17340
J. Phys. Chem. C 2007, 111, 17340-17345
Unravelling the Role of Steps in Cu2O Formation via Hyperthermal O2 Adsorption at
Cu(410)
Michio Okada,*,†,‡ Luca Vattuone,§ Andrea Gerbi,§ Letizia Savio,§ Mario Rocca,§
Kousuke Moritani,| Yuden Teraoka,| and Toshio Kasai†
Department of Chemistry, Graduate School of Science, Osaka UniVersity, 1-1 Machikaneyama-cho, Toyonaka,
Osaka 560-0043, Japan, PRESTO, Japan Science and Technology Agency, 4-1-8 Honcho Kawaguchi,
Saitama 332-0012, Japan, Dipartimento di Fisica dell’UniVersità di GenoVa and CNISM Unità di GenoVa,
Via Dodecaneso 33, GenoVa, Italy, Synchrotron Radiation Research Center, Japan Atomic Energy Agency,
1-1-1 Kouto, Mikazuki, Sayo, Hyogo 679-5148, Japan, and Dipartimento di Fisica dell’UniVersità di GenoVa
and IMEM-CNR, GenoVa, Italy
ReceiVed: June 11, 2007; In Final Form: September 7, 2007
We studied the oxidation of Cu(410) using high-resolution electron energy loss spectroscopy and X-ray
photoemission spectroscopy performed with synchrotron radiation. Cu2O formation starts above half a
monolayer oxygen coverage, and the oxidation rate is larger than for the parent low Miller index Cu(100)
surface. Open steps favor therefore the process by opening pathways for subsurface migration and oxygen
incorporation. Oxidation occurs only above 500 K when dosing O2 by backfilling, but the ignition temperature
can be lowered to room temperature by dosing O2 via a supersonic molecular beam at hyperthermal energy
indicating the opening of a different pathway. The oxidation rate is maximal at normal incidence; at grazing
incidence, it is higher when O2 impinges nearly normal to the (100) nanofacets than when it hits the surface
close to the normal to the step rises. A collision induced absorption mechanism can explain the experimental
findings.
Introduction
Since the discovery of high-Tc superconductivity in oxygendoped cuprate materials, the study of Cu-O bonding has
attracted enormous interest.1 In particular, cuprous oxide (Cu2O)
is regarded as a meaningful benchmark material for theories
and experiments. Cu2O, an industrially important direct-gap
semiconductor with a band gap energy of 2.17 eV, is considered
as one of the most promising materials for application in
photovoltaic cells.2-4 Both high carrier density and low leakage
currents are required to increase the performance in terms of
energy conversion in Cu2O-based photovoltaic devices. It is thus
necessary to improve the quality of the Cu2O films which are
usually produced by plasma deposition.4 The use of hyperthermal oxygen molecular beams (HOMB) may improve the quality
of the grown thin film as it is, e.g., the case for organic films
when the molecules are deposited via supersonic beams.5
Recently, some of us reported on HOMB-induced oxidation
of room-temperature Cu(100),6 Cu(111),7 and Cu(110)8 surfaces.
A collision-induced absorption (CIA) mechanism was proposed
both for Cu(100) and Cu(111), whereas an additional mechanism
involving mobile Cu adatoms was suggested for Cu(110).8 In
these studies, however, the role of defects in the oxidation
process, and especially in Cu2O formation, was neglected.
Unravelling their possible role is fundamental for a more
efficient fabrication of Cu2O thin films. Such information can
* To whom correspondence should be addressed. E-mail: okada@chem.
sci.osaka-u.ac.jp.
† Osaka University.
‡ PRESTO, Japan Science and Technology Agency.
§ Università Genova and CNISM Unità di Genova, + Università Genova
and IMEM-CNR.
| Japan Atomic Energy Agency.
be gained studying the oxidation of stepped Cu surfaces and
comparing the oxidation rate with the one found for low Miller
index surfaces. Furthermore, such information may shed light
on the role of Cu2O nanoparticles, an efficient catalyst for the
oxidation of CO.9
Herein we report a detailed high-resolution electron energyloss spectroscopy (HREELS) and X-ray photoemission spectroscopy (XPS) investigation on the initial oxidation and Cu2O
formation on Cu(410). The latter process is shown to take place
only at high temperature for O2 exposure at thermal energy;
the efficiency of Cu2O formation is, however, quite low
(although higher than for Cu(100)) as demonstrated by XPS
measurements performed using synchrotron radiation (SR). The
efficiency increases strongly with O2 translational energy,
allowing the oxide to be produced even at room temperature
for 2.2 eV O2 beams. The characteristic signatures of Cu2O are
then clearly visible in O(1s) and valence-band XPS spectra.
