Вы находитесь на странице: 1из 16

Applied Surface Science 210 (2003) 206–221

Surface chemistry and microstructural analysis of


CexZr1xO2y model catalyst surfaces$
Alan E. Nelsona, Kirk H. Schulzb,*
a
Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alta., Canada T6G 2G6
b
Dave C. Swalm School of Chemical Engineering, Mississippi State University, P.O. Box 9595, Mississippi State, MS 39762, USA
Received 30 May 2002; received in revised form 30 May 2002; accepted 27 January 2003

Abstract

Cerium-zirconium mixed metal oxides are widely used as promoters in automotive emissions control catalyst systems (three-
way catalysts). The addition of zirconium in the cubic lattice of ceria improves the redox properties and the thermal stability,
thereby increasing the catalyst efficiency and longevity. The surface composition and availability of surface oxygen of model
ceria-zirconia catalyst promoters was considered to develop a reference for future catalytic reactivity studies. The microstructure
was characterized with X-ray diffraction (XRD) to determine the effect of zirconium substitution on crystalline structure and
grain size. Additionally, the Ce/Zr surface atomic ratio and existence of Ce3þ defect sites were examined with X-ray
photoelectron spectroscopy (XPS) and Auger electron spectroscopy (AES) for samples with different zirconium concentrations.
The surface composition of the model systems with respect to cerium and zirconium concentration is representative of the bulk,
indicating no appreciable surface species segregation during model catalyst preparation or exposure to ultrahigh vacuum
conditions and analysis techniques. Additionally, the concentration of Ce3þ defect sites was constant and independent of
composition. The quantity of surface oxygen was unaffected by electron bombardment or prolonged exposure to ultrahigh
vacuum conditions. Additionally, XRD analysis did not indicate the presence of additional crystalline phases beyond the cubic
structure for compositions from 100 to 25 at.% cerium, although additional phases may be present in undetectable quantities.
This analysis is an important initial step for determining surface reactions and pathways for the development of efficient and
sulfur-tolerant automotive emissions control catalysts.
# 2003 Elsevier Science B.V. All rights reserved.

Keywords: Cerium; Zirconium; Oxide; Model; Catalyst; Emissions; Automotive

1. Introduction of oxygen near the catalyst surface [1,2]. Despite its


widespread use and application, pure cerium dioxide
Cerium oxide (CeO2) is used in automotive emis- has poor thermal stability and is known to sinter at
sions control catalysts to regulate the partial pressure 1123 K [3]. In order to increase its thermal stability
and ability to store and release oxygen during opera-
$
A portion of this work was performed at Department of tion, zirconium is substituted into the cubic structure
Chemical Engineering, Michigan Technological University, 1400 of ceria. The addition of zirconium to the cubic
Townsend Drive, Houghton, MI 49931, USA.
*
Corresponding author. Tel.: þ1-662-325-2480;
structure of ceria is reported [4–6] to increase the
fax: þ1-662-325-2482. oxygen storage capacity of the system while enhan-
E-mail address: schulz@che.msstate.edu (K.H. Schulz). cing the thermal stability under high temperatures [7],

0169-4332/03/$ – see front matter # 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0169-4332(03)00157-0
A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221 207

as compared to pure ceria. Indeed, a 10 at.% zirconium surface area mixed oxide powders suitable for auto-
substitution in cerium oxide markedly increases the motive catalyst emulation, including hydroxide, acetate,
oxygen storage capacity compared to ceria-only sys- and surfactant-assisted chloride precipitation [8,10]. In
tems [3]. Ceria-zirconia solid solutions are also reported this study, a complete range metastable cerium-zirco-
to have three to five times the oxygen storage capacity nium mixed metal oxide powders (CexZr1xO2y,
(per gram of catalyst) than ceria-only systems [8]. 1  x  0) were prepared through a hydroxide
Bulk reduction of pure ceria is reported to occur at precipitation technique reported by Hori et al. [8].
approximately 1173 K, while ceria-zirconia reduction Predetermined quantities of cerium(IV) ammonium
begins at lower temperatures near 853 K. Vliac et al. nitrate (Alfa Aesar, CAS #16774-21-3) and zirconium
[9] suggest the enhanced oxygen storage property of oxynitrate (Alfa Aesar, CAS #13826-66-9) precursors
the ceria-zirconia system is the result of enhanced are completely dissolved in deionized water with mild
diffusion of bulk oxygen anions to the surface. heat. Upon dissolution of the precursors, excess ammo-
Additionally, Raman spectroscopy, X-ray diffraction nium hydroxide (100 vol.%) is added to precipitate
(XRD), and EXAFS analysis indicates zirconium has the cerium-zirconium mixed metal oxide powder.
a lower coordination number within the ceria lattice, The resultant precipitate is filtered in a vacuum funnel
and consequently increases bulk oxygen mobility [9]. and thoroughly washed with excess distilled water
The near-neighbor zirconium oxygen atoms are not (2 l/10 g precipitate). The ceria-zirconia powder is
observed in ceria-zirconia cubic matrices and their allowed to dry in the hood overnight and is subse-
absence is attributed to a high degree of structural quently annealed at atmospheric conditions. Sample
disorder. The oxygen atoms are positioned at a weak annealing stabilizes the metal oxides and eliminates
bonding distance, resulting with increased oxygen carbonates and nitrate compounds remaining from the
mobility. This is thought to be a mechanism for release preparation technique [7]. The annealed powders are
of the stress generated by the insertion of a smaller milled and stored in a dry environment. A sufficient
zirconium ion into the lattice [9]. By moving the quantity of ceria-zirconia powder was initially pre-
oxygen ions to a non-bonding distance, mobile oxygen pared and utilized through this research to eliminate
anions are generated within the lattice. compositional inconsistencies that arise from multiple
In order to improve ceria-zirconia low-temperature preparations. The cerium-zirconium mixed oxides
performance and resistance to deactivation (SO2), a were formed into model wafers using a standard
fundamental understanding of the structure and sur- 13 mm diameter FT-IR pellet die and hydraulic press.
face chemistry is required. This paper presents infor- The oxide powders (approximately 100 mg) are loaded
mation on the characterization of cerium-zirconium into the die and subjected to a pressure of 670 MPa for
mixed metal oxide powders and model catalysts pre- a duration of 10 min. The prepared samples have a
pared via co-precipitation routines. The surface seg- thickness of approximately 100–150 mm, depending on
regation of cerium and zirconium, as well as the the initial quantity of oxide power used.
oxidation state of the metals, was investigated with
Auger electron spectroscopy (AES) and X-ray photo- 2.2. Equipment and analysis
electron spectroscopy (XPS). This information will be
useful for comparing reaction and deactivation Surface compositional analysis was performed with
mechanisms under oxidizing and reducing conditions AES and XPS. Auger electron spectroscopic analysis
in subsequent investigations. was performed with a Physical Electronics 545
Scanning Auger Microprobe (SAM). The Auger spec-
trometer was operated with an electron gun filament
2. Experimental (tungsten) current of 2.0 nA and an incident electron
energy of 2.0 keV with a corresponding resolution of
2.1. Materials and synthesis 0.6%. The spectrometer is operated at a background
pressure of approximately 2  1010 Torr obtained
Several cerium-zirconium solid solution preparation with a 220 l/s ion-pump and titanium sublimation
methods have been reported which produce medium pump. Sample positioning was performed with a
208 A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221

