Different CO2 collimation on stepped Pt(1 1 2): A comparison of NO(a) + CO(a) and O(a) + CO(a) reactions

Article abstract of DOI:10.1016/S0039-6028(02)02597-9

The collimation of the desorbing product CO₂ is very different in the NO(a) + CO(a) and O(a) + CO(a) reactions on Pt(1 1 2) = [(S)3(1 1 1) × (0 0 1)]. In the NO(a) + CO(a) reaction, CO₂ desorption collimated along the local normal of the (0 0 1) facets, whereas in the O(a) + CO(a) reaction, it sharply collimated close to the (1 1 1) terrace normal. This site switching is explained by different rate-determining steps.

Full text of DOI:10.1016/S0039-6028(02)02597-9

Surface Science 526 (2003) 159–165
www.elsevier.com/locate/susc
Different CO2 collimation on stepped Pt(112): a comparison
of NOðaÞ þ COðaÞ and OðaÞ þ COðaÞ reactions
Yu-Hai Hu , SongHan , Hideyuki Horino , Bernard Egbert Nieuwenhuys
a
b
b
a,c
,
a
Atsuko Hiratsuka , Yuichi Ohno , Kobal Ivan , Tatsuo Matsushima
a
d
a,
*
a
Catalysis Research Center, Hokkaido University, Sapporo 060-0811, Japan
Division of Material Science, Graduate School of Environmental Earth Science, Hokkaido University, Sapporo 060-0810, Japan
b
c
Leiden Institute of Chemistry, Gorlaeus Laboratories, Leiden University, Einsteinweg 55, P.O. Box 9502,
2300RA Leiden, The Netherlands
Jozef Stefan Institute, 1000 Ljubljana, Slovenia
d
Received 29 August 2002; accepted for publication 18 November 2002
Abstract
The collimation of the desorbingproduct CO 2 is very different in the NO(a) þ CO(a) and O(a) þ CO(a) reactions on
Ptð112Þ ¼ ½ðSÞ3ð111Þ Â ð001ފ. In the NO(a) þ CO(a) reaction, CO2 desorption collimated alongthe local normal of
the (0 0 1) facets, whereas in the O(a) þ CO(a) reaction, it sharply collimated close to the (1 1 1) terrace normal. This site
switchingis explained by different rate-determiningsteps.
Ó 2002 Elsevier Science B.V. All rights reserved.
Keywords: Angle resolved DIET (including electron stimulated desorption ion angular distribution (ESDIAD)); Surface chemical
reaction; Platinum; Carbon monoxide; Carbon dioxide; Nitrogen molecule; Nitrogen oxides
1. Introduction
Ptð112Þ ¼ ½ðSÞ3ð111Þ Â ð001ފ. The collimation
angle of desorbing CO2 is very different from that
found for the O(a) þ CO(a) reaction, whereas that
of N2 desorption is not changed by the presence of
CO(a).
The spatial distributions of desorbingproducts
provide structural information on their formation
site when the products carry a high excess of
translational energy, even if their emission is not
rate-determining[1]. For example, the distribution
in the CO oxidation changes largely when the
rate-determiningstep switches over [2]. This pa-
per is the first to report the angle-resolved
The spatial distribution of the desorbingCO
2
produced in the O(a) þ CO(a) reaction has already
been reported for various platinum surfaces in-
cludingstepped structures based on angle-resolved
thermal desorption spectroscopy (AR-TDS) and
angle-resolved steady-state desorption measure-
ments [2,3]. Reactive CO2 desorption always col-
limates closely alongthe flat terrace normal,
indicatingthe presence of active oxyegn atoms
on flat (1 1 1) terraces. On Ptð113Þ ¼ ½ðSÞ2ð111Þ Â
ð001ފ, CO2 desorption collimated close to the
measurements of the desorbingproducts CO and
2
N2 in the NO(a) þ CO(a) reaction on stepped
* Correspondingauthor. Fax: +81-11-706-3695.
E-mail address: tatmatsu@cat.hokudai.ac.jp (T. Matsu-
shima).
0039-6028/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved.
PII: S0039-6028(02)02597-9

160
Y.-H. Hu et al. / Surface Science 526 (2003) 159–165
(0 0 1) step normal. Originally, this result was in-
terpreted in terms of reactive desorption from one-
atom high step sites [4]. However, later LEED [5]
and STM [6] results showed that the surface
was reconstructed into the ð1 Â 2Þ structure as
½ðSÞ3ð111Þ Â 3ð001ފ.
