Nanoscale Heterogeneity in Ceria Zirconia with Low-Temperature

A spray drying system for synthesis of rare-earth doped cerium oxide nanoparticles. Vaneet Sharma , Kathryn M. Eberhardt , Renu Sharma , James B. Adam...
0 downloads 0 Views 356KB Size
Download PDF
18278

J. Phys. Chem. B 2006, 110, 18278-18285

Nanoscale Heterogeneity in Ceria Zirconia with Low-Temperature Redox Properties Ruigang Wang,†,‡ Peter A. Crozier,*,‡ Renu Sharma,‡ and James B. Adams†,§ Science and Engineering of Materials Program and Center for Solid State Science, Arizona State UniVersity, Tempe, Arizona 85287-1704, and Chemical and Materials Engineering, Arizona State UniVersity, Tempe, Arizona 85287-6006 ReceiVed: May 22, 2006; In Final Form: July 13, 2006

We have investigated nanoscale compositional and structural variations in Ce0.5Zr0.5O2 samples with different redox properties. Different samples were prepared using a spray freezing technique, and the synthesis conditions were varied to yield materials with reduction temperatures in the range of 400-750 °C. X-ray diffraction and thermal gravimetric analysis were used to characterize the average structures and redox properties of these materials. The nanoscale structural and compositional variations in individual nanoparticles of high activity were determined with atomic-scale electron imaging and nanometer-resolution electron energy loss spectroscopy. During the early stage of particle formation, the crystallization process is initiated via the nucleation of ceria-rich nanodomains. This results in the formation of high-surface-area materials that exhibit nanoscale compositional heterogeneity consisting of Ce-rich cores surrounded by Zr-rich shells. The effect of high-temperature redox cycling on the nanoscale structure, composition, and low-temperature redox properties was also determined. Our analysis suggests that our most active material exhibits significant compositional and structural heterogeneity at the nanometer level.

Introduction The ability of cerium to reversibly exist in mixed +3 and +4 valence states in oxides makes it ideally suited for applications where high redox activity is required, but pure cerium oxide is not suitable for many applications due to its high reduction temperature (700 °C and above) and problems with sintering. Moreover, reduction can result in ordering of oxygen vacancies and lowering of the oxygen mobility. These drawbacks can be overcome by mixing ceria with other oxides such as zirconia.1-4 Such solid solutions exhibit lower reduction temperatures and are more resistant to sintering, resulting in improved oxygen storage capacity (OSC). The excellent OSC found in cerium-based solid solutions has made them critical for maximizing the performance of the environmentally important automotive three-way catalyst (TWC).5,6 Ceria-based solid solutions are also being considered for other important catalytic processes such as the water-gas shift reaction7,8 and solid oxide fuel cell anodes.9 High redox activity (low-temperature reducibility and high percentage of Ce with an easily reversible oxidation state) is one of the most attractive properties of CexZr1-xO2 solid solutions. High-surface-area and nanoscale CexZr1-xO2 solid solutions have been widely studied by various characterization techniques to determine the optimal or most active oxygen storage component of the system. The redox properties of CexZr1-xO2 are strongly dependent on the structure,10,11 chemical composition,10,12-13 and pretreatment.14-16 The origin of this improved redox property is not yet resolved. Mamontov et al.17,18 have suggested that the desirable redox properties are related to the existence of compositional nanoscale domains (detected with neutron scattering) in which the local Ce/Zr composition changes * To whom correspondence should be addressed. E-mail: [email protected]. † Science and Engineering of Materials Program. ‡ Center for Solid State Science. § Chemical and Materials Engineering.

but the overall crystal structure remains the same. They hypothesize that the enhanced activity is associated with the interfacial regions between the domains. Recent work has shown enhanced low-temperature reducibility and high mobility in ceria films deposited on single-crystal zirconia substrates and also points to the role of interfacial interactions in the lowtemperature redox properties.19-21 An alternative hypothesis for low-temperature reducibility asserts that pyrochlore-type cation ordering is the key to understanding the origin of lowtemperature reduction. Montini et al.22 subjected ceria zirconia to redox cycles under high-temperature reducing conditions and found evidence for cation ordering at least on the surface of their material from Raman spectroscopy. Using transmission electron microscopy (TEM), Kang23 showed the presence of cubic, tetragonal, and pyrochlore-type phases in material subjected to high-temperature reduction. Even though most of the recent results reported above were for samples of average composition Ce0.5Zr0.5O2, Aneggi et al.24 pointed out that the samples came from different synthesis and treatment procedures, which complicates the interpretation. It is clear that some form of nanoscale heterogeneity may play an important role in explaining the redox properties of ceria zirconia. Considerable insight into this heterogeneity can be obtained by correlating nanoscale measurements of the composition and structure of individual nanoparticles with macroscopic redox properties. Despite the importance of nanoscale characterization of high activity in CeO2-ZrO2, the number of published reports remains limited.25,26 We have employed atomic-scale imaging and nanospectroscopy techniques available in modern transmission electron microscopy to characterize the nanoscale structure and composition of individual nanoparticles of high-activity Ce0.5Zr0.5O2. A series of solid solution samples were prepared under different conditions to optimize the lowtemperature redox properties. X-ray diffraction and thermal gravimetric analysis were used to characterize the average

