Surface defect characterization in oxygen-dosed nickel surfaces and in NiO thin films by CO adsorption-desorption experiments

Article abstract of DOI:10.1021/jp971854h

Oxygen-covered Ni(110) as well as thin NiO films grown by oxidation of that surface were characterized under ultrahigh vacuum by using low-energy electron diffraction (LEED), X-ray photoelectron (XPS), Auger (AES), and ion scattering (ISS) spectroscopies together with CO temperature-programmed desorption (TPD) titrations. Results from NiO(100) films with surface defects induced via Ar⁺ ion irradiation at room temperature were compared with those from partially oxidized ordered Ni(110) surfaces. CO TPD proved to be a useful local probe for the investigation of defective NiO surfaces, since its adsorption energy varies by over 20 kcal/mol in going from a metallic Ni(110) clean surface to NiO. The CO-probing experiments also revealed that Ar⁺ bombardment of thin NiO films leads to the formation of Ni-O phases similar to those found during the early oxidation stages of the Ni metal surface.

Full text of DOI:10.1021/jp971854h

J. Phys. Chem. B 1997, 101, 9069-9076
9069
Surface Defect Characterization in Oxygen-Dosed Nickel Surfaces and in NiO Thin Films
by CO Adsorption-Desorption Experiments
Helmut O2 fner and Francisco Zaera*
Department of Chemistry, UniVersity of California, RiVerside, California 92521
ReceiVed: June 5, 1997X
Oxygen-covered Ni(110) as well as thin NiO films grown by oxidation of that surface were characterized
under ultrahigh vacuum by using low-energy electron diffraction (LEED), X-ray photoelectron (XPS), Auger
(AES), and ion scattering (ISS) spectroscopies together with CO temperature-programmed desorption (TPD)
titrations. Results from NiO(100) films with surface defects induced via Ar+ ion irradiation at room temperature
were compared with those from partially oxidized ordered Ni(110) surfaces. CO TPD proved to be a useful
local probe for the investigation of defective NiO surfaces, since its adsorption energy varies by over 20
kcal/mol in going from a metallic Ni(110) clean surface to NiO. The CO-probing experiments also revealed
that Ar+ bombardment of thin NiO films leads to the formation of Ni-O phases similar to those found
during the early oxidation stages of the Ni metal surface.
1. Introduction
(EELFS),32 and Auger electron spectroscopy (AES).33 It has
been established that the oxidation of the Ni(110) sample above
room temperature begins with the development of three
sequential ordered submonolayers, namely, a low coverage (3
Metal oxides form a very important class of compounds for
heterogeneous catalysis.1-3 Although they are often used only
as high surface area supports to increase the dispersion of the
active component, in some cases they are also known to
participate directly in the catalytic reaction.4-7 Despite their
importance, however, a microscopic understanding of the role
of oxides as active catalytic components is still missing. Since
real oxide catalysts usually consist of polycrystalline samples
containing a variety of different lattice sites and metal oxidation
states, it is difficult to derive a detailed picture about the
contribution of each individual component to the overall
reaction. It is therefore desirable to start the atomic-level
characterization of oxides with well-defined surfaces in order
to develop a better understanding of the role of both structure
and oxidation states on their chemical properties. We have
chosen NiO as a model oxide both because of its interesting
properties as a catalytic material4,8-12 and because it can be
grown as thin crystalline films on various Ni foils and single-
crystal surfaces (Ni(110),13-16 Ni(100),17-21 Ni(111),22,23 and
others24-27). The latter allows for the use of electron spectro-
scopic methods without having to deal with the charging
problems that are typical in the study of insulating materials.
By characterizing NiO films with standard surface spectro-
scopic techniques, changes in chemical activity may be cor-
related to specific surface sites. The role of the electronic
structure of NiO in determining its reactivity is of particular
importance, since there seems to be, at least in some instances,
a straightforward correlation between the two.1,4,8-10,12,28,29 One
interesting aspect of this interdependence is the fact that the
electronic behavior of the oxide can be tuned by introducing
defects in the crystal. Our research focuses on the creation and
characterization of specific defects on NiO crystalline films by
ion bombardment of the surface. The films studied here were
grown by high-temperature exposures of Ni(110) to molecular
oxygen.
1
× 1)O structure with a coverage of /3 ML (which will be
denoted here as “(3 × 1)O-low”), a (2 × 1)O phase at an oxygen
1
coverage of /2 ML, and a high coverage (3 × 1)O phase at a
2
coverage of /3 ML. Finally, after higher doses of oxygen, a
thin layer of NiO(100) grows on top of the metal substrate. The
structures relevant to this study are shown schematically in
Figure 1.
In this work the use of CO adsorption for the titration of
defects on NiO surfaces was tested and contrasted against results
from oxygen-treated Ni(110) surfaces. Carbon monoxide was
chosen because of its drastically different desorption charac-
teristics from NiO and metallic Ni surfaces. Since in both cases
CO adsorbs on Ni sites, the weak base character of the carbon
monoxide molecule is ideal for probing the chemical state of
the surface Ni atoms. We found that the defects on NiO induced
by Ar+ ion bombardment can indeed be easily identified by
CO desorption and that those defects may be linked to specific
atomic configurations similar to those associated with the
oxygen submonolayer coverages during O2 uptake.
2. Experimental Section
The experiments reported here were performed in an ultra-
high vacuum chamber evacuated with a turbomolecular pump
to a base pressure below 5 × 10-10 Torr (mostly H2). The
chamber is equipped with a quadrupole mass spectrometer for
temperature-programmed desorption (TPD), a 100 mm hemi-
spherical energy analyzer used either as an electron energy
analyzer in X-ray photoelectron spectroscopy (XPS) or as an
ion energy analyzer for ion scattering spectroscopy (ISS), a
nonmonochromatized dual-anode Mg/Al X-ray source for XPS,
a differentially pumped rasterable ion gun for sample cleaning
and for ISS, and four-grid retarding field optics for LEED.
The TPD experiments were performed by ramping the temper-
ature of the crystal at a rate of about 10 K/s and by using a
collecting cone with a small opening in close distance to the
surface of the sample in order to enhance the collection
efficiency for the desorbing species and to also reduce back-
ground signals. The Mg anode was used as the excitation source
for XPS, and the analyzer was set so the overall resolution for
The interaction of O2 with the Ni(110) surface has already
been studied extensively by scanning tunneling microscopy
(STM),15,16 photoemission spectroscopy (PES),13,27 reflection
high-energy electron diffraction (RHEED),30 low-energy elec-
tron diffraction (LEED),31 electron energy loss fine structure
* Corresponding author.
