Page 1
Surface Science 555 (2004) 43–50
www.elsevier.com/locate/susc
Temperature-dependent work function shifts ofhydrogenated/deuteriated palladium: a new
theoretical explanation
S. Halas a,*, T. Durakiewicz a,b, P. Mackiewicz a
a Mass Spectrometry Laboratory, Institute of Physics, Maria Curie–Sklodowska University, 20-031 Lublin, Polandb Los Alamos National Laboratory, Condensed Matter and Thermal Physics, MST-10 Group, Los Alamos, NM 87545, USA
Received 6 June 2003; accepted for publication 1 March 2004
Abstract
We explain the phenomena of work function (WF) variations of polycrystalline palladium film due to adsorption
and absorption of hydrogen. A small increase of the WF observed at temperatures above 120 K is an indication of a
spontaneous formation of H� ions at the surface, subsequently dissociating to electrons and neutral atoms which
completely desorb at temperatures above 400 K. A large lowering of the WF at low temperatures (about 2 eV at 78 K) is
associated with the formation of PdH. This process is treated quantitatively in the frame-work of the metallic plasma
model. The mechanism of the isotope effect on the lowering of the WF is explained by the vibrational frequency dif-
ference of H and D atoms confined in the palladium lattice.
� 2004 Elsevier B.V. All rights reserved.
Keywords: Chemisorption; Deuterium; Hydrogen atom; Isotopic exchange/traces; Palladium; Work function measurements
1. Introduction
Palladium atoms have 10 d electrons and no s
electron in the valence shell, unlike other transition
metals. This unique electronic structure of the Pd
results in very specific interactions between the Pd
and the H atoms in Pd–H systems. The most
striking experimental fact is that a thin Pd layer
readily and reversibly converts itself to PdHx invery low pressure of H2. Du�s et al. [1] have found
that x attains 0.84 when H2 pressure (in equilib-
* Corresponding author. Tel.: +48-815-376-275; fax: +48-
815-376-91.
E-mail address: [email protected] (S. Halas).
0039-6028/$ - see front matter � 2004 Elsevier B.V. All rights reserv
doi:10.1016/j.susc.2004.03.001
rium) is as low as 10�6 Torr. The process of pal-ladium hydride formation may be reversed by a
decrease of the gas pressure. Both the ambient
hydrogen pressure and temperature strongly affect
the electronic work function (WF) of the metallic
layer. For this reason Pd and perhaps other tran-
sition metals which reversibly absorb hydrogen,
may be useful in future devices where a continuous
adjustment of WF will be desirable.In a number of experiments performed by Du�s
and his coworkers [1–4] an increase of the WF by
ca. 0.3 eV was noticed in temperatures above 120
K compared to clean Pd-films. Below 87 K these
authors have observed a new interesting phenom-
enon, namely a gradual decrease of the WF with
an increase of the H/Pd ratio in bulk metal. The
ed.
Page 2
44 S. Halas et al. / Surface Science 555 (2004) 43–50
most pronounced effect is reported at T ¼ 78 K in
the case of deuteriated Pd. For PdDx, where
x ¼ 0:7, the latter authors observed lowering of the
WF by 2.32 eV. Assuming that the experimental
WF for the pure Pd-film is 5.27 eV (see Section 4)
the deuteriated palladium may have a WF of lessthan 3.0 eV.
The primary motivation of this paper is to give
an explanation of the dual behavior of the Pd–H
system in a more simplistic way than it was pro-
posed by Grimley [5]. According to his model, the
increase in hydrogen adsorbate concentration
causes an increase in the magnitude of splitting
between induced localized states. There is a criticalconcentration above what the lower state merges
into the conduction band of the metal. Thereafter
the donation of electrons from hydrogen adatoms
into unoccupied states in the conduction band
occurs. This leads to formation of positively
charged hydrogen adspecies, resulting in a de-
crease of the WF.
In this paper we attempt to explain the behaviorof Pd–H and Pd–D systems by means of the con-
cept of the spontaneous plasma polarization near
metal surface and WF as the work against the
image force [6].
