists in the case of supports synthesized by ME method. Very small gold particles (1e2 nm), in particular highly dispersed gold clusters (<1 nm), are the prerequisite for high WGS activity [30]. The presence of very small, not detectable by XRD gold particles play the key role for the very similar WGS activity of the fresh gold catalysts on ME and IM supports. The existence of big gold particles together with very small ones evidenced by FTIR but not detectable by XRD and even HRTEM, was reported by Tabakova et al. studying gold on Fe-modified ceria [31]. In all cases, after WGS reaction, only metallic gold was registered by XPS. In Fig. 3 the comparison of experimental and fitted Au 4f XPS spectra of fresh and used Au5FeCeIM and Au10PrCeME catalysts is shown. These data are in accordance with the results obtained by Rodriguez et al. showing, through X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS), that the active phase of Au/ceria catalyst under WGS conditions, consists of metallic gold nanoparticles on a partially reduced ceria [ [32] and references therein]. The drastic drop in the WGS activity of Au10PrCeME during the second catalytic run could be explained by the agglomeration of the very small gold particles responsible for the higher initial activity. The Ce 3d and Pr 3d core level XP spectra are rather complex resulting from the different transition final states of the Ce3þ/Pr3þ

500

500

450

450

400

400

350

350

300

300

HC, a.u.

HC, a.u.

electropositive and electronegative (Aud) state can be distinguished. The negatively charged gold reflects a charge transfer from the support to the gold particles, whereas the positive may be attributed to the presence of such a small Au particles to be close to the atomic state rather than to the metallic. There are publications describing such phenomenon [27]. Besides metallic gold (Au0), both Audþand Aud states of gold were distinguished by some of us

250 200 150

200 150 100

50

50

0

0

-50

-50

50

100

150

200

250

Temperature, oC

300

350

0

50

100

150

200

250

300

350

Temperature, oC 500

500

450

450

400

400

350

350 300

HC , a.u.

HC , a.u.

250

100

0

143

250 200 150

300 250 200 150

100

100

50

50

0

0

-50 0

50

100

150

200

250

o Temperature, C

300

350

-50 0

50

100

150

200

250

300

350

o

Temperature, C

Fig. 4. TPR profiles of the studied gold catalysts on supports prepared by IM: continuous lineeexperimental curve, dotted lines e fitted calculated curve and peaks of individual reduction processes.

144

L. Ilieva et al. / Materials Chemistry and Physics 157 (2015) 138e146

similar to the observations found by other authors [34,35 and references therein]. For all gold catalysts with the exception of Au10PrCeME, the Au/Ce and Au/Pr ratios did not differ significantly in the fresh and used samples. However, for the fresh Au10PrCeME Au/Ce ¼ 0.068 and Au/Pr ¼ 0.316, while in the spent catalyst the corresponding values were only 0.029 and 0.164. The possible reason could be the sintering of finely dispersed gold in agreement to the already given explanation for the WGS activity of the fresh Au10PrCeME catalyst followed by a drastic drop in activity during the second catalytic run. The agglomeration leads to a decrease of the areal density of Au nanoparticles accessible to XPS analysis [36]. The effect of gold as well as of the doping ions on the Ce4þ / Ce3þ transition in ceria based materials was widely investigated by means of TPR measurements. The role of gold in the substantial lowering of the reduction temperature of surface ceria had already been proved in many studies. Relevant to the catalytic performance in low temperature WGS reaction was the influence of dopant, in particular Pr, not only on the reducibility but also on the re-oxidation ability of ceria-based supports. In the present study TPR measurements of fresh gold catalysts and TPR after reoxidation of the samples tested in WGSR were carried out. The TPR spectra were rather complex, as both Ce4þ4Ce3þ and Pr4þ4Pr3þ transformations were possible during the reduction and oxidation, respectively. Each complex TPR process had been simulated considering the equation for the rate of reduction [22]. The TPR profiles (Figs. 4 and 5) were deconvoluted into peaks related to individual reduction processes (dotted lines) with a good fit between the experimental TPR curve (continuous line) and the

500

500

450

450

400

400

350

350

300

300

HC, a.u.