X Abstract published in AdVance ACS Abstracts, October 1, 1997.
S1089-5647(97)01854-3 CCC: $14.00 © 1997 American Chemical Society

9070 J. Phys. Chem. B, Vol. 101, No. 44, 1997
O¨ fner and Zaera
Figure 1. Schematic representation of the ordered oxygen-nickel surfaces used in this study. Oxygen adsorption on a clean Ni(110) surface leads
to the buildup of low-coverage (3 × 1)O (not shown), (2 × 1)O, and (3 × 1)O structures before a NiO(100) thin film develops. Notice that the (2
× 1)O and (3 × 1)O layers display oxygen rows in the [001] direction and that are formed via an added-row mechanism.
the Ni 2p and O 1s spectra were 0.6 and 0.8 eV, respectively.
The XPS signal was collected in a normal emission geometry,
and the binding energy scale was referenced to the reported
value of 852.3 eV for the Ni 2p3/2 peak in metallic Ni.34 ISS
was performed at a backscattering angle of 135° by using 4He+
ions and a constant energy analyzer resolution of 4 eV. The
Ni(110) single crystal was mounted on a manipulator set for
resistive heating to up to 1300 K and for cooling to 90 K via
heat transfer to a liquid nitrogen reservoir. The sample
temperature was measured by a chromel-alumel thermocouple
spot-welded to the side of the crystal. The surface was cleaned
by mild oxidation in oxygen followed by sputtering-annealing
cycles until no contaminants could be detected by XPS, AES,
or ISS and the LEED showed a sharp Ni(110) (1 × 1) pattern.
Thin layers of NiO were grown by exposing the clean Ni(110)
surface to 300 langmuirs of oxygen (99.99% purity) at a sample
temperature of 450 K, at which point LEED showed a weak (1
× 1) pattern matching the spot positions expected for NiO-
(100). Defective NiO films were prepared by exposing the NiO
thin-film surface to Ar+ ions at an incidence angle of about
45°, an energy of 2.5 keV, and a current density of 0.5 µA/
cm2.
Figure 2. Oxygen 1s X-ray photoelectron spectroscopy (XPS) data
from Ni(110) after various (3, 12, 30, and 300 langmuirs) oxygen doses
at either 400 K (for the first three cases) or 450 K (for the 300 langmuir
exposure). The smooth lines through the top (300 langmuirs) and the
3. Results
3.1. Formation and Characterization of NiO Films. The
oxidation sequence of Ni(110) surfaces upon exposures to
molecular oxygen was characterized first. Figure 2 displays
the oxygen 1s photoelectron spectrum obtained from a thin film
of NiO(100) grown on Ni(110) by exposure to 300 langmuirs
of O2 at a substrate temperature of 450 K (top trace) along with
spectra of the (2 × 1)O-Ni(110) and (3 × 1)O-Ni(110) phases
which form at 400 K and lower (3.0 and 12-30 langmuirs)
oxygen doses. Both the ordered nature of the oxygen sub-
monolayers and the (100) orientation of the NiO film were
confirmed by LEED. A smooth polynomial background has
been subtracted from all the XP spectra shown in Figure 2.
Except for the 30 langmuir (3 × 1)O trace, all other spectra
could be closely fitted by a single Gaussian line with a half-
width of about 1.7 eV, indicating the presence of essentially
bottom (3 langmuirs) spectra are Gaussian fits to the data. The corre-
sponding LEED patterns seen in each case are also listed to the left of
the peaks. A slight shift toward lower binding energies is seen with
increasing oxygen coverages, from 529.7 eV for the low-coverage (2
× 1)O submonolayer to 529.4 eV in the case of the NiO thin film. A
small (<15%) amount of surface defects is also identified in the case
of the NiO surface by the small high-energy shoulder in that spectrum.
one single oxidation state; the results of this fit are shown as
solid lines through the data for the NiO and (2 × 1)O surfaces.
The peak maxima do shift slightly toward lower binding energy
with increasing oxygen dose, from a value of about 529.7 eV
for the (2 × 1)O surface to 529.6 and 529.4 eV for the (3 ×
1)O and NiO cases. The spectrum of the 30 langmuir (3 ×
1)O surface displays a broader full width at half-maximum
(about 2.1 eV) and a slight asymmetry toward higher binding

NiO Thin Films
J. Phys. Chem. B, Vol. 101, No. 44, 1997 9071
Figure 3. Ni 2p XP spectra from Ni(110): clean (a) and after a 300
langmuir oxygen dose at 450 K (b). The third spectrum (bottom trace)
is the result of subtracting 56% of spectrum a from spectrum b and
shows the stoichiometric character of the thin NiO layer grown by the
oxidation process. The oxide film is estimated to be approximately 3
ML thick.
Figure 4. Ion scattering spectroscopy (ISS) data from clean and
oxygen-dosed Ni(110) surfaces. The spectra were obtained by using
4He+ ions with a primary energy of 1 keV. Shown are traces for three
different oxygen coverages representative of the (2 × 1)O, (3 × 1)O,
and NiO film surfaces, which were prepared by dosing the clean Ni-
(110) surface with 3, 30, and 300 langmuir doses of O2, respectively,
at either 400 K (first two cases) or 450 K (oxide surface). The oxygen
uptake is manifested here both by a decrease in the intensity of the Ni
peak at 810 eV and by the growth of an O feature around 435 eV.
energies which are likely to be the result of NiO patches on
this surface, as also inferred from TPD results (see below).
Lastly, a careful inspection of the NiO spectrum reveals a weak
shoulder at about 2 eV higher binding energy from the main
peak.