Furthermore, we find Cu2O formation to be anisotropic with
respect to the polar direction of the incident HOMB, i.e., to
depend on whether the beam strikes the surface close to the
normal to terraces or to step rises, respectively (see the crystal
geometry in Figure 1).
Experimental Section
HREELS experiments were performed in an ultrahigh vacuum
apparatus equipped with a supersonic molecular beam located
in Genoa, Italy.10 Spectra were recorded in-specular at 60°
incidence and 2.7 eV electron energy. SR-XPS experiments were
performed with the surface reaction analysis apparatus (SUREAC
2000) constructed at BL23SU in SPring-8 in Japan.6-8,11 The
photon energy for recording XPS spectra was set to 890.4 eV.
The Cu(410) sample was cleaned by repeated cycles of 1-keV-
10.1021/jp074520t CCC: $37.00 © 2007 American Chemical Society
Published on Web 10/30/2007
Steps in Cu2O Formation
Figure 1. (a) HREEL spectra recorded in-specular after preparation
(I) (bottom: solid line) and (II) (upper: dotted line). The inset shows
the Cu(410) surface geometry. (b) SR-XPS spectra of O 1s after
preparation (I) (bottom: solid line) and (II) (upper: dotted line).
Ar+ sputtering and annealing at 870 K until no impurity was
detected by HREELS or by SR-XPS. Low-energy electron
diffraction (LEED) showed then the sharp pattern characteristic
of the stepped surface. The O2 molecules are deposited on Cu(410) either by backfilling or by HOMB seeded in He. The
estimated incident energy of the HOMB reads 2.2 eV at a nozzle
temperature of 1400 K. In a previous work7,11 some of us
measured the time-of-flight (TOF) spectra of the direct HOMB
for several nozzle temperatures and seeding ratios. The kinetic
energy of 2.2 eV of incident O2 was then estimated from the
previously determined relations between kinetic energy and
beam conditions (nozzle temperature and seeding ratio). This
value is however an approximate estimate affected by a large
error of (0.2 eV, since our time window of the gate function
for TOF is too broad to determine the precise stream velocity
and velocity distribution. The resulting error in the beam energy
determination is however systematic and does not affect our
qualitative conclusions. The HOMB was directed along the
surface normal for most measurements. Angular effects were
investigated for 30° incidence along respectively the highsymmetry directions [1h40] (on terrace) and [14h0] (on step rise).
The flux density of O2 molecules dosed by HOMB was
estimated experimentally12 to be 1.72 × 1015 molecules‚cm-2‚s-1
at the sample position.
Results and Discussions
Figure 1a shows HREELS spectra recorded after two different
preparation procedures for which O2 dosing is performed by
backfilling: (I) 108 L O2 dose at 573 K (solid line) and (II)
2250 L at 500 K (dotted line) (1L ) 1.33 × 10-4 Pa‚s). A
J. Phys. Chem. C, Vol. 111, No. 46, 2007 17341
single sharp loss is observed after preparation (I), which
corresponds to the c(2 × 2) structure at 0.5 ML (monolayer, 1
ML ) 1.58 × 1015 atoms‚cm-2 in ML of Cu(410)) oxygen
coverage previously investigated by Vlieg et al.13 by density
functional theory (DFT) and X-ray diffraction. In the accepted
structural model, the metal substrate is unreconstructed. Half
of the oxygen occupies the step-edge sites, organized in Cu-O
chains with the O nearly collinear with the Cu atoms, whereas
the other half sits at mid-terrace hollow sites. Two dipole active
vibrations would therefore be expected, whereas only one sharp
loss is observed by HREELS. This contradiction is only apparent
since recent theoretical calculations14 based on a five parameters
Morse potential model showed that the two losses differ in
energy by less than 1 meV and hence cannot be resolved.