[001]
-35.3
[112]
[111]
19.5
0
˚
˚
˚
The difficulty in observingreactive CO
de-
2
sorption from one-atom high step sites is due to
the low reactivity of oxygen on the steps. Oxygen
molecules preferentially dissociate on step sites or
structural defects because of the higher binding
energy [7,8]. The resultant oxygen atom is likely to
be removed as CO2 after movingto the flat terrace
area. This is reasonable when O(a) is distributed
on both step and flat terrace sites because mobile
CO(a) [9], which has a surface residence time in
the millisecond order [10], can visit oxygen on
both sites and interact with more reactive species
at ordinary steady-state reaction conditions [11].
Thus, reactive CO2 desorption from the one-atom
high step sites is expected when most of the O(a) is
located on the steps. Such conditions are expected
for the NO(a) þ CO(a) reaction because NO(a)
dissociation takes place on the steps at tempera-
tures high enough to proceed the CO oxidation
quickly, i.e., NO dissociation is rate-determining,
beingfollowed by the fast removal of N(a) and
O(a) [12–14].
_
[111]
Fig. 1. A side view of Pt(1 1 2).
normal, i.e., the [1 1 1] direction, is þ19.5° and the
(0 0 1) step normal is )35.3°.
The crystal was cleaned by repeated cycles of
Arþ bombardments and heatingin oxygen. After
flashingto 1200 K, the surface showed a sharp
ð1 Â 1Þ LEED pattern without higher-order spots.
No reactive oxygen was left on the surface as
judged from the absence of CO2 formation in the
TDS after CO exposure. 15N16O and 13C16O were
introduced through separate gas dosers when the
surface temperature ðTSÞ was down to 200 K.
Hereafter, the isotope 15N and 13C are mostly
designated as N and C in the text because only
13C16O (mass/charge ratio ¼ 29), 15N2 (30), 15N16O
(31), 16O2 (32) and 13C16O2 (45) were desorbed in
the subsequent heatingprocedures. The coverage
of each species was defined as the AI-TDS spec-
trum peak area relative to its maximum value. In
TDS experiments, the surface was heated at a rate
of 10 K/s.
2. Experimental
A UHV system with three chambers was used
[15]. The reaction chamber was equipped with fa-
cilities for LEED-AES, an Arþ gun, and a mass
spectrometer for angle-integrated (AI) desorption
analysis. The collimator house had a slit on each
end and the analyzer had another mass spec-
trometer for AR-TDS measurements. The crystal
was set on the top of a rotatable manipulator to
change the desorption angle (h, polar angle). This
angle was scanned in the normally directed plane
perpendicular to the surface troughs because the
CO2 reactive desorption was concentrated in this
plane [1–3]. Here, the sign of the desorption angle
is defined positive in the upward direction of the
steps (Fig. 1). In this definition, the (1 1 1) terrace
3. Results
3.1. General features
N2 desorption was observed above 400 K and
yielded a single c-N2 peak at around 500 K in TDS
experiments after the clean surface was exposed to
only NO (Fig. 2(a)) [16]. In the presence of CO(a),
it was sharply enhanced at around 425 K (d-N2),
indicatingthat removal of O(a) by CO is effective
for NO decomposition to proceed. A concomitant

Y.-H. Hu et al. / Surface Science 526 (2003) 159–165
161
1.8
1.5
1.2
0.9
0.6
0.3
0.0
K. Above this CO coverage, N2 desorption shifted
to higher temperatures, and the c-N2 peak at
around 500 K was suppressed as well.
ΘΝΟ=0.52
ΘCΟ
0.30
(a)
3.2. CO2 angular distribution
0.20
0.10
The AR-spectra of both N2 and CO2 are sen-
sitive to the desorption angle. The desorption of
both species was sharply collimated around
h ¼ À30°.