xyz-rotary ultrahigh vacuum sample manipulator and a samples was prepared from annealed oxide powder,
custom UHV terminal [11]. XPS was performed with which was formed into the model catalyst systems.
a Physical Electronics 1600 XPS Surface Analysis The SEM micrographs of each preparation method
System located in a dedicated stainless steel chamber are located in Fig. 1. The images clearly indicate the
used exclusively for XPS surface characterization. samples prepared with annealed oxide powder directly
The analyzer is fitted with an Omni Focus III formed into the model catalysts have less surface
small-area lens that produces 800 um diameter surface roughness and irregularity (Fig. 1b), compared to
analysis area. The XPS spectra were obtained by using samples prepared with dry mixed oxide powder fol-
an incident achromatic Mg Ka X-ray source lowed by wafer formation and subsequent annealing
(1253.6 eV) operated at 300 W with a corresponding (Fig. 1a). As a result of the SEM analysis, the pre-
voltage of 15 kV. Survey spectra are an average of 10 paration method of annealing the oxide powders at
scans over a range from 0 to 1100 eV with a pass 773 K for 1 h followed by subsequent milling and two-
energy of 26.95 eV. The high-resolution XPS spectra dimensional wafer formation was adopted. Addition-
are a composite average of 15 scans with a pass ally, the apparent bulk density of the model catalysts
energy of 23.5 eV at an incident sample angle of was characterized with a modified Archimedes density
458. The background pressure during XPS analysis measurement method to ensure adequate density and
was approximately 8  109 Torr. Additionally, the validate the model catalyst systems further (Table 1).
microstructure of the ceria-zirconia model catalysts The measured density of CeO2 model systems is 95%
was characterized with XRD and scanning electron of theoretical compared to a ZrO2 density of 85% of
microscopy (SEM). theoretical. The global average for all prepared and
analyzed samples is 89% of theoretical. Considering
alumina with a bulk density of 97% of theoretical is
3. Results and discussion sufficient for ultrahigh vacuum applications [13], the
densities of the model systems are adequate to limit
3.1. Effects of catalyst annealing the extent of diffusional processes which arise in
desorption spectrocscopies.
Several methods of model catalyst preparation were
considered to produce a model catalyst suitable for 3.2. Crystalline structure and grain size
ultrahigh vacuum spectroscopic characterization,
while maintaining attributes similar to automotive The phase diagram for intermediate cerium-zirco-
catalyst systems. Specifically, the effect of annealing nium oxides is the subject of debate due to reported
conditions on the final properties of the two-dimen- phases of metastable tetragonal crystallinity [14,15].
sional model catalysts was investigated. The ceria- Indeed, several publications have indicated the
zirconia samples require annealing to eliminate car- presence of three tetragonal phases, including t
bonates and nitrates remaining from the preparation (20–40 at.% cerium), t0 (40–65 at.% cerium), and t00
technique [8]. However, excessive annealing could (65–80 at.% cerium) over the intermediate composi-
result in particle sintering and undesirable crystallite tional range [5,16–18]. However, the t00 phase exhibits
migration [12]. Published research of ceria-zirconia no tetragonality and the t0 phase is a metastable structure
suggests particle sintering occurs at temperatures in formed through a diffusionless transition [18]. Hori
excess of 773 K at prolonged exposure times. As a et al. [8] reported cerium-zirconium mixed metal oxides
result, annealing conditions of 1 h at 773 K in atmo- in excess of 50 at.% cerium prepared via similar
sphere were used throughout the experiments. Two co-precipitation routines and aged at 1273 K exist as
distinct preparation routines were investigated to dis- a cubic (fluorite) solid solution. They also report the
cern the effects of annealing on the final two-dimen- presence of a separate tetragonal phase (zirconium rich)
sional model catalyst properties. The first set of model for compositions less than 49 at.% cerium.
catalysts was prepared from the dry cerium-zirconium The XRD patterns of the prepared cerium-zirco-
mixed oxide powder followed by wafer formation nium oxides are shown in Fig. 2 for comparison. The
and subsequent sample annealing. The second set of reference pattern (PDF 34-0394) for a standard cubic
A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221 209