HC , a.u.

and Ce4þ/Pr4þ cations. Concerning cerium, the procedure for the calculation of Ce3þ concentration was, like in many studies, using two sets of spineorbit multiplets corresponding to 3d5/2 and 3d3/2, usually labeled as v and u peaks, correspondingly. Although the Pr 3d spectrum has a rather similar shape as the Ce 3d, the precise analysis of Pr oxidation states has been less elaborated, and often only row spectra, are reported [24]. In the present study, like in our very recent investigation of PROX over gold catalysts on Prmodified ceria [13], the Pr 3d spectra of fresh and used in WGS catalysts were interpreted by deconvolution into five pairs of 3d5/2 and 3d3/2 spineorbit components (by the analogy to Ce 3d XP spectra) following the procedure given Konysheva et al. [33]. In all studied fresh catalysts the contribution from both Pr4þ and Pr3þ was observed, the concentration of Pr3þ being higher in the samples containing higher Pr amount [Table 2, in Ref. 13]. The latest results (after considering also the XPS measurements of Pr6O11 and Pr2O3) obtained for Au10PrCeME fresh sample were put together with some existing literature data in the Table S1, section Supplementary data. The applied fitting procedure was illustrated in Fig. S2. The results intended to extend the very scarce information in the literature and to improve the quantitative analysis of Pr3þ and Pr4þ oxidation states. In Table 2 the Pr/Ce, Au/Ce and AuPr intensity ratios calculated on the basis of Au 4f, Ce 3d and Pr 3d peak areas of the fresh and used in WGS gold catalysts are compared. In all cases Pr/Ce ratio was approximately twofold higher as compared to the theoretical values of 0.053 and 0.111 for doping with 5 at% and 10 at% Pr, respectively. These results indicated a surface enrichment of Pr ions

250 200 150 100

200 150 100

50

50

0

0

-50 0

50

100

150

200

Temperature,

250

o

300

-50

350

0

C

50

100

150

200

250

300

350

300

350

Temperature, oC

500

500

450

450

400

400

350

350

300

300

HC, a.u.

HC, a.u.

250

250 200 150 100

250 200 150 100

50

50

0

0

-50 0

50

100

150

200

250

o Temperature, C

300

350

-50 0

50

100

150

200

250

Temperature, oC

Fig. 5. TPR profiles of the studied gold catalysts on supports prepared by ME: continuous lineeexperimental curve, dotted lines e fitted calculated curve and peaks of individual reduction processes.

