Overpotential deposition of Ag monolayer and bilayer on Au(1 1 1) mediated by Pb adlayer underpotential deposition/stripping cycles

Article abstract of DOI:10.1016/S0039-6028(03)00792-1

Ultra-thin Ag films on the Au(1 1 1) surface were prepared via overpotential deposition (OPD) in the presence of Pb²⁺ ions. By carrying out repetitive Pb adlayer underpotential deposition (UPD) and stripping cycles during Ag bulk deposition, th

Full text of DOI:10.1016/S0039-6028(03)00792-1

Surface Science 540 (2003) 230–236
www.elsevier.com/locate/susc
Overpotential deposition of Ag monolayer and bilayer on
Au(1 1 1) mediated by Pb adlayer underpotential
deposition/stripping cycles
a,
b
a
J.X. Wang *, B.M. Ocko , R.R. Adzic
a
Department of Materials Science, Brookhaven National Laboratory, Building 555, Upton, NY 11733, USA
b
Department of Physics, Brookhaven National Laboratory, Upton, NY 11973, USA
Received 27 February 2003; accepted for publication 31 May 2003
Abstract
Ultra-thin Ag films on the Au(1 1 1) surface were prepared via overpotential deposition (OPD) in the presence of Pb2þ
ions. By carrying out repetitive Pb adlayer underpotential deposition (UPD) and stripping cycles during Ag bulk de-
position, the two-dimensional growth of Ag films was significantly enhanced in high OPD. The Ag monolayer sample
was made by comparing the voltammetry curves, in which the signatures for Pb adlayer UPD on Au(1 1 1) changed to
that on Ag(1 1 1). As demonstrated by the X-ray specular reflectivity measurements, nearly complete monolayer and
bilayer films can be made with optimized deposition procedures. On subatomic scale, however, we found that these
films have significant higher root-mean-square displacement amplitudes than those underpotentially deposited Ag
monolayer and bilayer on either Au(1 1 1) or Pt(1 1 1).
ꢀ 2003 Elsevier B.V. All rights reserved.
Keywords: Growth; Surface structure, morphology, roughness, and topography; X-ray scattering, diffraction, and reflection
1. Introduction
altered to a layer-by-layer growth mode by adding
Sb, which acts as a surfactant. Alternatively, the
2D growth of Ag thin films on Ag(1 1 1) can be
induced by ion bombarding the surface with low-
energy Arþ pulses, thereby creating defects [5].
Both procedures, adding surfactant and creating
defects, are believed to enhance the nucleation
density and/or reduce the magnitude of the edge
barrier so that the interlayer transport of depo-
sited adatoms is rapid enough to prevent nucle-
ation of islands on the as yet uncompleted growing
monolayers.
The growth of smooth layers of one material on
foreign substrates has long been an important goal
of surface science investigations. It has been found
that incorporating certain metal atoms during the
deposition process can promote layer-by-layer
growth, yielding smooth deposits [1,2]. For ex-
ample, three-dimensional (3D) island formation of
Ge on Si(1 1 1) [3] or Ag on Ag(1 1 1) [4] can be
Applying the results of the vacuum studies of
surfactant and defect mediated growth to electro-
deposition, Sieradzki et al. demonstrated that the
* Corresponding author. Tel.: +1-631-344-2515; fax: +1-631-
344-5815.
E-mail address: jia@bnl.gov (J.X. Wang).
0039-6028/$ - see front matter ꢀ 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0039-6028(03)00792-1

J.X. Wang et al. / Surface Science 540 (2003) 230–236
231
2D growth of Ag on Au(1 1 1) can be achieved by
using either lead or copper as the mediator [6].
Similar effects have also been observed in the
electrodeposition of Au in the presence of lead [7]
and antimony [8] ions. Although the scanning
tunneling microscopy (STM) images provided a
clear picture for the evolution of surface mor-
phology during the deposition of hundreds of
monolayers [6], an important unresolved issue has
been whether ultra-thin films exhibit layer-by-layer
growth. Can a monolayer or bilayer film be elec-
trodeposited by using these methods in the over-
potential deposition (OPD) regime? The answer
will lead to a better understanding on an atomic
scale of this growth, and the method may be ap-
plied in electrocatalysis. In the latter case, the
formation of a complete monolayer or bilayer is
interesting because electronic effects on catalytic
properties, due to the interaction between the ad-
metal and substrate, decay with the film thickness
rather rapidly.
