Surface Science 437 (1999) 386–396
www.elsevier.nl/locate/susc
a-Cr O (101: 2): surface characterization
2 3
and oxygen adsorption
Steven C. York, Mark W. Abee, David F. Cox *
Department of Chemical Engineering, Virginia Polytechnic Institute & State University, Blacksburg, VA 24061-0211, USA
Received 23 November 1998; accepted for publication 11 May 1999
Abstract
The composition and ordering of the Cr O (101: 2) surface has been studied with X-ray photoelectron spectroscopy,
2 3
Auger electron spectroscopy, and low-energy electron diffraction. It has been found that a nearly-stoichiometric,
(1×1) surface can be prepared by ion bombardment and annealing in vacuum to 900 K. The results are consistent
with a simple non-polar surface termination giving predominantly five-coordinate Cr3+ surface cations. Oxygen
exposures at 163 K lead to both dissociative and molecular adsorption. Dissociative adsorption dominates, giving an
O-terminated surface with a saturation coverage of nearly one O atom per surface Cr3+ cation. Dissociativelyadsorbed oxygen is stable to over 1100 K and is attributed to a terminal chromyl oxygen species (i.e. CrNO) that
caps the single coordination vacancy of the surface cations. On the O-terminated surface formed by dissociative
oxygen adsorption, molecular adsorption of O occurs, giving a weakly-bound species that desorbs at 220 K. The
2
saturation coverage of the molecular species is low at 163 K and corresponds to 2% or less of the available chromium
sites on the ideal, stoichiometric surface. Because of the low coverage, the adsorption sites are attributed to cations
at defect sites in the terminating oxide layer. © 1999 Elsevier Science B.V. All rights reserved.
Keywords: Auger electron spectroscopy; Chemisorption; Chromium oxide; Low energy electron diffraction (LEED); Single crystal
surfaces; X-ray photoelectron spectroscopy
1. Introduction
Amorphous, supported and microcrystalline
Cr O powders have been used in a wide variety
2 3
of catalytic reactions including SO oxidation,
2
alcohol dehydration and dehydrogenation, methanol synthesis, double bond migration in olefins,
ethylene polymerization, and the catalytic synthesis
of chlorofluorocarbon alternatives [1,2]. The study
of chemistry on well-ordered Cr O surfaces has
2 3
* Corresponding author. Fax: +1-540-231-5022.
E-mail address: [email protected] (D.F. Cox)
received far less attention. There are reports of the
interaction of a number of small molecules, including O , on oriented Cr O (0001) thin films [3,4]
2
2 3
but no reports of chemistry over the non-polar
(101: 2) surface which is the predominant exposed
crystal plane on microcrystalline Cr O powders
2 3
[5].
Zecchina and coworkers [5,6 ] used infrared
(IR) spectroscopy to study the interaction of
oxygen with Cr O powders that exhibit predomi2 3
nantly (101: 2) natural growth faces. They observed
that oxygen was strongly adsorbed and used IR
assignments to argue that the adsorbed oxygen
0039-6028/99/$ – see front matter © 1999 Elsevier Science B.V. All rights reserved.
PII: S0 0 39 - 6 0 28 ( 99 ) 0 06 7 6 -7
S.C. York et al. / Surface Science 437 (1999) 386–396
was linked to the surface via covalent double
bonds to chromium cations (i.e. a terminal CrNO
species). Carrott and Sheppard [7] made a similar
assignment for both amorphous and crystalline
Cr O powders. However, neither of these two
2 3
studies clearly addressed the oxidation state of
surface chromium cations. Both groups reported
a single adsorption state for oxygen, the stronglybound atomic species having a bond order of two.
Neither group observed molecularly bound O .
2
Davydov and coworkers [8–10] also investigated the O –Cr O system using IR spectroscopy
2
2 3
and found two states of adsorbed oxygen, one
corresponding to CrNO and the other to a molecular adsorbate suggested to be O2−. They were able
to prepare samples selectively to obtain either
dissociative or molecular adsorption, demonstrating that the preferred type of oxygen adsorption
is a function of sample pretreatment. Dissociative
oxygen adsorption was proposed to occur at
Cr2+ sites, while molecular adsorption at Cr3+
sites was proposed to form Cr4+–O− complexes.
2
Interestingly, Davydov and coworkers [8] also
proposed that five-coordinate surface Cr3+ cations
[i.e. the cation type at the ideal, stoichiometric
(101: 2) surface] are unreactive towards oxygen
adsorption.
Two groups have examined oxygen adsorption
over geometrically-well-defined Cr O (0001) thin
2 3
films grown on Cr(110) single crystal surfaces.
Foord and Lambert [4] give no description of the
coordination numbers associated with surface cations on the Cr O (0001) thin film surface, but
2 3
report molecular and dissociative adsorption of
oxygen with heats of adsorption of 110 and
210 kJ mol−1, respectively. They also report that
oxygen adsorption at the Cr O (0001) surface
2 3
does not cause any pronounced changes in the
Cr 2p levels in X-ray photoelectron spectroscopy
( XPS), indicating that oxidation of the surface to
CrO can be ruled out [4]. Freund and coworkers
3
describe both polar and non-polar terminations of
the Cr O (0001) thin film surface [11–14].