Fig. 1. SEM micrographs of prepared CeO2 model catalyst surfaces. The micrographs represent two evaluated preparation methods, including
preparation with dry mixed oxide powder followed by wafer formation and subsequent calcination (a), and preparation with calcined oxide
powder directly formed into the model catalysts (b). The samples were imaged at 1000 magnification.

CeO2 sample, shown as vertical dotted lines, is super- cubic solid solution is still maintained up to 25 at.% of
imposed on the XRD patterns. The pattern of the cerium oxide (75 at.% zirconium oxide). There is no
100 at.% CeO2 sample matches with the reference indication from XRD analysis of a separate tetragonal
data, verifying the cubic crystalline structure (CaF2) ZrO2 phase under any composition. In previous studies,
for the sample. As the amount of CeO2 in the mixed mixed oxides prepared via the same route could form a
oxide decreases, the positions of the peaks of spectra solid cubic solution up to only 50 at.% cerium [8].
shift farther away from those of pure CeO2 and the These results could be attributed to the zirconium oxide
peaks also broaden (FWHM) as a function of decreased precursor used in the preparation of the mixed oxides
cerium loading. Duwez and Odell [14] and other and the preparation method itself, as the crystalline
researchers have also reported similar results. Another structure of the final mixed oxide is highly sensitive to
important observation from the XRD patterns is that a these two factors. The pure zirconium oxide pattern is

Table 1
Bulk density and XRD

Bulk density X-ray diffraction (XRD)

Theoretical Archimedes Cubic lattice Scherrer grain


(g cm3) (g cm3)  0.08 parameter (Å)  0.005 size (Å)  0.05

Ce1.0Zr0.0O2y 7.13 6.75 5.413 109.2


Ce0.9Zr0.1O2y 7.00 6.36 5.390 70.62
Ce0.8Zr0.2O2y 6.86 6.15 5.371 61.28
Ce0.7Zr0.3O2y 6.71 5.89 5.353 50.89
Ce0.6Zr0.4O2y 6.57 5.80 5.328 51.32
Ce0.5Zr0.5O2y 6.42 5.68 5.301 46.82
Ce0.25Zr0.75O2y 6.02 5.23 5.219 58.70
Ce0.0Zr1.0O2y 5.60 4.83 Monoclinic 148.8
210 A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221

Fig. 2. XRD patterns of cerium-zirconium oxides. The reference pattern (PDF 34-0394) for a standard CeO2 cubic structure is superimposed
on the XRD pattern as vertical lines. The data indicates a cubic solid solution from 100 to 25 at.% cerium and a monoclinic structure for pure
zirconia.

identified as the monoclinic ZrO2 spectra. This is in structure over the intermediate composition range
agreement with the reported phase diagrams [14,15] (25–100 at.% cerium), we acknowledge the limitations
for a cerium-zirconium mixed oxide system at ambient of XRD and the possibility of intermediate tetragonal
conditions. The shifted peak position as a function of phases at high zirconium concentrations.
composition indicates a change in the cubic lattice The lattice parameters and grain sizes were calcu-
parameter of the system [8,19]. As a result, X-ray lated [19] from the diffraction patterns as a function of
powder diffraction characterization of the fresh oxide sample composition (Table 1). The data indicates the
powders indicates a cubic solid solution from 100 to lattice parameter, a0, follows a downward trend with
25 at.% cerium and a monoclinic structure for pure increasing amount of zirconium incorporated into
zirconia. While our XRD analysis indicates a cubic the mixed oxides. This agrees well with Vegard’s
A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221 211