It is well known that for certain metals mono-
layer or bilayer films can be deposited on many
noble metal surfaces at potentials positive of the
bulk deposition potential. This occurs when the
admetal–substrate bonding is stronger than ad-
metal–admetal bonding and is referred to as the
underpotential deposition (UPD). In some cases,
the UPD metal thin films exhibit properties dif-
fering from the pure metals of either the admetal
or the substrate, depending on the film thickness.
For example, the Ag monolayer on the Pt(1 1 1)
surface adsorbs bisulfate ions in a manner similar
to the Pt(1 1 1) surface, while the Ag bilayer ad-
sorbs sulfate in a manner similar to the Ag(1 1 1)
surface [9]. Modification of surface properties
through thin film deposition is thus of a great in-
terest for electrocatalysis [10]. Since the UPD of a
foreign admetal is a thermodynamic property, it is
not always applicable. Therefore, a general meth-
od of metal adlayer growth, based on kinetic ra-
ther than thermodynamic factors, which may be
applicable to a wide range of systems where UPD
is not a viable option, is of a considerable interest.
In this paper, we describe a deposition proce-
dure for controlled making of well ordered Ag
monolayers and bilayers on the Au(1 1 1) surface
at potentials far below the Ag bulk deposition
potential. It involves the use of Pb UPD/stripping
potential cycles to modify the growth kinetic of the
Ag film. Since Ag monolayer and bilayer can be
formed by UPD on Au(1 1 1) [11,12], the procedure
itself may not have much practical value. Our goal
in this X-ray reflectivity study is using it as an ex-
ample to explore the potential of kinetic controlled
2D growth for ultra-thin film. The system is chosen
because the monolayer formation can be moni-
tored in situ by using the Pb UPD voltammetry
feature and Ag deposits can be easily removed from
gold so that various deposition procedures can be
tested in reasonable time. Unlike microscopic
scanning techniques, which monitor the micro-
scopic surface morphology during the deposition,
X-ray reflectivity measurements, as a global probe
for atomic distribution along the surface normal
direction, are taken after a certain amount depo-
sition is made. A layer-by-layer growth is judged
by whether only one monolayer is found when
the whole surface is fully covered by Ag, and
whether a bilayer forms when the amount of de-
posit is twice as much as that for a monolayer. This
is more strict definition than that based on the
lack of small clusters and multi-atomic height
islands in STM images. On the basis of this defi-
nition, we found that a kinetic controlled Ag layer-
by-layer OPD on Au(1 1 1) can be achieved under
optimized conditions, probably only up to two
monolayers.
2. Experimental
The Au(1 1 1) single crystal (10 mm in diameter),
was electropolished after orientation within 0.15 ꢁC
of the crystalline direction. The crystal was flame
annealed and cooled to below 100 ꢁC when a drop
of MilliQ UV-plus (Millipore) water was placed on
to the crystal surface. The water drop protected the
surface from contamination during the transfers to
an electrochemical cell. A Ag/AgCl/(3 M NaCl)
electrode was used as the reference electrode and a
Pt wire was used as the counter electrode. All po-
tentials reported here have been referenced to the
normal hydrogen electrode (NHE). Solutions were
prepared from Ag2SO4, PbO (99.999%, Aldrich),
and ultra-pure HClO4.

232
J.X. Wang et al. / Surface Science 540 (2003) 230–236
A hanging meniscus configuration was used in
voltammetry measurements and for controlled
electrodeposition of Ag in the presence of Pb2þ
ions. After Ag electrodeposition, the crystal was
removed from the solution underpotential control,
dipped into Millipore water, and then transferred
to the X-ray cell with a drop of water on top of the
crystal surface. This sample transfer procedure has
been verified in our previous works using Ag single
crystals to be effective to prevent Ag surface oxi-
dation in air. Using this procedure, instead of de-
positing Ag in the X-ray cell, ensured that the
change of the fine feature in the voltammetry curve
during Ag deposition could be observed clearly
and used for monitoring the amount of Ag de-
posited. After sealing the X-ray cell with a thin
plastic window, the cell was filled with deoxygen-
ated water. An outer chamber was filled with ni-
trogen gas to keep the cell free of oxygen during
the measurements. Under these conditions, re-
peated measurements show no change of measured
X-ray intensity with time.