2 3
Following O adsorption, they report the removal
2
of molecularly adsorbed oxygen at temperatures
between 110 and 320 K, and the removal of atomic
(dissociatively adsorbed ) oxygen at temperatures
>800 K [3].
387
2. Surface investigated: a-Cr O (101: 2)
2 3
a-Cr O is an electrical insulator (band gap=
2 3
3.4 eV ) with the corundum structure [15,16 ]. The
bulk chromium coordination geometry is a distorted octahedron, and oxygen anions are coordinated by a distorted tetrahedral arrangement of
cations. In the corundum structure, one third of
the possible cation sites are vacant, with the vacancies located along the (101: 2) and other crystallographically-equivalent planes [17]. A ball model
representation of the ideal, stoichiometric surface
is shown in Fig. 1. The (101: 2) surface is slightly
corrugated due to the alternating tilt of incomplete
octahedra relative to the macroscopic (101: 2)
plane [17].
The ideal (101: 2) surface is non-polar and has
the lowest energy of any perfect low-index surface
of Cr O [18]. The topmost atomic layer of the
2 3
ideal surface is composed entirely of oxygen
anions. One full stoichiometric repeating unit
normal to the surface contains five atomic layers
arranged as [O, Cr, O, Cr, O]. The surface has a
rectangular (almost square) periodicity with a ratio
of sides of a/b=0.94. At the (101: 2) surface, all
O2− anions in the top atomic layer are threecoordinate, and the Cr3+ cations contained in the
second atomic layer are five-coordinate. Both ions
have one degree of coordinate unsaturation relative to their bulk counterparts [19]. All ions below
the top two atomic layers are fully-coordinated.
3. Experimental
Experiments were conducted in two different
ultrahigh vacuum ( UHV ) systems. One system is
a turbo-pumped, dual-chamber vacuum system
equipped with a Leybold EA-11 hemispherical
analyzer and a dual anode Mg/Al X-ray source
for XPS. The second system is an ion-pumped,
Physical Electronics chamber, equipped with a
Model 15-155 single-pass CMA for Auger electron
spectroscopy (AES ). Both systems also include an
Inficon Quadrex 200 mass spectrometer for thermal desorption spectroscopy ( TDS ) and a set of
vacuum generators three-grid reverse view low
energy electron diffraction (LEED) optics. The
388
S.C. York et al. / Surface Science 437 (1999) 386–396
ished to a final mirror finish with 0.25 mm diamond
paste. The sample was mechanically clamped
onto a tantalum stage that was fastened to
LN -cooled copper electrical conductors. A Type
2
K thermocouple was attached through a hole in
the stage to the back of the single crystal using
Aremco No. 569 ceramic cement. This arrangement allowed direct measurement of the sample
temperature.
4. Results
4.1. Ion bombardment and annealing experiments
Fig. 1. Ball model illustrations of the ideal, stoichiometric,
Cr O (101: 2) surface assuming no relaxation. The small black
2 3
balls represent Cr3+ cations, while the large gray balls represent
O2− anions. The top view is normal to the surface along the
[101: 2] direction and shows the rectangular (nearly square) periodicity expected for a stoichiometric surface. The bottom view
gives a perspective across the surface looking down the [022: 1]
direction and shows the corrugation due to the alternating tilt
of the incomplete octahedra coordinating the surface Cr3+ cations. Increased shading of the anions represents increasing distance away from the surface. For clarity, only the top five
atomic layers are shown.
base operating pressure in both systems is
1×10−10 Torr. XPS experiments were run at pass
energies of 60 eV, resulting in electron energy
resolutions (DE) of 0.9 eV. All XPS ratios have
been corrected with Leybold atomic sensitivity
factors.
Matheson
research-grade
oxygen
(99.997%) was used as received. Gas doses were
done by backfilling the vacuum chamber through
a variable leak valve. Oxygen exposures were conducted at 163 K, unless stated otherwise.
The crystal was oriented to within 1° of the
(101: 2) surface using Laue backreflection and pol-
The sample was bombarded with 2 keV Ar+
ions for 30 min and then annealed to successively
higher temperatures. Thermal treatments ranged
from 300 to 1100 K in 100 K increments. The
sample was heated at ca 2 K s−1 to the appropriate
annealing temperature then held for two minutes.
XPS spectra were collected, and LEED observations were made, following each annealing step
after the sample was cooled to room temperature.
LEED experiments at room temperature were conducted with beam energies of 150 eV or higher
because charging problems were encountered at
lower beam energies. XPS spectra could be collected at room temperature with an adjustment for
steady state charging. In addition to collecting
data at room temperature, observations were also
made at elevated temperatures (as described below)
where sample charging is minimized.