L. Ilieva et al. / Materials Chemistry and Physics 157 (2015) 138e146

computed one (dotted line). Fig. 4 illustrates the reduction behavior of fresh and re-oxidized gold catalysts on CeePr oxide supports synthesized by IM method and Fig. 5 e the reduction behavior in the case of ME preparation method. Since only the ceria surface reduction peaks were of interest for the WGS reaction, the temperature range up to 350  C was considered. The reduction of IM catalysts proceeded at lower temperatures (the end of TPR peaks was ~250  C) as compared to that of ME samples (the Tmax of that peaks was located at ~200  C). According to the literature data higher reducibility was expected for smaller ceria particles [37], however the XRD results (Table 1) showed bigger ceria particles using IM method of preparation. The observed enhanced reduction could be related to the higher oxygen mobility in the more defective structure of ceria doped with Pr by impregnation than in the better ordered one, expected for the more homogeneous CeePr solid solutions, synthesized by ME. The lattice microstrain eps confirmed the less defective structure obtained in the case of the ME prepared samples. Another reason could be connected to gold particles with smaller average size on the supports prepared by IM. The mentioned difference in the reducibility correlated with the relatively higher WGS activity of fresh gold catalysts on IM supports as compared to the gold catalysts on ME supports. A good fit of the experimental profiles for the fresh gold catalysts on supports prepared both by IM and ME method was obtained considering two individual reduction processes (dotted lines). Zhou and Gorte [7] have concluded that oxygen could be more easily removed from praseodymia compared to ceria. Studying Rh/Ce0.8Pr0.2O2x catalysts, Lopez-Cartes et al. [38] have suggested that the reduction of Pr4þ species took place at much lower temperature then Ce4þ reduction. The first TPR peak has to be attributed to the low temperature Pr4þ / Pr3þ and Ce4þ / Ce3þ reduction influenced by gold. Having in mind the low amounts of Pr-doping and the XPS results proving the presence of Pr3þ in the initial samples, the main process should be related to the easiest reducible ceria surface influenced by the presence both of gold and the dopant. The reduction of the positively charged gold took place as well but the corresponding hydrogen consumption was negligible. The second TPR peak could be due to the reduction of the consecutive surface layers of modified ceria. For the explanation of the observed WGS activity after reoxidation following the first catalytic run, the handling of individual reduction processes could be helpful. The TPR profiles after reoxidation of gold catalysts on IM supports were wider and with more intense low temperature part as compared to ME supports. A good fit between theoretical and experimental TPR profiles registered after the re-oxidation of gold catalysts on ME supports was achieved by two individual processes. However, for the much broader TPR peaks in the case of IM method, a consideration of three individual processes was needed. In Table 3 the Tmax and the calculated hydrogen consumption (HC) for every individual reduction process are summarized. The lattice oxygen of ceria supports with the highest mobility is of greatest importance for the WGS activity. After its extraction and oxygen vacancy formation, it enables to be recovered by the participation of water which is a weaker oxidant in comparison with oxygen. Only for Au5PrCeIM catalyst the Tmax of the first peak after re-oxidation was even lower and the HC was the same as compared to the initial sample. The first TPR peak for the re-oxidized Au10PrCeIM was with a slightly higher Tmax and lower HC than the first peak of the as prepared catalyst. For the re-oxidized Au5PrCeME the Tmax of the first TPR peak was relatively lower (168 compared to 182  C for the fresh sample) but it was significantly higher as compared to the gold catalyst on supports prepared by IM. In the case of re-oxidized Au10PrCeME sample practically the process of lower temperature reduction did not exist.

145

Table 3 Tmax and HC of the individual processes related to the TPR of fresh and re-oxidized gold catalysts. Sample

Tmax ( C)

HC (mmol g1)

Au5PrCeIM fresh

94 136 86 132 180 102 139 105 152 193 182 208 168 191 180 202 218 224

0.15 0.34 0.15 0.14 0.16 0.16 0.29 0.10 0.11 0.14 0.18 0.34 0.07 0.36 0.15 0.28 0.07 0.36

Au5PrCeIM reox

Au10PrCeIM fresh Au10PrCeIM reox

Au5PrCeME fresh Au5PrCeME reox Au10PrCeME fresh Au10PrCeME reox

These results revealed that the re-oxidation of Pr3þ and Ce3þ cations in ceria surface layers is depending on the preparation method of the mixed supports. The better ordered solid solution structure caused by ME preparation method could hinder both the reduction and the re-oxidation processes, leading to lower LT WGS activity. Gold catalysts on Pr-doped ceria supports synthesized by ME method also exhibited lower activity in PROX as compared to the preparation by IM [13]. The present study showed that the use of the IM method favored the catalyst re-oxidation. Especially for the lower Pr amount (5 at%) in IM-prepared sample the most easily removable oxygen could be fully recovered. The easier re-oxidation of this catalyst by steam was consistent with the smallest expansion of ceria lattice obtained for the sample after the catalytic reaction. In addition, the increased WGS activity during the second catalytic run was in accordance with the assumption of better WGS performance in the presence of well crystallized support as compared to the case of partially amorphous ceria in the fresh Au5PrCeIM sample. The redox and the WGS activity were strongly affected by the presence of gold, which dispersion was a crucial factor. The proposed agglomeration of the very small gold particles after the first test in WGSR over Au10PrCeME catalyst can satisfactorily explain the reduction behavior and the significant drop in activity during the sequence of catalytic test.