After preparation II, two additional losses are observed at 78
and 19 meV, respectively. Although the latter is only a shoulder
of the elastic tail, its presence is evident when comparing the
spectrum with the one recorded after preparation I. The
appearance of these peaks indicates that Cu2O has formed, since
in bulk Cu2O two infrared-active modes are present at 18.1 and
75.5 meV15 and similar vibrations were reported also for Cu2O
films grown on Cu(111)16 and on Cu(110).17 However, much
larger exposures were needed for oxide formation on these
surfaces at similar crystal temperatures, whereas no oxide
formation was detected under UHV conditions for Cu(100).18
This indicates that the open steps favor the oxidation process.
It should be noted that no Cu2O losses are observed on
Cu(410) when the O2 exposure occurs by backfilling at roomtemperature even after doses greatly exceeding those of the
experiment of Figure 1.19
Figure 1b shows the O1s SR-XPS peaks measured after
preparations I and II, respectively. For preparation I, a broad
symmetric peak appears at a binding energy (EB) of 529.8 eV
(Gaussian width G ) 0.93 eV). It consists possibly of two
components corresponding to adsorption at step edge and midterrace. For preparation II, the O1s peak is, on the contrary, at
the slightly higher EB value of 529.9 eV and is narrower (G )
0.79 eV). The integrated intensity has increased by 17%
compared to that of preparation I. Additional O2 doses induce
no further growth of the O1s SR-XPS intensity. Since the O1s
XPS peak of bulk Cu2O is located at 530.2∼530.5 eV20 and
HREELS shows Cu2O formation, we conclude that the shift to
higher EB is due to cuprous oxide. A possible explanation for
the narrowing of the O1s signal is that the formation of oxide
patches influences also the EB value of the O adatoms in both
sites, shifting the peak globally by 0.1 eV. The 17% increase
in O coverage tells us that Cu2O formation has occurred in
islands and that the process is kinetically limited.21 The dipole
scattering cross section of Cu2O patches must be quite high since
HREELS shows a robust signal while the amount of oxide
formed is quite low. Thus, high pressures of O2 and high surface
temperatures are required for the growth of thick Cu2O films
under thermal O2 exposure conditions.
For nanoscale device technology, thermal O2 exposure at high
temperature and pressure is of limited use because of contamination problems. It is therefore mandatory to find new processes
allowing Cu2O films to be fabricated at room or even lower
temperatures. One possible way of overcoming the problem is
using HOMB, as recently demonstrated by some of us for the
low Miller index Cu surfaces.6-8
Figure 2a shows the evolution of the O1s spectral region for
2.2 eV HOMB doses performed on Cu(410) along the surface
normal and at room temperature. The O1s SR-XPS peak shows
now a binding energy EB ) 529.9 eV at a coverage of 0.43
17342 J. Phys. Chem. C, Vol. 111, No. 46, 2007
Okada et al.