Fig. 1. Linear sweep voltammetry curves for Au(1 1 1) in 0.1 M
HClO4 containing 0.1 mM Agþ with (solid line) and without
(dashed line) 10 mM Pb2þ. Sweep rate 10 mV/s.
peaks on Au(1 1 1) [15]. The smaller bulk Ag
stripping peak, compared to that one in the ab-
sence of Pb, indicates that the presence of the Pb
decreases the rate of Ag deposition.
X-ray specular reflectivity measurements were
ꢀ
performed at beam line X22A with k ¼ 1:20 A at
the National Synchrotron Light Source. Following
convention, a hexagonal coordinate system was
used for the Au(1 1 1) crystal in which the reci-
procal-sꢀpace vector along the surface direction was
Fig. 2 shows the voltammetry curves obtained
for the OPD of Ag monolayer and bilayer on a
freshly prepared Au(1 1 1) electrode in 0.1 M
HClO4 solution containing 0.1 mM Agþ ions and
10 mM Pb2þ ions. To mediate and control the 2D
growth of Ag monolayer in OPD potential region,
the electrode potential was continuously cycled at
30 mV/s between )0.1 (or )0.08 V) and 0.1 V, a
potential range where the Pb monolayer is alter-
natively deposited and removed. The potential
cycling was initiated at a potential just negative of
the UPD features and reversed at a potential
where the Pb UPD layer is completely stripped. As
shown in Fig. 2a, whereas the current peaks at
higher potential decrease with cycling, the low-
potential current peaks increase with the number
of cycles. Since the small Pb UPD peak at higher
potential occurs on Au(1 1 1), but not on Ag(1 1 1),
the decrease of this pre-peak and the increase of
the major peak at lower potential indicate that the
Au(1 1 1) surface becomes gradually covered by
Ag. When the peak at higher potential vanished
(after about the thirteenth positive potential
ꢀ
ꢀ
Qz ¼ Lc , where c ¼ 2p=c and c ¼ 7:064 A. A
2 · 2-mm detector slit was located 650 mm from
the sample.
3. Results and discussion
Linear sweep voltammetry curves for Ag UPD
and OPD on Au(1 1 1) in HClO4 solution con-
taining 0.1 mM Agþ ions without and 10 mM Pb2þ
ions are shown in Fig. 1. A low Agþ concentration
was chosen to make the rate of OPD of Ag slow.
The current peaks at potentials positive of 0.563 V
are due to the UPD of Ag. These peaks are not as
sharp and symmetric as the literature data [13,14],
obtained at higher concentrations where there is
no diffusion limitation for Ag deposition. The
presence of Pb2þ ions gives rise to the additional
peaks close to zero volts. These features are similar
to those observed for the case of the Pb UPD

J.X. Wang et al. / Surface Science 540 (2003) 230–236
233
Fig. 3. X-ray specular reflectivity profile for the Ag monolayer
on a Au(1 1 1) electrode. The dotted line is calculated for an
ideal Au(1 1 1) surface. The solid line shows the best fit and the
dashed line is calculated as described in the text.
Fig. 2. (a) Linear sweep voltammetry curves showing the de-
crease of the high potential peak and the increase of the low
potential peak during the overpotential deposition of a Ag
monolayer on Au(1 1 1) mediated by repeated Pb UPD/strip-
ping potential cycles in 0.1 M HClO4 containing 0.1 mM Agþ
and 10 mM Pb2þ. Sweep rate 30 mV/s. (b) Voltammetry curves
recorded during formation of the Ag bilayer sample, which was
carried out under the same condition as for the monolayer with
the number of potential cycles doubled.
sweep) the electrode was disconnected, rinsed with
pure water, and transferred to the X-ray cell for
the X-ray specular reflectivity measurement.
Samples for monolayer deposition were also made
by potential cycling at 10 and 50 mV/s, or by
pulsing potential between )0.1 and 0.1 V. The
most complete and flat monolayer was obtained by
the potential cycles at 30 mV/s, as described above.
For the bilayer sample, as shown in Fig. 2b, the
number of potential cycles was 26, twice as many
as that for the monolayer deposition.