4. Conclusions Gold catalysts on Pr-doped ceria supports synthesized by IM method exhibited higher WGS activity than those on supports prepared by ME method but the differences were not very significant. However, after re-oxidation following the first catalytic run, the WGS activity of gold catalyst on IM support with 5 at% Pr increased; while a significant drop in activity of the gold catalysts on ME prepared support containing 10 at% Pr was registered. The explanation was connected to the dispersion of gold and the supports features. In the fresh Au10PrCeME sample in addition to big gold particles the presence of very small ones (XPS results of partially charged gold) was responsible for the initial WGS activity. The agglomeration of these small particles after the catalytic reaction caused a substantial decrease of the activity in the second test. The higher WGS activity during the second catalytic run over Au5PrCeIM catalyst could be explained on one hand by the important role of the redox behavior. Only in the case of this gold

146

L. Ilieva et al. / Materials Chemistry and Physics 157 (2015) 138e146

catalyst the most easily removable oxygen could be fully recovered. On the other hand the morphology of the support before and after reaction affected the WGS performance: in the presence of well crystallized support the WGS activity is better than in the case of partially amorphous ceria in the fresh gold catalyst. The study established a relationship between the catalytic behavior in WGS reaction and redox properties connected to the most mobile oxygen, depending on the preparation method of doped ceria supports and Pr-dopant amount. Acknowledgments L.I. is grateful for the financial support via Bulgarian project BG051PO001/3.3.-05.0001 (contract D03-197) “Science and Business” funded under the programme “Human Resources Development” of the European Social Fund. The study was performed in the frame of COST ActionsCM0903 and CM1104. Bulgarian team thanks the National Science Fund of Bulgaria for a financial support through the Projects DFNI E-01/7 and E-02/2. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.matchemphys.2015.03.028. References [1] W. Deng, J. De Jesus, H. Saltsburg, M. Flytzani-Stephanopoulos, Low-content gold-ceria catalysts for the wateregas shift and preferential CO oxidation reactions, Appl. Catal. A 291 (2005) 126e135. [2] Q. Zhang, Y. Zhan, X. Lin, Q. Zheng, The effect of La on Au-ceria catalyst for water gas shift reaction, Catal. Lett. 115 (2007) 143e147. [3] Y. Wang, S. Liang, A. Cao, R.L. Thompson, G. Veser, Au-mixed lanthanum/ cerium oxide catalysts for water-gas-shift, Appl. Catal. B 99 (2010) 89e95. [4] D. Andreeva, I. Ivanov, L. Ilieva, M.V. Abrashev, R. Zanella, J.W. Sobczak, W. Lisowski, M. Kantcheva, G. Avdeev, K. Petrov, Gold catalysts supported on ceria doped by rare earth metals for water gas shift reaction: influence of the preparation method, Appl. Catal. A 357 (2009) 159e169. [5] A. Trovarelli, C. Leitenburg, G. Dolcetti, Design of better cerium-based oxidation catalysts, Chemtech 27 (1997) 32e37. [6] V.G. Keramidas, W.B. White, Raman spectra of oxides with the fluorite structure, J. Chem. Phys. 59 (1973) 1561e1562. [7] G. Zhou, R.J. Gorte, Ceriaeterbia solid solution nanobelts with high catalytic activities for CO oxidation, J. Phys. Chem. B 112 (2008) 9869e9875. [8] B.M. Reddy, G. Thrimurthulu, L. Katta, Y. Yamada, S.E. Park, Structural characteristics and catalytic activity of nanocrystalline ceria-praseodymia solid solutions, J. Phys. Chem. C 113 (2009) 15882e15890. [9] K. Ahn, D.S. Yoo, D.H. Prasad, H.-W. Lee, Y.-C. Chung, J.-H. Lee, Role of multivalent Pr in the formation and migration of oxygen vacancy in Pr-doped ceria: experimental and First-Principles investigations, Chem. Mater. 24 (2012) 4261e4267. [10] E.S. Putna, J.M. Vohs, R.J. Gorte, G.W. Graham, An examination of praseodymia as an oxygen-storage component in three-way catalysts, Catal. Lett. 54 (1998) 17e21. [11] A.D. Logan, M. Shelef, Flexural strength of feldspathic porcelain treated with ion exchange, overglaze, and polishing, J. Mater. Res. 9 (1994) 468e472. [12] W. Chun, G.W. Graham, J.A. Lupescu, R.W. McCabe, M.M. Koranne, R. Brezny, The effect of pretreatment procedures on the activities of Fe- and Mnpromoted sulfated zirconia catalysts, Catal. Lett. 95 (2006) 95e100. [13] L. Ilieva, P. Petrova, T. Tabakova, G. Pantaleo, V. Montes, J.W. Sobczak, W. Lisowski, Z. Kaszkur, M. Boutonnet, A.M. Venezia, Pure hydrogen production via PROX over gold catalysts supported on Pr-modified ceria, Fuel 134 (2014) 628e635. [14] K. Bakhmutsky, G. Zhou, S. Timothy, R.J. Gorte, The WatereGas-shift reaction on Pd/ceriaePraseodymia: the effect of redox thermodynamics, Catal. Lett. 129 (2009) 61e65. [15] E. Poggio-Fraccari, F. Marino, M. Laborde, G. Baronetti, Copper and nickel catalysts supported on praseodymium-doped ceria (PDC) for the water-gas shift reaction, Appl. Catal. A 460e 461 (2013) 15e20.