Figure 2. (a) Evolution of O1s SR-XPS spectra for 2.2-eV HOMB exposure along the surface normal at room temperature. The Shirley background
has already been subtracted. (b) Line-shape analysis of the representative O1s SR-XPS spectra. The four components, corresponding to chemisorbed
O on Cu, Cu2O, CuO and chemisorbed O on Cu2O, are indicated by thick gray, thick dashed, thick solid, and thick dotted lines, respectively. The
background is indicated by the thin dashed line. (c) Valence-band SR-XPS spectra of Cu(410) clean (gray) and covered by 2.07-ML-O (black)
dosed by 2.2 eV HOMB at room temperature. The thin dashed line corresponds to the XPS spectrum of bulk Cu2O.20 The spectra are measured at
70° from the surface normal. The calculated valence band peak positions of Cu2O are marked by vertical bars at 7.95, 6.70, 2.78, and 1.29 eV.20
ML and increases in intensity and EB with increasing O
coverage. The line shape of the O1s peak depends also on O
coverage. Figure 2b shows the results of the peak-shape analysis
of representative O1s spectra of Figure 2a. The symmetric peak
of O1s at 0.43 ML is fitted with a Voigt function with
parameters G ) 0.79 eV and Lorentzian width Γ ) 0.32 eV. It
corresponds mainly to chemisorbed O atoms, because peak
position and shape are the same as for thermal O2 exposure at
room temperature where HREELS shows no Cu2O loss.22 At
1.05 ML, the O1s peak can be separated into three components;
chemisorbed O atoms at 530.0 eV (G ) 0.79 eV, Γ ) 0.32
eV), Cu2O (G ) 0.71 eV, Γ ) 0.21 eV) at 530.5 eV, and a
small CuO peak (G ) 0.71 eV, Γ ) 0.10 eV) at 529.2 eV.20 A
small amount of the latter phase was also reported for Cu2O
formation on Cu(110).17 The formation of a CuO phase strongly
depends on the surface temperature during the HOMB dose.23
Increasing the O coverage causes an increment of the Cu2O
and CuO peaks and induces the formation of an additional small
peak at 531.4 eV (G ) 0.80 eV, Γ ) 0.77 eV), possibly due to
chemisorbed O on Cu oxide.20 The relative weight of the
component due to oxygen adatoms (530.0 eV) decreases slowly
above 0.5 ML and is no longer present at 2.07 ML, indicating
that complete surface oxidation has occurred.
Cu2O formation is confirmed by inspection of the valenceband spectra in Figure 2c: after the 2.2 eV HOMB dose at room
temperature, the main feature gets narrower and becomes very
similar to the one of bulk Cu2O.20 Since at 900 eV photon energy
the photoionization cross-section ratio is σ(O 2p)/σ(Cu 3d) ∼
0.0324 the d spectral weight of Cu2O, concentrated at 1-4 eV,
dominates the valence band spectrum, whereas the O 2p
character around 6-7 eV is hardly visible. Since the spectra
are measured in the surface-sensitive condition, the prominent
decrease of the density of state (DOS) at the Fermi level
indicates the opening of a band gap and the formation of a Cu2O
thin film. We could not determine the exact value of this gap
because we have no data for the unoccupied states.
In Figure 3a, we compare the O uptake curve during 2.2eV-HOMB oxidation on Cu(410) with those for Cu(100)6 and
Cu(110).8 A clear dependence of the uptake curves on crystal
face symmetry is present for coverages exceeding 0.5 ML. The
uptake curve on Cu(410) falls in-between those of its constituent
low Miller index planes.
The sticking probability on Cu(100) decreases suddenly
around ΘO ∼ 0.5 ML and reads thereafter 10-5-10-6 corresponding to the Cu2O formation6 rate. The relatively low value
of S is consistent with the proposed CIA process.6 On the other
hand, S is by 1-2 orders of magnitude higher during the
formation of Cu2O on Cu(110) than on Cu(100).7 This difference
was speculated to be due to migrating Cu adatoms acting as
additional dissociation centers for the impinging O2.8
Steps in Cu2O Formation
Figure 3. (a) O uptake curves for HOMB at normal incidence on Cu(110) (open squares),8 Cu(410) (solid circles) and Cu(100) (open
triangles)6 at room temperature. Incident energies are 2.3 eV for (110)
and (100), and 2.2 eV for (410). (b) Dependence on incidence angle of
the O uptake curves for 2.2 eV HOMB on Cu(410) at room temperature.
The angle of incidence θ is either normal or at 30° from the surface
normal. Incidence near normally to (100) terraces (•, θ ) -30°) and
to step rises (O, θ ) +30°) (see inset) is denoted by solid and open
circles, respectively. Full triangles correspond to θ )0°.
In order to elucidate which mechanisms determine Cu2O
formation on Cu(410), we performed experiments dosing O2 at
off normal incidence (see inset of Figure 3b for details on the
angles of incidence). If oxide formation were to proceed via
detachment of copper atoms by impinging O2, the oxidation
rate should be largest when the molecules impinge from the
upper side of the step edge at large angles (corresponding to •
symbols in Figure 3b), i.e., when the momentum transfer is
directed toward the lower terrace. As evident from the data,
this scattering condition is indeed slightly more efficient for
oxidation than when the molecules collide near normally against
the step rise. However, the highest efficiency for oxide formation
is attained when O2 molecules impinge along the surface normal.
