Formation of surface oxides on Pt(100) during CO oxidation in the millibar pressure range

Article abstract of DOI:10.1016/S0039-6028(00)00189-8

CO oxidation on Pt(100) at oxygen pressures of 9.0×10^-2 mbar was studied. An unusual decrease in catalytic activity with time was observed, which has not been observed at lower pressures. Post-reaction Auger and thermal desorption spectroscopy measurements showed the formation of a surface oxide species, which deactivated the catalyst. The surface oxide is compared to subsurface oxygen and to surface oxide studied previously on low index Pt single crystal surfaces. An adsorbate induced surface reconstruction is proposed as a possible driving force for oxide formation.

Full text of DOI:10.1016/S0039-6028(00)00189-8

Surface Science 454–456 (2000) 352–357
www.elsevier.nl/locate/susc
Formation of surface oxides on Pt(100) during CO oxidation
in the millibar pressure range
*
J. Dicke a,b, , H.H. Rotermund a, J. Lauterbach b
a Fritz-Haber-Institut of the Max-Planck Society, Faradayweg 4–6, 14195 Berlin, Germany
b School of Chemical Engineering, Purdue University, West Lafayette, IN 47907-1283, USA
Abstract
CO oxidation on Pt(100) at oxygen pressures of 9.0×10−2 mbar was studied. An unusual decrease in catalytic
activity with time was observed, which has not been observed at lower pressures. Post-reaction Auger and thermal
desorption spectroscopy measurements showed the formation of a surface oxide species, which deactivated the catalyst.
The surface oxide is compared to subsurface oxygen and to surface oxide studied previously on low index Pt single
crystal surfaces. An adsorbate induced surface reconstruction is proposed as a possible driving force for oxide
formation. © 2000 Elsevier Science B.V. All rights reserved.
Keywords: Auger electron spectroscopy; Carbon monoxide; Catalysis; Oxidation; Platinum; Solid–gas interfaces; Surface chemical
reaction; Thermal desorption spectroscopy
1. Introduction
geneously catalyzed reactions at higher pressures,
CO oxidation on Pt(100) was carried out at oxygen
pressures of 9.0×10−2 mbar and a surface temper-
ature of 473 K. Surprisingly, the catalytic activity
of the freshly cleaned Pt crystal decreased rapidly
with time. Auger and thermal desorption spectro-
scopy (TDS) measurements showed the sample to
be covered by a stable form of oxygen, which
causes the surface to become non-catalytic.
In the literature the terms subsurface oxygen
and surface oxide have been introduced [9–17] to
distinguish oxygen species with behavior different
from regular chemisorbed oxygen. Both names
have been used interchangeably, while in this paper
we want to clearly differentiate between subsurface
oxygen and surface oxide. Subsurface oxygen is
located directly underneath the first atomic layer
of the surface, as deduced by Lauterbach et al.
from spatially resolved work function measure-
ments and TDS [9]. Surface oxide is also believed
CO oxidation on Pt has been a model reaction
for many years [1]. Under certain experimental
conditions the catalytic activity of the surface can
be oscillatory in time [2,3] and the formation of
spatio-temporal adsorbate patterns has frequently
been observed [4]. Several detailed studies have
been performed by photoemission electron micro-
scopy (PEEM) at oxygen pressures of 4.0×
10−4 mbar on Pt(100) [5] and Pt(110) [6]. In
contrast to low-pressure surface science experi-
ments, industrial catalysis is typically performed
at atmospheric pressures and above. Recently,
increased efforts have been made to bridge this
pressure gap [7,8]. In a step towards the under-
standing of non-linear phenomena during hetero-
* Corresponding author. Fax: +49-30-84135106.
E-mail address: dicke@fhi-berlin.mpg.de (J. Dicke)
0039-6028/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S0039-6028 ( 00 ) 00189-8