δ N2
–
0.05
0.01
0.003
0
γ
–N2
Typical CO2 spectra in the AR-form observed
from HNO ¼ 0:52 and HCO ¼ 0:05 are summarized
in Fig. 3(a). The signal was maximized at around
h ¼ À30° and mostly suppressed in the posi-
tive desorption angle region. The peak height at
425 K is plotted against the desorption angle in
the polar coordinates in Fig. 3(b). CO2 desorption
3.0
3.5
4.0
5.0
5.5
6.0
Ts /41.05 0 K
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
ΘΝΟ=0.52
ΘCO
(b)
0.30
0.20
0.10
ΘNO
=0.52
θ
10
8
(a)
-60o
Θ
CO=0.05
δ CO
–
2
0.05
0.01
-50o
0.003
0
6
3.0
3.5
4.0
4.5
5.0
5.5
6.0
-40o
Ts /100 K
δ—CO
2
4
-30o
-10o
Fig. 2. Typical AI-TDS spectra of (a) 15N2 and (b) 13CO2 in a
15NO(a) þ 13CO(a) reaction at various 13CO coverages. The
surface was exposed to 15NO to the amount of HNO ¼ 0:52 at
200 K and further to 13CO in various amounts. The apparent
peak areas of 13CO2 were much fewer than those of 15N2 be-
cause of the large difference in the pumping rate.
2
0o
10o
90o
0
3.0
3.5
4.0
4.5
5.0
5.5
6.0
T /100 K
s
Step down
Up
θ
narrow CO2 desorption peak (d-CO2) was ob-
served in the same temperature range (Fig. 2(b)),
-20o
20o
[111]
40o
-40o
0o
[112]
[001]
confirmingthe quick removal of O(a) as CO
.
2
(b)
Desorption of non-reacted oxygen was found
above 700 K when the amount of CO(a) was small.
CO desorption was observed only when the rela-
tive CO coverage was higher than 0.10, being
consistent with the fast reaction of CO(a) with
O(a) above 400 K [17].
60o
80o
-60o
-80o
cos7(
θ
+ 30)
δ—CO
2
The shape of the N2 desorption curve at
HNO ¼ 0:52 was sensitive to the amount of CO(a)
below HCO ¼ 0:10, i.e., a small increment of ad-
sorbed CO can enhance N2 desorption around 420
Fig. 3. (a) AR-TDS of 13CO2 at different desorption angles in
the 15NO(a) þ 13CO(a) reaction at HNO ¼ 0:52 and HCO ¼ 0:05
and (b) angular distribution of 13CO2 in the 15NO(a) þ 13CO(a)
reaction at the above condition.

162
Y.-H. Hu et al. / Surface Science 526 (2003) 159–165
collimated sharply along h ¼ À30°. The signal can
be approximated as cos7 ðh þ 30Þ displayed by the
solid curve. The collimation angle was estimated to
be À30 Æ 2°. It is shifted only about 5° from the
[0 0 1] direction. The reactive desorption is likely to
proceed on (0 0 1) facets.
This is very different from the desorption of
CO2 in the O(a) þ CO(a) reaction on Pt(1 1 2) in
our previous work [17]. For a direct comparison,
the angular distribution of desorbing CO2 in the
O(a) þ CO(a) reaction was also examined after the
experiments described above. Typical AR-CO2
spectra are reproduced in Fig. 4(a), where the
surface was initially covered to HO ¼ 0:40 and
then to HCO ¼ 0:30. The CO2 formation started
already at around 210 K and was completed at
around 470 K, yieldingthree desorption peaks, b3-
CO2 at 260 K, b2-CO2 at 310 K and b1-CO2 at
around 400 K. The sharp b2-peak was maximized
around h ¼ þ10° as shown in Fig. 4(b). The signal
intensity can be approximated as cos13 ðh À 10Þ.
On the other hand, b3-CO2 was maximized at
h ¼ þ5° and followed a cos20 ðh À 5Þ distribution.
The broad peak of b1-CO2 was maximized at
around h ¼ þ10° and approximated as cos12 ðh À
10Þ. These results agree well with our previous
work [17]. CO2 is formed on the terrace since its
desorption is closely collimated alongthe (1 1 1)
terrace normal.
3.3. N2 angular distribution
The AR-spectra of N2 at different desorption
angles induced from HNO ¼ 0:52 and HCO ¼ 0:05
are summarized in Fig. 5(a). The signal was max-
imized at around h ¼ À30°. It should be noted
that there is a significant signal intensity at the
desorption angle of h ¼ 90°. This is due to N2
molecules that did not pass through the slits di-
rectly from the surface but first desorbed into the
reaction chamber and then penetrated the analyzer
chamber. This extraneous signal was subtracted
from the observed intensity. The apparatus was
originally designed for the angle-resolved mea-
surements of desorbingCO 2. The pumpingrate for
CO2 is about 2000 l/s, high enough to reduce the
CO2 penetration from the reaction chamber to a
negligible level. On the other hand, the pumping
rate for N2 was estimated to be around one
twentieth of that for CO2 based on the signal ratio
of CO2 versus N2 found in the AI-form. In other
words, the N2 penetration from the reaction
chamber to the analyzer may increase by a factor
of about 20 compared with that for CO2. This is
why a significant N2 signal is observed in the AR-
form even at h ¼ 90° [18].