4.1.1. Low-energy electron diffraction
Following 30 min of Ar+ ion bombardment at
room temperature, no LEED pattern was visible
for any beam energy; only a diffuse background
was seen. The first faint LEED beams were
observed after annealing to ca 700 K. After the
800 K annealing step, the rectangular periodicity
of the surface could be recognized, although the
spots were broad. Annealing at 900 K produced a
sharp, rectangular (1×1) LEED pattern. This
(1×1) pattern remained unchanged by additional
annealing treatments up to 1100 K.
Because of increased conductivity of the Cr O
2 3
sample at elevated temperatures, LEED observa-
S.C. York et al. / Surface Science 437 (1999) 386–396
Fig. 2. A typical (1×1) rectangular LEED pattern
(E =62 eV ) following ion bombardment and annealing in
b
vacuum to 900 K. The pattern was taken at a sample temperature of 775 K to avoid the charging problems encountered at
room temperatures for low beam energies.
389
Fig. 3. Cr 2p and O 1s XPS spectra following (a) ion bombardment and annealing at 900 K to give a (1×1) LEED periodicity,
and (b) saturation with irreversibly adsorbed oxygen. All
spectra taken at a sample temperature of 300 K.
tions made at 725 K can be performed at beam
energies of 50 eV or less. A typical (1×1) LEED
pattern observed at a sample temperature of 775 K
is shown in Fig. 2. This pattern was obtained
following 30 min of ion bombardment and 2 min
of annealing at 900 K.
4.1.2. X-ray photoelectron spectroscopy
It was found that XPS spectra could be collected
from the insulating Cr O (101: 2) surface at room
2 3
temperature because of uniform steady state charging. To reference the binding energy scale, short
XPS runs were made at a sample temperature of
900 K where the conductivity of the material is
sufficient to prevent charging. At 900 K, the
Cr 2p binding energies fall at 576.9±0.2 eV for
3/2
an ion-bombarded and 900 K-annealed (ordered )
surface. This value is within the range typically
attributed to Cr3+ in Cr O [4,20,21]. Fig. 3a
2 3
shows Cr 2p and O 1s XPS data taken at room
temperature for an ordered surface. The binding
energy scales have been shifted to align the
Cr 2p peak to 576.9 eV to compensate for steady
3/2
state charging. The 530.8 eV binding energy for
Fig. 4. XPS O/Cr ratios as a function of annealing temperature
following ion bombardment. All measurements were made at
300 K following annealing at the indicated temperatures.
O 1s falls within the range expected for oxide
lattice oxygen [21], and it is typical for O 1s in
Cr O [20]. The 9.8 eV splitting between the
2 3
Cr 2p
and 2p
features is also typical of
1/2
3/2
Cr O [20,21].
2 3
In the ion bombardment and annealing study,
XPS was used to determine the O/Cr ratio
following each annealing step. As shown in Fig. 4,
the XPS data indicate no significant change in the
O/Cr ratio as the sample is progressively annealed
to higher temperatures to order the surface. The
390
S.C. York et al. / Surface Science 437 (1999) 386–396
average O/Cr ratio determined using XPS is
1.52±0.05. The XPS compositions and binding
energies suggest nearly-stoichiometric surface compositions for ion bombarded and annealed
surfaces.
4.2. Auger electron spectroscopy of the ordered,
(1×1) surface
Auger electron spectroscopy measurements are
problematic due to the insulating nature of the
sample [22–24]. However, it was possible to collect
room-temperature AES spectra from this bulk
insulator in cases where the primary beam and the
secondary emission are nearly balanced and
differential charging is minimal. To compensate
for the low signal-to-noise ratio in AES at room
temperature, sampling times of up to 2 h were
used. It was also found that charging problems in
AES are eliminated at sample temperatures of
875 K and above. Sufficient thermal promotion of
charge carriers occurs at this temperature to make
the sample conductive [25], leading to an improved
signal-to-noise ratio. An additional benefit of using
the higher temperature is that spectra can be
collected faster. Spectra taken at 300 K require a
sampling time of 1–2 h, while spectra taken at
875 K can be collected in a single 30 s scan.
However, significant differences occur between
spectra collected at room temperature and 875 K.
Fig. 5 shows AES spectra of identically-prepared, ion-bombarded and 900 K-annealed (1×1)
surfaces taken at 300 and 875 K. Differences in
peak-to-peak heights and peak positions are apparent between the 300 and 875 K spectra. For spectra
collected at 875 K, the oxygen KLL line (which
occurs at 520 eV at 300 K ) is shifted upwards in
kinetic energy by ca 5 eV. Also, the chromium
L M M line at 525 eV in the 875 K spectrum
3 2,3 4,5
is decreased in kinetic energy by ca 2 eV relative
to the 300 K spectrum, while the chromium
L M M feature at 499 eV shifts downward
2,3 2,3 2,3
to 490 eV. The Cr L M M
peak is almost
3 4,5 4,5
completely obscured by noise in the 300 K
spectrum but is easily seen in AES spectra collected
at 875 K.