J. Dicke et al. / Surface Science 454–456 (2000) 352–357
353
to be located underneath the surface [10], but
might not necessarily be confined between the first
and second layer of Pt atoms. Although the exact
location of the different types of oxygen is very
difficult to pinpoint, they can be distinguished
using TDS due to their different properties of
thermal decomposition. Adsorbed oxygen desorbs
from Pt(100) at temperatures between 600 and
730 K [18]. Subsurface oxygen formed on Pt(100)
is slightly more stable and decomposes between
650 and 770 K with a desorption maximum at
750 K [11]. Surface oxide is the strongest bound
oxygen species and is stable below 1000 K [12].
Both adsorbed and subsurface oxygen can be
reduced by CO exposure [9], whereas surface oxide
and could be reduced by CO, it could fall more
into the category of subsurface oxygen.
It has become clear that the differentiation
between adsorbed oxygen, subsurface oxygen, and
surface oxide is difficult, if one wants to categorize
the different types of oxygen reported on Pt single
crystals. In this paper, we report a stable surface
oxide formed exclusively under reaction conditions
on Pt(100) at elevated pressures.
2. Experimental
All experiments were carried out in an ultrahigh
vacuum system consisting of a main chamber and
a reaction chamber, in which a base pressure
below 2×10−10 mbar could be achieved. The
sample can be transferred between chambers via a
transfer line [25]. The Pt(100) crystal was spot-
welded to two tungsten wires, through which it
could be heated resistively. It was cleaned in the
main chamber by alternating sputtering and annea-
ling cycles. AES measurements were performed
using a cylindrical mirror analyzer (Perkin-Elmer)
and were recorded as differentiated spectra. TDS
measurements were performed using a mass spec-
trometer (Stanford Research Systems) by heating
the sample in the line of sight of the mass spectrom-
eter at a controlled rate of 4 K/s.
is shown to be stable with respect to CO or H
exposure [12].
2
The formation of subsurface oxygen on Pt(100)
[9] and Pt(110) [14] was attributed to the restruc-
turing of the surface. Low energy electron diffrac-
tion (LEED) and scanning tunneling microscopy
(STM) have shown that
a
clean Pt(100)
reconstructs to a (5×20) surface structure [19–
21]. The adsorption of CO or oxygen can lift the
reconstruction, after which the Pt(100) surface has
a (1×1) structure [22,23]. The restructuring of
the surface from a (5×20) to a (1×1) surface
allows oxygen to penetrate underneath the first
layer of Pt atoms, producing subsurface oxygen.
The formation of surface oxide on Pt(110) and
Pt(111) was observed during oxygen exposure for
surface temperatures above 800 K at oxygen pres-
sures of 1×10−6 mbar [10,13,15–17]. Below 800 K
oxygen could only chemisorb on the surface. The
formation of surface oxide on Pt(111) and Pt(110)
was attributed to Si impurities [15,24]. Si is difficult
to detect in Auger electron spectroscopy (AES),
because Pt masks its main peak located at 94 eV.
The presence of Si impurities becomes visible
through the chemical shift of the oxidized Si
to 76 eV.
The reaction chamber can be closed off from
the main chamber by a gate valve. It allows for
ellipsomicroscopy for surface imaging (EMSI) of
the sample [26]. The reactants (O and CO) are
2
fed into the reaction chambers with mass flow
controllers (MKS). Reaction products can be
detected with the mass spectrometer through a
leak valve connecting the reaction chamber to the
main chamber.
3. Results and discussion
The formation of surface oxide during CO
oxidation has only been reported for the Pt(111)
surface [8]. Oscillatory reaction rates, observed at
oxygen pressures of 2.6×10−3 mbar, were
explained through the periodic formation and
reduction of surface oxide. Because the observed
oxygen species only existed during CO oxidation
Before each experiment the sample was cleaned
in the main chamber until no impurities could be
detected in the Auger spectrum (see Fig. 2a). The
numbers shown in the spectra indicate the energy
of the peak position. They all are characteristic
for Pt and are in agreement within 3 eV with the

