Electron Spectroscopy of Single Crystal and Polycrystalline Cerium Oxide Surfaces
Surface Science , 409 (1998) 307.
Valence band photoemission (XPS), Ce 3d and 4d XPS, and O 1s x-ray
absorption (XAS) have been investigated for oxidized and sputtered single
crystal CeO2 films and for oxidized Ce foil.
Features were identified that distinguish between the Ce+4 or
Ce+3 oxidation states.
Ce 3d spectra are the most commonly used for studying CeOX because it is readily accessible using a laboratory Mg ka source. The Ce 3d spectrum for CeO2 does not consist of a single spin orbit doublet as expected but instead has 6 peaks. This multiplet of peaks results from a final state "shake down" which places a different occupation in the O 2p and Ce 4f valance orbitals. The principal labels, u and v, correspond to spin orbit pairs split by 16.5 eV while the superscripts denote a common final state occupation. The peaks labeled u''' and v''' are unique to CeO2 and result from a Ce 3d9 O 2p6 Ce 4f0 final state. The spectrum for Ce(III) oxide contains only four peak because it lacks the Ce 4f0 final state component. As can be seen in the middle spectrum. The degree of oxidation of a partially reduced sample can be determined from a linear combination of the spectra from fully oxidized and fully reduced CeOX samples.
The Ce 4d spectra are similar to the Ce 3d spectra. Using a laboratory source they are much weaker than the the Ce 3d spectra, however using synchrotron radiation they are quite intense. The spectra at left were recorded using 500 eV radiation from Beamline X 1B at the National Synchrotron Light Source. The peaks labeled w''' and x''' are a spin-orbit pair split by 3.3 eV and have a final state of Ce 4d9 O 2p6 Ce 4f0. The other peaks can be roughly assigned to mixtures of shake states between the O 2p and Ce 4f level. The spectra are more complex than the Ce 3d spectra. The peaks are broader and cannot be fit using a simple combination of six or four peaks. A fifth peak is clearly evident in the Ce(III) oxide spectrum. The additional complexity in the Ce 4d spectra results from an interaction between the unpaired electron in the Ce 4d9 level and an unpaired electron in the Ce 4f1 or Ce 4f2 levels. The oxidation state of a partially reduced sample can again be determined from a linear combination of spectra from Ce+3 and Ce+4 oxides (blue curve).
The
next series of spectra are from the highest occupied states in the CeOX
valence band. These spectra were recorded using 300 eV radiation from Beamline
X 1b. The Ce 4f signal is resonance enhanced above the Ce 4d excitation
edge (~120 eV). Therefore these spectra are particularly useful when
excited by synchrotron radiation. The valence band spectra are perhaps the
easiest to interpret. The difference between Ce+3 and Ce+4
is the presence or absence of an electron in the Ce 4f state,
respectively. This is readily apparent in the spectra where the Ce 4f peak
is absent from CeO2 but is very intense in Ce(III) oxide.
Changes in the O 2p portion of the spectra are indicative of a change in the
hybridization between O 2p and Ce 4f. The reference spectra can again be
used to determine the degree of oxidation between of a partially reduced
sample. In practice, the intensity of the Ce 4f peak is an ideal measure
of the percent of Ce+3.
The O 1s x-ray absorption spectra are analogous to the valence band spectra except that they probe the lowest unoccupied state rather than the highest occupied states. The intense peak at the lowest photon energy in CeO2 results from an excitation from the the O 1s level into the unoccupied Ce 4f level. This peak is absent in the Ce(III) spectrum but a new peak emerges near 535 eV from an excitation of the O 1s electron into a Ce 4f level that is already occupied by an electron. Although dipole selection rules dictate that only excitations from s to p levels are allowed, the Ce 4f and Ce 3d levels are accessible due to hybridization with the O 2p level. The fit to the spectrum for a partially reduced sample using Ce+4 and Ce+3 reference spectra is not very satisfactory and does not agree with the other spectroscopic techniques. The reason is not clear but may be due to change in hybridization during the change in oxidation state and to the different depth dependencies of the probes. X-ray absorption probes deeper than photoemission.