7
13CO
13CO(a)+O(a
)
(a)
2
6
5
4
3
2
1
0
θ
-30o
-10o
0o
β3-CO2
β1-CO2
10o
30o
90o
β
-CO2
2
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
Ts /100 K
Step down
Up
θ
-20o
0o
20o
40o
(b)
-40o
[112]
[111]
1.0
[001]
13
cos ( -10)
θ
60o
80o
-60o
-80o
The angular distribution of N2 peaks is plot-
ted against the desorption angle in Fig. 5(b). The
d-N2 desorption collimated at h ¼ À29 Æ 2° as
cos9 ðh þ 29Þ. The distribution became sharper
from cos9 ðh þ 29Þ to cos14 ðh þ 30Þ with increasing
the CO coverage although the collimation angle
did not shift. In the absence of CO(a), N2 desorbed
in the range of 450–550 K, yielding the c-N2 peak
at around 510 K. Its desorption followed a
-β1—CO
2
cos20( -5)
θ
-β2—CO
2
-β3—CO
cos12( -10)
2
θ
Fig. 4. (a) AR-TDS of 13CO2 at different desorption angles in
the O(a) þ 13CO(a) reaction at HCO ¼ 0:30 and HO ¼ 0:40 and
(b) angular distribution of 13CO2 in the O(a) þ 13CO(a) reac-
tion. (Ã) b1-13CO2, ( ) b2-13CO2 and ( ) b -13CO2. The signal
}
ꢀ
3
intensity was normalized to the maximum value of b2-13CO2.

Y.-H. Hu et al. / Surface Science 526 (2003) 159–165
163
16
14
12
10
8
We reported a sharp collimation of CO2 de-
sorption along h ¼ À30° in the NO(a) þ CO(a)
reaction. This angle is close to the local normal of
(0 0 1) steps. As described in Section 1, when O(a)
is supplied from gaseous O2 in advance, desorption
of the product CO2 in the O(a) þ CO(a) reaction
collimates closely alongthe local normal of the
(1 1 1) terrace on Ptð113Þ ¼ ½ðSÞ3ð111Þ Â 3ð001ފ,
Ptð112Þ ¼ ½ðSÞ3ð111Þ Â ð001ފ, Ptð335Þ ¼ ½ðSÞ4-
ð111Þ Â ð001ފ and Ptð557Þ ¼ ½ðSÞ6ð111Þð001ފ
[3]. In general, CO2 is formed on oxygen adsorp-
tion sites [22], most likely on threefold hollow sites
on (1 1 1) facets or on fourfold hollow sites on
(0 0 1) facets [11]. In the O(a) þ CO(a) reaction, CO
is mostly oxidized on the threefold hollow sites
because O(a) on (1 1 1) terraces has a higher reac-
tivity [11]. The fact that the CO2 reactive desorp-
tion mostly takes places on (0 0 1) facets indicates
the absence of oxygen on (1 1 1) facets. This situ-
ation holds when O(a) is provided from NO dis-
sociation on step sites at high temperatures and the
subsequent removal of O(a) is very fast.
In the NO(a) þ CO(a) reaction, NO(a) must be
dissociated before the formation of either N2 or
CO2. The reaction is triggered by NO(a) dissocia-
tion yieldingO(a) and N(a). It is well known that
NO(a) dissociates preferably on step sites [16,19].