law, which states that the lattice parameter of a solid quantification of the oxygen KLL transition ampli-
solution is directly proportional to the atomic percent tude, it is evident that surface oxygen remains constant
solute present [19]. Zirconium is reported to have an for excess exposures to ultrahigh vacuum conditions.
atomic radius ratio (rcation/ranion) of 0.59 compared to Additionally, no appreciable reduction is surface
0.68 for cerium [20]. This implies that zirconium has a oxygen content occurs as a result of electron bombard-
smaller ionic radius and would result in a reduction ment. If any surface oxygen is removed as a result of
of the lattice parameter of the crystalline system as beam bombardment or ultrahigh vacuum exposure, it
Ce4þ ions are being substituted by Zr4þ ions. This occurs during initial degassing or sample positioning.
observation is consistent with the calculated cubic The zirconium MNN transition amplitudes for each
lattice parameters. The lattice parameter of pure cer- composition were numerically averaged (n ¼ 7) and
ium oxide, CeO2, is found to be 5.413 Å, a value which graphed as a function of bulk composition (Fig. 3).
is comparable to that reported in a previous study [14]. Qualitative analysis of the zirconium peak amplitudes
The data also indicates the grain size decreases gra- indicates a linear correlation with respect to bulk
dually with increasing amounts of zirconium. composition (R2 ¼ 0:9924). This regressed slope
value indicates that cerium is more sensitive to the
3.3. AES analysis of surface composition Auger electron process than zirconium for the com-
positions considered and analyzer used. This observa-
The chemical composition of the prepared model tion is in agreement with published sensitivity factors
catalysts was analyzed with AES to characterize sur- for Auger spectroscopic analysis [22]. The Auger
face species segregation compared with the bulk analysis of the mixed oxides indicates the surface
composition [21]. Initially, the effects of ultrahigh composition is representative of the bulk regardless
vacuum conditions and electron beam bombardment of cerium-zirconium concentrations. In addition to
were examined to determine the potential impacts on zirconium MNN peak intensity correlation, additional
surface oxygen and surface reduction. A 100 at.% numerical analysis was performed using the oxygen
cerium oxide sample was degassed under high vacuum KLL transition. Similar to the zirconium peak inten-
conditions for 1 h and subsequently transferred to the sity analysis, the amplitude of the oxygen KLL
ultrahigh vacuum chamber for Auger spectroscopic (513 eV) transition was quantified and correlated to
characterization. Following the collection of an initial bulk mixed oxide composition (Fig. 4). The oxygen
Auger spectrum (1 h), the sample remained is position KLL peak intensity analysis indicates a non-linear
for an additional 3 h, as to bombard the surface with relationship with composition, decreasing with increas-
the incident electron source. The sample was again ing zirconium concentration. This decreasing trend
analyzed at intervals of 28 and 96 h of ultrahigh in surface oxygen concentration with increasing zirco-
vacuum exposure. The specimen was not exposed nium substitution is suggested to result from surface
to additional electron bombardment except for sample contamination, as opposed to surface oxidation state
alignment and data collection. The collected Auger and oxygen diffusion.
spectra were normalized with respect to the cerium
MNN transition and the amplitude of the oxygen KLL 3.4. XPS analysis of surface composition
transition was determined (Table 2). Based on the
The catalyst systems were further analyzed with
XPS to verify surface composition and elemental
Table 2
oxidation states. Due to observed charging effects
Effect of ultrahigh vacuum on ceria surface oxygen
during XPS analysis, the binding energy scale was
Time (h) Oxygen KLL (513 eV) transition amplitude calibrated using adventitious carbon (285.4 eV) [23].
1 5570  50 This resulted in a binding energy correction of 6 to
4a 5500  50 10 eV for each spectrum. The spectral features were
28 5520  50 fitted with Gaussian distributions and the peak posi-
96 5530  50 tions and areas were determined (Table 3). The Ce 3d
a
Sample was bombarded with incident electron beam for 3 h. Gaussian peak fits corresponding to Ce3þ and Ce4þ
212 A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221

Fig. 3. Zirconium AES peak intensity analysis. The zirconium MNN transition amplitudes for each composition were numerically averaged
over seven data points. Linear regression analysis (zero constant) performed with the zirconium peak amplitude as a function of composition
produces a correlation with an R2 value of 0.9924.

states are based on published CeO2 XPS analysis by satellite features. Specifically, the band (u0, u) located
Burroughs et al. [24] and Pfau and Schierbaum [25]. at (901.0–901.4 eV) is the Ce 3d3/2 ionization and
However, due to the complex electronic structure, the band (v0, v) located at (882.5–882.8 eV) is the Ce
the absolute distinction between 3d94f2Vn1 and 3d5/2 ionization for Ce3þ and Ce4þ [24–26]. The bands
3d94f2Vn2 final states for Ce3þ (v0, u0) and Ce4þ labeled v0 (885.5–885.8 eV), v00 (889.0–889.3 eV) and
(v, u) could not be resolved. As a result, the high- v000 (898.3–898.6 eV) are satellites arising from the Ce
resolution spectra for the Ce 3d3/2 and Ce 3d5/2 ioni- 3d5/2 ionization, while bands u0 (904.0–904.3 eV), u00
zation features were numerically fitted with eight (907.4–907.7 eV), u000 (916.7–916.9 eV) are satellites
Gaussian distributions representing the initial and final arising from the Ce 3d3/2 ionization [24,25]. The Ce
states in Ce 3d core level X-ray photoelectron spectra 3d3/2 and Ce 3d5/2 peak areas and amplitudes increased
(Fig. 5). A composite spectrum of the Ce 3d3/2 and Ce as a function of increasing cerium concentration.
3d5/2 ionization features as a function of composition This is attributed to increasing cerium concentration,
is shown in Fig. 6. as the amplitude and area of XPS photoemission
The bands labeled v collectively represent the Ce features are proportional to surface composition
3d5/2 ionization, while bands labeled u represent the [27]. The difference in Ce 3d3/2 and Ce 3d5/2 binding
Ce 3d3/2 ionization. The bands with unprimed labels energies is also in agreement with an expected value
represent the primary Ce 3d5/2 and Ce 3d3/2 transi- of 18.6 eV [24]. Examination of the Ce 3d3/2 and
tions, while the primed labels represent ionization Ce 3d5/2 photoemission features indicates a slight
A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221 213