2. Method
The idea of using the image potential as the
measure of WF is quite old, e.g. it may be found in
a paper by Langmuir [7]. The image force potential
of a conducting plane is usually expressed in terms
of the potential energy of an electron located at
distance x from the plane
uðxÞ ¼ e2
16pe0x; ð1Þ
Table 1
Electronic properties of Pd and PdH
Pd
Fermi energy (FE) (eV) 5.75
Number of free electrons per Pd atom (Z) 2
Wigner–Seitz radius (rs) 2.279 bohr
Lattice constant (a) 3.8824 �A
Work function (U) 5.27 eV
where e is elementary charge, e0 is the electric
constant. The above formula is used for explana-
tion of the Schottky effect, see e.g. [8], because the
barrier lowering occurs at relatively large distances
from a conducting plane. However the formula (1)cannot be used directly for WF calculation be-
cause uðxÞ tends to infinity, when x ! 0. On the
basis of the metallic plasma concept, we have
demonstrated previously that the expression for
the WF of a metals is [6]
U ¼ e2
16pe0d; ð2Þ
where d is the length of plasma polarization in the
direction perpendicular towards metal surface.
The magnitude of d depends on the free electrondensity (or Wigner–Seitz radius, rs) and the Fermi
energy, FE, and is in the order of 1 �A. The final
formula for WF of the transition metal is [6]
U½eV� ¼ 43:46
r3=2s FE1=2; ð3Þ
where rs and FE are expressed in bohr and eV,
respectively. The necessary data for Pd and PdH
exist in the literature and they are collected in
Table 1.
The evaluation of the rs value for a pure metal
may be made by the following formula [12]
rs ¼ 1:3882AZq
� �1=3
; ð4Þ
where A is atomic mass in grams, q is the bulk
density in g cm�3 and Z is the number of free
electrons per atom (which constitute the metallic
plasma). In the case of hydrogenated metal, the
lattice is somewhat expanded compared to pure
metal and each lattice cell occupied by hydrogen
PdH Remarks
7.05 Ref. [9]
1 See Section 4
Calculated by Eq. (4),
b � a b ¼ 1:035 after Ref. [10]
After Ref. [11]
Page 3
S. Halas et al. / Surface Science 555 (2004) 43–50 45
atom may contain a lower number of free elec-
trons, than the Z for pure Pd, see Section 4.
Fig. 1. Schematic representation of the location of H� ion with
respect two neighbouring Pd atoms on the palladium (1 0 0)
surface, a is the lattice constant, d is the cut-off distance of the
free electron density distribution. Dashed area denotes the
space available for free electrons at 0 K.
3. Increase of WF at high T
At high temperatures, above 450 K, the equi-
librium state of a Pd–H2 system is shifted towards
the complete degassing of Pd from hydrogen,
hence no surface phenomena will influence the WF
of the hot Pd metal immersed in a low pressure H2
atmosphere. However at room temperature Du�set al. [1] noticed an increase of the WF by about0.2 eV compared to pure Pd-film at the equilib-
rium with H2 gas under pressures ranging from
10�6 to 10�2 Torr. A broad maximum of WF in-
crease was observed at temperature around 120 K.
The maximum increase of the WF was 0.40 eV at
that temperature, in which a nearly complete layer
of PdH (or PdD) was formed on the surface. Sim-
ilar magnitude of the WF increase was reported inthe single crystal experiments [13,14].
This phenomenon may be understood as a re-
sult of a spontaneous conversion of H atoms at the
surface into negative ions, H�. The presence of
negative ions gives rise to a lowering of the electric
potential of Pd surface. The potential lowering,
detected by the vibrating capacitor method, is
small because the thickness of the dipole layerformed by the negative charge of adsorbed H�
ions and positive image charge is relatively small.
A schematic representation of the location of H�
ion with respect to the neighbouring Pd atoms is
shown in Fig. 1. It is clear from this figure that in
the zero vibrating state the center of H� ion is
shifted by a small distance, d, away from the cut-
off distance, d, of electron density distribution.This ‘‘d’’ is the same as d in Eq. (2). Hence the
presence of H� ion on the surface will result in
local lowering of the surface potential, which is
equivalent to an increase of the WF. We may
estimate the average change in WF, DU, using the
model capacitor comprising two square plates of
surface area, S, being spaced by a distance, d, andcharged by elementary charge, e.