In fact, no dissociation of NO has been found on
Pt(1 1 1). Furthermore, on platinum surfaces, the
combinative reaction of 2N(a) to N2(g) is fast
around 400 K [13] and the reaction of CO(a) with
O(a) is also very fast above 400 K [3]. This is
consistent with the observed concomitant desorp-
tion of d-N2 and d-CO2. The resultant products of
N(a) and O(a) are likely to be first trapped on step
sites and then immediately removed as CO2 or N2
before movingto (1 1 1) terraces. The site switch-
Θ
Θ
CO=0.05
NO=0.52
(a)
θ
-60o
δ—N2
γ—N2
-50o
-40o
-30o
-10o
6
0o
4
10o
20o
90o
2
3.0
3.5
4.0
4.5
5.0
5.5
6.0
Ts /100 K
Step down
Up
θ
-20o
20o
40o
-40o
0o
[112]
[111]
[001]
(b)
60o
-60o
-80o
80o
δ—N2
cos9(
θ+ 29)
Fig. 5. (a) AR-TDS of 15N2 at different desorption angles in the
15NO(a) þ 13CO(a) reaction at HNO ¼ 0:52 and HCO ¼ 0:05. (b)
Angular distribution of 15N2 in the 15NO(a) þ 13CO(a) reaction
at the above condition.
cos7–8 ðh þ 30Þ form. This is similar to the colli-
mation of around )30° for N2 in NO decomposi-
tion on Pd(1 1 2) reported by Ikai and Tanaka [19].
4. Discussion
ingfor CO formation is controlled by the low
2
4.1. Desorption mechanism
O(a) density and its fast removal. In other words,
the preference of CO2 desorption from (0 0 1) steps
is governed by the fact that NO dissociation on
steps is rate-determining. The above mechanism
can also be applied to N(a) removal.
Both c and d-N2 desorption are commonly due
to the process of 2N(a) ! N2(g) and collimate
alongthe (0 0 1) step normal. However, the
desorption temperature range is different. This
combinative desorption was reported to take place
A Pt(112) surface is rather stable, compared
with those of Pt(1 1 0) and Pt(1 1 3) which are
easily reconstructed into a missing-row form by
heatingand the resultant reconstruction is lifted by
adsorption of CO above a critical value. Pt(1 1 2) is
stable duringheating. The ð1 Â 1Þ structure is kept
after CO or oxygen adsorption in a wide range of
coverages [20,21].

164
Y.-H. Hu et al. / Surface Science 526 (2003) 159–165
on stepped Pt(3 3 5) in the range of 350–450 K
after N(a) was deposited by electron bombard-
ments [13]. This is consistent with that the d-N2
desorption at around 425 K is retarded by the slow
NO dissociation. The desorption would be faster
when N(a) is supplied more quickly. On the other
hand, the c-N2 desorption must be more activated
to proceed, suggesting that oxygen blocks sites
for NO dissociation or the combination itself is
retarded by the presence of O(a) because O2 de-
sorption started at around 700 K after N2 de-
sorption. This comparison suggests that d-N2
desorption takes place on oxygen-free sites formed
by the fast removal of O(a).
reaction of N(a) because no N2O is found in NO
decomposition. However, the contribution from
the N2O decomposition pathway is not completely
ruled out since N2O dissociation proceeds on
Pt(1 1 2) at 120–140 K [26].
Acknowledgements
Y.-H. Hu is supported by JSPS (The Japanese
Society for the Promotion of Science) Fellowship
Programs for Foreign Researchers for research in
Japan in the period 2001–2003. S. Han is indebted
to the Clark Memorial Foundation in the period
2002–2003. I. Kobal was supported by the scientist
exchange program between JSPS and the Ministry
of Education, Science, and Sport of Slovenia in
2002. This work was partly supported by Grant-
in-Aid no. 13640493 for General Scientific Re-
search from JSPS.
4.2. Comparison with other surfaces
Ikai and Tanaka reported similar results of N2
desorption in NO decomposition on Pd(1 1 2) in
the absence and presence of gaseous hydrogen [19].
N2 desorption peaked at 480–500 K and colli-
mated at around )30° off the surface normal.
Usinglabeled N, it was shown that N 2 was formed
14
as 15N(a) þ NO(a) ! 15N14N. Although the au-
References
thors did not argue this point, their result strongly
suggests that N2O(a) is the intermediate of de-
sorbingN in a similar way to Pd(1 1 0) [23] be-
[1] I. Kobal, T. Matsushima, Trends Chem. Phys. 7 (1999)
169.