As shown in Fig. 5, the peak-to-peak height of
the oxygen KLL line is dramatically reduced relative to the two chromium peaks for spectra col-
Fig. 5. AES spectra taken at 300 K (top panel ) and 875 K
(bottom panel ) following ion bombardment and annealing in
vacuum to 900 K.
lected at 875 K. The smaller size of the oxygen
signal suggests a decrease in the amount of oxygen
at the surface between 300 and 875 K, but the
smaller oxygen peak-to-peak height is actually
the result of a change in the shape of the signal
envelope involved in the overlapping of the O KLL
and Cr L M M signals. XPS spectra collected
3 2,3 4,5
at 875 K definitively show these differences are
only artifacts of the AES measurement conditions
and not evidence of a temperature-induced change
in the O/Cr ratio at the surface. Additionally, the
(1×1) LEED pattern at 875 K does not indicate
the occurrence of a surface reconstruction.
The Cr L M M
peak is the largest chro3 2,3 4,5
mium Auger signal and is normally chosen as the
signal upon which to base AES measurements of
chromium concentration. However, the overlap
between this primary chromium peak and the
O KLL peak in spectra collected at 875 K makes
using the Cr L M M peak at 490 eV a more
2,3 2,3 2,3
reasonable choice for tracking the chromium
signal. To use the Cr L M M
peak for
2,3 2,3 2,3
quantification purposes, the Cr L M M /
3 2,3 4,5
Cr L M M peak ratio from a standard chro2,3 2,3 2,3
mium metal spectrum (1.40 [36 ]) can be applied
to the usual sensitivity factor for the L M M
3 2,3 4,5
peak to estimate a sensitivity factor for the
Cr L M M
peak. This scaling contains the
2,3 2,3 2,3
inherent assumption that the relative peak-to-peak
S.C. York et al. / Surface Science 437 (1999) 386–396
391
Fig. 6. Variation in (a) the integrated O desorption intensity and (b) AES O/Cr ratio for consecutive doses of O on a surface
2
2
prepared by ion bombardment and annealing to 900 K. All exposures were at 163 K. The AES data were collected at 875 K following
a thermal desorption run after each exposure.
heights of the two chromium AES features are the
same in both chromium metal and in Cr O . Given
2 3
the overlap in the chromium and oxygen AES
signals, the relative peak-to-peak heights of the
chromium AES features in Cr O cannot be inde2 3
pendently established in the absence of the
O KLL signal.
Because the oxygen peak-to-peak intensity in
the AES derivative spectra is reduced by the
O KLL and Cr L M M signal overlap, AES
3 2,3 4,5
O/Cr ratios measured at 875 K do not yield a true
value of the O/Cr ratio at the surface. An ordered
(1×1) surface yields an O/Cr ratio of 0.67±0.03
if the spectrum is collected at 875 K, rather than
the 1.52 value obtained by XPS. The effect of
heating upon the apparent AES O/Cr ratio is,
however, consistent between spectra. While the
absolute ratio is not correct, variations in the
875 K AES O/Cr ratio give an accurate indication
of changes in surface composition as demonstrated below.
4.3. Oxygen uptake
Thermal desorption spectra, AES, and XPS
experiments were conducted on an ion-bombarded
and 900 K-annealed, (1×1) surface that was subjected to repeated doses of molecular oxygen.
Oxygen exposures were conducted at 163 K, and
AES spectra were collected at 875 K following
each TDS run. Fig. 6a shows the variations in the
integrated O desorption signal as a function
2
of total dose for a set of consecutive 0.25 L
(1 L=10−6 Torr s−1) exposures. No oxygen
desorbs following the first two 0.25 L doses, suggesting that the initial oxygen doses are irreversibly
adsorbed under the conditions of the TDS experiments with a maximum temperature of 650 K. For
successive exposures totaling >0.5 L, an oxygen
desorption feature is observed at 220 K in TDS.
In subsequent TDS runs, the amount of O desorp2
tion first increases, then levels out to a nearly
constant value. As demonstrated below with AES,
the integrated desorption signal levels out once the
surface is saturated with irreversibly-adsorbed
oxygen. The 220 K peak desorption temperature
for O is unchanged with coverage, a characteristic
2
of first-order desorption kinetics. Because of the
low desorption temperature and first-order kinetics, the O desorption observed in TDS is assigned
2
as originating from a molecular oxygen (O ) sur2
392
S.C. York et al. / Surface Science 437 (1999) 386–396
face species. An activation energy for desorption
of 57 kJ mol−1 is estimated for this species using
the Redhead Equation and assuming a normal
first-order pre-exponential factor of 1013 s−1 [26 ].
The amount of surface O associated with the
2
220 K desorption feature in TDS is small. A set
of O TDS experiments was run for an oxygen2
saturated surface for different dose sizes up to 1 L
(not shown). It was found that the 220 K desorption feature saturates for the lowest dose investigated (0.03 L). Assuming a unity sticking
coefficient for the molecular surface species up to
a saturation dose of 0.03 L, an upper-limit estimate
of 1.08×1013 O molecules per cm2 can be made.