The electrostatic energy of the capacitor is equal
to twice of the WF increase. It is because the ori-
ginal surface potential plane of pure metal divides
the space between capacitor plates into equal
parts. Hence, from the electrostatic formula we
have
2DU ¼ 1
2e
ee0S
d: ð5Þ
The surface area, S, related to a single adsorp-
tion site is equal to a2 for the Pd(1 0 0) plane, where
a is the Pd-lattice constant, see Table 1. For
practical calculation of DU the above formula may
be rewritten as follows
DU ¼ e2
4e0a2� d
¼ e2
8pe0a0
2pa0
a2� d
¼ 13:6 eV � 0:22 �A�1
� d; ð6Þ
where a0 is Bohr radius, while the d distance may
be calculated from the a and d values and theatomic radii of Pd and H� (R and r, respectively,see Fig. 1) as follows
d ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðRþ rÞ2 � a2=4
q� R� d: ð7Þ
From formula (7) one obtains d ¼ 0:115 �A,
employing a d value calculated from Eq. (2) for
pure Pd and the Pd and H� radii after CRC
Handbook [15]. Substituting the calculated d value
into (6), one obtains DU ¼ 0:34 eV, which is close
to the average WF increase detected by the
Page 4
Fig. 2. A schematic plot of the potential energy versus distance
of H0 or H� nuclei from the first (1 0 0) plane of palladium
nuclei in the crystal lattice. W is the work of H� ion extraction
from the minimum vibration state to infinity, U is the work
function, EA is the electron affinity, DE is dissociation energy of
H� ion at the surface, Ed is the desorption energy of neutral
hydrogen atom, a is the lattice constant. Note that both the avalue and height of the potential barrier between the neigh-
bouring wells depend on the presence of H0 atom in the inner
(left) well.
46 S. Halas et al. / Surface Science 555 (2004) 43–50
vibrating capacitor method for nearly totally
covered Pd surfaces [1,13].
At certain temperature the thermal dissociation
of H� ion starts. The dissociation reaction
H� () H0 þ e;
may be considered as the first step of desorption of
H atoms from the surface. The dissociation energy
of H� ion in vacuum is EA¼ 0.75 eV (electronaffinity) [16]. In the case of H� ion at the Pd sur-
face the dissociation energy, DE, is significantly
lower than EA because the ion is confined in a
quantum well of limited size, see Fig. 2.
4. Decrease of WF at low T
At 78 K a strong lowering of WF has been re-
ported [1–4]. The lowering in the case of Pd satu-ration by deuterium and isotopically light
hydrogen is 2.30 and 1.59 eV, respectively [4]. We
can explain these effects by use the metallic plasma
model. For a lowering of the WF of PdHx at
sufficiently low temperatures one has to assume a
lowering of the density of free electrons, i.e. in-
crease of the rs parameter. The existence of nega-
tive hydrogen ions in the Pd lattice should be
excluded because their diameter of 3.08 �A exceeds
the available space in the interior of the lattice cell.
At low temperatures the H atoms are incorporated
into the lattice forming a weakly bounded H–Pd
systems.
We postulate a covalent character of the H–Pdbonding because it causes the localization of one
free electron in each occupied lattice cell. That
covalent bonding, however, by no means stops
the nearly free motion of the hydrogen atom
within the cell volume. This motion is much slower
than the motion of the Pd free electrons with the
kinetic energy of order of the Fermi energy.
The covalently bounded H atom can move fromone to another Pd atom with the kinetic energy
comparable to that of a hypothetic noninteracting
hydrogen atom confined in a small-size cell of the
Pd lattice. This motion produces a high pressure
within the cell and consequently causes an increase
of the lattice constant. The relative increase of the
lattice constant and rs parameter is the same.
Another reason for the increase of rs is a reductionof the number of free electrons per Pd atom, Z, dueto covalent bonding between hydrogen and palla-
dium atoms inside the lattice. This results in
ZPdH < ZPd, below we demonstrate that the best
choice for these Z values is as given in Table 1.
Let us calculate WF by means of formula (3)
taking into account the increase of rs and the shift
of Fermi energy. We will assume in calculation ofrs and FE of PdHx that these values are closely
approximated by the linear combination of the
end-member values
rsðPdHxÞ ¼ rsðPdÞ � ð1� xÞ þ rsðPdHxÞ � x; ð8Þ
and
FEðPdHxÞ ¼ FEPdð1� xÞ þ FEPdH � x; ð9Þwhere x denotes the molal fraction of hydrogen in
hydrogenated (deuteriated) palladium lattice. The
WF of PdHx may be expressed in the following
form
U ¼ UPd
1� xþ xZPd
ZPdH
� �13
b
" #32
1� xþ xFEPdH
FEPd
� �12
;
ð10Þ
Page 5
Table 2
WF as a function of molal H/Pd ratio, the difference between
columns demonstrates the influence of the increase of the lattice
constant due to hydrogen absorption
x WF [eV]
b ¼ 1:035 b ¼ 1
0 5.27 5.27
0.1 4.98 5.02
0.2 4.72 4.78
0.3 4.48 4.56
0.4 4.25 4.35
0.5 4.04 4.16
0.6 3.85 3.98
0.7 3.67 3.81
0.8 3.50 3.66
0.9 3.34 3.51
1.0 3.20 3.37
Fig. 3. Lowering of WF of Pd exposed to hydrogen gas in low
temperatures calculated according to formula (10) and experi-
mental data from Du�s and Nowicka [4], where their ratio of the
‘‘positive hydrogen’’ to palladium at surface is converted to our
x ¼ H=Pd in bulk metal by a scaling factor of 3.0.