The specular reflectivity profiles obtained for the
monolayer and bilayer Ag deposits on Au(1 1 1)
are shown in Figs. 3 and 4. Both curves show
distinct features between the Bragg peaks that are
different from those obtained for the ideally ter-
minated Au(1 1 1) surfaces. In the kinematic ap-
proximation, the specular reflectivity is related to
Fig. 4. X-ray specular reflectivity profile for the Ag bilayer on a
Au(1 1 1) electrode. The dotted line is calculated for an ideal
Au(1 1 1) surface. The solid and dashed lines are the fits de-
scribed in the text.
the sum over atomic layers with the appropriate
atomic form and phase factors for each layer. In
our model, the scattering factor for the surface

234
J.X. Wang et al. / Surface Science 540 (2003) 230–236
layers incorporates a nearly ideally terminated
Au(1 1 1) surface with up to three adlayers, given
by
supports the model with a pure Ag monolayer.
The possibility of surface alloying, i.e., an Au–Ag
place exchange, can also be ruled out based on the
unity coverage for both the Au top layer and the
Ag adlayer. This is because the fitted Ag coverage
would be higher than unity and that of the top Au
layer would be lower than unity as the electron
density shifts from the top layer of the substrate to
the adlayer by exchanging Au (electron-rich,
Z ¼ 79) with Ag (having a considerably smaller
number of electrons, Z ¼ 47).
2
2
0
F ðLÞ ¼ fAuðLÞeꢁq r =2e2piLZ
0
3
X
2
2
j
j
þ
hjfjðLÞeꢁq r =2e2piLZ =c
;
ð1Þ
j¼1
where f is atomic factor and q is the reciprocal-
space vector along the surface direction, which is
ꢀ
equal to 2pL=c with c ¼ 7:064 A for Au(1 1 1). The
adjustable parameters are the layer spacings rep-
resented by the position along the surface normal
direction, Zj, the vertical root-mean-square (rms)
displacement amplitudes, rj, and the coverage
with respect to the atomic density of the Au(1 1 1)
surface, hj. The density of the top Au layer is fixed
at unity because we have checked by in-plane
diffraction that the Au(1 1 1) surface reconstruc-
tion is lifted by Pb UPD.
To facilitate a comparison between the X-ray
specular reflectivity profiles for our monolayer
sample and for that obtained for the Ag UPD
monolayer [12], a dashed line is shown in Fig. 3
which is calculated with the same parameters as
those for the solid line except a lowered rms dis-
placement amplitude for the Ag adlayer (from 0.45
ꢀ
to 0.15 A). This curve is rather similar to that for
the Ag UPD monolayer and thus indicates that
there is a significant difference in subatomic scale
between the two monolayer-films. We have found
that the rms value for the Ag UPD monolayer on
For the monolayer sample, as shown in Fig. 3,
the measured specular reflectivity is well repro-
duced (solid line) with one nearly complete Ag
monolayer (i.e., h1 ¼ 0:98 and h2 ¼ h3 ¼ 0). Other
fitted parameters are the Ag–Au layer spacing
ꢀ
Pt(1 1 1) is 0.09 A and in general the rms value for
ꢀ
a UPD adlayer is less than 0.2 A. Therefore, the
high rms displacement amplitude likely originates
from the deposition method.
ꢀ
(Z1 ꢁ Z0 ¼ 2:36 A), the top Au layer spacing
ꢀ
(Z0 ¼ 2:35 A) and the rms amplitudes for the Ag
ꢀ
adlayer (r1 ¼ 0:45 A) and the top Au layer
For the Ag bilayer sample (Fig. 4), the charac-
teristic feature in the specular reflectivity curve is
the double dip between the origin and the (0 0 3)
Bragg positions. The best fit (shown by the solid
line) was obtained with h1 ¼ 0:95, h2 ¼ 0:79, and
h3 ¼ 0. The layer spacings are 2.35, 2.36, and 2.39
ꢀ
(r0 ¼ 0:14 A). Since X-ray reflectivity measures the
electron density profile along the surface normal
direction, and hence is not directly atomic sensi-
tive, there are unlimited possibilities for a mixed
Ag–Pb adlayer to have the electron density
equivalent to that for a 0.98 monolayer Ag. They
are, however, unlikely because the adlayer–Au
spacing clearly supports Ag atoms as the adsor-
bates. Pb atoms are about 20% larger than Au
atoms, while Ag has nearly the same lattice con-
stant as Au. The adlayer–substrate spacing ob-
ꢀ
A for the top Au layer, the first and second Ag
adlayers, respectively. Fittings with fixed nonzero
third Ag layer coverage were carried out to see if
other models also reproduce the reflectivity curve.
