Electroless deposition of metal nanoislands on aminothiolate-functionalized Au(111) electrodes

Article abstract of DOI:10.1021/jp981684o

We show that selective electroless deposition (ELD) can be used as a tool for electrochemical nanostructuring. In a first step, we bound palladium ions to an aminothiolate (AT) layer on an Au(111) surface. Chemical reduction then served for the fabrication of metallic palladium islands of monatomic height that can in turn activate the ELD of cobalt. We studied the growth of cobalt islands as a function of the oxygen concentration in the deposition bath and of the aminothiolate concentration in mixed amino-/alkanethiolate self-assembled monolayers (SAMs). In situ scanning tunneling microscopy (STM) and X-ray photoelectron spectroscopy (XPS) measurements showed that palladium and cobalt form islands of 1-3 nm up to 60 nm diameter, the size being controlled by the oxygen concentration in the bath.

Full text of DOI:10.1021/jp981684o

7582
J. Phys. Chem. B 1998, 102, 7582-7589
Electroless Deposition of Metal Nanoislands on Aminothiolate-Functionalized Au(111)
Electrodes
Hannes Kind,* Alexander M. Bittner, Ornella Cavalleri, and Klaus Kern
Institut de Physique Expe´rimentale, EÄ cole Polytechnique Fe´de´rale de Lausanne,
CH-1015 Lausanne, Switzerland
Thomas Greber
Physik-Institut, UniVersita¨t Zu¨rich, CH-8057 Zu¨rich, Switzerland
ReceiVed: March 30, 1998; In Final Form: June 26, 1998
We show that selective electroless deposition (ELD) can be used as a tool for electrochemical nanostructuring.
In a first step, we bound palladium ions to an aminothiolate (AT) layer on an Au(111) surface. Chemical
reduction then served for the fabrication of metallic palladium islands of monatomic height that can in turn
activate the ELD of cobalt. We studied the growth of cobalt islands as a function of the oxygen concentration
in the deposition bath and of the aminothiolate concentration in mixed amino-/alkanethiolate self-assembled
monolayers (SAMs). In situ scanning tunneling microscopy (STM) and X-ray photoelectron spectroscopy
(XPS) measurements showed that palladium and cobalt form islands of 1-3 nm up to 60 nm diameter, the
size being controlled by the oxygen concentration in the bath.
1. Introduction
From an electrochemist’s point of view, small aggregatess
such as the Pd or Co islands that we investigatedsform an
ultramicro- or even nanoelectrode array. Such electrodes of
geometrical dimension smaller than the Nernst diffusion layer
have important advantages pointed out earlier by Ewing et al.11
They can be used to study fast electron-transfer reactions,12,13
to perform in vivo voltammetric measurements inside living
brain tissue14 and inside single cells,14-16 to observe changes
in the conductivity of single ion channels,17 and to perform
small-scale etching and lithography.18 All these applications
are mainly based on the fact that redox reaction kinetics and
spatial resolution of electrodes scale inversely with the electrode
radius, i.e., small currents can be sampled on a small area.13,19,20
Electroless deposition (ELD) is an autocatalytic redox process
in which metal ions are chemically reduced to metal at a surface
in absence of any external current source.1-6 A cation of the
metal to be deposited is reduced by receiving electrons from
the surface of a metal substrate or from the surface of the
catalysts used to initiate the deposition. The reductant in turn
delivers electrons to this surface and is thereby oxidized. This
redox process generally takes place only on catalytically active
metal surfaces: Noncatalytic surfaces first have to be activated
with an appropriate catalyst before the metal deposition can take
place. The growing metal surface itself has of course to be
catalytically active for the process.