S. Halas et al. / Surface Science 555 (2004) 43–50 47
where b denotes the relative increase of the lattice
constant of PdH with respect to Pd and
UPd ¼43:46
½rsðPdÞ�32FE
12
Pd
; ð11Þ
is the calculated value of the WF of pure poly-
crystalline palladium.
In the first step of the calculation we will con-sider UPd as a function of ZPd and FEPd. The for-
mula (11) should yield a value close to that
determined experimentally. However, the experi-
mental values for polycrystalline Pd range from
4.8 to 5.15 eV [17,18], while new determination for
(1 1 1) surface yield 5.60 and 5.55 eV [19]. The
reason for such a large spread of the experimental
values is a strong interaction of Pd with traces ofhydrogen, oxygen and water [10,20]. Instead we
choose the value 5.27 eV obtained for iridium by
the thermionic emission method [21]. This value
should be a better choice because in recent exper-
iments with Pd deposited on Ir layers, and vice
versa, no difference in the measured WFs was de-
tected within the experimental error of 0.05 eV for
these two metals [11].Employing FEPd ¼ 5.75 eV and ZPd ¼ 2 (Table
1) we calculate UPd ¼ 5:27 eV using Eq. (11).
Excellent agreement with the experimental WF
value suggests that ZPd ¼ 2 is acceptable, and this
value is therefore our choice for the pure Pd lattice.
Consequently for the PdH lattice we have to assume
ZPdH ¼ 1, because one free electron per Pd atom is
covalently bounded with hydrogen. Using theb ¼ 1:035 and FE ¼ 7:05 eV for PdH crystal (Table
1), we calculated the WF for the PdHx system as a
function of x. The results of this calculation are
shown in Table 2 and they are plotted in Fig. 3.
As it is seen from Fig. 3, our results are in
excellent agreement with the experimental values
reported by Du�s and Nowicka [4] when we employ
a scaling factor of 3.0 to their ratio of the ‘‘positivehydrogen’’ to palladium in order to convert that
ratio to the H/Pd ratio in bulk PdHx. It would be
ideal if the rise of H/Pd ratio would be directly
proportional to the surfacial concentration of the
‘‘positive hydrogen’’. However, according to Fig. 5
in Ref. [4], the relationship between bulk and
surfacial ‘‘positive hydrogen’’ concentration is
somewhat nonlinear.
5. Isotope effect
According to Urey [22] the isotope effect results
primary from the difference of the zero-energy
levels of the isotope species. In the quantum wells
Page 6
48 S. Halas et al. / Surface Science 555 (2004) 43–50
shown in Fig. 2 the zero-energy levels of D0 and
D� are located below respective levels of H0 and
H�, while the profiles of the quantum wells are
identical because no differences in the size and
electronic structure of the isotopic species are
encountered. Hence the heavy isotope species D0
and D� are released at higher temperatures than
H0 and H� species. Reversely, when temperature is
lowered then PdDx will be formed earlier than
PdHx. This implies a stronger WF lowering in
isothermic experiments with deuterium than with
light hydrogen.
The quantum well in which the particles H0 and
D0 are confined is relatively wide. Taking intoaccount that the radius of H0 (or D0) atom inside
the lattice cell is 2/3 bohr, we estimated the size of
the quantum well as 1.94 �A and the shift of the
zero-energies of the isotopic species as 0.027 eV.
This value was estimated from the quantum-
mechanical formula for the energy levels of a
square well. If we express this zero-energy shift in
Kelvins, then we have DT ¼ 32 K. It means thatthe same lowering of the WF due to Pd hydroge-
nation by H requires a temperature of 32 K lower
than in the case of D.