2
Based on the unit cell parameters for a-Cr O , an
2 3
ideal, stoichiometric (101: 2) surface exposes
4.86×1014 five-coordinate Cr3+ cations per cm2.
The coverage of reversibly-adsorbed molecular
O , therefore, corresponds to 2% or less of the
2
available chromium sites on the ideal, stoichiometric surface.
Fig. 3b shows the Cr 2p and O 1s XPS data
collected from a surface oxygenated during consecutive TDS runs as described above. The adsorption
of oxygen on the (1×1) surface results in no
broadening of the Cr 2p peaks, and no new features attributable to an oxidation state greater
than Cr3+ are observed. The primary differences
of note are a slight broadening (0.2 eV ) and an
increase in the intensity of the O 1s feature
following oxygen adsorption. The low-temperature
adsorption and TDS treatments described above
cause an increase in the XPS O/Cr ratio from 1.52
for the ordered (ion-bombarded and 900 Kannealed) surface to 1.74, following oxygen
adsorption.
The extent of irreversible uptake during the
TDS treatments can be estimated from the XPS
data using a simple model based on the exponential
decay of signal with sampling depth1 [27].
Estimates of the expected O/Cr ratio for an ideal,
unrelaxed, stoichiometric surface (as shown in
1 Estimates of the expected XPS O/Cr ratios were calculated
assuming: (1) an exponential decay of signal intensity with distance for normal emission; (2) no diffraction effects; and (3)
inelastic mean free paths of 10.0 Å for Cr 2p (E =670 eV )
kin
and 10.1 Å for O 1s (E =720 eV ) photoelectrons. Mean free
kin
paths were estimated from the ‘universal’ curve in Ref. [27].
Fig. 1) yield a value of 1.52 compared to a measured value of 1.52±0.05 for an ion-bombarded
and 900 K-annealed surface. If a coverage of one
oxygen atom per five-coordinate surface Cr cation
is assumed, an estimated O/Cr ratio of 1.71 is
found from the model, compared to a measured
value of 1.74 for an oxygen saturated surface. The
similarity in the measured and estimated values
suggests an oxygen uptake yielding essentially a
1:1 ratio of oxygen adatoms to surface chromium
cations.
Auger electron spectroscopy also confirms the
irreversible uptake of oxygen by the surface.
Fig. 6b shows the increase in the AES O/Cr ratio
as the surface changes from a nearly-stoichiometric, (1×1) surface to an oxygen-saturated surface
that only adsorbs O molecularly. The stoichiomet2
ric surface has an O/Cr ratio of 1.52 as measured
by XPS and 0.67 from AES (875 K ). After fully
oxygenating the surface, the O/Cr ratio measured
using XPS is 1.74, and AES measurements yield
0.76, an increase in the surface O/Cr ratio of 13–
14% relative to the 900 K-annealed surface for
each technique. The data indicate that AES taken
at elevated temperatures is a poor technique for
measuring the surface O/Cr ratio, although it can
be used to track changes in composition accurately.
Thermal desorption experiments above 650 K
were not possible with the experimental setup used,
but the thermal stability of the ‘irreversibly’
adsorbed oxygen deposited during the TDS experiments was investigated at higher temperatures. An
oxygen-saturated surface was annealed at 1100 K
for 2 min and then cooled. AES O/Cr ratios collected from an oxygen-saturated surface before
and after annealing to 1100 K were comparable,
indicating no significant loss of oxygen up to
sample temperatures of 1100 K. Following the high
temperature treatment, O was also dosed onto
2
the sample at 163 K, and the resulting TDS spectra
were identical to those collected from the fullyoxygenated surface. The integrated desorption
areas before and after annealing were similar,
again indicating that the composition of the
oxygen-saturated surface was unchanged by the
thermal treatment.
Low energy electron diffraction was used to
investigate the ordered (unoxygenated ), partially-
S.C. York et al. / Surface Science 437 (1999) 386–396
oxygenated, and fully-oxygenated surfaces. LEED
observations were made at 300 and 725 K. Only a
(1×1) LEED periodicity was observed following
any degree of oxygen exposure at 163 K. The
LEED beams appeared sharper following the initial oxygen exposures.
393
5. Discussion
the nearly-stoichiometric composition of this
ordered surface suggests a simple non-polar termination, exposing predominantly five-coordinate
Cr3+ cations as expected for an ideal termination
of the bulk structure. Similarly, the repeated exposure of oxygen to an ordered, nearly-stoichiometric
surface at low temperature results in an ordered
uptake approaching one oxygen atom per surface
chromium cation.