Metal island nanopatterning requires a selective deposition
that can be achieved either by a selective deactivation of a
catalytic substrate or by selective activation of an inactive
surface by a catalyst.21 Several methods of producing patterned
catalysts are known, often based on photolithographic tech-
niques.22,23 Other approaches to selective metallization are
chemical vapor deposition (CVD)24 and microcontact printing
(µCP).25-28 Recently direct patterning with a catalyst by MCP
of colloids has been demonstrated by Hidber et al.29
The method of ELD has long been used in the electronic
industry for the metallization of plastics, ceramics, and other
insulating substrates. However, this was based more or less
on a trial and error approach. Although several fundamental
studies of ELD have been conducted (see, e.g., [refs 2, 3, and
7-9]), the mechanisms of ELD are not fully understood yet.
In this paper, we propose ELD on a thiolate surface as model
for the metallization of organic substrates.
Our interest in ELD is also related to the patterning of
nanometer-sized magnetic objects placed on an organic spacer.
The study of ferromagnetism of ultrathin films and small
aggregates has shown a very interesting correlation between
elastic and magnetic properties.10 A better study of the
magnetism as a function of size and geometry of nanostructures
on top of a spacer will allow an interesting comparison with
systems without spacers. In our case the thiolate layer should
help to minimize any electronic and magnetic interaction
between the substrate and magnetic objects on top of the spacer.
It will also be possible to study the long-range interaction
between different objects.
The resolution of submicrometer metallic features obtained
by a selective ELD approach depends on the size of the catalyst
particle.30 Minimal attainable feature line widths will be limited
by the largest catalyst particle on the surface.31 So far the
dimensional polydispersity has limited the resolution at which
a particular size/shape-dependent characteristic of metal particles
may be probed.
Our approach to circumvent this problem is based on the
special properties of doubly functionalized molecules. Doubly
functionalized self-assembling molecules are very interesting
candidates to pattern substrates by MCP.25-28 While one
functional group selectively binds to the surface, the other can
be tailor-made to control the external interfacial properties.
* Corresponding author. E-mail: hannes.kind@epfl.ch. Fax: (++41) 21
693 36 04.
S1089-5647(98)01684-8 CCC: $15.00 © 1998 American Chemical Society
Published on Web 09/09/1998

Electroless Deposition of Metal Nanoislands
J. Phys. Chem. B, Vol. 102, No. 39, 1998 7583
Particularly attractive are functionalized thiols that can bind with
the sulfur to metal surfaces and form strongly adsorbed, ordered
monolayers.32-34 A variety of metal ion ligands such as
pyridine, 2,2′-bipyridine, amine, and ethylenediamine can be
used as external functional groups.22,35,36 Metal ions such as
Pd2+ (which upon reduction becomes the catalyst Pd0) can bind
to these functional groups. The binding mechanism has been
shown to be highly selective for different surface ligands.37
Hence, the ability to control the strength of the bond between
the self-assembled monolayer (SAM) and a catalyst precursor
at the molecular level can offer control of the adhesion of the
catalyst and hence of metal structures deposited on the catalyst.
We describe here a new method for selective ELD of only
one-monolayer thin Pd metal films in which binding of the Pd2+
solution species by amine ligands anchors the Pd2+ to the
organic surface. The reductant (dimethylamine-borane, DMAB)
then reduces Pd2+ to Pd0. In a second step we focus on the
ELD of nanosized islands of Co on these catalytic sites. Our
experimental methods are in situ electrochemical scanning
tunneling microscopy (STM) at 300 K, UV spectroscopy of the
Pd2+ solution and X-ray photoelectron spectroscopy (XPS) of
emersed samples.
SCHEME 1: Pd(II) Species in Aqueous Solution
and then annealed for several hours at ∼600 K (2 × 10-6 mbar).
Directly before the self-assembly was started, the gold films
were flame-annealed at very dark red glow and cooled in
ethanol. For the self-assembly, the samples were then quickly
transferred into 20 µM ethanolic solutions of the aminothiol.