6. Discussion
In this section we discuss the advantages andthe shortcomings of the metallic plasma approach
applied to the Pd–H system. Originally this
method was applied by Halas and Durakiewicz [6]
to the WF calculation of polycrystalline metals,
soon after the WF of lanthanides and actinides
have been calculated [23]. Later on the ionization
potentials of small metallic clusters [24] and face-
dependent WF have been calculated by thismethod [25]. In paper [26] Halas and Durakiewicz
have presented the quantitative description of the
WF changes of polycrystalline tungsten the surface
of which is covered by Cs atoms. The initial slopedUdH, where H is the coverage, and the minimum Uvalue were found to be in excellent agreement with
the classical experiments made by Taylor and
Langmuir [27].In this paper we describe the cause of the large
variations of the WF of thin palladium film
deposited on a glass surface in UHV and subjected
to various doses of hydrogen gases (H2 or D2) in
the frame-work of the metallic plasma model. Such
experiments were performed by Du�s et al. [1–4] fortemperatures varying from room temperature
down to 78 K (liquid nitrogen cooling). Otherstudies on Pd–H system were performed recently
for several planes of the single crystal [13,28], but
only for the high temperatures, where the WF of
the covered surface was higher than that of pure
surface.
The calculations performed in Section 3 refer to
the Pd(1 0 0) surface with the single layer of H�
ions. Such a layer may be formed at the beginningof H2 dosing to the evacuated system. It is formed
due to dissociative adsorption of hydrogen and its
spontaneous conversion to H� at room tempera-
ture (or at lower temperatures). This process may
be described in terms of charge transfer. The cal-
culated WF increase due to the formation of a H�
layer is in a good agreement with the experiments
[1–4,13,28] and with the ab initio theoretical cal-culations [29,30]. The simple model presented in
Section 3 has a drawback: the ionic radius (of H�
in this case) is not defined as strictly as the lattice
constant or the atomic radii. It should be also
noted that the d value may be modified due to
absorption of H atoms. We do realize that com-
plex phenomena of hydrogen adsorption on Pd
surface cannot be quantitatively described bysimple electrostatics and geometrical consider-
ations.
In Section 4 we have extended the metallic
plasma model to the Pd metal which has confined
hydrogen atoms in the lattice cells. The absorbed
hydrogen may influence on the WF of the PdHx
because the Fermi energy of PdH is higher than
that of the Pd metal and because of the hydrogenbonding with the neighbouring Pd atoms, thereby
one of the two free electrons (assessed for pure Pd
lattice) becomes localized as the remaining eight
electrons in the valence shell. Hence the Wigner–
Seitz radius, rs, is shifted to high values. Another
reason of the increase of the rs value is well-known
lattice increment of hydrogenated Pd in compari-
son to the pure metal. The calculated results for arealistic b ¼ 1:035 and for b ¼ 1 indicate that the
effect of the lattice increment is relatively small
Page 7
S. Halas et al. / Surface Science 555 (2004) 43–50 49
(Table 2). It seems therefore that the metallic
plasma model for the WF calculation in Section 4
is fully justified. Note, however, that the plasma
model requires the knowledge of Fermi energy
what may be found from the density of the elec-
tronic states of a system. The distribution of theelectronic states may be successfully calculated by
the ab initio methods [29,30] or it may be deter-
mined experimentally by the scanning tunneling
microscopy employing the differential voltage
contrast method [31].
The theoretical explanation presented in Sec-
tion 4 does not require adsorption of a ‘‘positive
hydrogen’’ which was assessed by Du�s and Now-icka [4]. However, recently such an adsorption site
on the Pd(2 1 0) surface for molecular H2 was
discovered by Schmidt et al. [32], were the H2
molecule gets to be highly polarized, thereby a WF
decrease of 0.35 eV was observed at temperatures
below 50 K. This effect disappeared at tempera-
tures above 100 K. Therefore in a very specific
conditions a layer of ‘‘positive hydrogen’’ maybe formed, which further may lower the WF
slightly.
In Section 5 we explain the D/H isotope effect of
different absorption rate with temperature on the
basis of well-known theory of isotope effects [22],
which bases solely on statistical mechanics.
7. Conclusions
In the palladium–hydrogen system two effects
may shift WF simultaneously:
(1) A small WF increase is due to adsorption of H
atoms, the fraction of which is converted to the
negative ions at temperatures below 400 Kaccording to statistical mechanics.
(2) A large WF lowering at temperatures below
120 K are due to PdHx formation by a weak
covalent bonding.
Both effects may be simply explained in the
frame-work of the metallic plasma model. The
large isotope effect of D substitution may be ex-plained by zero-energy difference in the quantum
well in which H or D atoms are confined.