10.1021/jp063113z CCC: $33.50 © 2006 American Chemical Society Published on Web 08/31/2006

Nanoscale Heterogeneity in Ceria Zirconia structures and redox properties of the materials. Atomicresolution phase-contrast imaging, Z-contrast imaging, and electron energy loss spectroscopy (EELS) were used to determine the nanoscale structure, morphology, and chemistry of the materials. We found that our most active materials exhibit significant compositional and structural heterogeneity at the nanometer level. Experimental Section Powder Synthesis and Thermal Processing in Air. Spray freezing is a wet chemical route which can be used to prepare nanoscale homogeneous solid solution powders. The objective of spray freezing is to create an intimate mixture of Ce/Zr precursors in a form suitable for calcination by rapidly freezing a homogeneous liquid solution. Most of the water is removed from the system by pumping the frozen solid in a vacuum system, resulting in the formation of a viscous gel. The resulting gel can then be calcined, resulting in the formation of an oxide. The procedure involves three main steps: freeze granulation, vacuum-drying, and heat treatment. We employed this method to prepare nanoscale CeO2-ZrO2 solid solution powders starting with 0.15 M aqueous solutions of Ce(NO3)3‚6H2O (cerium(III) nitrate hexahydrate; Aldrich, 99.999%) and ZrO(NO3)2‚2H2O (zirconium dinitrate oxide; Alfa Aesar, 99.995%). These solutions were mixed in the appropriate ratio to give a Ce/Zr atomic concentration ratio of unity and stirred for 1 h immediately before spraying. The solution was then sprayed using an airbrush onto a surface which was kept at liquid nitrogen temperature. Most of the water was removed from the resulting homogeneous ice granules by pumping in a vacuum system for a period of 24 h (while the sample was kept cold). The resulting gel was thinned in alcohol (to facilitate removal from the container) and calcined in air at 360 °C for 0.5 h. The resulting solid precursor was then ground and mixed for 30 min and further calcined at different temperatures (360, 500, 700, and 1000 °C) and holding times (0.5, 2, and 5 h) to synthesize a series of samples with a nominal composition of Ce0.5Zr0.5O2 (50Ce50Zr). All the synthesized materials were pale yellow in color. For the purposes of comparison, a pure CeO2 nanoparticle sample was also prepared by a standard precipitation method27 and calcined in air at 700 °C. X-ray Diffraction. The effect of the calcination temperature on the phase change, solid solution composition, and crystal size was investigated by X-ray diffraction (XRD; Rigaku D/Max-IIB) using Cu KR radiation (λ ) 1.54056 Å). The samples were loaded onto a zero background quartz plate, and data were collected over a 2θ range of 10-90°. The crystal size was estimated from X-ray peak broadening using the Scherrer equation (computer software Jade 6.0).28 Thermal Processing in Hydrogen, Redox Activity, and Thermal Gravimetric Analysis. Some of the samples were also exposed to a high-temperature treatment in hydrogen to run the material through redox cycles. Simultaneous thermal gravimetric analysis (TGA) was performed under 5% H2/95% He at a scan rate of 2 or 10 °C/min during treatment or heating in a Setaram TG92 system to detect and characterize the reduction process. All samples were subjected to the same cleaning pretreatment step to remove loosely bound adsorbates (such as water) which may complicate the interpretation of the TGA profiles. In this cleaning treatment, the samples were heated to 300 °C (at 10 °C/min) and held at this temperature for 5 h in the hydrogen. The material was then cooled to 150 °C and held at this temperature for a further 5 h. The mass of the sample at the end of the cleaning procedure was used as