The substrates were left immersed between 17 and 48 h at ∼325
K, thereafter several more days at 300 K. For the experiments
with mixed monolayers, the samples were transferred into
different mixtures of 20 µM ethanolic solution of the aminothiol
and 20 µM ethanolic solution of decanethiol. These substrates
were only left immersed 17 h at ∼325 K. All samples were
rinsed with ethanol and used immediately.
2. Experimental Techniques
Sample Preparation. The samples were dried in air and
then mounted in a PCTFE STM cell, covered by a drop of the
Pd2+ catalyst solution for 2 min, rinsed three times with H2O,
and finally covered by the Co metallization bath. For XPS
analysis the thiolate-covered substrates were rinsed with ethanol
and then dipped into the Pd2+ catalyst solution for 2 min, rinsed
with H2O, and introduced via a fast entry lock into the vacuum
system. All measurements were made at room temperature.
Instrumentation. The STM measurements were performed
with a Besocke beetle type STM modified for in situ electro-
chemical measurements.39,40 Additionally the microscope was
insulated against vibration and electromagnetic fields by an
acoustic cabin which at the same time serves as a Faraday cage.
The microscope was suspended in the cabin by four elastic
strings to suppress the vibrations of the building. All STM
images are shown derivatized. Note that the height difference
between Au(111), Pd(111), and Co(001) monatomic steps is
too small to be detected with our STM. The X-ray photoelec-
tron spectra were collected in a VG ESCALAB 220. To
minimize X-ray tube induced damage, the spectrometer trans-
mission was maximized with lowest angular resolution and an
overall energy resolution of 1.4 eV fwhm on the Au 4f7/2 peak
at 84.0 eV binding energy. The X-ray twin anode that provides
nonmonochromatized Mg KR (1253.6 eV) radiation was
retracted and run with 70 W input power. For all spectra the
Au 4f7/2 peak was set to 84.0 eV binding energy. The emission
angles were 0° (normal emission) and 80° (grazing emission).
UV-absorption spectra of the Pd2+ solutions were obtained using
a Shimadzu UV-260 spectrometer.
Chemicals. Milli-Q water (18 MΩ resistivity; Millipore
Systems) was used for all experiments. All reagents were p.a.
grade or better and were used as received. Ethanol, NaCl,
Na4EDTA, NaOH, NH4Cl, dimethylamine-borane complex
(DMAB, (CH3)2NHBH3), decanethiol, H2SO4, and HCl were
all from Fluka. Na2PdCl4 and CoCl2 were both from Aldrich
Chemical Co. The aminothiol H2N(CH2)2NHCO(CH2)10SH was
used as received from H. Vogel (Institut de Chimie Physique,
EÄ cole Polytechnique Fe´de´rale de Lausanne). Oxygen and
nitrogen (48 res. 50 grade from CarbaGas, CH) gases were
purified by passage through H2SO4 and saturated with H2O by
bubbling through H2O.
Solutions. Glassware was cleaned with Nochromix solution
(Thomas Scientific, U.S.), which reacts violently with many
organic materials and should be handled with care. The
electroless Co metallization bath was prepared immediately prior
to use by mixing three parts of stock Co solution, two parts of
stock solution of DMAB, and five parts of H2O. The Co stock
solution was prepared by adjusting a solution of 1 g of NH4Cl,
0.6 g of CoCl2, and 0.9 g of Na4EDTA in 100 mL of H2O to
pH ) 8 using a 2 M NaOH (aqueous) solution. Stock reductant
consisted of a solution of 1.7 g of DMAB in 50 mL of H2O.
Pd2+ catalyst solution was prepared of 10 mg of Na2PdCl4 and
1.75 g of NaCl in 50 mL of H2O adjusting the solution with
37% HCl to pH ) 1. This yielded the desired [Pd2+
]
of 0.7
tot
mM and [Cl-]tot of 0.7 M and showed the typical pale yellow
2-
color of dilute PdCl4 solutions.38 All solutions were used
within 4 weeks.