Acknowledgements
Thanks are due to Professor R. Du�s, Institute
of Physical Chemistry of the Polish Academy of
Sciences, for his cordial encouragement, stimulat-ing discussion with the representative of Mass
Spectrometry Laboratory (S. Halas) and help in
collecting of the literature. We appreciate con-
structive criticism of two unknown reviewers of
the manuscript.
References
[1] R. Du�s, E. Nowicka, Z. Wolfram, Surf. Sci. 216 (1989) 1.
[2] E. Nowicka, Z. Wolfram, R. Du�s, Surf. Sci. 247 (1991) 248.
[3] E. Nowicka, R. Du�s, Prog. Surf. Sci. 48 (1995) 3.
[4] R. Du�s, E. Nowicka, Langmuir 16 (2000) 584.
[5] T.B. Grimley, in: W.E. Garner (Ed.), Chemisorption,
Butterworths, London, 1957.
[6] S. Halas, T. Durakiewicz, J. Phys.: Condens. Matter 10
(1998) 10815.
[7] I. Langmuir, Phys. Rev. 43 (1933) 224.
[8] H. Ibach, H. L€uth, An Introduction to Principles of
Materials Science, Springer-Verlag, Berlin, Heidelberg,
1995, Polish edition PWN Warszawa 1996.
[9] Landold-B€ornstein, in: J.L. Olsen, K.-H. Hellwege (Eds.),
Electron States and Fermi Surfaces of Alloys, vol. 13a,
Springer-Verlag, Berlin, 1981.
[10] R. Nowakowski, P. Grzeszczak, R. Du�s, Surf. Sci. 507–510
(2002) 813.
[11] U. Harms, R.B. Schwarz, Phys. Rev. B 65 (2002) 085409-1.
[12] A. Kiejna, K.F. Wojciechowski, Metal Surface Electron
Physics, Elsevier Science, Oxford, 1996.
[13] H. Conrad, G. Ertl, E.E. Latta, Surf. Sci. 41 (1974) 435.
[14] R.J. Behm, K. Christmann, G. Ertl, Surf. Sci. 99 (1980)
320.
[15] CRC Handbook of Chemistry and Physics, 61th, CRC
PRESS Inc., Boca Raton, FL, 1981.
[16] H.E. Gove, Nukleonika 25 (1980) 31.
[17] V.S. Fomenko, Handbook of Thermionic Properties,
Plenum Press Data Division, New York, 1966.
[18] H.B. Michaelson, J. Appl. Phys. 48 (1977) 4729.
[19] Landold-B€ornstein, New Series III/24b, Springer, Berlin,
1994.
[20] J.M. Heras, G. Esti�u, L. Viscido, Appl. Surf. Sci. 108
(1997) 455.
[21] R.G. Wilson, J. Appl. Phys. 37 (1966) 3170.
[22] H.C. Urey, J. Chem. Soc. (1947) 562.
[23] T. Durakiewicz, S. Halas, A1. Arko, J.J. Joyce, D.P.
Moore, Phys. Rev. B 64 (2001) 045101-1.
[24] T. Durakiewicz, S. Halas, Chem. Phys. Lett. 34 (2001) 195.
[25] K.F. Wojciechowski, A. Kiejna, H. Bogdan�ow, Mod.
Phys. Lett. 13 (1999) 1081.
[26] S. Halas, T. Durakiewicz, Annales Universitatis Mariae
Curie-Sklodowska Sectio AAA 54 (1999) 63.
Page 8
50 S. Halas et al. / Surface Science 555 (2004) 43–50
[27] J.B. Taylor, I. Langmuir, Phys. Rev. 44 (1933) 423.
[28] M.G. Cattania, V. Penka, R.J. Behm, K. Christmann, G.
Ertl, Surf. Sci. 126 (1983) 382.
[29] W. Dong, V. Ledentu, Ph. Sautet, A. Eichler, J. Hafner,
Surf. Sci. 411 (1998) 123.
[30] S.Wilke, D. Hennig, R. L€ober, Phys. Rev. B 50 (1994) 2548.
[31] S. Milshtein, E. Thurkins, N. Merritt, K. Novaris, J.
Therrrien, Scanning 23 (2001) 232.
[32] P.K. Schmidt, K. Christmann, G. Kresse, J. Hafner,
M. Lischka, A. Groß, Phys. Rev. Lett. 87 (2001) 096103-1.