J. Phys. Chem. B, Vol. 110, No. 37, 2006 18279 the baseline for determining the mass loss due to reduction. After completion of the cleaning step, the samples were heated at 2 °C/min to 1000 °C and held at this temperature for 2.75 h. The negative of the differential mass with respect to temperature (-dM/dT) was plotted with respect to temperature to visualize the reduction process during the ramp-up period. The total mass loss due to reduction was calculated from the mass determined just before the start of the cooling step. Nanoanalysis. The structural and compositional heterogeneity between and within individual nanoparticles was determined using TEM-based techniques. The TEM characterization was performed using a JEOL 2010F operated at 200 kV and equipped with a Gatan Enfina electron energy loss spectrometer and annular dark-field detector. The point-to-point resolution of this instrument is 0.19 nm, and the electron probe can be focused to 0.2 nm. The synthesized powders were dispersed over standard holey carbon films supported on Cu grids. Relatively low resolution imaging was used to determine the grain size and sample densification. Atomic-resolution phasecontrast imaging was used to determine the degree of crystallinity and disorder, surface structure, and local ordering in individual nanoparticles. The average Ce/Zr composition and composition profiles of individual nanoparticles were determined using EELS. To measure the composition profiles, the microscope was operated in scanning mode (STEM), and a subnanometer focused electron beam was stepped across the grains so an energy loss spectrum could be recorded every 0.5 or 1 nm with a dwell time of 1 s per spectrum. The Z-contrast STEM image was used to locate the grains and position the probe for the start of the scan. We used the Zr M45 edge at 180 eV and Ce M45 at 883 and 901 eV signals to determine the Ce/Zr concentration ratio. To convert the integrated signal intensities into concentration ratios, it is necessary to remove the background under the ionization edges and to know the ratio of the relevant partial ionization cross sections. The ionization cross section for the Ce M45 edge is not well-known, so instead we used an empirical approach to determine the appropriate cross section ratio.29,30 For thin samples in the absence of plural scattering,31 the Ce/Zr concentration ratio CCe/Zr can be determined from the integrated intensities ICe and IZr through the equation

CCe/Zr ) kICe/IZr

(1)

where k is a constant that depends on the cross section ratio. In this case, we can determine the constant k because we know the average composition of the sample. We evaluate the constant by calculating the average signal ratio for 24 particles and use eq 1 with CCe/Zr set to unity. The advantage of this approach is that it avoids the need to precisely know values of ionization cross sections and also makes a correction for systematic errors associated with the background correction. We estimate the error in our measurement of the Ce/Zr concentration to be ∼4%. Results Characterization of Calcined Samples. Figure 1 shows the (111) and (200) peaks in the X-ray diffraction pattern for the samples calcined under different conditions. The pattern for pure ceria is also displayed for reference. The X-ray lines are very broad at least in part because of the small crystallite size. The crystallite size determined from the line broadening is given in Table 1. The grain size for the pure ceria sample was much larger (25 nm) than the grain size for the solid solution samples (5-12 nm), confirming that the addition of Zr inhibits sintering.

18280 J. Phys. Chem. B, Vol. 110, No. 37, 2006

Wang et al.

TABLE 1: XRD and TGA Data for CeO2 and 50Ce50Zr Calcined in Air at Different Temperatures for (a) 2 h and (b) 5 h (a) 2 h sample conditions

(111) d spacing (Å)

crystal size (nm)

500 °C, 2 h 700 °C, 2 h 1000 °C, 2 h

3.0920 3.0810 3.0666

5.4 ( 0.2 6.2 ( 0.2 10.1 ( 0.2 (b) 5 h

sample conditions

(111) d spacing (Å)

crystal size (nm)

mass loss (%)

reduction fraction (%)

reduction temp (°C)

CeO2, 700 °C, 2 h 50Ce50Zr, 500 °C, 5 h 50Ce50Zr, 700 °C, 5 h 50Ce50Zr, 1000 °C, 5 h

3.1230 3.0813 3.0829 3.0741

25.2 ( 0.3 6.1 ( 0.2 5.8 ( 0.2 12.5 ( 0.2

1.880 2.682 2.502 1.539

40 99 92 57

757 486 506 731

For the solid solution samples, the grain size remains almost constant at approximately 6 nm for calcination temperatures up to 700 °C but increases to 12 nm for the samples calcined at 1000 °C. The XRD peaks were observed to shift to slightly higher angles with increasing calcination temperature for holding times of both 2 and 5 h. For example, for the 2 h sample, the (111) lattice spacing determined from XRD changed from 3.09 to 3.066 Å when the calcination temperature increased from 500 to 1000 °C. Moreover, the sample calcined at 1000 °C for 5 h showed evidence of the formation of an additional peak on the high-angle side of the (111) reflection. Figure 2 shows a typical high-resolution electron micrograph of an individual Ce0.5Zr0.5O2 nanoparticle (after the first TGA reduction cycle) and corresponding energy loss spectrum. The TEM images show that the particles are well crystallized and have somewhat irregular surfaces. The energy loss analysis showed the presence of Zr, O, and Ce inside each of the individual nanoparticles examined, confirming that a solid solution has been formed. Reducibility of Calcined Materials. Figure 3 shows TGA mass derivative profiles (-dM/dT) for pure ceria and the solid

Figure 1. XRD patterns of CeO2 and the 50Ce50Zr sample calcined at different temperatures and holding times: (a) 2 h and (b) 5 h. A small peak shift was observed with an increase of temperature.