Figure 3 shows Ni 2p XP spectra from clean Ni(110) and
from the same NiO(100) surface as in Figure 2 (that obtained
by exposing clean Ni(110) to 300 langmuirs of oxygen at 450
K). The third trace shown at the bottom of this figure, which
is the result of subtracting the clean Ni spectrum after appropri-
ate scaling from that of the oxidized surface, is in reasonably
good agreement with literature spectra for NiO.35-37 The
thickness of the NiO film can be estimated from the ap-
proximately 50% reduction of the Ni 2p3/2 signal from metallic
Ni at 852.3 eV upon oxidation by assuming a homogeneous
sample, Beer’s law, and a mean free path for the photoelectrons
of about 4 monolayers (ML) at a kinetic energy of 400 eV.37,38
Employing this model, one arrives at a layer thickness of about
3.0 ML, the same as reported in other studies.15,26
4He+ ISS was used to get additional information on the
chemical and structural composition of the clean and CO-
covered oxygen-treated nickel surfaces. Figure 4 is a compila-
tion of IS spectra taken from clean the Ni(110), (2 × 1)O-
Ni(110), (3 × 1)O-Ni(110), and NiO surfaces. The Ni signal,
which is centered at 810 eV in our setup, is reduced by
approximately a factor of 21/2 when going from clean Ni(110)
to the (2 × 1)O structure and then remains almost constant as
the (3 × 1)O and NiO phases develop. A sharp oxygen peak
grows at the same time around a kinetic energy of 435 eV, but
a quantitative distinction between the oxygen concentrations in
the (2 × 1)O, (3 × 1)O, and NiO surfaces is not possible from
these data.
Figure 5. CO temperature-programmed desorption (TPD) spectra from
different oxygen-dosed Ni(110) samples. The surfaces were prepared
by initially dosing different amounts (0.0, 0.5, 1.0, 3.0, 30, and 300
langmuirs) of O2 at 400 K (450 K for the 300 langmuir case) and then
cooling to 90 K and dosing 3.0 langmuirs of CO. At least four distinct
CO desorption states can be seen around <150, 190, 240-300, and
350-400 K, identifiable with NiO, (3 × 1)O, (2 × 1)O, and clean
Ni(110) surfaces, respectively.
3.2. CO Desorption from Partially Oxidized Ni(110)
Surfaces. The use of carbon monoxide as a probe for different
adsorption sites on the O/Ni(110) surfaces was explored next.
Figure 5 displays CO TPD from a Ni(110) crystal first exposed
to the oxygen doses indicated in the figure (at 450 K for the
300 langmuir oxygen dose and at 400 K for all other cases)
and then dosed with 3.0 langmuirs of CO at 90 K. It is clear
from those data that increasing O surface concentrations lead
to a drastic shift of the main CO desorption peak, from about
400 K for clean Ni(110) to about 120 K in the case of NiO (the
300 langmuir O2 dose). Furthermore, the shifts appear to be
well-defined, not gradual. For low oxygen doses the CO
desorption feature from clean Ni(110) around 400 K decreases
in size at the expense of some additional intensity that develops
between 200 and 300 K. After a 3.0 langmuir oxygen dose, at
which point a (2 × 1)O phase is seen by LEED, CO desorption

9072 J. Phys. Chem. B, Vol. 101, No. 44, 1997
O¨ fner and Zaera
Figure 7. CO TPD from (2 × 1)O-Ni(110) (left) and (3 × 1)O-
Ni(110) (right) surfaces as a function of CO dose at 90 K. The oxygen-
covered surfaces were prepared by dosing 3.0 and 30 langmuirs of O2
at 400 K, respectively. Three CO desorption states fill in sequentially
about 300, 240, and 180 K during the CO uptake on the (2 × 1)O
surface, but only one broad peak develops around 190 K for the (3 ×
1)O case.
Figure 6. CO saturation coverages versus oxygen predose on Ni(110)
surfaces, as determined by the areas under the TPD peaks in Figure 5.
Small amounts of oxygen are quite effective at blocking CO adsorption
sites, but the subsequent development of the (2 × 1)O ordered structure
is accompanied by a significant increase in CO uptake.
of the mass spectrometer to hot molecules when measuring
fluxes instead of pressures. A full inclusion of this effect (which
is not appropriate in our setup, where the ionizer is placed inside
an enclosed cylinder with small openings for pumping43) would
lead to CO saturation coverages of 0.47, 0.60, 0.68, 0.57, and
0.23 ML for the 0.5, 1.0, 3.0, 30, and 300 langmuirs of O2,
respectively. Notice that even though the correction is signifi-
cant for the highly oxidized surfaces, the general trends seen in
Figure 6 still remain qualitatively the same.
Figure 7 shows the coverage dependence of CO desorption
for the (2 × 1)O (left) and (3 × 1)O (right) Ni(110) surfaces.
In the case of the (2 × 1)O surface CO initially desorbs around
300 K, but at high CO coverages a second peak develops around
240 K, and at saturation the main desorption feature in the TPD
traces is seen at 180 K. The (3 × 1)O surface, on the other
hand, displays a single desorption peak at about 190 K at all
coverages, even though there is a slight shoulder at higher
temperatures in those features. A second small low-temperature
shoulder appears around 120 K for CO coverages near satura-
tion.
Figure 8 shows ISS data from clean Ni(110), (2 × 1)O-Ni-
(110), and (3 × 1)O-Ni(110) surfaces after saturation (18
langmuir exposure) with CO at 90 K. In the case of clean Ni-
(110) the spectrum is dominated by the signal due to oxygen,
which is seen to peak at a kinetic energy of about 435 eV and
to have a long low-energy tail. The carbon peak, expected at
about 310 eV with our experimental setup, cannot be resolved
in this case but is in fact not expected as long as the CO remains
molecular and adsorbed in an upright position with the carbon
end down.40,41,44 The nickel signal is also completely suppressed
by the adsorbed CO, so only a slight background intensity can
be seen at kinetic energies below the position where the Ni
feature is to be expected. For the (2 × 1)O-Ni(110) and (3 ×
1)O-Ni(110) surfaces CO adsorption leads to qualitatively
similar IS spectra as for the clean Ni(110) surface, but there
are noticeable changes in the signal intensities: the O signal,
which contains contributions from both CO and surface O,
amounts to about 66 and 60% of the intensity of that seen after
dosing the clean Ni(110) surface with CO, respectively, and in
addition, a small Ni signal appears in the spectra for the oxygen-
dosed samples.
occurs in two or three regimes between 190 and 300 K.
Following an oxygen dose of 30 langmuirs, which results in
the formation of the (3 × 1)O surface, CO desorbs in one main
peak at 190 K with a shoulder at lower temperatures. Finally,
from the Ni(110) surface exposed to 300 langmuirs of oxygen
at 450 K, which leads to the formation of NiO(100), CO desorbs
in one sharp peak around 120 K, in agreement with recent
results.39 Therefore, it could be concluded that to a first
approximation CO desorption around 400, 250-300, 190, and
100 K corresponds to the clean, (2 × 1)O, (3 × 1)O, and NiO-
(100) phases, respectively.