The angle dependence of S indicates therefore that CIA is the
main mechanism for oxide formation on Cu(410) by HOMB.
Detachment of Cu atoms plays only a minor role, contrary to
the case of HOMB oxidation of Cu(110) where it was suggested
to be important.8
In CIA what matters is the energy transfer to the oxygen
adatom in the direction pointing toward the subsurface region.
This argument naturally predicts the efficiency for collisions
to be highest along the surface normal, in agreement with
experimental observation. According to DFT calculations,
subsurface adsorption on Cu(100) is energetically more favored
than on-surface adsorption above 0.5 ML.25 For Cu(410) this
condition corresponds to 0.25 ML of adatoms at fourfold hollow
terrace sites and the remaining 0.25 ML at the step edges.
For an incidence angle θ ) -30° (off normal toward the
(100) nanoterrace) O2 molecules hit the pre-adsorbed fourfold
O moiety at nearly the same angle as for θ ) 0° (since the
surface normal to (100) nanoterraces is at -14°), whereas the
J. Phys. Chem. C, Vol. 111, No. 46, 2007 17343
adatoms at the step edge are pushed toward the bottom of the
(110) step against the underlying (100) nanoterrace. When
moving to θ ) +30° (off normal toward the step rise) the terrace
O moiety is hit at 45°, whereas the adatoms at step sites are
pushed against the step rise. The larger cross section for oxide
formation for θ ) -30° indicates therefore that subsurface
incorporation is easier when O adatoms move toward the
fourfold hollow below the step. This finding agrees with the
observation that on Ag(210) direct subsurface incorporation
occurs more easily for O2 molecules impinging normally to the
(100) than normally to the (110) nanofacet.26
The CIA mechanism is also supported by the fact that when
dosing at 0.5 eV no trace of oxide formation is detected in O1s
and valence-band SR-XPS spectra, leading to the conclusion
that the translational energy is not high enough to induce oxygen
incorporation.27
For HOMB exposures exceeding 5 × 1018 molecules/cm2,
the efficiency of Cu2O formation is independent of the incidence
polar direction: this result is reasonable since for such large
exposures the coverage exceeds 1 ML and the oxide covers a
large fraction of the surface. At this stage, memory of the initial
stepped geometry is lost and further oxide formation occurs by
growth and coalescence of already existing copper oxide islands
rather than by nucleation of new oxide patches. The oxide free
areas are moreover in the valleys between oxide islands and
the direction of impingement of the oxygen molecules is no
longer relevant. No ordered patterns were observed by LEED
inspection upon such high HOMB exposures.
Coming back to the comparison between (410) and low Miller
index surfaces, we note that the hyperthermal oxidation mechanism, whatever its origin, is more efficient for Cu(110) than
for Cu(100) and Cu(111); since Cu(410) consists of (100)
terraces separated by (110) steps it is not surprising that its
oxidation rate falls in between. In fact, taking into consideration
the density of step O atoms, the efficiency of Cu2O formation
on O-Cu(410) comes out to be comparable to that of the added
row reconstructed O-Cu(110). Indeed O atoms at the step edges
of Cu(410) form Cu-O- chains similar to the ones present on
added rows of O-Cu(110). The reduced efficiency for the
closed packed (111) surface, for which subsurface incorporation
is likely to be more difficult, further supports the CIA mechanism. Conversely we notice that the opposite holds true for
oxidation under thermal dosing conditions for which Cu(111)
shows the highest oxidation rate.28 The mechanism underlying
HOMB oxidation is thus entirely different from the one
operating under thermal conditions, which is dominated by
surface diffusion of the oxygen adatoms.29
Rate Equation Model. When O2 collides with an oxygen
adatom it has a certain probability to push it subsurface. Three
different fates are then possible for the incoming O2 molecule:
(a) it may desorb;
(b) it may dissociate leaving one atom at the surface and
ejecting the second into the vacuum,30 a process which is
energetically possible for hyperthermal molecules;
(c) it may dissociate leaving both atoms at the surface.