[16] G.K. Williamson, W.H. Hall, X-ray line broadening from filed Al and W, Acta Metall. 1 (1953) 22e31. [17] V.D. Mote, Y. Purushotham, B.N. Dole, Williamson-Hall analysis in estimation of lattice strain in nanometer-sized ZnO particles, J. Theor. Appl. Phys. 6 (2012) 6e28. [18] L. Armelao, D. Barecca, G. Bottaro, A. Gasparotto, E. Tondello, Plasmaenhanced CVD CeO2 nanocrystalline thin films analyzed by XPS, Surf. Sci. Spectra 8 (2001) 247e249. [19] A.Q. Wang, P. Punchaipetch, R.M. Wallace, T.D. Golden, X-ray photoelectron spectroscopy study of electrodeposited nanostructured CeO2 films, J. Vac. Sci. Technol. B 21 (2003) 1169e1175. [20] N. Kotzev, D. Shopov, Thermodesorption study of the system olefin-NiO, J. Catal. 22 (1971) 297e301. [21] D.A.M. Monti, A. Baiker, Temperature-programmed reduction. Parametric sensitivity and estimation of kinetic parameters, J. Catal. 83 (1983) 323e335. [22] G. Munteanu, L. Ilieva, R. Nedyalkova, D. Andreeva, Influence of gold on the reduction behaviour of AueV2O5/CeO2 catalytic systems: TPR and kinetic parameters of reduction, Appl. Catal. A 277 (2004) 31e40. [23] O.H. Laguna, M.A. Centeno, M. Boutonnet, J.A. Odriozola, Fe-doped ceria solids synthesized by the microemulsion method for CO oxidation reactions, Appl. Catal. B 106 (2011) 621e629. [24] S. Somacescu, V. Parvulescu, J.M. Calderon-Moreno, S.-H. Suh, P. Osiceanu, B.L. Su, Uniform nanoparticles building Ce1xPrxO2v mesoarchitectures: structure, morphology, surface chemistry, and catalytic performance, J. Nanopart. Res. 14 (2012) 885e902. [25] T. Tabakova, V. Idakiev, D. Andreeva, I. Mitov, Influence of the microscopic properties of the support on the catalytic activity of Au/ZnO, Au/ZrO2, Au/ Fe2O3, Au/Fe2O3eZnO, Au/Fe2O3eZrO2 catalysts for the WGS reaction, Appl. Catal. A 202 (2000) 91e97. [26] D. Andreeva, V. Idakiev, T. Tabakova, R. Giovanoli, Low-temperature WGS reaction on Au/TiO2, Au/Fe2O3 and Au/Co3O4, Bulg. Chem. Comm. 30 (1998) 59e68. [27] M. Mihaylov, H. Knozinger, K. Hadjiivanov, B.C. Gates, Characterization of the oxidation states of supported gold species by IR spectroscopy of adsorbed CO, Chem. Ing. Tech. 79 (6) (2007) 795e806. [28] L. Ilieva, G. Pantaleo, I. Ivanov, R. Zanellac, J.W. Sobczak, W. Lisowskid, A.M. Venezia, D. Andreeva, Preferential oxidation of CO in H2 rich stream (PROX) over gold catalysts supported on doped ceria, Eff. Water CO2, Catal. Today 175 (2011) 