Figure 6 displays the saturation CO coverages on the oxygen-
treated nickel surfaces obtained by integration of the TPD traces
in Figure 5. The absolute coverages were normalized to that
on clean Ni(110), which has been established to be 1.0 ML.40-42
CO adsorption is blocked quite effectively by small amounts
of oxygen: the CO uptake becomes reduced to about half of
that on the clean surface after a dose of only 0.5 langmuir of
O2. However, increasing oxygen coverages after that lead to a
subsequent increase, not decrease, of the CO saturation surface
concentration, until reaching a maximum about 0.86 ML on
the (2 × 1)O-Ni(110) surface. Only beyond that point does
the capacity of the surface for CO adsorption start to be reduced
again: 0.78 and 0.41 ML are the maximum coverages that can
be attained on the (3 × 1)O-Ni(110) and NiO(100) structures.
As mentioned above, the changes in CO uptake are also
accompanied by shifts in adsorption energy, which means that
the variations in the surface population of particular adsorption
sites are more abrupt than what the overall coverage values lead
to believe. Indeed, the amount of CO in the 400 K desorption
state goes from 1.0 ML on clean Ni(110) to about approximately
0.5 and 0.3 ML for 0.5 and 1.0 langmuir O2 doses, respectively,
and disappears by exposures above 2.0 langmuirs. Also, the
approximately 0.1 ML in CO coverage decrease between the
3.0 and 30 langmuir O2-dosed surfaces is mainly due to the
complete suppression of the high-temperature (>250 K) states;
the peak at 190 K is in fact some 25% larger in the 30 langmuir
case. One word of warning: because of the significant
difference in desorption temperatures among the different CO
states, there could be a systematic error in the evaluation of
surface coverages from TPD traces due to the lower sensitivity
Time-resolved ISS measurements on the CO-saturated O-Ni-
(110) surfaces revealed that the signal intensities in the spectra
are strongly affected by sputtering of the surface. The two

NiO Thin Films
J. Phys. Chem. B, Vol. 101, No. 44, 1997 9073
Figure 8. 4He+ ISS data from CO-saturated clean Ni(110), (2 × 1)O-
Ni(110), and (3 × 1)O-Ni(110) surfaces. Helium ion beams with a
primary energy of 1 keV were used in these experiments. The oxygen-
covered surfaces were first prepared by dosing 3.0 and 30 langmuirs
of O2 at 400 K, respectively, and then saturated by exposure to 18
langmuirs of CO at 90 K. Upon CO adsorption, the O peak around
435 eV develops a low-energy tail, and the Ni signal almost completely
disappears.
Figure 10. CO TPD spectra from NiO/Ni(110) as a function of CO
exposure at 90 K. The inset shows the dependence of the CO surface
coverage on the initial CO dose. The NiO film was prepared by dosing
a clean Ni(110) surface with 300 langmuirs of oxygen at 450 K. Only
one peak is seen in all spectra, around at 140 K at low coverages and
about 110 K at saturation. By quantitative comparison with similar data
from a clean Ni(110) surface, it is concluded that CO adsorption on
the oxide film is precursor-mediated, displays an initial sticking
coefficient of about half of that on the clean metal, and reaches a
saturation value of approximately 0.45 ML.
) 0. Taking the extrapolated O ISS intensity as a measure of
the CO coverage (not taking into account the contribution from
the substrate oxygen, since that is estimated to amount to less
than 25% of the total peak by comparison with the ISS from
clean NiO), one arrives at a coverage of about 40% of the value
for CO-saturated clean Ni(110). Also, extrapolation of the Ni
signal intensity for the undamaged CO-saturated NiO surface
yields a value of about 25% of that found for clean NiO.
3.3. CO Desorption from NiO Films. Figure 10 shows the
CO TPD traces obtained from a NiO(100) film grown on the
Ni(110) crystal after CO dosing at 90 K. They display a single
desorption peak that shifts from about 140 K for a CO dose of
0.3 langmuir to about 110 K for saturation coverages (>3.0
langmuirs). The additional CO desorption signal below 100 K
for low CO doses is due to desorption from the sample
mounting. A comparison of the CO TPD peak areas between
CO-saturated NiO(100) and Ni(110) yields a ratio of 0.45, in
excellent agreement with that estimated from the ISS data in
Figure 9. The insert in Figure 10 shows the uptake behavior
of CO. (The coverages were again calculated from the CO
desorption peak area from NiO after normalizing to the
corresponding peak area from the CO-saturated Ni(110) surface.)
This figure shows an almost linear dependence of CO coverage
on dose up to saturation, indicating the approximately constant
sticking coefficient typical of precursor-mediated adsorption
kinetics.45 The sticking coefficient, estimated from the slope
of that curve, comes up to be about 0.2 ML/langmuir, a bit
over half of that on the clean Ni(110) metal.
Figure 9. 4He+ ISS data from both clean and CO-saturated NiO films
showing the sputtering effect of the ion beam. The bottom trace
corresponds to the first scan taken after CO adsorption and the middle
to the one taken right after. The acquisition time of each scan was 1
min. The inset displays the time evolution of Ni and O ISS signal
intensities during the exposure to the 4He+ ion beam. This figure
provides evidence for CO adsorption on the nickel centers of the NiO
film with an upright geometry and the carbon end down.
bottom spectra of Figure 9 show two successive 1 min scans
taken from the same CO-saturated NiO film. It was found that
on this surface the damage induced by the helium ion beam,
namely, the time-dependent reduction of the O signal and the
concomitant increase in the Ni signal, is more severe than on
the clean or partially oxidized cases. However, since the time
dependence of the O and Ni ISS signal intensities can be
approximately fitted by linear decay and growth curves in this
time range, respectively (see inset of Figure 9), their values for
the undamaged surface can be estimated by extrapolation to t
3.4. CO Desorption from Sputtered NiO. Finally, our CO
titration methodology was employed to characterize the sites
produced by ion bombardment of NiO thin films. Figure 11
provides a compilation of CO TPD traces from NiO/Ni(110)
surfaces sputtered at room temperature for different times (as
indicated in the figure) and then exposed to 3.0 langmuirs of
CO at 90 K. The sputtering was done by using an Ar+ ion
current density of approximately 0.5 µA/cm2 and an ion energy

9074 J. Phys. Chem. B, Vol. 101, No. 44, 1997
O¨ fner and Zaera
less than 5% of the total O 1s XPS signal intensity in our
experiments, which represents a defect density in the top NiO
layer of about 15%.22 The Ni 2p XP spectra shown in Figure
3 also agree well with literature data35,36 and further confirm
the presence of a thin stoichiometric NiO layer. Finally, the
development of a weak (1 × 1) LEED pattern with spots in the
positions expected for NiO, which has a unit cell size of 2.95
× 2.95 Å2, evidences the formation of NiO(100) with at least
some degree of long-range ordering.16 The thickness of the NiO
layer was calculated at about 3 ML.