In case a, since the subsurface location is unstable the atom
can either return to the surface (and in this case the initial
situation is restored) or move to another subsurface site below
another O adatom. In this case an oxide nucleus (Oad + Osub)
is formed and the number of adatoms not having a subsurface
companion decreases by two units.
In case b, an oxide nucleus (Oad + Osub) is formed and the
number of adatoms without subsurface companion decreases
by one unit.
17344 J. Phys. Chem. C, Vol. 111, No. 46, 2007
Okada et al.
Figure 4. Comparison between the experimental total coverage of O atoms vs O2 dose (at 2 eV for normal and off normal dosing conditions) and
the total coverage obtained by numerically solving the rate equations model described in the text.
In case c, an oxide nucleus (Oad + Osub) is formed by the
adatom pushed subsurface and by one of the atoms produced
by dissociation. Since the second atom arising from the
dissociation cannot stay at the surface (because the coverage
of oxygen atoms would then exceed 0.5 ML), it can only move
subsurface below another (already present) O adatom. The net
balance is then the formation of two oxide nuclei and the loss
of two sites with adatoms without subsurface companion.
If n is the number of new oxide nuclei being formed and q
is the number of lost adatoms, it is possible to model semiquantitatively the experimental uptake curves using rate equations.
dΘO/dt ) 2ΦSadatom - qΦPoxideΘO
dΘoxide /dt ) 2nΦPoxideΘO
where Sadatom ) S0(1 - ΘO/ΘO sat - Θoxide/Θmax)m
where ΘO and Θoxide are, respectively, the coverage of oxygen
adatoms (without subsurface companion) and of atoms in the
oxide, Φ the beam flux, SO the O2 initial sticking probability,
ΘO sat the O saturation coverage (0.5 ML for Cu(410)), Poxide
the probability that an O atom ends up subsurface in a moleculeadatom collision.
The factor 2 in the second formula arises by the definition
of the oxide nucleus as a Oad + Osub couple, so that the number
of oxygen atoms is twice the number, n, of such couples.
The factor (1 - ΘO/ΘO sat - Θoxide/Θmax)m accounts for two
effects:
(a) the sticking probability for on surface (adatom) sites
vanishes at ΘO sat
(b) when oxide growth starts some surface sites are occupied
by oxide nuclei and are thus no longer available for new
adatoms.
The free parameter Θmax (larger than 0.5 ML, typically 2.1
ML) accounts for the non-layer-by-layer growth of the oxide:
its value determines the total coverage for which no free Cu
sites are available for further oxygen adsorption. At Θmax, oxide
formation by CIA stops and another mechanism is needed to
describe the growth of a thicker oxide layer. Due to the euristic
nature of the present model, Θmax has to be included as an input
of the model itself; a much more sophisticated modeling would
be necessary to account for the description of the island growth
mechanism in three dimensions and the parameters needed to
build it (diffusion coefficients on the substrate and at the edges
of the islands, interaction energy between particles etc. could
be obtained only by ab initio methods or possibly and partially
also by extensive analysis of experimental data).
The exponent m accounts for the rapidity of the decrease of
the sticking probability just below 0.5 ML: m ) 2 would be
expected for dissociative adsorption yielding noninteracting
adatoms. In our model this parameter is obtained by fitting the
decrease of the sticking probability with coverage below 0.5
ML and reads ≈4, in agreement with the existence of strong
lateral repulsive interactions between oxygen adatoms. The latter
are responsible also for subsurface site occupation becoming
energetically favored above 0.5 ML.
We note that for mechanism (a) both a significant temperature
and flux dependences are expected since the atoms pushed
subsurface can diffuse back to the surface if the vacancy is not
rapidly filled by a further adatom produced by dissociating
another incoming O2. On the contrary, in cases (b) and (c) a
purely collisional mechanism is operative so that neither flux
nor temperature dependence is expected. At the temperatures
of the present experiment, recombinative desorption is not
possible and is thus disregarded.