5.1. Ordered surface structure and composition
5.2. Ion bombardment and annealing experiments
The (1×1) rectangular LEED pattern shown
in Fig. 2 for an ion-bombarded and annealed
surface is consistent with a simple termination of
the bulk structure of Cr O along a (101: 2) plane
2 3
[19]. Surface energy considerations dictate that the
non-polar, stoichiometric termination should be
the most stable [28]. However, oxide surfaces
prepared by ion bombardment and annealing can
exhibit non-stoichiometric terminations, particularly if one of the components is preferentially
sputtered.
Insight into the composition of the ordered
surfaces prepared in this study can be gained by
comparing XPS composition measurements to
expected values of the O/Cr ratio for different
surface terminations estimated from a simple exponential decay model for the photoemission
intensity1. For an ideal, non-polar stoichiometric
surface termination, the expected XPS O/Cr ratio
is 1.52, in good agreement with the value of
1.52±0.05 observed experimentally for ion bombarded and annealed surfaces. If the stoichiometric
surface is modified by the addition of an extra
terminating layer of oxygen anions (i.e. one additional oxygen atom for each five-coordinate surface cation on the stoichiometric surface), the
exponential decay model 1 gives an estimated XPS
O/Cr ratio of 1.71. This value agrees well with the
experimentally measured ratio of 1.74 for a fullyoxygenated surface.
The observed (1×1) LEED periodicity and the
similarity between the experimental and anticipated XPS O/Cr ratios suggest that a nearlystoichiometric surface can be prepared by ion
bombardment and high temperature annealing.
While the surface cannot be deemed defect free,
X-ray photoelectron spectra of the surface
following ion bombardment and throughout subsequent annealing steps show no significant change
in the O/Cr ratio for temperatures up to 1100 K.
The O/Cr ratios measured using XPS (300 K )
remain essentially constant at 1.52±0.05, a ratio
similar to that expected from an ideal stoichiometric (101: 2) surface. The results suggest that oxygen
and chromium are removed from the surface at
close to a stoichiometric rate during ion bombardment, in agreement with some previous work
[29,30]. It should be noted that no evidence for
surface oxidation during ion bombardment, as
seen by Li and Henrich [31], was observed to
within the error of our measurements (±0.05).
However, XPS spectra were collected in normal
emission which is the least surface sensitive geometry. Therefore, small changes in the surface O/Cr
ratio may have gone undetected.
5.3. Oxygen uptake
The lack of an oxygen desorption signal in TDS
following the initial doses and the accompanying
increase in the surface O/Cr ratio seen in both
XPS and AES demonstrate that a stoichiometric
Cr O (101: 2) surface irreversibly adsorbs oxygen
2 3
within the temperature range of our study. This
state of adsorbed oxygen is stable up to at least
1100 K, consistent with observations of strongly
bound surface oxygen made over Cr O powders
2 3
and films [3,4,7,9]. Since the desorption temperature for molecularly-adsorbed oxygen is expected
to be low [3,32], the ‘irreversibly’-adsorbed species
is attributed to the dissociative adsorption of
oxygen.
394
S.C. York et al. / Surface Science 437 (1999) 386–396
Retention of the (1×1) LEED pattern
following oxygen exposure demonstrates that the
periodicity of the oxygen adlayer is commensurate
with the stoichiometric surface, and XPS provides
an estimate of the saturation coverage of nearly
one oxygen atom per five-coordinate Cr3+ cation.
The periodicity and composition suggest that the
dissociative adsorption of oxygen completes the
coordination octahedra of the surface cations,
effectively capping the single vacant coordination
site of the cations at the stoichiometric surface to
give six-coordinate surface chromium cations.
Based on IR studies, the state of dissociativelyadsorbed oxygen most commonly proposed in the
literature on Cr O powders is a terminal chromyl
2 3
species (i.e. CrNO) [6,7] with a bond order near
two [6,9]. In the original study by Zecchina and
coworkers [6 ], the predominant crystal face on
microcrystalline powders was misidentified as the
(0001) surface. In a later electron diffraction study,
however, the (101: 2) surface (i.e. the same as that
under consideration here) was correctly identified
as the predominant exposed crystal face on microcrystalline powders [5]. The original misidentification led to confusion about the nature of the
surface cations available over Cr O microcrystal2 3
line powders. Because of this misidentification, the
surface cations were assumed to be predominantly
four- and five-coordinate Cr3+ prior to oxygen
adsorption [6 ]. Davydov and coworkers suggested
CrNO formation occurs at initially four-coordinate chromium cations [8] most likely in a 2+
oxidation state [9] and concluded that five-coordinate chromium ions are inert toward dissociative
oxygen adsorption [8]. It is clear from the current
work, however, that five-coordinate Cr3+ cations
on the (101: 2) surface actively dissociate oxygen.
For the ion-bombarded and annealed (1×1)
surface in this study, LEED and XPS indicate the
expected non-polar termination of the crystal, suggesting predominantly five-coordinate Cr3+ surface cations. In agreement with the literature on
powders, the form of dissociated oxygen on the
(101: 2) single crystal surface is attributed to terminal chromyl species (CrNO). The large (~4.5 Å)
separation between nearest neighbor cations on
the ideal (101: 2) surface suggests the occurrence of
bridging oxide ions or peroxide species is unlikely.