solution samples (calcined for 5 h) recorded while the samples were heated in hydrogen to 1000 °C. The main peak is the predominant reduction peak, and its position is given in Table 1. For pure ceria, this peak is located at 760 °C with a minor peak at 400 °C possibly due to surface reduction. For the solid solution samples calcined at 500 and 700 °C, the main reduction peak is at about 500 °C, rising to 730 °C for the sample calcined at 1000 °C. The mass loss percentage is also displayed in Table 1. If we assume that all of the mass loss is due to loss of oxygen associated with reduction of Ce from the +4 to +3 oxidation state (equivalent to assuming a composition change from Ce0.5Zr0.5O2 to Ce0.5Zr0.5O1.75 for the solid solution), we can calculate the total percentage of Ce reduced during heating in hydrogen. The analysis shows that the addition of zirconia dramatically increases the percentage of Ce atoms reduced during heating in H2. For the samples calcined at 500 and 700 °C, over 90% of the Ce is reduced compared to only 40% in the pure ceria sample. Increasing the calcination temperature to 1000 °C causes the percentage of reduced Ce to drop to 57%. An irreversible change in color, from pale yellow to gray, was also observed after the hydrogen treatment. Redox Cycling. Figure 4 shows a series of differential TGA profiles for Ce0.5Zr0.5O2 over three consecutive redox cycles. The main reduction peak positions and fractions of Ce reduced in each case are listed in Table 2. The data show that the position of the main reduction peak drops with each redox cycle starting at 486 °C for the first cycle and finishing at 430 °C for the last cycle. The second cycle shows the presence of two peaks whose positions were determined using a standard Gaussian peak fitting routine to be at 448 and 510 °C. The fraction of Ce reduced during each of the redox cycles steadily falls from close to 100% for the first cycle to almost 40% for the last cycle. Figure 5 shows the XRD patterns of the material obtained after each reduction cycle (after cooling to room temperature and exposure to air). Table 2 also shows the effect of reduction cycles on the (111) d spacing and crystal size as determined by XRD. The particle size approximately doubles after the first redox cycle but changes much more slowly after subsequent cycles. Although the position of the (111) peak does not significantly change during the redox cycles, a small shoulder on its low-angle side indicated the formation of a different phase after heating at high temperature in hydrogen. High-spatial-resolution electron energy loss spectroscopy was used to carry out a detailed exploration of compositional heterogeneity at the nanometer level. Typical data are given in Figure 6, which shows a Z-contrast image and compositional line scan across two adjacent individual nanoparticles in the sample after the first redox cycle was completed. One grain has an average Ce/Zr composition of around 0.95, while the

Nanoscale Heterogeneity in Ceria Zirconia

J. Phys. Chem. B, Vol. 110, No. 37, 2006 18281

Figure 2. (a) Background-subtracted electron energy loss spectrum from one nanoparticle. (b) High-resolution image of an individual nanoparticle showing {111} fluorite fringes.

Figure 3. TGA characterizations: effect of the calcination temperature on the reduction temperature of 50Ce50Zr samples.

Figure 4. Effect of TGA cycles on the reduction temperature of the 50Ce50Zr sample calcined at 500 °C for 5 h.

other has a Ce/Zr ratio of about 0.75, showing that the second particle has a lower Ce content. While the average composition for all the grains analyzed was close to unity (by definition from our calibration method), EELS analysis showed substantial variation in the composition between individual nanocrystals of the sample. The standard deviation in the average composition from all the individual particles analyzed was about 0.16. We also found evidence for compositional heterogeneity within individual grains. Figure 7 shows a Z-contrast image and EELS line scan recorded from an individual nanoparticle. The Ce/Zr atomic ratio profile clearly shows a composition variation from the center to the edge of the particles. The line scan shows that the projected Ce/Zr concentration ranges from 0.8 ( 0.1 on the outer perimeter of the particles to 1.4 ( 0.1 in the core. This particle consists of a Ce-rich core surrounded by a Zr-rich shell. The signal coming from the center of the line scan measures the projected concentration through both the outer surface and the particle’s interior. If we assume that the Zrrich shell surrounds the entire particle, we can make a correction for this projection effect and obtain a more accurate estimate of the concentration in the center of the particle. In this case, the average corrected Ce/Zr concentration in the center is 1.8 ( 0.3. This core-shell composition variation was found in almost all particles although the degree of variation changed from particle to particle. The change in the Ce/Zr concentration ratio between the core and shell was less than 0.5, in the range 0.51.3 and greater than 1.3 for 40%, 30%, and 30% of the particles, respectively. There appeared to be no correlation between the degree of compositional heterogeneity observed in the particle and the average overall composition or size of the particle. In all cases, the center of the particle had the highest Ce concentration and the edge showed the highest Zr concentration.