The pH of the EL Co deposition bath measured over 24 h at
room temperature showed that the bath is stable at these
conditions. Bubbling of clean nitrogen through the EL plating
solution and simultaneous scratching of a pipet against the
beaker wall started the bath. A Co layer deposited at the beaker
wall was visible to the naked eye. The ELD was always
accompanied by the formation of small hydrogen bubbles at
the solid/liquid interface.
Substrate Preparation and Monolayer Formation. The
gold samples were prepared by evaporation of 120-150 nm
99.99% gold (Advent; Halesworth, U.K.) on 570 K-preheated
cleaved mica (Balztec; Vaduz, LIE) in a 10-5 mbar vacuum
3. Pd2+ Catalyst Experiments
Pd2+ Catalyst Solution. Understanding of the Pd2+ activa-
tion is based on knowledge of the chemical composition of the
activation bath. The chemistry of Pd2+ halogen complexes in
aqueous solution is dominated by hydrolysis above pH ∼ 241-44
where hydroxo-and/or chloro-bridged Pd2+ oligomers precipi-
tate. The mechanisms that dominate the chemistry in an
aqueous solution of PdCl42- are shown in Scheme 1. The Cl-
ligands are replaced one after another by H2O.45-48 Deproto-
nation of the hydrolyzed species PdCl3(H2O)- and PdCl2(H2O)2
is described by eqs 3 and 4. The resulting hydroxo complexes

7584 J. Phys. Chem. B, Vol. 102, No. 39, 1998
Kind et al.
Figure 1. Schematic representation of the ELD of Pd islands and Co/Pd islands. Pd2+ binds to the amino groups in the thiolate layer. Dimethylamine-
borane (DMAB) reduces Pd2+ and Co2+ to the respective metals. Possible configurations of the Pd2+ bonds with the AT-SAM molecules are
shown.
easily lead to the formation of hydroxo- and/or chloro-bridged
oligomers.37,41-44
it was seen from the O ls detail spectra (not shown) that the
untreated as well as the Pd2+-treated samples show only minute
amounts of oxygen although the second had been exposed to
air for more than 30 s. As described elsewhere39 for alkanethi-
olate films, we performed X-ray aging experiments for the AT-
SAM. Intensity curves for the Au and Pd signal indicate a
similar destruction of the AT-SAM with X-ray exposure time.
For this reason exposure time was minimized to less than a
few minutes.
Our purpose was to bind only one layer of Pd2+ ions (see
Figure 1) to the amine-terminated surface. This necessitates
that the concentration of hydroxo-and/or chloro-bridged oligo-
mers (for simplicity we call them “Pd nodules” from here on)
is kept as low as possible. For this reason we calculated from
the reaction constants the required concentration of H+ and Cl-
to have almost only PdCl4 present in the solution.42,43,47,49,50
2-
In Figure 2 the normalized concentrations of the species are
shown for pH ) 1 and 5 as a function of the Cl- concentration.
For our experiments we chose a Pd2+ solution at pH ) 1 with
[Cl-] ) 0.7 M. In this case the solution contains 86.6%
By varying the Pd2+ activation time between 1 and 15 min,
we found in STM measurements that the amount of unwanted
Pd nodules (not shown) increases with activation time. These
nodules are bigger than 20-30 nm in diameter and more than
5-10 monolayers high. Nevertheless we always found flat areas
of more than 2.3 × 104 nm2 free of these nodules. Rinsing of
the Pd2+-activated samples with a solution of [Cl-] ) 0.7 M at
pH ) 1 instead of water reduced the number and size of the
remaining Pd nodules further. Pd2+ activation times between
1 and 2 min were finally determined as best compromise for a
sufficient activation without too many Pd nodules. In passing
we note that careful rinsing should be sufficient to eliminate
all excess Pd species that are not chemically bound to the
surface.