4.2. CO Adsorption on the NiO Films Grown by Ni(110)
Oxidation. To increase the surface sensitivity in our charac-
terization of the state of the outermost NiO layer, CO was used
as a probe molecule. This titration method proved to be quite
useful in identifying adsorption sites, since it was found that
the shape and position of the CO desorption peaks depend
critically on the preparation conditions of the surface prior to
oxidation. In particular, insufficient annealing of the Ni(110)
crystal prior to oxidation leads to the development of a
significant high-temperature tail in the CO desorption from the
oxidized surface (data not shown). Also, different oxygen
coverages on the surface yield significantly different CO TPD
traces (see below). However, the sequence of CO TPD spectra
on the NiO film as a function of CO coverage displayed in
Figure 10 shows only one single peak at 110-140 K, in good
agreement with another recent investigation.39 This corresponds
to a desorption activation energy of between 7 and 9 kcal/mol,
close to the 9-12 kcal/mol range reported for the isosteric heat
of adsorption of CO on NiO(100).48 Also, a quantitative
analysis of the CO TPD data indicates that the CO saturation
coverage on the oxide film is about 0.4 ML and that the initial
sticking coefficient on the NiO(100) surface at 90 K is about a
factor of 2 lower than on the Ni(110) clean surface. The
estimate of 0.4 ML for the saturation coverage, also confirmed
by the ISS results presented in Figure 9, is not far from the
value of 0.5 ML obtained by microgravimetric measurements.49
As an aside, the excellent agreement between the saturation
coverage estimates for CO on NiO obtained with TPD and ISS
vouch for the validity of using ISS for quantitative determina-
tions in this system.50-52
Figure 11. CO TPD spectra from NiO/Ni(110) surfaces after being
damaged by room temperature Ar+ ion sputtering. The initial NiO film,
prepared by dosing 300 langmuirs of O2 at 450 K, was sputtered with
an Ar+ ion beam of jAr ) 0.5 µA/cm and Ep(Ar ) ) 2.5 keV for the
times specified in the figure and subsequently exposed to 3.0 langmuirs
of CO at 90 K. As the sample is sputtered, not only is oxygen removed
from the surface, but also CO TPD peaks similar to those seen
throughout the oxygen uptake on clean Ni(110) develop. The results
in this figure argue for the reversibility of the formation of the different
ordered oxygen overlayers and for the coexistence of more than one
phase on the surface even after long (15 min) sputtering times.
2
+
+
of 2.5 keV. After ion bombardment for 3 min (approximately
0.5 Ar+ ions per nickel atom in the topmost layer of the initial
Ni(110) surface), the intensity of the main CO TPD peak at
110 K (typical of desorption from NiO) decreases, and the
intensity of the high-temperature tail extends beyond 200 K.
That tail increases in size until it evolves into a distinctive
second CO desorption peak at about 190 K after 6 min (1 Ar+/
Ni) of sputtering. Additional exposures to the ion beam lead
to a further reduction of the CO desorption at 110 K and to an
increase of the desorption intensity at higher temperatures, to
the point that after sputtering for 15 min (2.5 Ar+/Ni) CO
desorbs mainly around 390 K, the desorption position of the
clean Ni(110). Notice that distinctive features at 190 and 240-
300 K similar to those from the (2 × 1)O and (3 × 1)O ordered
overlayers are seen after 6 and 9-12 min of sputtering,
respectively. Nevertheless, some NiO seems to remain on the
surface even after high sputtering doses, possibly because of
either inhomogeneities in the ion beam across the sample surface
or selective sputtering; there is also a significant number of metal
defects capable of dissociating CO, as indicated by the recom-
binative peak at 580 K.
With respect to the adsorption orientation of the CO molecules
on the NiO surface, both theoretical considerations53,54 and
experiments39,49 suggest a bonding with the C-O axis perpen-
dicular to the surface and the carbon end pointing toward the
surface. Moreover, the same theoretical calculations53,54 and
additional experiments48,49,55 favor on-top bonding on the Ni
atoms. Our ISS results are compatible with such a picture, since
the adsorption of CO onto the NiO surface reduces the Ni signal
selectively while leading to an increase in the oxygen peak
height.
4.3. CO Adsorption on Partially Oxidized Ni(110). The
main goal of this investigation was to characterize specific
defects sites on NiO surfaces. Unfortunately, the O 1s XP
spectra of the oxidized Ni surface show only a small and gradual
shift in binding energy between early oxidation stages, i.e., the
low oxygen coverages, and the formation of a NiO layer (Figure
2 and refs 26 and 39). Since these shifts result from a
combination of both initial and final state effects as the nature
of the surface is transformed from metallic to oxide-like,56 it is
quite difficult to identify distinct oxidation states for oxygen
from these data. Likewise, the Ni 2p XP spectra do not display
any detectable changes when going from low oxygen coverages
to the formation of the NiO layer (Figure 3). In this case the
problem is mainly one of low surface sensitivity at those kinetic
energies, which causes the Ni bulk contribution to dominate
the spectra and to obscure changes in the oxidation state of
4. Discussion
4.1. Growth of NiO Films by Oxidation of Ni(110)
Surfaces. The oxidation of the Ni(110) surface has been studied
extensively, and the growth of NiO(100) films has been
discussed in detail in several previous publications.13-16,27,30-33
The O 1s photoelectron spectrum from the NiO grown under
the conditions of this study essentially displays one single line
in the position expected for NiO;35 only a very small amount
of additional intensity is seen at about 2 eV higher binding
energy. According to similar investigations on thin films of
NiO(100) and NiO(111) over Ni(100) surfaces, this latter feature
is likely to be due to the OH groups that result from the
adsorption of water on defects;22,36,46,47 its intensity amounts to

NiO Thin Films
J. Phys. Chem. B, Vol. 101, No. 44, 1997 9075
surface Ni atoms. CO desorption from partially oxidized Ni
surfaces, on the other hand, proved to be very sensitive to the
oxygen coverage of the surface, as seen in Figures 5 and 7.