Fig. 4. Oxygen AES peak intensity analysis. The amplitude of the oxygen KLL (513 eV) transition was averaged (n ¼ 7) for each
composition and correlated to bulk mixed oxide composition. The oxygen KLL peak intensity analysis indicates a non-linear relationship with
composition, decreasing with increasing zirconium concentration.

chemical shift to a higher binding energy (0.3–0.4 eV) surface electronic structure independent of oxidation
from 100 to 25 at.% cerium. Although a slight binding state.
energy shift is observed with changing composition, To verify the surface oxidation state as a function
the shift is suggested to arise from the changing of composition, the bands representative of Ce3þ and

Table 3
Summary of principle XPS binding energies

Photoemission binding energy (eV)

Ce 3d5/2 Ce 3d3/2 Zr 3d5/2 Zr 3d3/2 O 1s C 1s

Ce1.0Zr0.0O2y 882.5 901.0 – – 529.6 285.4


Ce0.9Zr0.1O2y 882.6 901.0 181.5 184.0 529.6 285.4
Ce0.8Zr0.2O2y 882.7 901.2 182.4 184.9 529.8 285.4
Ce0.7Zr0.3O2y 882.6 901.1 182.4 184.8 529.7 285.4
Ce0.6Zr0.4O2y 882.7 901.3 182.4 184.8 529.9 285.4
Ce0.5Zr0.5O2y 882.7 901.2 182.3 184.8 529.9 285.4
Ce0.25Zr0.75O2y 882.8 901.4 182.4 184.8 530.0 285.4
Ce0.0Zr1.0O2y – – 182.4 184.9 530.1 285.4
214 A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221

Fig. 5. Ce 3d photoemission Gaussian peak fit for CeO2. The analyzed region of cerium illustrates the complex satellite structure arising from
a mixture of multielectron and multiplet interactions. The bands labeled v collectively represent the Ce 3d5/2 ionization, while bands labeled u
represent the Ce 3d3/2 ionization. Additionally, the bands with unprimed labels represent the primary Ce 3d5/2 and Ce 3d3/2 transitions, while
the primed labels represent ionization satellite features.

Ce4þ ions were further studied. The bands labeled u0 XPS spectra [25], the intensities of the Ce4þ emissions
and v0 represent the 3d104f1 initial electronic state are consistent with surface oxygen stoichometry of
corresponding to Ce3þ, while the peaks labeled u000 (y ¼ 0:00:08) for CexZr1xO2y.
and v000 represent the 3d104f0 state of Ce4þ ions [25]. The narrow scan spectra of the Zr 3d3/2 and Zr 3d5/2
Analysis of the peak areas associated with the repre- features were deconvoluted with two Gaussian dis-
sentative Ce3þ (u0 , v0 ) and Ce4þ (u000 , v000 ) features tributions representing the primary Zr 3d5/2 and Zr 3d3/
indicates a constant concentration of Ce3þ defect sites 2 features. A composite spectrum of the Zr 3d3/2 and
on the surface independent of zirconium substitution. Zr 3d3/2 ionization features as a function of composi-
Because the surface composition is representative of tion is shown in Fig. 7. The ionization features are
the bulk, this suggests a constant cerium surface distinct and uninfluenced by satellite features. The
oxidation state independent of composition. Addition- Zr 3d5/2 ionization feature increases from 181.5 to
ally, the intensities of the Ce3þ emissions (u0 , v0 ) are 182.4 eV, while the Zr 3d3/2 ionization feature
relatively small compared to Ce4þ emissions (u000 , v000 ), increases from 184.0 to 184.9 eV as zirconium con-
which are in relative proportion to highly oxidized centration increases. However, unlike the gradual
cerium oxide. Considering previously published ceria increase in binding energy observed with cerium
A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221 215

Fig. 6. Ce 3d High resolution XPS analysis. The Ce 3d photoemission features are represented as a function of composition. The analysis
indicates a slight shift to higher cerium oxidation state as zirconium substitution increases, as evidenced by a chemical shift of the Ce 3d3/2 and
Ce 3d5/2 bands to a higher binding energy (0.3–0.4 eV) from 100 to 25 at.% cerium. Additionally, the Ce 3d3/2 and Ce 3d5/2 peak areas and
amplitudes linearly increase as a function of increasing cerium concentration.

ionization features, the zirconium binding energy independent of composition. This value is in excellent
shifts markedly (0.9 eV) between 90 and 80 at.% agreement with the reported values of 2.43 [28] and
cerium, followed by a region of constant binding 2.4 eV [23]. The constant binding energy separation
energy independent of composition. In addition to between the Zr 3d3/2 and Zr 3d5/2 features suggests a
the noted photoemission features, additional bands constant zirconium oxidation state independent of
near 174 eV are suggested to correspond to satellite composition. Analysis of the Zr 3d5/2/Zr 3d3/2 peak
features resulting from the use of an achromatic amplitudes and areas is consistent with expected
X-ray source. The average delta between the Zr 3d3/2 results. The peak amplitude ratio average is 1.47
and Zr 3d5/2 photoemission features is 2.4 eV and (0.1):1, which is in good agreement with the 3:2
216 A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221

Fig. 7. Zr 3d high resolution XPS analysis. The spectra for Zr 3d photoemission includes two distinct features corresponding to Zr 3d3/2 and
Zr 3d5/2 ionization bands. The analysis indicates an initial shift to a higher zirconium binding energy between 90 and 80 at.% cerium, followed
by a region of constant binding energy.