PdCl42-, 12.4% PdCl3(H2O)-, 0.9% PdCl2(H2O)2, 1.3 × 10-5
%
PdCl3(OH)2-, and 5 × 10-4% PdCl2(H2O)(OH)- (calculated).
A comparison of the corresponding UV spectra (not shown)
with spectra obtained by other groups37,45-48 provides further
2-
support that under these conditions PdCl4 is indeed the
majority species.
Pd2+-Activation Experiments. Let us first focus on the
chemical composition of the Pd2+-activated aminothiolate SAM
(AT-SAM). An XPS analysis was performed to investigate the
vertical position of the Pd2+ relative to the monolayer. In Figure
3 Au 4d and Pd 3d XP spectra are shown for grazing (80°) and
normal (0°) emission from a Pd2+-activated AT-SAM. The Pd
3d5/2 peak is clearly visible in grazing emission at ∼338.2 eV
and corresponds to that of Pd2+ bound to nitrogen and/or
chlorine.37 A component at 336 eV that should represent the
oxo- or hydroxo-bridged Pd sites in the Pd nodules37 was never
found. In normal emission the Au signal is substantially
increased while the Pd signal has disappeared. This can only
be explained by a model in which the Pd resides on top of the
aminothiolate film.37
Pd2+ Reduction Experiments. It is known that only metallic
Pd0 can act as the catalyst for the ELD but not Pd2+. This means
that prior to ELD of the metal (in our case Co) the Pd2+ species
have to be reduced by the reductant (DMAB).
We performed in situ STM measurements to estimate the size
and the surface density of such Pd0 particles. After the
activation with Pd2+, the catalyst solution was removed with a
pipette; the samples were then carefully rinsed three times with
H2O and were immediately covered by a mixture of two parts
of the DMAB stock solution and eight parts of H2O. A few
minutes later the cell was mounted in the STM and measure-
ments were started.
A comparison of the overview spectra (not shown) of pure
AT-SAM showed that the samples are clean and not damaged
during the 2 min of Pd2+ activation at pH ) 1. Furthermore,

Electroless Deposition of Metal Nanoislands
J. Phys. Chem. B, Vol. 102, No. 39, 1998 7585
Figure 4. STM image of Pd islands (bright spots) on a Pd2+-activated
AT-SAM electrode. The island height is one Pd monolayer. Areas of
different gray shades (i.e., different heights) are substrate terraces.
Bath: 50 mM DMAB. Tunneling parameters: I ) 1 nA, U ) 70 mV.
particles form islands that are homogeneously distributed on
the surface (see Figure 4). We only found Pd islands of
monatomic height. (The interpretation of a Pd island’s height
is highly complex since the tunneling mechanism on thiolates
and Pd species should be very different. However, a step
between a first and a second Pd layer (that we never found)
should exhibit the real Pd step height. Further, the steps between
the first Pd and the following layers of Co should give an
accurate estimate for the Co step height.) The average diameter
of the islands was between 1.5 and 6 nm (30-500 Pd atoms
per island). Owing to some variability during the Pd2+
activation (quality of rinsing with H2O, contact time with Pd2+
solution), the coverage of the surface by the islands varied
between 1 and 4%. Assuming a thiolate coverage of typically
1/3 on Au(111) the Pd/NH2 ratio is between 3 and 12%.
The reason for the low surface coverage of Pd islands has
two origins: First, at pH ) 1 a non-negligible amount of amino
groups will be protonated which reduces the capability to bind
Pd2+. Second, the reduction of the Pd2+ species to Pd0 decreases
the strength of the bond to the amino group, and some Pd0 is
probably lost.
Figure 2. Normalized composition of the Pd2+ activation solution at
(a) pH ) 1 and (b) pH ) 5 as a function of [Cl-]tot. The aqueous
solution contains Na2PdCl4, NaCl, and HC1.