2
cause of the concomitant formation of the
byproduct N2O. This byproduct is formed within
the same temperature range as N2 desorption and
is quickly decomposed at around the temperature
of NO dissociation, yieldingN 2. In fact, N2 is
merely desorbed through N2O dissociation in the
thermal NO decomposition on Pd(1 1 0) at lower
temperatures [23]. The N2 emission on palladium
may proceed through N2O dissociation on (0 0 1)
steps. A similar N2 emission is also possible on
Rh(3 3 5) where N2 desorption was reported to
collimate at )15° off the surface normal [24]. This
is fairly at the middle of the (1 1 1) terrace normal
of þ14.1° and the (0 0 1) step normal of )40.3°.
N2O(a) is likely to be easily dissociated at around
100–150 K on palladium and rhodium surfaces
[25]. Spatial distribution measurements of N2 in
N2O decomposition might be useful to differen-
tiate the above associative mechanism from the
N2O decomposition pathway. On Pt(1 1 2), N2
desorption may proceed through the associative
[2] I. Kobal, I. Rzeznicka, T. Matsushima, in: Recent Devel-
opments of Physical Chemistry, Transworld Research
Network, in press.
[3] T. Matsushima, Heterogen. Chem. Rev. 2 (1995) 51.
[4] T. Yamanaka, C. Moise, T. Matsushima, J. Chem. Phys.
107 (1997) 8138.
[5] R. Kose, D.A. King, Chem. Phys. Lett. 313 (1999) 1.
[6] T. Yamanaka, Q.K. Xue, K. Kimura, T. Matsushima,
Y. Hasegawa, T. Sakurai, Jpn. J. Appl. Phys. 39 (2000)
3562.
[7] A. Rar, T. Matsushima, Surf. Sci. 318 (1994) 89.
[8] H. Wang, R.G. Tobin, D.K. Lambert, C.L. DiMaggio,
G.B. Fisher, Surf. Sci. 372 (1997) 267.
[9] J.E. Reutt-Robey, D.J. Doren, Y.J. Chabal, S.B. Christ-
man, Phys. Rev. Lett. 61 (1988) 2778.
[10] C.T. Campbell, G. Ertl, H. Kuipers, J. Segner, J. Chem.
Phys. 73 (1980) 5862.
[11] G.Y. Cao, G. Moula, Y. Ohno, T. Matsushima, J. Phys.
Chem. B 103 (1999) 3235.
[12] Y.O. Park, W.F. Banholzer, R.I. Masel, Surf. Sci. 155
(1985) 341.
[13] H. Wang, R.G. Tobin, C.L. DiMaggio, G.B. Fisher, D.K.
Lambert, J. Chem. Phys. 107 (1997) 9569.
[14] R.D. Ramsier, Q. Gao, H.N. Waltenburg, J.T. Yates Jr., J.
Chem. Phys. 100 (1994) 6837.

Y.-H. Hu et al. / Surface Science 526 (2003) 159–165
165
[15] T. Matsushima, Surf. Sci. 127 (1983) 403.
[16] T. Sugisawa, J. Shiraishi, D. Machihara, K. Irokawa, H.
Miki, C. Kodama, T. Kuriyama, T. Kubo, H. Nozoye,
Appl. Surf. Sci. 169–170 (2001) 292.
[21] A. Szabo, M.A. Henderson, J.T. Yates Jr., J. Chem. Phys.
96 (1992) 6191.
[22] A. Alavi, P.J. Hu, T. Deutsch, P.L. Silvestrelli, J. Hutter,
Phys. Rev. Lett. 80 (1998) 3650.
[17] A. Rar, H. Sugimura, A. Barrera, Y. Ohno, T. Matsu-
shima, Surf. Sci. 348 (1996) 77.
[23] Y. Ohno, K. Kimura, M. Bi, T. Matsushima, J. Chem.
Phys. 110 (1999) 8221.
[18] M. Kobayashi, Y. Tuzi, J. Vac. Sci. Technol. 16 (1979)
685.
[24] M. Ikai, N.M.H. Janssen, B.E. Nieuwenhuys, K. Tanaka,
J. Chem. Phys. 106 (1997) 311.
[19] M. Ikai, K. Tanaka, J. Chem. Phys. 110 (1999)
7031.
[25] H. Horino, I. Rzeznicka, A. Kokalj, I. Kobal, Y. Ohno, A.
Hiratsuka, T. Matsushima, J. Vac. Sci. Technol. A 20
(2002) 1592.
[20] M.A. Henderson, A. Szabo, J.T. Yates Jr., J. Chem. Phys.
91 (1989) 7245.
[26] Y. Hu, unpublished data.