Indeed, Figure 5 shows a drastic decrease in the CO desorption
temperature with increasing oxygen concentration on the surface.
It is in principle possible for CO to change its bonding geometry
as the nature of the surface evolves toward the oxide, but as
discussed above, this does not seem to be the case: CO appears
to always bond to the Ni atoms and with the carbon end down.
It is then tempting to associate the decrease in the CO desorption
temperature with changes in either the surface nickel valence
state or the repulsive nature of the adjacent adsorbed oxygen
atoms, at least in a qualitative way.
surface coverage changes during the TPD runs. An alternative
model could be proposed where all CO molecules adsorb on
bridge sites between the atoms in the Ni[011] orientation
exposed by the Ni-O added rows (by analogy to the adsorption
site identified on the clean Ni(110) surface40,41) and where the
different desorption states are associated with changes in
repulsion energies as the CO layer compression changes as the
CO coverage is reduced. We nevertheless favor the former
interpretation and point to a roughly linear correlation between
the CO adsorption energy and the number of oxygen atoms
adjacent to the Ni adsorption site.
4.4. CO Adsorption on Defective NiO Films. Despite the
complications mentioned above, the data presented here do
provide information that helps in the understanding of the
surface structure of the NiO surface after ion bombardment.
The initial changes seen after Ar+ sputtering times of up to about
One of the observations of this study that may appear puzzling
at first is the fact that while small doses of oxygen are quite
efficient at blocking CO adsorption sites, further increases in
oxygen surface coverage lead to an increase in CO uptake by
the surface (Figure 6). To explain this observation, a few words
need to be said first about the uptake of the oxygen itself on
the Ni(110). As mentioned above, three ordered structures
develop sequentially on that surface with increasing oxygen
dosing, namely, the (3 × 1)O-low, (2 × 1)O, and (3 × 1)O
overlayers described in the Introduction.15 Recent atom-resolved
STM investigations have suggested that those phases all consist
of Ni-O rows running along the Ni[001] direction, one for every
three, one for every two, and two for every three [11h0] lattice
spacings, respectively, as shown in Figure 1.16 The nucleation
and growth of those reconstructions proved to be via an added-
row mechanism, which means that nickel atoms are removed
from steps and added to the new terrace Ni + O mixed layer.
The oxygen adsorbed during the very early stages of the uptake,
on the other hand, appears to create strings running along the
Ni[11h0];16 since the oxygen atoms seem to reside in 4-fold
hollow sites in that case, they block a large ensemble of nickel
atoms.57
+
3 min under the conditions used in our experiments (jAr ) 0.5
µA/cm2, Ep(Ar+) ) 2.5 keV) are manifested by a broadening
of the original sharp CO desorption peak with additional signal
intensity around 190 K (Figure 11). In light of the results
obtained from the partially oxidized Ni surfaces discussed above,
this may be interpreted as due to a slight decrease in surface
oxygen concentration. Indeed, the concomitant reduction of
Ni2+ surface ions induced by sputtering has also been observed
in other NiO films,37 in bulk NiO samples,60-62 and in powder
catalysts.63 Increasing sputtering times then leads to an evolu-
tion of the surface in a way that is essentially the reverse of
that caused by increasing the oxygen dose on clean Ni(110),
namely, to a gradual shift in the CO TPD desorption peak toward
higher temperatures. Distinct desorption peaks can indeed be
resolved for the sputtered samples around 190, 240, and 390
K, coinciding with the desorption peaks seen for the (3 × 1)O,
(2 × 1)O, and clean Ni(110) surfaces, respectively. A similar
effect in terms of the reversal of the order of the reconstructions
seen with increasing oxygen concentrations when sputtering the
fully oxidized surface has also been reported for NiO(100) films
on Ni(100).62 Apparently, sputtering of the NiO surface at room
temperature leads to oxygen depletion and to a surface local
reconstruction to the ordered Ni-O phases. It could therefore
be argued that the oxygen overlayers that form during the
oxygen uptake on Ni(110) above room temperature are those
favored thermodynamically and that sputtering of the NiO thin
films not only removes some oxygen atoms from the surface
but also provides enough energy for the remaining surface
species to rearrange into their most stable configuration. Having
said that, it is interesting to note that the long-range spread of
oxygen concentrations within the nickel crystal after the
sputtering seems to be particularly high, as suggested by the
bottom spectrum of Figure 11, where CO desorption in both
<150 and 300-400 K regimes is seen, indicating the coexist-
ence of NiO and clean Ni patches on the surface. This
inhomogeneity in surface oxygen coverage may be due either
to inhomogeneities in the sputtering beam profile or to a
selective sputtering of O atoms in some areas.
The next experimental result to be explained is the fact that
each ordered structure appears to induce a distinct adsorption
state for CO, as indicated by the well-defined peaks seen in the
CO TPD traces. These states represent significantly different
adsorption energies: by using a simple Redhead analysis58 and
a preexponential factor of 1015 s-1, the peaks at 120, 190, 240,
300, and 400 K are estimated to correspond to activation
energies of about 8, 13, 17, 21, and 28 kcal/mol, respectively.