4. Electroless Co Deposition Experiments
A prerequisite to understand the ELD on Pd2+-activated AT-
SAM is the ELD on a bare metal and a nonactivated AT-SAM
surface. After presenting relevant experiments, we focus on
Co deposition studies at Pd2+-activated AT-SAMs. Finally we
discuss the influence of aminothiolate concentration in mixed
aminothiolate/alkanethiolate monolayers and show the depen-
dence of the Co plating on the presence of oxygen.
Electroless Co Deposition on Bare Gold. To test the
activity of the Co plating bath we performed an experiment on
a gold wire. The gold wire (99.99%) was first cleaned in
Nochromix solution, rinsed with H2O, and finally immersed in
a Co plating bath. After 10-20 min, the Co deposit was visible
to the naked eye. The deposition could be accelerated either
by bubbling nitrogen through the bath or by increasing the bath
Figure 3. Normal and grazing emission XPS of the Au 4d and Pd 3d
core level region of a Pd2+-activated AT-SAM. The increase of the
ratio Pd 3d/Au 4f at grazing emission points toward Pd2+ on top of the
AT-SAM. At normal emission the Pd 3d signal is too low to be
resolved.
It has been shown by Dressick et al. that Pd2+ particles in
the presence of DMAB are completely transformed to Pd0.37 In
our work we determined shape and density of such Pd0 particles
by STM. Images of the activated surface show that the Pd0

7586 J. Phys. Chem. B, Vol. 102, No. 39, 1998
Kind et al.
Figure 6. STM image of Co islands on a mixed Pd2+-activated
aminothiolate/alkanethiolate-covered Au(111) electrode for a low
aminothiolate concentration. Bath: 10 mM Co2+ + 50 mM DMAB.
The holes are typical for SAMs on Au(111). Tunneling parameters: I
) 0.8 nA, U ) 40 mV.
andsdepending on the samplesone to two (in some rare cases
three) Co monolayers height (see Figure 5). Usually the islands
were randomly distributed over the terraces and not attached to
step edges. In rare cases islands were mainly deposited at step
edges. It is very probable that such samples had a lower quality
of the SAM. The coverage of the surface by the metallic islands
varied between 2 and 20%, again depending on the sample. The
variation in island size and surface coverage is mainly attributed
to the well-known variability of the SAMs,39,52,53 the activation
time, and the contact to air. Measurements after a few hours
and even 24 h later never showed any further growth of the
islands. Even a strong increase of either Co2+ or DMAB
concentration in the plating bath did not influence the island
size.
We always found structures similar to that of the above-
mentioned Pd nodules. We are sure that these nodules are
covered with Co, because also here Pd2+ is easily reduced to
Pd0, albeit of unknown structure.
Upon increasing the tunneling current above 2 nA, we could
move the islands over the surface, and even downward over
steps. Sometimes they coalesced. Such a mobility has never
been found for Cu islands electrodeposited on alkanethiolate
SAMs, obviously since the latter penetrate into the SAM.39,54
This gives us further support that the islands are located on top
of the monolayer. However, the adhesion of the islands to the
surface was strong enough that rinsing with water after the
deposition did not remove the islands.
Influence of the Amine Surface Concentration in the
SAM. Further proof that the Pd2+ species actually bind to the
amine ligands was given in experiments where we changed the
amino group concentration in the SAM. For this the freshly
annealed Au(111) samples were transferred into different
mixtures of a 20 µM ethanolic solution of the aminothiol and
a 20 µM ethanolic solution of decanethiol. It is well-known55-57
that in this way the amino group concentration in the SAM can
be varied. We changed the ratio aminothiol/alkanethiol in the
solution between 2/1 and 1/100. Figure 6 shows a typical image
of samples with low or medium amine density. It is obvious
Figure 5. (top) STM image of Co islands on a Pd2+-activated AT-
SAM electrode. Bath: 10 mM Co2+ + 50 mM DMAB. The brightest
islands (e.g. in the right upper corner) contain two or three Co layers.
(bottom) Line scan through islands with one (Pd only) and two layers
(Pd and Co) measured on an absolute height image (not shown here).