A typical result is shown by the dashed line, which
is obtained by fixing the coverage for the third Ag
adlayer at 0.1, while allowing other parameters to
ꢀ
tained from the best fit is 2.36 A, which is close to
ꢀ
ꢀ
the Au(1 1 1) bulk spacing of 2.355 A and far less
vary. The third layer spacing (3.05 A) and rms
ꢀ
than the Pb–Au layer spacing of 2.65 A, found for
ꢀ
amplitude (1.2 A) are both much higher than those
the Pb UPD adlayer on Au(1 1 1) [16]. If a signi-
ficant amount of Pb were in the adlayer, the ad-
layer–substrate layer spacing would be larger than
the Au(1 1 1) bulk spacing, which would cause
visible asymmetry in the reflectivity curve near the
Bragg positions. The absence of this feature clearly
for the first two layers, suggesting that there is no
tightly bounded third Ag adlayer. The quality of
the bilayer film is, however, quite different from
that made by UPD. The fitted rms displacement
ꢀ
amplitudes are 0.38 and 0.76 A for the first and
second Ag layers, respectively, significantly higher

J.X. Wang et al. / Surface Science 540 (2003) 230–236
235
than the those found for the Ag UPD bilayer
on the Au(1 1 1) and Pt(1 1 1) surfaces [12,17]. The
notable feature related to the large rms values
is the absence of the double dip in the specular
reflectivity curve between the (0 0 3) and (0 0 6)
Bragg positions. In order to explore the reason
for the higher rms values, a model that allows a
small amount of Pb (which should have larger
spacing than Ag) to be in the second adlayer was
also used. With the layer spacings fixed at the ideal
not surprising because the OPD of Ag results in 3D
islands in the absence of Pb ions and the layer-by-
layer growth relies on a kinetic manipulation in-
volving Pb UPD/stripping cycles. Since the smaller
a 2D island is, the easier the adatom diffusion to its
edge is, the mechanism of surfactant mediated
growth is believed to involve a restraining the size
of 2D nuclei to prevent growth of a new layer on
the uncompleted growing monolayers. For the Pb
mediated Ag deposition studied here, formation of
large 2D Ag islands is blocked by co-deposition of
Pb in the negative potential sweep. Stripping of the
Pb monolayer in the positive potential sweep lib-
erates those Pb-covered sites to allow further Ag–
Pb co-deposition in the next potential cycle. Due to
the continuous Ag OPD, Pb adatoms may be
buried into the Ag film if the potential was kept
negative of the UPD current peaks for an extended
time. By using potential cycles to repeatedly de-
posit and remove Pb, Ag–Pb place exchange is ef-
fectively enhanced. All these effects depend on the
experimental detail, so does the quality of the
mediated 2D growth.
ꢀ
values, Au–Ag and Ag–Ag at 2.355 A and Ag–Pb
ꢀ
at 2.65 A, a reasonable fit was obtained with the
ꢀ
rms displacement amplitudes of 0.4 and 0.7 A
for the first Ag layer (0.97 ML) and the second
Ag (0.76 ML)–Pb (0.03 ML) mixed layer, respec-
tively. Since the rms value for the second layer
is only slightly reduced in this model compared
ꢀ
to that without any Pb (from 0.76 to 0.70 A),
the large rms value for the second adlayer cannot
be fully accounted by a possible Ag–Pb mixed
second layer. Similarly to the monolayer case,
the rougher bilayer film compared to those ob-
tained by UPD, is likely due to the lack of ‘‘an-
nealing’’ effect of reversible adsorption–desorption
under thermodynamic equilibrium, which helps
the UPD adlayers to achieve a highly ordered
state.