High-resolution imaging of the samples subjected to hightemperature reduction also showed the presence of structural heterogeneity. Domains of a crystallographically distinct second phase were coherently embedded in the fluorite-like matrix as seen in the high-resolution image (Figure 8). The second-phase domain shows up as a superstructure with a lattice spacing of about 6.2 Å, which is parallel to and approximately twice the {111} fluorite spacing. By using a longer scan time, we were also able to detect this reflection at 14.6° in the XRD pattern (inset of Figure 5). Discussion Nanoparticle Formation and Compositional Heterogeneity. The XRD patterns show that a fluorite-like phase is formed after calcination in air. An unambiguous identification of the synthesized phase is difficult from the XRD patterns because of the very significant line broadening associated with the small particle size and compositional heterogeneity. We know from the ceria zirconia phase diagram that, close to this 50Ce50Zr composition, a cubic structure and two types of tetragonal phases (t′ and t′′) have been reported.32-35 The phase is usually detected from the splitting in the {200} and {400} lines in the XRD patterns. However, the tetragonal distortion is only 1% and is not detectable for the nanophase materials synthesized here because of the line broadening. Our simulations of the XRD patterns showed that it would be very difficult to differentiate the cubic and tetragonal forms for particle sizes of 10 nm or less. For most samples calcined for 2 and 5 h, the (111) peak shifts to slightly higher angle with increasing calcination temperature. For the sample calcined at 1000 °C for 5 h, the (111) peak also splits and a second peak at higher angle appears, indicating the formation of a second phase. Our EELS measurements show that there is a ceria-rich component in the central region of most nanocrystals. The TEM

18282 J. Phys. Chem. B, Vol. 110, No. 37, 2006

Wang et al.

TABLE 2: Effect of the TGA Cycle on the Redox Behavior of the 50Ce50Zr Sample

first TGA cycle second TGA cycle third TGA cycle

(111) d spacing (Å)

crystal size (nm)

mass loss (%)

reduction fraction (%)

reduction temp (°C)

3.0797 3.0790 3.0763

11.4 ( 0.3 12.7 ( 0.3 13.6 ( 0.3

2.682 1.963 1.182

99 72 43

486 448 (510, small peak) 431

analysis of the samples calcined at lower temperatures (not shown) exhibited considerable evidence of the presence of a highly disordered component which may contribute mostly background to the XRD profile. This suggests that the shift of the XRD peaks may be caused by variations in the degree of mixing between the ceria and zirconia phases in the larger wellcrystallized component of the solid solution. Applying Vegard’s law to the (111) line position suggests that the well-crystallized component is Ce-rich, in agreement with the EELS measurements (the cubic zirconia lattice is ∼6% smaller than the cubic ceria lattice). Monte and Kaspar36 pointed out that the presence of a nanodomain structure can lead to deviations from the wellknown “Vegard law”. However, this approach can still be used to give a qualitative indication of changes taking place within the material during calcination in air. Increasing the calcination temperature (at least for temperatures below 1000 °C) and holding time increases the degree of mixing of the ceria and zirconia phases in the well-crystallized component of the solid solution. However, calcination at 1000 °C results in the nucleation of a second distinct phase, and EELS analysis of this material confirms the presence of Zr-rich crystallites, indicating that the high-angle peak of the split (111) XRD line is coming from this Zr-rich phase. These TEM and XRD observations suggest that, for our samples, the nucleation and growth mechanism for the ceria zirconia nanoparticles initiates with the formation of a ceriarich seed crystal during calcination of the amorphous gel. This seed crystal initially grows preferentially from the ceria component of the gel. Eventually the concentration of the ceria component is reduced in the gel, and the growth of the final layers is completed with the remaining Zr-rich component. For temperatures up to 700 °C, increasing the calcination temperature and time seems to increase the degree of mixing within the solid solution. However, calcining at 1000 °C for 5 h changes the nucleation process and gives rise to the nucleation and growth of two distinct phases. Calcination in Air and Reducibility. The samples calcined in air at 500 and 700 °C showed the lowest reduction temperature and the highest reduction fractions. However, the

Figure 5. Comparison of XRD patterns from untreated material and after TGA cycles (redox cycle) for the 50Ce50Zr sample. TGA conditions: 1000 °C, 2.75 h, and 5% H2/95% He. Untreated materials were calcined in air at 500 °C for 5 h. Shown in the inset is the lowangle diffraction peak after the third TGA cycle for the sample showing a new superstructure phase.

sample calcined in air at 1000 °C showed redox properties that were similar to those of pure ceria in terms of both reduction temperature and reduction fraction. The sintering and compaction which takes place at 1000 °C may be expected to slow the reduction process. Similar sintering effects observed in the materials heated in hydrogen had no adverse effect on the reducibility of Ce, and consequently, sintering cannot explain the observed difference in redox properties. It appears that a fundamental change has taken place in the material calcined at high temperature, causing the change in the redox properties. Part of this change may be associated with the more pronounced phase separation as observed from XRD. However, we also know from our EELS measurements that compositional homogeneity is not a requirement for low-temperature redox activity. It is possible that the active or reducible component is an inherently heterogeneous phase and increasing the calcination temperature and time (similar to a traditional ceramic-processing oxidation treatment) results in a more fully oxidized and crystallographically less defective phase in which strain relaxation has occurred. This improved compositional uniformity and crystal perfection may render the material more stable and thus less easily reduced. This suggests that low-temperature calcina-

Figure 6. STEM image (a) and chemical profile (b) between two individual nanoparticles by an EELS line scan.