Figure 4 shows the coverage vs exposure obtained for cases
(a), (b), and (c). It shows a semiquantitative agreement with
the data thus proving that the model captures the essential
physics although it does not allow to determine unambiguously
which is the actual path followed. Indeed the same curve can
be obtained, e.g., by mechanism (c) with Poxide ) 3 × 10-4 or
by mechanism (a) with Poxide ) 6 × 10-4. The model predicts
a lower growth rate than experiment below 1 ML and a faster
growth rate above it. The existence of an asymptote in the oxide
thickness is supported by the decrease in the oxide growth rate
above 1.5 ML. Recent unpublished experiments on Cu(511),31
for which longer exposures were performed, indicate that a
limiting coverage is indeed approached, thus supporting the
existence of a limiting thickness for the oxide film grown by
CIA.
Steps in Cu2O Formation
The data at grazing incidence seem to be better described by
a lower value of Θmax (≈1.6 ML instead of 2.1 ML). Although
such a conclusion would require performing much longer O2
exposures, we note that since Θmax is a phenomenological
parameter accounting for the morphology of the oxide islands,
a dependence on the angle of incidence is not too surprising.
The model also predicts that at a total O coverage of 1 ML
a significant fraction of adatoms is still present and that their
contribution should disappear close to 2 ML. The prediction is
confirmed by experiment (Figure 2b): for a 2.0 ML coverage
the contribution of O adatoms to the XPS spectra (at 530.0 eV)
is no longer visible, while a significant fraction is still present
at 1.0 ML. At 2.0 ML, all adatoms have been used up to form
Cu2O (component at 530.4 eV) or have been turned into oxygen
adatoms on copper oxide (component at 531.4 eV).
The simulation shows moreover that uptake curves following
closely the experimental data can be obtained for Poxide ≈ 3 ×
10-4, corresponding to a sticking probability of 3 × 10-4 at
0.5 ML. This value is about 1 order of magnitude larger than
for Cu(100) (10-5, see ref 6) and only slightly smaller, than for
Cu(110).6 Since it does not scale with the relative area of (110)
and (100) nanofacets, the result indicates unambiguously that
the presence of the defect enhances the sticking probability in
the oxidation region.
Conclusions
In summary, we demonstrated by SR-XPS and by HREELS
that Cu2O thin films grow more easily on Cu(410) than on Cu(100) both when dosing by backfilling and when exposing to
HOMB. In the former case the growth process is determined
by the availability of Cu atoms, which may detach via thermal
activation from the open step or by the availability of channels
for subsurface migration at the step. In the latter case oxide
formation is caused by collision induced subsurface incorporation of preadsorbed oxygen adatoms for Ei ≈ 2.2 eV and occurs
also at room temperature.
Acknowledgment. The authors are thankful to Y. Saitoh
and S. Fujimori for their help with the operation of the
monochromatic system at the beam line. The synchrotron
radiation experiments were performed at the BL23SU
(SUREAC2000) in the SPring-8 with the approval of JAEA as
Nanotechnology Support Project of the Ministry of Education,
Culture, Sports, and Technology (MEXT) (Proposal No.
2005B0049 and 2006A1609). M.O. gratefully acknowledges the
Hyogo Science and Technology Association, and also MEXT
for a Grant-in-Aid for Scientific Research (No. 17550011). The
Genova group thanks Compagnia san Paolo for financial
support.
J. Phys. Chem. C, Vol. 111, No. 46, 2007 17345
References and Notes
(1) Waldram, J. R. SuperconductiVity of Metals and Cuprates; IOP
Publishing Ltd.: London, UK, 1996.
(2) Olsen, L. C.; Addis, F. W.; Miller, W. Solar Cells 1982/1983, 7,
247.
(3) Pollack, G. P.; Trivich, D. J. Appl. Phys. 1975, 46, 163.
(4) Okamoto, Y.; Ishizuka, S.; Kato, S.; Sakurai, T.; Fujiwara, N.;
Kobayashi, H.; Akimoto, K. Appl. Phys. Lett. 2003, 82, 1060.
(5) Casalis, L.; Danisman, M. F.; Nickel, B.; Bracco, G.; Toccoli, T.;
Iannotta, S.; Scoles, G. Phys. ReV. Lett. 2003, 90, 206101.