ratio expected on the basis of J multiplicities [24,27]. of the O 1s photoemission feature results in the
The peak areas and amplitudes increase as a function deconvolution of three O 1s ionization binding ener-
of increasing zirconium (decreasing cerium) concen- gies, as supported by independent analysis of zirco-
tration. nium oxide [23]. The O 1s Gaussian peak fit analysis
The high-resolution spectrum for the O 1s ioniza- for pure ZrO2 is shown in Fig. 8, and the O 1s feature
tion feature was numerically fitted with three Gaussian as a function of composition is located in Fig. 9. The
features representing the primary O 1s ionization primary band (529.6–530.1 eV) represents the O 1s
feature and chemically shifted O 1s features from ionization for oxygen associated with the cerium-
chemisorbed surface species [23,28,29]. The analysis zirconium complex [23,28,29], while the additional
A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221 217

Fig. 8. O 1s photoemission Gaussian peak fit for ZrO2. The spectra for the O 1s ionization as a function of composition is numerically fitted
with three Gaussian features. The features include the primary O 1s feature corresponding to the metal oxide and two additional features
chemically shifted to higher binding energies suggesting the presence of surface contaminants.

bands (531.6–532.0 and 532.9–533.8 eV) are O 1s [29]. The frequently observed O 1s shoulder shifted
features with slightly higher binding energy assign- to higher binding energies is the result of chemisorbed
ments. The difference between the secondary O 1s oxygen [29]. The binding energy for the O 1s ioniza-
assigned features is compositional dependent, where tion arising from adsorbed water is 533.1 eV
the binding energies are shifted to a lower binding [27,30,31]. This would suggest one secondary band
energy as a function of decreasing cerium (increasing (533.3–532.5 eV) is the result of adsorbed surface
zirconium) concentration. In addition, the estimated water or –OH species. Analysis of the remaining O
area of the O 1s primary feature is approximately 1s feature lends assignment to a carbon species. The O
constant as a function of decreasing cerium concen- 1s shift from carbon bonding is highly variable and
tration, as is the total calculated area for the O 1s strongly dependent on the chemical compound.
feature. Indeed, the O 1s shift arising from carbon bonding
The assignment of the O 1s bands beyond the has been reported to occur between 534.9 and
cerium-zirconium oxide is suggested through com- 542.6 eV [31] depending on carbon containing com-
parative analysis of similar studies [23,30]. The posi- pound. As a result, the final band (535.2–533.8 eV) is
tion of the primary O 1s feature (ca. 529.8 eV) is suggested to arise from surface C–O species, as the
characteristic of metal oxides [29]. However, it is presence of surface carbon has previously been deter-
impossible to distinguish between oxygen anions mined. The O 1s primary and secondary photoemis-
associated with cerium and zirconium complexes sion chemical shift as a function of composition can be
218 A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221

Fig. 9. O 1s high resolution XPS spectra. The O 1s photoemission features are represented as a function of composition. The spectra suggest a
slight compositional shift in binding energy, and also indicate secondary O 1s bands suggested to arise from surface (–OH) and (–CO) species.

accounted for by the changing surface electronic oxygen atoms (ca. 529.8 eV) indicates a relatively
structure and the binding energies associated with low concentration of Ce3þ ions, suggesting a near-
adsorbed surface water and C–O species, respectively. stoichometric surface.
Additional analysis of the O 1s feature also suggests
a low concentration of Ce3þ surface defect sites, 3.5. XPS peak area correlation
further supporting the conclusion of a highly oxidized
surface. The presence of Ce3þ surface defect sites The peak areas for the dominant photoemission
results in an additional O 1s band with a core level features were calculated using an adjusted baseline
shift of þ2.4 eV [25]. The absence of an O 1s band relative to the signal background. In this analysis, the
2.4 eV higher than the binding energy for lattice primary photoemission features included the Ce 3d3/2,
A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221 219

Table 4
XPS elemental surface concentrations

Atomic concentrations ( 0.5) Atomic ratios

Ce (at.%) Zr (at.%) O (at.%) C (at.%) Ce/Zr O/(CeþZr)

Ce1.0Zr0.0O2y 19.4 0.0 49.1 31.5 – 2.5


Ce0.9Zr0.1O2y 17.1 1.6 49.6 31.6 10.7 (9.0) 2.7
Ce0.8Zr0.2O2y 14.2 3.1 46.5 36.1 4.6 (4.0) 2.7
Ce0.7Zr0.3O2y 16.8 4.9 48.3 30.0 3.4 (2.3) 2.2
Ce0.6Zr0.4O2y 12.8 8.3 50.2 28.6 1.5 (1.5) 2.4
Ce0.5Zr0.5O2y 10.3 10.2 50.2 29.3 1.0 (1.0) 2.4
Ce0.25Zr0.75O2y 8.0 15.0 51.3 25.6 0.5 (0.3) 2.2
Ce0.0Zr1.0O2y 0.0 24.6 56.8 18.6 0.0 (0.0) 2.3