Nanoscale Heterogeneity in Ceria Zirconia

J. Phys. Chem. B, Vol. 110, No. 37, 2006 18283

Figure 8. Local superstructure (circled region) embedded in an individual nanoparticle.

Figure 7. STEM and EELS line scan profile from an individual nanoparticle: (a) STEM image, (b) Ce/Zr ratio. The direction of the EELS line scan is from the vacuum to the particle.

tions in air yield highly active material and high calcination temperatures should be avoided during synthesis. Effect of High-Temperature Reduction Treatment. Our TGA results clearly show that treatment in hydrogen at 1000 °C has a completely different effect on the reducibility of the material compared to treatment in air at 1000 °C. Both treatments result in a significant sintering of the particles, but the treatment in hydrogen caused the reduction temperature to decrease by up to 60 °C (by the third cycle), whereas the treatment in air caused the redox temperature to increase by almost 250 °C. Our nanospectroscopy results show that individual particles of the more active material are compositionally heterogeneous and consist of Ce-rich cores surrounded by Zrrich shells. Mamontov and Egami17,18 also investigated the nanoscale heterogeneities and oxygen storage capacity of active Ce0.5Zr0.5O2 by neutron diffraction. On the basis of pair distribution function analysis of neutron diffraction data, they found the structure of the nanocrystallites to consist of 25-30 Å domains of Ce0.4Zr0.6O2 in a matrix of Ce0.7Zr0.3O2. Our EELS measurements show similar average core and shell compositions within the particles. However, in our case, there are large variations between particles, the composition changes continuously within the particles, and the relationship between the compositional domains is different from that found by Mamontov and Egami. The TGA data show not only that the redox cycling in hydrogen decreases the onset of the reduction temperature but also that the reduction fraction is apparently decreased with each reduction cycle, dropping from nearly 100% to 43% after the third cycle. This suggests that most of the Ce is in some inactive

form during the third cycle. After the first hydrogen treatment, some of the Ce does not completely return to the +4 oxidation state on cooling to room temperature. (This will occur if, for example, some fraction of the sample is converted to pyrochlorerelated phases which contain Ce3+, which is stable at room temperature.22,23) During a subsequent reduction in hydrogen, a smaller weight change occurs because a significant fraction of Ce in the sample is already in a +3 oxidation state. This interpretation suggests that more than 50% of the Ce is already reduced at the beginning of the third TGA cycle and will essentially be inactive during subsequent hydrogen treatments. (A high-temperature oxidation treatment may convert this Ce back to the +4 state.) We do have experimental evidence to suggest that a substantial fraction of the sample subjected to high-temperature reduction remains changed after cooling and exposure to air. The most obvious evidence for this is that the samples change from yellow to gray after reduction and the color does not change upon cooling to room temperature in air. This color change is not caused by changes in particle size because we know that the samples treated in air and hydrogen both had particle sizes on the order of 10 nm. Some researchers37 have suggested the color evolution is due to the formation of oxygendeficient defects in the fluorite structure at high temperature and in a reducing atmosphere. It is possible for such defects or cations to order and give rise to the formation of a superstructure. We have observed the existence of superstructures both in electron microscopy and in XRD. These ordered phases may be either pyrochlore-related or fluorite-like phases consisting of ordered oxygen vacancies.22-23,38 The cation ordering that takes place in the pyrochlore phase would give rise to the observed doubling in the periodicity of the (111) diffraction peak. The formal oxidation state of Ce in this system is +3, so it would certainly be an inactive phase during subsequent reduction cycles. It is also possible that the superstructure may simply be associated with ordering of oxygen vacancies without requiring cation ordering. Such phases have been observed in pure ceria and they may also form in solid solutions.25 Ordering of oxygen vacancies may make the structure stable on exposure to air at room temperature. Further work is required to differentiate between these two superstructures. However, from the work presented here it seems clear that these superstructures may account, at least in part, for the decreased reduction fraction observed during the second and third redox cycles. High Redox Activity. The structure and composition of the most active cerium component is still not well defined. It is well established that an association with zirconia is somehow