411e419. [29] T. Tabakova, L. Ilieva, I. Ivanov, R. Zanella, J.W. Sobczak, W. Lisowski, Z. Kaszkur, D. Andreeva, Influence of the preparation method and dopants nature on the WGS activity of gold catalysts supported on doped by transition metals ceria, Appl. Catal. B 136e137 (2013) 70e80. [30] T. Tabakova, M. Manzoli, F. Vindigni, V. Idakiev, F. Boccuzzi, CO-Free hydrogen production for fuel cell applications over Au/CeO2 catalysts: FTIR insight into the role of dopant, J. Phys. Chem. A 114 (2010) 3909e3915. [31] T. Tabakova, G. Avgouropoulos, J. Papavasiliou, M. Manzoli, F. Boccuzzi, K. Tenchev, F. Vindigni, T. Ioannides, CO-free hydrogen production over Au/ CeO2eFe2O3 catalysts: Part 1. Impact of the support composition on the performance for the preferential CO oxidation reaction, Appl. Catal. B 101 (2011) 256e265. [32] J.A. Rodriguez, Gold-based catalysts for the wateregas shift reaction: active sites and reaction mechanism, Catal. Today 160 (2011) 3e10. [33] E.Y. Konysheva, M.V. Kuznetsov, Fluctuation of surface composition and chemical states at the hetero-interface in composites comprised of a phase with perovskite structure and a phase related to the RuddlesdenePopper family of compounds, RSC Adv. 3 (2013) 14114e14122. [34] H. Borchert, Y.V. Frolova, V.V. Kaichev, I.P. Prosvirin, G.M. Alikina, A.I. Lukashevich, V.I. Zaikovskii, E.M. Moroz, S.N. Trukhan, V.P. Ivanov, E.A. Paukshtis, V.I. Bukhtiyarov, V.A. Sadykov, Electronic and chemical properties of nanostructured cerium dioxide doped with praseodymium, Phys. Chem. B 109 (2005) 5728e5738. n-Hurtado, R. Lo pez-Fonseca, F. Coloma-Pascual, [35] B. de Rivas, N. Guille rrez-Ortiz, A. Bueno-Lo  pez, Activity, selectivity and A. García-García, J.I. Gutie stability of praseodymium-doped CeO2 for chlorinated VOCs catalytic  n, Appl. Catal. B 121e122 (2012) 162e170. combustio [36] A. Mihaylova, A. Naydenov, D. Kovacheva, E. Ivanova, D. Stoyanova, P. Stefanov, Silver-based storage catalyst for neutralization of nitrogen oxides, Catal. Commun. 10 (2009) 1288e1291. [37] Q. Fu, S. Kudriavtseva, H. Saltsburg, M. Flytzani-Stephanopoulos, Goldeceria catalysts for low-temperature water-gas shift reaction, Chem. Eng. J. 93 (2003) 41e53. pez-Cartes, S. Bernal, J.J. Calvino, M.A. Cauqui, G. Blanco, J.A. Pe rez-Omil, [38] C. Lo J.M. Pintado, Helveg, P.L. Hansen, in situ transmission electron microscopy investigation of Ce(IV) and Pr(IV) reducibility in a Rh (1%)/Ce0.8Pr0.2O2x catalyst, Chem. Commun. (2003) 644e645.