Zr 3d3/2, O 1s and C 1s bands. Surface atomic con- 3.6. Effect of electron bombardment on
centrations and atomic ratios calculated from these surface composition
core level transitions are located in Table 4. The
cerium to zirconium atomic ratio is in agreement The effect of electron bombardment on ceria-
with the expected ratios calculated form the bulk zirconia model catalysts was examined to determine
composition, as indicated by the values in parentheses. impact on surface oxygen and surface reduction. A
Additional qualitative information regarding the sur- 100 at.% cerium oxide sample was initially degassed
face oxidation state was also extracted from the XPS under high vacuum for 1 h and transferred to the
atomic concentration analysis. The ratio of the oxygen ultrahigh vacuum chamber for XPS characterization.
concentration to the summation of the cerium and Following the collection of initial XPS data, the
zirconium concentrations was calculated and com- sample was bombarded with an electron source for
pared to the expected ratio from the bulk composition. 2 h and subsequently analyzed with XPS. The sample
The value ranged from 2.2 to 2.7, with a value of 2 was bombarded for an additional 16 h before a final set
indicating a fully oxidized surface and a correspond- of XPS data was collected.
ing oxidation state of Ce4þ. Analysis clearly indicates Two sets of XPS data were collected at different
an excess quantity of surface oxygen with regards to spectrometer angles to discern between surface and
stoichometric cerium and zirconium concentrations. A bulk species. The data included a 458 and 158 acquisi-
similar observation of excess surface oxygen was also tion angle to determine elemental compositions to
reported in an independent study of ceria-zirconia depths of approximately 25–35 and 10–15 Å, respec-
systems [32] and is attributed to the high concentration tively. The 458 acquisition angle data is more repre-
of surface oxygen as an adsorbed layer of oxidized sentative of the bulk composition, while the 158
carbon species (CO, CO2) or water. This is further acquisition angle is limited to surface compositions.
supported in this analysis by the detection of surface The data obtained at each spectrometer angle were
carbon species. analyzed with peak fit routines and the surface atomic
The area of the Zr 3d3/2 peak was also considered to concentrations were calculated (Table 5). Based on the
estimate the surface composition of the model catalyst quantification of the oxygen and carbon photoemis-
systems. The calculated Zr 3d3/2 peak areas were sion features, it is evident that electron bombardment
linearly regressed as a function of zirconium concen- is responsible for removing adsorbed surface carbon
tration to correlate zirconium photoemission peak contaminants. This is clearly evidenced by the reduc-
areas to bulk composition. The analysis produced tion of surface carbon at a 458 analysis angle, and
a linear relationship with bulk composition, as indi- enhanced at a 158 spectrometer angle. The decrease in
cated by an R2 value of 0.9852. The analysis provides the carbon to oxygen and carbon to cerium ratios as
excellent correlation between XPS analysis of surface a function of electron beam exposure also supports
compositions and the actual bulk composition. the observation of a reduction of adsorbed surface
220 A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221

Table 5
Effect of electron bombardment of surface species

Atomic concentrations ( 0.5) Atomic ratios

Ce (at.%) O (at.%) C (at.%) O/Ce C/O C/Ce

458 As prepared 19.1 50.5 30.4 2.6 0.60 1.59


2 (h) 17.8 58.0 24.2 2.1 0.42 1.34
16(h) 23.2 54.6 22.2 2.3 0.41 0.97
158 As prepared 10.9 50.1 39.0 4.6 0.78 3.58
2 (h) 22.5 55.4 22.2 2.5 0.40 0.99
16 (h) 32.1 54.6 13.4 1.7 0.25 0.42

carbon contaminants. The ratio of oxygen to cerium is structure for compositions from 100 to 25 at.% cerium,
relatively constant as a function of electron beam although the existence of an additional phase is pos-
exposure at 458, while 158 angle analysis indicates sible. This analysis is an important initial step for
a decrease as a function of electron beam bombard- determining surface reactions and pathways for the
ment. This might suggest a slight cerium reduction at development of efficient and sulfur-tolerant automo-
the gas phase boundary, although the 458 acquisition tive emissions control catalysts.
indicates a relatively constant near surface oxygen to
cerium ratio. A constant cerium oxidation state as a
function of electron beam exposure is also supported Acknowledgements
by the analysis of the Ce3d3/2 binding energy. As a
result, it is suggested the surface composition with The authors gratefully acknowledge the assistance
respect to cerium oxidation state is uninfluenced by of Melissa K. Graves in the Department of Chemical
electron bombardment, while the removal of adsorbed Engineering at Mississippi State University for XPS
surface carbon species is evidenced. analysis. The authors would also like to acknowledge
financial support from the Ford Motor Company.
Additionally, A.E. Nelson would like to acknowledge
4. Conclusions additional support from the Interdisciplinary Center for
Advanced Propulsion (ICAP) at Michigan Technolo-
The chemical composition and availability of sur- gical University through the Department of Energy
face oxygen of ceria-zirconia model automotive emis- (DOE) Graduate Automotive Education Technology
sions control catalyst promoters were considered to (GATE) program.
develop a reference for future catalytic reactivity ana-
lysis. The XPS and AES analysis of ceria-zirconia
model catalysts clearly indicate (1) no significant sur- References
face species segregation occurs, confirming the surface
composition is representative of the bulk; (2) the [1] K.C. Taylor, Catal. Rev. Sci. Eng. 35 (1993) 457.
concentration of Ce3þ surface defect sites is constant [2] A. Trovarelli, Catal. Rev. Sci. Eng. 38 (1996) 439.
[3] J.-P. Cuif, S. Deutsch, M. Marczi, H.-W. Jen, G.W. Graham,
and independent of composition, suggesting incorpora-
W. Chun, R.W. McCabe, SAE Technical Paper Series 980668,
tion of zirconium into the ceria lattice does not inher- 1998.
ently create additional defect sites; (3) the surface [4] J.R. González-Velasco, M.A. Gutiérrez-Ortiz, J.-L. Marc, J.A.
chemistry of the oxides is unaffected by prolonged Botas, M.P. González-Marcos, G. Blanchard, Appl. Catal. B.
exposure to ultrahigh vacuum or electron bombard- Environ. 22 (1999) 167.
[5] P. Fornasiero, G. Balducci, R. Di Monte, J. Kaspar, V. Sergo,
ment; (4) the model catalyst surface stoichometry is
G. Gubitosa, A. Ferrero, M.J. Graziani, J. Catal. 164 (1996) 173.
consistent with (y ¼ 0:00:08) for CexZr1xO2y. [6] H.-W. Jen, G.W. Graham, W. Chun, R.W. McCabe, J. Culf, S.
Additionally, XRD analysis did not indicate the pre- Deutsch, M. Marczi, SAE Technical Paper Series 980668,
sence of additional crystalline phases beyond the cubic 1998.
A.E. Nelson, K.H. Schulz / Applied Surface Science 210 (2003) 206–221 221