18284 J. Phys. Chem. B, Vol. 110, No. 37, 2006 responsible for producing the active low-temperature phase. Our work and the work of others suggest that simply making a solid solution of ceria and zirconia will not give material with the highest activity. There is still considerable uncertainty about the possible detailed mechanism by which ZrO2 enhances the oxygen storage and release capacity for CeO2. Such mechanisms include facilitating oxygen vacancy formation,39 lattice distortion induced by cubic and tetragonal transformation,1 forming easily reducible intermediate phases,22-23,40 and local strain by lattice mismatch, etc.20 In our sample, the lattice spacing will decrease from the center to the edge of the particles as the Zr content increases. Thus, this compositional gradient that we observe may lead to the existence of differential strain fields across the nanoparticle which may have a destabilizing effect on the material and may give rise to the low-temperature redox properties. XRD of materials annealed in air at higher temperature suggests the formation of phases with more uniform Zr and Ce components. This improved uniformity within individual grains will reduce the strain gradient across the particle and may lead to greater stability and lower redox activity. This may be one reason calcinations in air at higher temperature may not yield the most active material. Enhancing low-temperature reducibility by subjecting the sample to high-temperature reduction treatments has received considerable attention recently.22-23,40 High-temperature reduction in hydrogen will introduce many oxygen vacancies and may also cause high degrees of localized lattice strain. We know that when the samples are cooled, many of the changes associated with the high-temperature reduction remain, leading to the formation of superstructure phases coherently embedded in the fluorite-like matrix. These structural domains will introduce additional strain especially in the interfacial regions and further destabilize the materials, leading to higher activity. Our nanoanalysis shows that both the compositional and structural heterogeneity leads to the existence of coherent associations between crystallographically related phases. However, the lattice spacings of these phases are not identical, and the coherent association will give rise to the creation of strain fields especially in the interfacial regions or in the zones where the composition is changing. Thus, the key to creating active materials may be to generate nanoscale heterogeneity on a scale that maximizes the strain in the fluorite-type lattice. We are currently performing nanoscale in situ measurements of the reducibility of individual nanoparticles and density functional calculations to further explore the relationship among reducibility, strain, and nanoscale heterogeneity. Conclusions We have used a spray freezing method to synthesize highsurface-area Ce0.5Zr0.5O2, which shows low-temperature redox activity during both the initial and subsequent redox cycles. XRD and nanospectroscopy of samples calcined over a range of different temperatures suggest that, during the early stage of particle formation, the crystallization process is initiated via the nucleation of ceria-rich nanodomains. This results in the formation of a high-surface-area material that exhibits nanoscale heterogeneity consisting of Ce-rich cores surrounded by Zrrich shells. With this synthesis approach, the optimum temperature for calcination of the amorphous gel is in the range of 500-700 °C to yield material that exhibits an initial reduction temperature of about 500 °C. Subsequent high-temperature reduction cycles in hydrogen lower the reduction temperature to about 430 °C, although the