It can be speculated that the CO desorption below 200 K seen
for both high CO coverages on the (2 × 1)O surface (Figure
7a) and all CO coverages on the (3 × 1)O surface (Figure 7b)
are due to desorption from Ni atoms in the added rows. The
CO desorbing above 200 K seen for low CO coverages on the
(2 × 1)O case (Figure 7a) might then stem from Ni atoms in a
more metallic environment, namely, those between the added
rows. The coverage evolution of the different states with oxygen
dose supports this assignment: by assuming a 1:1 CO:Ni
stoichiometry in all cases, the 190 K CO TPD state would be
expected to amount to 0.50 and 0.67 ML for the (2 × 1)O and
(3 × 1)O surfaces, and they were found experimentally to be
approximately 0.5 and 0.6 ML, respectively. It is, however,
difficult to analyze these TPD data in a quantitative way, since
it is known from STM16 and LEED31 investigations that local
fluctuations in the oxygen concentration on the surface may
lead to the coexistence of several Ni-O phases. Also, there
are nonequivalent adsorption sites within and at the end of Ni-O
rows. The end sites have in fact been found to be a key factor
in determining the chemical reactivity of this surface.59 Perhaps
some of the CO desorption around 240 K in the (2 × 1)O case
could be associated with those row-end states. Lastly, CO
mobility could in itself create different desorption states as the
Finally, the possibility of CO2 formation from adsorbed CO
and surface oxygen will be briefly addressed. Even though CO
+ O surface recombination is well-known on many late
transition metals,64-67 no significant formation of CO2 could
be detected either on partially oxidized surfaces or on NiO by
the TPD experiments presented in this study. CO desorbs
molecularly before the oxidation step can take place, most likely
because oxygen adsorption is particularly strong on nickel
surfaces.
5. Conclusions
The adsorption of carbon monoxide on different oxygen-
treated Ni(110) surfaces was characterized by TPD, XPS, and

9076 J. Phys. Chem. B, Vol. 101, No. 44, 1997
O¨ fner and Zaera
(22) Cappus, D.; Xu, C.; Ehrlich, D.; Dillmann, B.; C. A. Ventrice, J.;
Shamery, K. A.; Kuhlenbeck, H.; Freund, H.-J. Chem. Phys. 1993, 177,
533.
(23) Cappus, D.; Hassel, M.; Neuhaus, E.; Heber, M.; Rohr, F.; Freund,
H.-J. Surf. Sci. 1995, 337, 268.
(24) Demuth, J. E.; Rhodin, T. N. Surf. Sci. 1974, 45, 249.
(25) Holloway, P. H.; Hudson, J. B. Surf. Sci. 1974, 43, 123.
(26) Brundle, C. R. In The Chemical Physics of Solid Surfaces and
Heterogeneous Catalysis; King, D. A., Woodruff, D. P., Eds.; Elsevier:
Amsterdam, 1990; Vol. 3A (Chemisorption Systems), pp 132-388.
(27) Norton, P. R.; Tapping, R. L.; Goodale, J. W. Surf. Sci. 1977, 65,
13.
(28) Wolkenstein, T. AdV. Catal 1960, 12, 189.
(29) Badyal, J. P. S.; Zhang, X.; Lambert, R. M. Surf. Sci. 1990, 225,
L15.
(30) Mitchell, D. F.; Sewell, P. B.; Cohen, M. Surf. Sci. 1977, 69, 310.
(31) Norton, P. R.; Bondner, P. E.; Jackman, T. E. Surf. Sci. 1986, 175,
313.
(32) Caputi, L. S.; Chiarelle, G.; Amoddeo, A. Surf. Sci. 1987, 188, 63.
(33) Holloway, P. H.; Outlaw, R. A. Surf. Sci. 1981, 111, 300.
(34) Handbook of X-ray Photoelectron Spectroscopy; Wagner, C. D.,
Riggs, W. M., Davis, L. E., Moulder, J. F., Muilenberg, G. E., Eds.; Perkin-
Elmer Corp.: Eden Prairie, MN, 1978.
(35) Kuhlenbeck, H.; Odo¨rfer, G.; Jeager, R.; Illing, G.; Menges, M.;
Mull, T.; Freund, H.-J.; Po¨hlchen, M.; Staemmler, V.; Witzel, S.; Scharf-
schwerdt, C.; Wennemann, K.; Liedtke, T.; Neumann, M. Phys. ReV. B
1991, 43, 1969.
(36) Langell, M. A.; Nassir, M. H. J. Phys. Chem. 1995, 99, 4162.
(37) de Jesu´s, J. C.; Pereira, P.; Carrazza, J.; Zaera, F. Surf. Sci. 1996,
369, 217.
ISS. As reported before, the oxidation of the clean Ni(110)
single-crystal surface was found to start with the formation of
a series of well-ordered oxygen submonolayers with (3 × 1),
(2 × 1), and (3 × 1) periodicity and to reach the buildup of a
3 ML NiO(100) film after an exposure to about 300 langmuirs
of O2 at 450 K. The adsorption of CO was shown to depend
strongly on the nature of the O/Ni surface: distinct desorption
states were identified by TPD peaks around <150, 190, 240-
300, and 400 K associated with the NiO, (3 × 1)O, (2 × 1)O,
and clean Ni(110) surfaces, respectively. CO adsorption on the
NiO thin films was found to take place on the nickel atoms, to
display an initial sticking coefficient of about half of that seen
on the clean nickel surface, and to reach saturation at a coverage
of approximately 0.4 ML. Argon ion sputtering of the NiO
films leads to the removal of oxygen atoms and to the formation
of surface structures similar to the ordered phases seen during
the oxygen uptake, at least as far as CO adsorption is concerned.
Significant overall inhomogeneities are nevertheless seen after
ion bombardment, as manifested by the coexistence of NiO and
clean nickel phases. In a broad sense, CO was found to be an
excellent probe for the identification of different binding sites
on nickel oxide surfaces.
Acknowledgment. Financial support for this research was
provided by a grant from the Department of Energy, Basic
Energy Sciences Division, under Contract DE-FGO3-94ER14472.
H.O¨ . gratefully acknowledges additional financial support from
the Welch Foundation.
(38) Seah, M. P.; Dench, W. A. Surf. Interface Anal. 1979, 1, 2.
(39) Cappus, D.; Klinkmann, J.; Kuhlenbeck, H.; Freund, H.-J. Surf.
Sci. 1995, 325, L421.
(40) Behm, R. J.; Ertl, G.; Penka, V. Surf. Sci. 1985, 160, 387.
(41) Bertel, E.; Memmel, N.; Rangelov, G.; Bischler, U. Chem. Phys.
1993, 177, 337.
(42) Klier, K.; Zettlemoyer, A. C.; Leidheiser, H., Jr. J. Chem. Phys.
1970, 52, 589.
References and Notes
(1) Thomson, S. J.; Webb, G. Heterogeneous Catalysis; John Wiley
& Sons: New York, 1968.
(2) Haber, J. In Surface Properties and Catalysis by Non-Metals;
Bonnelle, J. P., Delmon, B., Derovane, E., Eds.; D. Reidel: Dordrecht,
1983; Vol. NATO ASI C105.