4. Conclusions
In addition to the potential cycling method de-
scribed above, Pb mediated Ag depositions were
also carried out at a fixed potential, where a partial
Pb monolayer exists, and with potential pulses
between the low and high limits of the Pb UPD
potential region. These two methods were suc-
cessfully employed in 2D growth of Ag films on
Au(1 1 1) monitored with STM [6]. In the present
study, we have had a limited success in making a
complete overpotential deposited Ag monolayer by
these two methods. The specular reflectivity pro-
files obtained from the three samples made by
pulsing potential between )0.1 and 0.1 V failed to
show layer-by-layer growth at monolayer level. For
these deposits, two partial layers were required to
fit the specular reflectivity profiles even the sum of
the coverage for the two layers was equal to or less
than one monolayer. These observations demon-
strated that the uniformity of the film on an atomic
scale is sensitive to the details on how the potential
was controlled in the deposition procedure. This is
Layer-by-layer OPD of Ag monolayer and bi-
layer on Au(1 1 1) mediated by Pb UPD/stripping
cycles was investigated in order to explore the
methods for enhancing 2D growth of ultra-thin
metal films under OPD conditions. As demon-
strated by the X-ray specular reflectivity mea-
surements, nearly complete monolayer and bilayer
films can be made with optimized deposition pro-
cedures. These results indicate that the 2D growth
is enhanced through blocking the formation of
large Ag islands at low coverage by the co-
adsorption of Pb, and through enforced place
exchange during Pb stripping processes. Both ef-
fects keep the interlayer transport of deposited Ag
adatoms rapid enough to prevent nucleation of
islands on the as yet uncompleted growing mono-
layer. On subatomic scale, however, for films ob-
tained using this procedure we found that it is
rather difficult to match the smoothness of thin
films made by UPD.

236
J.X. Wang et al. / Surface Science 540 (2003) 230–236
Acknowledgements
[7] J.D.E. McIntyre, W.F. Peck, J. Electrochem. Soc. 123
(1976) 1800.
ꢁ
ꢂ ꢁ
[8] D. Davidovic, R.R. Adzic, Electrochim. Acta 33 (1988)
This work is supported by US Department
of Energy, Divisions of Chemical and Material
Sciences, under the contract no. DE-AC02-98-
CH10886.
103.
[9] N.S. Marinkovic, J.X. Wang, J.S. Marinkovic, R.R. Adzic,
ꢂ ꢁ
J. Phys. Chem. B 103 (1999) 139.
ꢂ ꢁ
[10] R.R. Adzic, in: H. Gerischer (Ed.), Adv. Electrochem. &
Electrochem. Eng., vol. 13, J. Wiley & Sons, New York,
1984, p. 159.
References
[11] M.J. Esplandiu, M.A. Schneeweiss, D.M. Kolb, Phys.
Chem. Chem. Phys. 1 (1999) 4847.
[12] T. Kondo, J. Morita, M. Okamura, T. Saito, K. Uosaki, J.
Electroanal. Chem. 532 (2002) 201.
[1] M. Copel, M.C. Reuter, M. Horn-von Hoegen, R.M.
Tromp, Phys. Rev. B 42 (1990) 11682.
[13] C.H. Chen, S.M. Vesecky, A.A. Gewirth, J. Am. Chem.
Soc. 14 (1992) 451.
[2] M. Horn-von Howgen, F.K. LeGoues, M. Copel, M.C.
Reuter, R.M. Tromp, Phys. Rev. Lett. 67 (1991) 1130.
[3] G. Meyer, B. Voigtlander, N.M. Amer, Surf. Sci. 274
(1992) L541.
[14] D.M. Kolb, in: R.C. Alkire, D.M. Kolb (Eds.), Advances
in Electrochemical Science and Engineering, vol. 7, Wiley
VCH, Weinheim, Germany, 2002, p. 107.
[15] M.F. Toney, J.G. Gordon, M.G. Samant, G.L. Borges,
O.R. Melroy, D. Yee, L.B. Soresen, J. Phys. Chem. 99
(1995) 4733.
[4] H.A. van der Vegt, H.M. van Pinxteren, M. Lohmeier,
E. Vlieg, J.M.C. Thornton, Phys. Rev. Lett. 68 (1991)
3335.
[5] G. Rosenfeld, R. Servaty, C. Teichert, B. Poelsema, G.
Comsa, Phys. Rev. Lett. 71 (1993) 895.
[16] Unpublished results.
[17] J.X. Wang, N.S. Marinkovic, R.R. Adzic, B.M. Ocko,
Surf. Sci. 398 (1998) L291.
ꢂ ꢁ
[6] K. Sieradzki, S.R. Brankovic, N. Dimitrov, Science 284
(1999) 138.