Wang et al. fraction of reducible Ce drops by a factor of 2. Nanospectroscopy with EELS showed that these highly active materials still exhibited significant composition variation both within and between individual nanoparticles. Most of the nanoparticles have Ce-rich cores with an average composition of Ce0.65Zr0.35O2 and Zr-rich shells with an average composition of Ce0.45Zr0.55O2. After the initial reduction we observed the presence of a superstructure component in which the (111) fluorite spacing was doubled. This phase corresponds to a pyrochlore-type structure or an ordered oxygen vacancy phase, and it appears to coexist with a fluorite-like structure within individual nanoparticles. Both the compositional and structural heterogeneity may give rise to the creation of differential strain fields especially in the interfacial regions, resulting in low-temperature redox activity. Acknowledgment. We acknowledge the financial support from the NSF through Grant NSF-CTS-0306688 and the use of the TEM facilities at the John M. Cowley Center for High Resolution Microscopy at Arizona State University. We thank Dr. Z. C. Kang for valuable and stimulating discussions and Dr. Mike McKelvy and Tim Karcher for assistance with the TGA experiment. References and Notes (1) Fornasiero, P.; Balducci, G.; DiMonte, R.; Kaspar, J.; Sergo, V.; Gubitosa, G.; Ferrero, A.; Graziani, M. J. Catal. 1996, 164, 173. (2) Jen, H. W.; Graham, G. W.; Chun, W.; McCabe, R. W.; Cuif, J. P.; Deutsch, S. E.; Touret, O. Catal. Today 1999, 50, 309. (3) Shelef, M.; McCabe R. W. Catal. Today 2000, 62, 35. (4) Shelef, M.; Graham, G. W.; McCabe, R. W. In Catalysis by Ceria and Related Materials; Trovarelli, A., Ed.; Imperial College Press: London, 2002. (5) Marecot, P.; Pirault, L.; Mabilon, G.; Prigent, M.; Barbier, J. Appl. Catal., B 1994, 5, 57. (6) Trovarelli, A.; Zamar, F.; Liorca, J.; Deleitenburg, C.; Dolcetti, G.; Kiss, J. T. J. Catal. 1997, 169, 490. (7) Bunluesin, T. Appl. Catal., B 1998, 15, 107. (8) Wang, X. Q.; Rodriguez, J. A.; Hanson, J. C.; Gamarra, D.; Martinez-Arias, A.; Fernandez-Garcia, M. J. Phys. Chem. B 2006, 110, 428. (9) Park, S. D.; Vohs, J. M.; Gorte, R. J. Nature 2000, 404, 265. (10) Sugiura, M. Catal. SurV. Asia 2003, 7, 77. (11) Monte, R. D.; Kaspar, J. J. Mater. Chem. 2005, 15, 633. (12) Kaspar, J.; Fornasiero, P.; Balducci, G.; Monte, R. D.; Hickey, N.; Sergo, V. Inorg. Chim. Acta 2003, 349, 217. (13) Hirano, M.; Suda, A. J. Am. Ceram. Soc. 2003, 86, 2209 (14) Liotta, L. F.; Macaluso, A.; Pantaleo, G.; Longo, A.; Martorana, A.; Deganello, G.; Marci, G.; Gialanella, S. J. Sol-Gel Sci. Technol. 2003, 26, 235. (15) Kozlov, A. I.; Kim, D. H.; Yezerets, A.; Andersen, P.; Kung, H. H.; Kung, M. C. J. Catal. 2002, 209, 417. (16) Vidal, H.; Kaspar, J.; Pijolat, M.; Colon, G.; Bernal, S.; Cordo´n, A.; Perrichon, V.; Fally, F. Appl. Catal., B 2000, 27, 49. (17) Mamontov, E.; Brezny, R.; Koranne, M.; Egami, T. J. Phys. Chem. B 2003, 107, 13007. (18) Mamontov, E.; Egami, T.; Brezny, R.; Koranne, M.; Tyagi, S. J. Phys. Chem. B 2000, 104, 11110. (19) Punta, E. S.; Bunluesin, T.; Fan, X. L.; Gorte, R. J.; Vohs, J. M.; Lakis, R. E.; Egami, T. Catal. Today 1999, 50, 343. (20) Costa-Nunes, O.; Gorte, R. G.; Vohs, J. M. J. Mater. Chem. 2005, 15, 1520. (21) Costa-Nunes, O.; Ferrizz, R. M.; Gorte, R. J.; Vohs, J. M. Surf. Sci. 2005, 592, 8. (22) Montini, T.; Hickey, N.; Fornasiero, P.; Graziani, M.; Ban˜ares, M. A.; Martinez-Huerta, M. V.; Alessandri, I.; Depero, L. E. Chem. Mater. 2005, 17, 1157. (23) Kang, Z. C. J. Alloys Compd. 2006, 408-412, 1103. (24) Aneggi, E.; Boaro, M.; Leitenburg, C. de; Dolcetti, G.; Trovarelli, A. J. Alloys Compd. 2006, 408-412, 1096. (25) Sharma, R.; Crozier, P. A.; Kang, Z. C.; Eyring, L. Philos. Mag. 2004, 84, 2731. (26) Pe´rez-Omil, J. A.; Bernal, S.; Calvino, J. J.; Herna´ndez, J. C.; Mira, C.; Rodrı´guez-Luque, M. P.; Erni, R.; Browning, N. D. Chem. Mater. 2005, 17, 4282.

Nanoscale Heterogeneity in Ceria Zirconia (27) Li, J. G.; Ikegami, T.; Wang, Y. R.; Mori, T. J. Solid State Chem. 2002, 168, 52. (28) Suryanarayana, C. X-ray Diffraction: A Practical Approach; Plenum Press: New York, 1998. (29) Batson, P. E. Nature 1993, 366, 727. (30) Browning, N. D.; Pennycook, S. J. J. Microsc. 1995, 180, 230. (31) Egerton, R. F. Electron Energy-Loss Spectroscopy in the Electron Microscope; Plenum Press: New York, 1996. (32) Yashima, M.; Sasaki, S.; Yamaguchi, Y.; Kakihana, M.; Yoshimura, M.; Mori, T. Appl. Phys. Lett. 1998, 72, 182. (33) Yashima, M.; Arashi, H.; Kakihana, M.; Yoshimura, M. J. Am. Ceram. Soc. 1994, 77, 1067.

J. Phys. Chem. B, Vol. 110, No. 37, 2006 18285 (34) Yashima, M.; Morimoto, K.; Ishizawa, N.; Yoshimura, M. J. Am. Ceram. Soc. 1993, 76, 2865. (35) Yashima, M.; Morimoto, K.; Ishizawa, N.; Yoshimura, M. J. Am. Ceram. Soc. 1993, 76, 1745. (36) Monte, R. D.; Kaspar, J. J. Mater. Chem. 2005, 15, 633. (37) Trovarelli, A. In Catalysis by Ceria and Related Materials; Trovarelli, A., Ed.; Imperial College Press: London, 2002. (38) Kang, Z. C.; Eyring, L. J. Solid State Chem. 2000, 155, 129. (39) Esch, F.; Fabris, S.; Zhou, L.; Montini, T.; Africh, C.; Fornasiero, P.; Comelli, G.; Rosei, R. Science 2005, 309, 752. (40) Sasaki, T.; Ukyo, Y.; Suda, A.; Sugiura, M.; Kuroda, K.; Arai S.; Saka, H. J. Ceram. Soc. Jpn. 2002, 111, 382.