Following dissociative adsorption of oxygen, surface chromium cations are capped by the terminal
chromyl oxygen to give six-coordinate (i.e. fullycoordinated) surface cations. Additionally, LEED
observations show that the surface periodicity is
unchanged by oxygen uptake, suggesting that the
octahedral coordination environment of the cations is retained.
Formation of terminal CrNO may be formally
described as a two electron oxidation of surface
Cr3+ cations to Cr5+ [8]. Unfortunately, the XPS
results offer little insight into this oxidation process. Similar to the findings of Foord and Lambert
[4] for Cr O (0001) films, XPS following oxygen
2 3
adsorption on Cr O (101: 2) gives no indication of
2 3
a change in the oxidation state of surface chromium cations. Employing the same exponential
decay model used to estimate the surface
compositions1, the fraction of the Cr 2p XPS signal
originating from surface chromium cations is
expected to be ca 16%. If a significant chemical
shift of 16% of the Cr 2p signal occurred upon
oxidation, it should be easily observable as a
broadening or a shoulder in the Cr 2p XPS signal.
Complicating the interpretation of the XPS data
is the lack of standard spectra for Cr5+ [21], likely
due to the very restricted chemistry of Cr4+ and
Cr5+ compounds [33]. No clear indication is available from the literature as to the XPS binding
energy of Cr5+. One study of supported chromia
catalysts suggests the binding energy of Cr5+ falls
between that of Cr3+ and Cr4+ [34], although the
assignments are made on the basis of peak fits
rather than standard spectra. Even the binding
energy of Cr4+ oxides is unclear. Some references
suggest the binding energy is less than that of
Cr3+ in Cr O [21,35], while others suggest it is
2 3
greater [34].
In addition to the ‘irreversible’ dissociative
adsorption of oxygen, the observation of a low
temperature (220 K ) desorption feature in TDS
indicates that reversible molecular adsorption of
O also occurs. Since this desorption feature grows
2
in intensity as the surface concentration of irreversibly adsorbed oxygen increases ( Fig. 6), it seems
clear that molecular adsorption occurs on oxideterminated regions of the surface formed by the
initial dissociative adsorption of oxygen. This
S.C. York et al. / Surface Science 437 (1999) 386–396
interpretation is in agreement with the findings of
Davydov and coworkers [9,10] who observed that
dissociative and molecular adsorption of oxygen
occurs over reduced and oxidized Cr O samples,
2 3
respectively.
The adsorption of molecular oxygen at Cr3+
sites to form Cr4+–O− has been proposed pre2
viously in the literature on Cr O powders [1,9].
2 3
While the nearly-stoichiometric (101: 2) surface
exposes many such sites, the XPS results suggest
that nearly all of the surface cations are capped
by terminal chromyl oxygen species in the initial
stages of oxygen adsorption. The low uptake of
molecular oxygen (equivalent to a 2% coverage or
less based on the density of surface chromium
cations) is consistent with this idea. Because of the
low coverage of molecular oxygen, the corresponding adsorption sites are attributed to cations at a
low concentration of defects in the oxide-terminated surface. The current data shed no light on
the nature of these defect sites. They could originate from defective sites on the nearly-stoichiometric surface that are not amenable to dissociative
oxygen adsorption, or they could simply be isolated five-coordinate cations with no vacant neighbors capable of providing a site pair for O
2
dissociation.
6. Conclusions
It has been demonstrated that ion-bombarded,
annealed and oxygenated Cr O (101: 2) surfaces
2 3
are stable in UHV at temperatures of up to 1100 K.
LEED and XPS data suggest that the surface
periodicity and stoichiometry following ion bombardment and annealing are consistent with a
nearly-stoichiometric, non-polar (1×1) termination of the surface. The nearly-stoichiometric
Cr O (101: 2) surface interacts strongly with
2 3
oxygen. O adsorbs dissociatively and ‘irreversibly’
2
(for temperatures up to 1100 K ) with a (1×1)
periodicity and in an amount corresponding to
nearly one oxygen atom per surface chromium
cation. The oxygen adatoms are attributed to
terminal chromyl species (i.e. CrNO) that complete the coordination shell of the five-coordinate
Cr3+ cations at the stoichiometric surface. The
395
adsorption of small quantities of molecular oxygen
is observed on the oxide-terminated surface, and
the corresponding adsorption sites are assigned to
cations at defect sites in the terminating oxide
layer.
Acknowledgements
The authors gratefully acknowledge financial
support from the DOE Office of Basic Energy
Sciences (DE-FG02-97ER14751).
References
[1] R.L. Burwell Jr., L. Haller, K.C. Taylor, J.F. Read, Adv.
Catal. 20 (1969) 1.
[2] L.E. Manzer, V.N.M. Rao, Adv. Catal. 39 (1993) 329.