(3) Spencer, N. D.; Somorjai, G. A. Rep. Prog. Phys. 1983, 46, 1.
(4) Bielanski, A.; Haber, J. Oxygen in Catalysis; Marcel Dekker: New
York, 1991.
(5) Gates, B. C.; Katzer, J. R.; Schuit, G. C. A. Chemistry of Catalytic
Processes; McGraw-Hill: New York, 1979.
(6) Kung, H. H. Transition Metal Oxides: Surface Chemistry and
Catalysis; Elsevier: Amsterdam, 1989; Vol. 45.
(7) Sachtler, W. M. H.; Backx, C.; van Santen, R. A. Catal. ReV.sSci.
Eng. 1981, 23, 127.
(8) Deren, J.; Russer, B.; Nowotny, J.; Rog, G.; Sloczynski, J. J. Catal.
1974, 34, 124.
(43) Tjandra, S.; Zaera, F. Langmuir 1991, 7, 1432.
(44) Bertolini, J. C.; Tardy, B. Surf. Sci. 1981, 102, 131.
(45) Kisluik, P. J. Phys. Chem. Solids 1957, 3, 95.
(46) Ba¨umer, M.; Cappus, D.; Illig, G.; Kuhlenbeck, H.; Freund, H.-J.
J. Vac. Sci. Technol. A 1992, 10, 1407.
(47) de Jesu´s, J. C.; Carrazza, J.; Pereira, P.; Zaera, F. Surf. Sci., in
press.
(48) Vesecky, S. M.; Xu, X.; Goodman, D. W. J. Vac. Sci. Technol. A
1994, 12, 2114.
(49) Platero, E. E.; Coluccia, S.; Zecchina, A. Surf. Sci. 1986, 171, 465.
(50) Taglauer, E.; Heiland, W. Appl. Phys. Lett. 1974, 24, 437.
(51) Englert, W.; Taglauer, E.; Heiland, W. Surf. Sci. 1982, 117, 124.
(52) Gleason, N. R.; Zaera, F. Surf. Sci. 1997, 385, 294.
(53) Pacchioni, G.; Cogliandro, G.; Bagus, P. S. Surf. Sci. 1991, 255,
344.
(9) Miller, F. A.; Carlson, G. L.; Bentley, F. F.; Jones, W. H.
Spectrochim. Acta 1960, 16, 135.
(10) Hatano, M.; Otsuka, K. Inorg. Chim. Acta 1988, 146, 243.
(11) Zhang, X.; Ungar, R. K.; Lambert, R. M. J. Chem. Soc., Chem.
Commun. 1989, 473.
(12) Tong, Y.; Lunsford, J. H. J. Chem. Soc., Chem. Commun. 1990,
792.
(13) Komeda, T.; Sakisaka, Y.; Onchi, M.; Kato, H.; Masuda, S.; Yagi,
K. Surf. Sci. 1987, 188, 45.
(14) Ritter, E.; Behm, R. J. In The Structure of Surfaces II; van der
Veen, J. F., Van Hove, M. A., Eds.; Springer-Verlag: Berlin, 1988.
(15) Besenbacher, F.; Nørskov, J. K. Prog. Surf. Sci. 1993, 44, 5.
(16) Eierdal, L.; Besenbacher, F.; Laegsgaard, E.; Stensgaards, I. Surf.
Sci. 1994, 312, 31.
(17) Wang, W.-D.; Wu, N. J.; Thiel, P. A. J. Chem. Phys. 1990, 92,
2025.
(54) Allen, V. M.; Jones, W. E.; Pacey, P. D. Surf. Sci. 1989, 220, 193.
(55) Sanders, H. E.; Gardner, P.; King, R. A.; Morris, M. A. Surf. Sci.
1994, 304, 159.
(56) Fuggle, J. C. In Electron Spectroscopy: Theory, Techniques, and
Applications; Brundle, C. R., Baker, A. D., Eds.; Academic Press: London,
1981; Vol. 4, Chapter 2.
(57) van den Berg, J. A.; Verheij, L. K.; Armour, D. G. Surf. Sci. 1980,
91, 218.
(58) Redhead, P. A. Vacuum 1962, 12, 203.
(59) Ruan, L.; Stensgaard, I.; Lægsgaard, E.; Besenbacher, F. Surf. Sci.
1994, 314, L873.
(60) Langell, M. A. Surf. Sci. 1987, 186, 323.
(61) Kim, K. S.; Winograd, N. Surf. Sci. 1974, 43, 625.
(62) Kubiak, R.; Morgner, H.; Rakhovskaya, O. Surf. Sci. 1994, 321,
229.
(63) Gonzalez-Elipe, A. R.; Alvarez, R.; Olgado, J. P.; Espino, J. P.;
Munuera, G. Appl. Surf. Sci. 1991, 51, 13.
(64) Engel, T.; Ertl, G. In The Chemical Physics of Solid Surfaces and
Heterogeneous Catalysis; King, D. A., Woodruff, D. P., Eds.; Elsevier:
Amsterdam, 1982; Vol. 4 (Fundamental Studies of Heterogeneous Catalysis),
pp 73-93.
(65) Ertl, G. Surf. Sci. 1994, 299/300, 742.
(66) Nieuwenhuys, B. E. Surf. Sci. 1983, 126, 307.
(67) Zaera, F.; Liu, J.; Xu, M. J. Chem. Phys. 1997, 106, 4204.
(18) Pope, T. D.; Bushby, S. J.; Griffiths, K.; Norton, P. R. Surf. Sci.
1991, 258, 101.
(19) Ba¨umer, M.; Cappus, D.; Kuhlenbeck, H.; Freund, H.-J.; Wilhelmi,
G.; Brodde, A.; Neddermeyer, H. Surf. Sci. 1991, 253, 116.
(20) Saiki, R. S.; Kaduwela, A. P.; Sagurton, M.; Osterwalder, J.;
Friedman, D. J.; Fadley, C. S.; Brundle, C. R. Surf. Sci. 1993, 282, 33.
(21) Freitag, A.; Staemmler, V.; Cappus, D.; Ventrice, C. A., Jr.; Al
Shamery, K.; Kuhlenbeck, H.; Freund, H.-J. Chem. Phys. Lett. 1993, 210,
10.