[7] J.R. González-Velasco, M.A. Gutiérrez-Ortiz, J.-L. Marc, J.A. [20] R.P. Ingel, P. Lewis, B.A. Bender, R.W. Rice, Adv. Ceram. 12
Botas, M.P. González-Marcos, G. Blanchard, Appl. Catal. B. (1984) 408.
Environ. 35 (2000) 19. [21] A.E. Nelson, K.H. Schulz, Surf. Sci. Spectra 7 (2000) 281.
[8] C.E. Hori, H. Permana, K.Y.S. Ng, A. Brenner, K. More, [22] L.E. Davis, N.C. MacDonald, P.W. Palmberg, G.E. Riach,
K.M. Rahmoeller, D. Belton, Appl. Catal. B. Environ. 16 R.E. Weber, Handbook of Auger Electron Spectroscopy,
(1998) 105. Perkin-Elmer Corporation, Eden Prairie, MN, 1978.
[9] G. Vliac, R. Di Monte, P. Fornasiero, E. Fonda, J. Kaspar, M. [23] B.V. Crist, XPS Handbook of the Elements and Native
Graziani, Studies in Surface Science and Catalysis: Catalysis Oxides, XPS International, Ames, IA, 1999.
and Automotive Pollution Control IV, vol. 116, Elsevier, [24] P. Burroughs, A. Hamnett, A. Orchard, G. Thornton, J. Chem.
Amsterdam, 1998. Soc., Dalton Trans. (1976) 1686.
[10] D. Terribile, A. Trovarelli, J. Llorca, C. de Leitenburg, G. [25] A. Pfau, K.D. Schierbaum, Surf. Sci. 321 (1994) 71.
Dolcetti, Catal. Today 43 (1998) 79. [26] T.L. Barr, C.E. Fries, F. Cariati, J.C.J. Bart, N. Giordano, J.
[11] S.L. Peterson, K.H. Schulz, C.A. Schulz Jr., J.M. Vohs, Rev. Chem. Soc., Dalton Trans. (1983) 1825.
Sci. Instrum. 66 (1995) 3048. [27] D. Briggs, M.P. Seah, Practical Surface Analysis by Auger
[12] C.H. Bartholomew, Chem. Eng. 11 (1984) 96. and X-ray Photoelectron Spectroscopy, Wiley, New York, 1983.
[13] Accuratus Ceramic Technical Bulletin, Accuratus Ceramic [28] J. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, in: J.
Corporation, Washington, NJ, 1999. Chastain (Ed.), Handbook of X-ray Photoelectron Spectro-
[14] P. Duwez, F. Odell, J. Am. Ceram. Soc. 33 (1950) 274. scopy, second ed., Perkin-Elmer Corporation, Eden Prairie,
[15] E. Tani, M. Yoshimura, S. Somiya, J. Am. Ceram. Soc. 66 MN, 1992.
(1983) 506. [29] A. Platau, L.I. Johansson, A.L. Hagstrom, S.-E. Karlsson,
[16] M. Yashima, H. Arashi, M. Kakihana, M. Yoshimura, J. Am. S.B.M. Hagstrom, Surf. Sci. 63 (1977) 153.
Ceram. Soc. 77 (1994) 1067. [30] C.D. Wagner, D.A. Zatko, R.H. Raymond, Anal. Chem. 52
[17] M. Yashima, K. Morimoto, N. Ishizawa, M. Yoshimura, J. (1980) 1445.
Am. Ceram. Soc. 76 (1993) 1745. [31] D. Nordfors, A. Nilsson, N. Martensson, S. Svensson, U.
[18] M. Yashima, K. Morimoto, N. Ishizawa, M. Yoshimura, J. Gelius, H. Agren, J. Electron Spectrosc. Relat. Phenom. 56
Am. Ceram. Soc. 76 (1993) 2865. (1991) 117.
[19] B.D. Cullity, Elements of X-ray Diffraction, Addison-Wesley, [32] A. Galtayries, R. Sporken, J. Riga, G. Blanchard, R. Caudano,
Reading, MA, 1956. J. Electon Spectrosc. Relat. Phenom. 88 (1998) 951.

Вам также может понравиться