354
J. Dicke et al. / Surface Science 454–456 (2000) 352–357
literature [27]. TD spectra were recorded for satu-
rated adsorbate layers of CO and O on the Pt(100)
crystal. The measured spectra were in accordance
with the literature [18].
For CO oxidation experiments the sample was
transferred into the reaction chamber, where it
was heated to 473 K. Oxygen was fed into the
pressure decreases abruptly, and the sample
becomes catalytically inactive. From low-pressure
studies [28], we conclude that above a critical
CO/O ratio the sample is transformed from a
2
catalytically active, predominantly oxygen covered
state, to a solely CO covered surface. Because the
dense CO adlayer inhibits oxygen adsorption, the
surface is called CO poisoned. As CO is turned
off for the first time at 0.5 min (see Fig. 1c and d),
reaction chamber at
a partial pressure of
9.0×10−2 mbar. CO was dosed at a CO/O ratio
2
of 0.1 and was cycled via the flow controller with
the CO pressure at first remains low and then
2
a period of 1 min. The signal from the CO flow
controller was recorded as an indication of the CO
pressure in the reaction chamber (see Fig. 1a). The
quickly increases. After reaching a maximum it
shows a gradual decay. The peak height of the
second maximum is clearly lower than that of the
first. Because of the hysteresis observed for the
reactivity of Pt, the sample stays CO poisoned
until the CO pressure falls below a second critical
CO partial pressure in the reaction chamber was
2
measured with the mass spectrometer (see Fig. 1b).
Fig. 1c and d show the data for the first two
periods of forced oscillations. After 17 min, O
and CO were turned off and the reaction chamber
CO/O ratio. At this point the surface makes a
transition to a catalytically active surface, causing
2
2
was evacuated.
the CO pressure to quickly rise. Meanwhile, the
CO pressure continually decreases, which causes
2
At the beginning of the time series and with
increasing CO pressure CO is being produced as
expected, indicating that the sample is catalytically
the CO pressure to also decrease after reaching
its maximum value. Because the reactivity of the
2
2
active. Above a certain CO pressure the CO
sample is linear in the CO pressure and because
2
Fig. 1. Pt(100) was heated to 473 K and oxygen and CO were introduced at pressures of 9.0×10−2 mbar and 9.0×10−3 mbar,
respectively. CO was cycled with a period of 1 min. (a) The CO signal from the flow controller; (b) the partial CO pressure inside
2
the reaction chamber. (c, d) The expanded region of (a) and (b) between 2 and 4 min.

J. Dicke et al. / Surface Science 454–456 (2000) 352–357
355
the first critical ratio has a higher value than the
second, the maximum reactivity observed for the
first transition also shows a higher value than that
observed for the second transition. The maximum
value of the CO pressure observed when CO is
2
switched on in the cycle clearly decreases with each
period. On the contrary, the maximum CO pres-
2
sure observed whenever CO is switched off at first
increases with each period. Only after both peaks
reach a similar height do both decrease similarly.
With EMSI the spreading of CO or O fronts
can be observed whenever CO is switched on or
off, which is characteristic for Pt(100) [29]. With
each cycle, the front velocity observed increased
whereas the image contrast decreased. After the
seventh CO cycle, the fronts were becoming too
fast to be distinguished from uniform global trans-
itions. After the 12th period finally no change
could be observed.
Fig. 2. Auger spectrum of a clean Pt(100) surface after sputter-
ing and annealing (a) and after 20 min of cyclic CO oxidation
at 473 K (b). The numbers indicate the exact location of the
minimum peak position of clean Pt (a) and inactive oxide cov-
ered surface (b).
Both the change in CO peaks and the increase
2
in front speeds indicate a marked change in the
sample’s catalytic behavior. Clearly the change in
the surface is gradual, but continuous with each
cycle. The change in reactivity can be understood
in terms of the hysteresis measured with respect to
the CO pressure. In the low-pressure region, the
reactivity remains linear in the CO pressure, but
has a decreasing slope due to the deactivation of
the surface, causing the first maximum to decrease.
Simultaneously the CO pressure difference between
the two transition points of the hysteresis
decreases, causing the second maximum to
increase. After the seventh cycle this difference has
decreased to zero and both transition points show
the same maximum reactivity. It is interesting to
note that after the seventh period the CO pro-
duction stays finite while CO is switched on.
Consequently the surface cannot be CO poisoned
any longer.
Fig. 3. Oxygen TD spectra of the Pt(100) crystal after 20 min
of cyclic CO oxidation at 473 K. The individual curves show
the spectra of different runs.
2
After the reaction measurements, we performed
AES and TDS in order to study the surface
composition of the deactivated sample. AE spectra
showed two additional peaks at 490 eV and 510 eV,
which were identified as oxygen (see Fig. 2b).
Oxygen TDS showed two desorption maxima
located at approximately 1020 K and 1160 K (see
Fig. 3b). The formed oxygen could not be removed
by exposure to 9.0×10−3 mbar of CO at 473 K
over 30 min. Due to the high thermal stability of
the new oxygen species, we believe to have formed
a surface oxide.
The amount of surface oxide formed after CO
oxidation revealed a rather poor reproducibility,
which also has been noted in the literature [30,31].
The peak-to-peak O /Pt ratio varied between
510 233
values of 2 and 8. The integrated areas of the
oxygen TDS curves, which were normalized to the