[3] B. Dillmann, F. Rohr, O. Seiferth, G. Klivenyi, M. Bender,
K. Homann, I.N. Yakovkin, D. Ehrilich, M. Bäumer, H.
Kuhlenbeck, H.-J. Freund, Faraday Discuss. Chem. Soc.
105 (1996) 295.
[4] J.S. Foord, R.M. Lambert, Surf. Sci. 169 (1986) 327.
[5] D. Scarano, G. Spoto, S. Bordiga, G. Ricchiardi, A.
Zecchina, J. Electron Spectrosc. Relat. Phenom. 64
(1993) 307.
[6 ] A. Zecchina, S. Coluccia, L. Cerruti, E. Borello, J. Phys.
Chem. 75 (1971) 2783.
[7] P.J.M. Carrott, N. Sheppard, J. Chem. Soc. Faraday
Trans. 1 79 (1983) 2425.
[8] D. Klissurski, K. Hadjiivanoy, A. Davydov, J. Catal. 111
(1988) 421.
[9] A. Davydov, J. Chem. Soc. Faraday Trans. 87 (1991) 913.
[10] A. Davydov, Y.M. Shchekochikhin, N.P. Keirer, Kinet.
Catal. 13 (1972) 980.
[11] C. Xu, M. Hassel, H. Kuhlenbeck, H.-J. Freund, Surf. Sci.
258 (1991) 23.
[12] H. Kuhlenbeck, C. Xu, B. Dillmann, M. Haßel, B. Adam,
D. Ehrlich, S. Wohlrab, H.-J. Freund, U.A. Ditzinger, H.
Neddermeyer, M. Neuber, M. Neumann, Ber. Bunsenges.
Phys. Chem. 96 (1992) 15.
[13] M. Bender, D. Ehrlich, I.N. Yakovkin, F. Rohr, M.
Bäumer, H. Kuhlenbeck, H.-J. Freund, V. Staemmler,
J. Phys.: Condens. Matter 7 (1995) 5289.
[14] F. Rohr, M. Bäumer, H.-J. Freund, J.A. Mejias, V.
Staemmler, S. Müller, L. Hammer, K. Heinz, Surf. Sci.
372 (1997) L291.
[15] D. Adler, Solid State Phys. 21 (1968) 83.
[16 ] R.E. Newnham, Y.M. de Haan, Zeitschrift fur Kristallographie 117 (1962) 235.
[17] V.E. Henrich, P.A. Cox, in: The Surface Science of Metal
Oxides, Cambridge University Press, Cambridge, 1994,
pp. 49–51.
396
S.C. York et al. / Surface Science 437 (1999) 386–396
[18] P.J. Lawrence, S.C. Parker, P.W. Tasker, Commun. Am.
Ceram. Soc. 42 (1988) 389.
[19] R.J. Lad, V.E. Henrich, Surf. Sci. 193 (1988) 81.
[20] C.D. Wagner, W.M. Riggs, L.E. Davis, J.F. Moulder, G.E.
Muilenberg, Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer, Eden Prairie, MN, 1979.
[21] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben,
J. Chastain, Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer, Eden Prairie, MN, 1992.
[22] J. Cazaux, P. Lehuede, J. Electron Spectrosc. Relat.
Phenom. 59 (1992) 49.
[23] M.C. Kung, H.H. Kung, Surf. Sci. 104 (1981) 253.
[24] Q. Guo, L. Gui, P.J. Moller, K. Binau, Appl. Surf. Sci. 92
(1996) 513.
[25] D. de Cogan, G.A. Lonergan, Solid State Commun. 15
(1974) 1517.
[26 ] P.A. Redhead, Vacuum 12 (1962) 203.
[27] W.M. Riggs, M.J. Parker, in: A.W. Czanderna ( Ed.),
Methods of Surface Analysis, Elsevier, Amsterdam, 1975,
p. 103.
[28] P.W. Tasker, J. Phys. C 12 (1979) 4977.
[29] N.S. McIntyre, D.G. Zetaruk, J. Vac. Sci. Technol. 14
(1977) 181.
[30] K.S. Kim, W.E. Baitinger, J.W. Amy, N. Winograd,
J. Electron Spectrosc. Relat. Phenom. 64 (1993) 307.
[31] X. Li, V.E. Henrich, Solid State Commun. 98 (1996) 711.
[32] K. Wandelt, Surf. Sci. Rep. 2 (1982) 20.
[33] F.A. Cotton, G. Wilkinson, Advanced Inorganic Chemistry, 4th ed.Wiley, New York, 1980.
[34] K. Jagannathan, A. Srinivasan, C.N.R. Rao, J. Catal. 69
(1981) 418.
[35] I. Ikemoto, K. Ishii, S. Kinoshita, H. Kuroda, M.A.A.
Franco, J. Thomas, J. Solid State Chem. 17 (1976) 425.
[36 ] L.E. Davis, N.C. Macdonald, P.W. Palmberg, G.E. Riach,
R.E. Weber, Handbook of Auger Electron Spectroscopy,
Perkin-Elmer, Eden Prairie, MN, 1976.