(6) Okada, M.; Moritani, K.; Goto, S.; Kasai, T.; Yoshigoe, A.; Teraoka,
Y. J. Chem. Phys. 2003, 119, 6994.
(7) Moritani, K.; Okada, M.; Goto, S.; Sato, S.; Kasai, T.; Yoshigoe,
A.; Teraoka, Y. J. Vac. Sci. Technol. A 2004, 22, 1625.
(8) Moritani, K.; Okada, M.; Fukuyama, T.; Yoshigoe, A.; Teraoka,
Y.; Kasai, T. Eur. Phys. J. D 2006, 38, 111.
(9) White, B.; Yin, M.; Hall, A.; Le, D.; Stolbov, S.; Rahman, T.; Turro,
N.; O’Brien, S. Nano Lett. 2006, 6, 2095.
(10) Rocca, M.; Valbusa, U.; Gussoni, A.; Maloberti, G.; Racca, L. ReV.
Sci. Instrum. 1991, 62, 2172.
(11) Moritani, K.; Tsuda, M.; Teraoka, Y.; Okada, M.; Yoshigoe, A.;
Fukuyama, T.; Kasai, T.; Kasai, H. J. Phys. Chem. C 2007, 111, 9961.
(12) Teraoka, Y.; Yoshigoe, A. Jpn. J. Appl. Phys. 2002, 41, 4253.
(13) Vlieg, E.; Driver, S. M.; Goedtkindt, P.; Knight, P. J.; Liu, W.;
Lüdecke, J.; Mitchell, K. A. R.; Murashov, V.; Robinson, I. K.; de Vries,
S. A.; Woodruff, D. P. Surf. Sci. 2002, 516, 16.
(14) Wang, Z. X.; Tian, F. H. J. Phys. Chem. B 2003, 107, 6153.
(15) Dawson, P.; Hargreave, M. M.; Wilkinson, G. R. J. Phys. Chem.
Solids. 1973, 34, 2201.
(16) Dubois, L. H. Surf. Sci. 1982, 119, 399.
(17) Baddorf, A. P.; Wendelken, J. F. Surf. Sci. 1991, 256, 264.
(18) Sexton, B. A. Surf. Sci. 1979, 88, 299.
(19) Okada, M.; Vattuone, L.; Moritani, K.; Savio, L.; Teraoka, Y.;
Kasai, T.; Rocca, M. J. Phys.: Condens. Matter 2007, 19, 305022.
(20) Ghijsen, J.; Tjeng, L. H.; van Elp, J.; Eskes, H.; Czyzyk, M. T.
Phys. ReV. B 1988, 38, 11322.
(21) Zhou, G.; Yang, J. C. Surf. Sci. 2003, 531, 359.
(22) for a more detailed report on HREELS investigation of O/Cu(410)
see: Vattuone, L.; Savio, L.; Okada, M.; Moritani, K.; Rocca, M. J. Phys.
Chem. B 2007, 111, 1679.
(23) Okada, M.; Vattuone, L.; Moritani, K.; Savio, L.; Teraoka, Y.;
Kasai, T.; Rocca, M. Phys. ReV. B 2007, 75, 233413.
(24) Yeh, J. J.; Lindau, I. Atomic Data Nuclear Data Tables 1985, 32,
1.
(25) Kangas, T.; Laasonen, K.; Puisto, A.; Pitkänen, H.; Alatalo, M.
Surf. Sci. 2005, 584, 62.
(26) Vattuone, L.; Savio, L.; Rocca, M. Phys. ReV. Lett. 2003, 90,
228302.
(27) Vattuone, L.; Gambardella, P.; Cemic, F.; Valbusa, U.; Rocca, M.
Chem. Phys. Lett. 1997, 278, 245. Vattuone, L.; Gambardella, P.; Burghaus,
U.; Cemic, F.; Valbusa, U.; Rocca, M. J. Chem. Phys. 1998, 109, 2490.
(28) Zhou, Ph. Thesis, University of Pittsburgh: Pittsburgh, PA, 2003;
p 106.
(29) Zhou, G.; Yang, J. C. Phys. ReV. Lett. 2002, 89, 106101.
(30) Pazzi, V. I. P.; Philipsen, P. H. T.; Baerends, E. J.; Tantardini, G.
F. Surf. Sci. 1999, 443, 1.
(31) M. Okada et al., unpublished.