356
J. Dicke et al. / Surface Science 454–456 (2000) 352–357
amount of adsorbed oxygen, varied between values
of 1.5 and 4. No measurable amounts of impurities
could be detected in the Auger spectra either before
or after the reactions. On the other hand, minute
concentrations of impurities, which were below the
detection limit of AES, have been shown to alter
the surface’s behavior towards oxygen adsorption
[15,24]. Therefore we cannot completely rule out
the presence of any impurities and purposely called
the oxygen species surface oxide.
By changing the experimental conditions, we
tried to further elucidate the formation conditions
of the surface oxide. For example exposing the
clean surface at 473 K to oxygen and CO without
cycling of the reactant feed produced no measurable
amounts of surface oxide. Since the surface is
predominantly CO poisoned under these conditions,
obviously oxygen cannot adsorb in order to form
the surface oxide. Prolonged exposure to oxygen
pressure of 9.0×10−2 mbar at 473 K, as well as at
temperatures up to 900 K, also produced no detect-
able surface oxide. The formation of surface oxide
during CO oxidation hence must be caused by the
transition from the catalytically active to the CO
poisoned surface and vice versa. Since these trans-
itions induce the 1×1<5×20 restructuring of the
surface [22], we believe that the formation of surface
oxide follows a similar process as was proposed for
the formation of subsurface oxygen on Pt(100) [9].
In comparison to the surface oxides observed on
the low index Pt single crystals so far, the surface
oxide reported here is formed at much lower tem-
peratures, namely 473 K. Therefore, surface recon-
struction acts as a promoter towards the formation
of the surface oxide.
Pt(100) was observed. AES proved the presence
of large amounts of oxygen on the deactivated
catalyst. TD spectra with two desorption maxima
at 1020 K and 1160 K showed the oxygen to be
strongly bound to the surface. The oxygen was
stable towards CO exposure, but could be decom-
posed by annealing up to 1200 K. Due to its high
temperature stability it is believed to be a surface
oxide, which is in strong contrast to subsurface
oxygen and chemisorbed oxygen. The surface
oxide could not be formed by simply exposing the
surface to oxygen, but was formed only during
CO oxidation. To explain this, restructuring of the
surface is proposed as a possible mechanism, which
allows oxygen to penetrate underneath the surface.
Acknowledgement
JL acknowledges support from the National
Science Foundation, Grant No. CTS 9733821.
References
[1] G. Ertl, H.H. Rotermund, Curr. Opin. Solid State Mater.
Sci. 1 (1996) 617.
[2] M.M. Slinko, N. Jaeger, Oscillatoring Heterogeneous
Catalytic Systems, Elsevier, Amsterdam, 1994.
[3] R. Imbihl, G. Ertl, Chem. Rev. 95 (1995) 697.
[4] H.H. Rotermund, G. Haas, R.U. Franz, R.M. Tromp, G.
Ertl, Science 270 (1995) 608.
[5] J. Lauterbach, H.H. Rotermund, Surf. Sci. 311 (1994) 231.
[6] S. Nettesheim, A.v. Oertzen, H.H. Rotermund, G. Ertl,
J. Chem. Phys. 98 (1993) 9977.
[7] H.H. Rotermund, G. Haas, R.U. Franz, R.M. Tromp, G.
Ertl, Appl. Phys. A 61 (1995) 569.
[8] R.E.R. Colen, J. Christoph, F. Pena, H.H. Rotermund,
˜
Surf. Sci. 408 (1998) 310.
[9] J. Lauterbach, K. Asakura, H.H. Rotermund, Surf. Sci.
313 (1994) 52.
The formation of a stable surface oxide at
intermediate pressures, and its lack of formation
at pressures only two orders of magnitude lower
[9], indicates that the total reaction pressure can
indeed have a significant influence on the basic
surface mechanisms in heterogeneous catalysis.
The presented measurements are therefore only
one step towards building a bridge between surface
science and industrial catalysis.
[10] H. Niehus, G. Comsa, Surf. Sci. 93 (1980) L147.
[11] H.H. Rotermund, J. Lauterbach, G. Haas, Appl. Phys. A
57 (1993) 507.
[12] T. Matsushima, D.B. Almy, J.M. White, Surf. Sci. 67
(1977) 89.
[13] J.L. Gland, B.A. Sexton, G.G. Fischer, Surf. Sci. 95
(1980) 587.
[14] A.v. Oertzen, A. Mikhailov, H.H. Rotermund, G. Ertl,
Surf. Sci. 350 (1996) 259.
4. Summary
During CO oxidation with cycled CO feed, a
rapid decrease of the surface’s catalytic activity of
[15] H.P. Bonzel, A.M. Franken, G. Pirug, Surf. Sci. 104
(1981) 625.

J. Dicke et al. / Surface Science 454–456 (2000) 352–357
357
[16] B. Carriere, P. Legare, G. Maire, J. Chim. Phys. 71
(1974) 355.
[17] R. Ducros, R.P. Merril, Surf. Sci. 55 (1976) 227.
[18] M.A. Barteau, E.I. Ko, R.J. Madix, Surf. Sci. 102
(1981) 99.
[24] H. Niehus, G. Comsa, Surf. Sci. 102 (1981) L14.
[25] T.A. Jachimowski, J. Lauterbach, Rev. Sci. Instrum. 69
(1998) 2757.
[26] G. Haas, T.D. Pletcher, G. Bonilla, T.A. Jachimowski,
H.H. Rotermund, J. Lauterbach, J. Vac. Sci. Technol. A
16 (1998) 1117.
[19] K. Heinz, E. Lang, K. Strauss, K. Muller, Appl. Surf. Sci.
¨
11/12 (1982) 611.
[27] L.E. Davis, N.C. MacDonald, G.E. Riach, R.E. Weber,
Handbook of Auger Electron Spectroscopy, second ed.,
Perkin-Elmer Corporation, Eden Prairie, 1978.
[20] M.A.v. Hove, R.J. Koestner, P.C. Stair, J.P. Biberian, L.L.
Kesmodel, I. Bartos, G.A. Somorjai, Surf. Sci. 103
(1981) 189.
[28] G. Ertl, P.R. Norton, J. Rustig, Phys. Rev. Lett. 49
¨
[21] A. Borg, A.-M. Hilmen, E. Bergene, Surf. Sci. 306
(1994) 10.
[22] K. Griffiths, T.E. Jackman, J.A. Davies, P.R. Norton, Surf.
Sci. 138 (1984) 113.
[23] P.A. Thiel, R.J. Behm, P.R. Norton, G. Ertl, Surf. Sci. 121
(1982) L553.
(1982) 177.
[29] H.H. Rotermund, S. Jakubith, A.v. Oertzen, G. Ertl,
J. Chem. Phys. 91 (1989) 4942.
[30] R.W. McCabe, L.D. Schmidt, Surf. Sci. 60 (1976) 85.
[31] C.E. Smith, J.B. Biberian, G.A. Somorjai, J. Catal. 57
(1979) 426.