Tunneling parameters: I ) 0.6 nA, U ) 40 mV.
temperature. These Co depositssa few micrometers thicks
could always be dissolved by either HNO3 or HCl (both of pH
) 1) within minutes.
Electroless Co Deposition on Nonactivated AT-SAM. To
test the quality of our AT-SAM, we performed in situ STM
measurements in the Co plating bath without having activated
the AT-SAM with Pd2+ solution. We never found any Co
deposits on these samples. This agrees well with Dressick’s
proposition.22 Furthermore, it proves that we image Co and
not contamination. It is important to note that for a Co
deposition an activation with the Pd2+ solution was absolutely
necessary.
Electroless Co Deposition on Pd2+-Activated AT-SAM.
After the activation with Pd2+ solution, the solution was
removed by pipet, samples were then carefully rinsed three times
with H2O, and were immediately covered with the Co plating
bath. A few minutes later the cell was mounted in the STM
and measurements were started. During the induction time of
10-30 min for our system the Faradaic tip current was too high
to perform in situ STM measurements.6,51 After this period the
islands always reached their final size of 1-10 nm diameter

Electroless Deposition of Metal Nanoislands
J. Phys. Chem. B, Vol. 102, No. 39, 1998 7587
cause increased island sizes or even overgrowth of the Pd
islands.
5. Discussion
We shall start the discussion with a look at the binding
2-
2-
mechanism of the PdCl4 ions to the amino groups. PdCl4
is the most convenient starting material for many complexes
with N ligands and therefore a large collection of chemical and
physical data exists.58,59 Complexes are of the types [Pdam4]2+
,
[Pdam2X2], and [PdamX2]2 with am ) NH3, amine, 1/2 diamine
and X ) Cl-, Br-, I-, SCN-, CN-. Adding the amine to a
solution of [PdX4]2- under either neutral or acid conditions
usually produces [Pdam2X2]; hence, we expect a similar product
when am ) aminothiolate. In contrast [Pdam4]2+ is more
difficult to prepare.58
The inset of Figure 1 shows the different possible configura-
tions of the N-(2-aminoethyl)amide end group of the AT-SAM.
Mainly three factors determine the binding process at our
interface: chelate effect and steric and electrostatic repulsion.
Concentrating on these effects, case b is not very likely because
Pd2+ would have to penetrate into the layer, which would be
followed by a strong increase of steric and electrostatic repulsion
between the neighboring Pd2+-aminothiolate entities. For this
reason it is more likely that complexes in Figure 1d will exist
at the interface. This is supported by the fact that a chelate
effect occurs and that steric and electrostatic repulsion are
minimized.
Figure 7. STM image of Co islands on a Pd2+-activated AT-SAM
electrode in a bath saturated with nitrogen. Bath: 10 mM Co2+ + 50
mM DMAB. The increased island size is obvious from comparison
with Figure 5 (no nitrogen), which shows the same scan width.
Tunneling parameters: I ) 0.8 nA, U ) 40 mV.
that a decreased amine concentration leads to a lower Pd island
density and as a consequence also to a lower Co island density.
This strongly supports the model of amine-bound Pd2+ species.
Further we noticed a trend toward smaller islands for decreased
amine concentrations (e.g. for aminothiol/alkanethiol ) 1/100
the islands were only between 1 and 4 nm in diameter). We
deduce from this that we can control the island density and size
over a wide range.
For a better understanding of the island formation, we want
to focus now on the binding energies between Pd and N before
and after the reduction of the Pd2+ species. First, we calculated
approximately the Pd2+-N binding energy from thermodynami-
cal data60 to be ∼1.5 eV (the Pd2+-Cl binding energy was
estimated to be ∼2.1 eV).
Influence of the Oxygen Concentration in the Plating
Bath. For a better understanding of the Co growth stop, we
performed experiments with plating baths containing either more
or less dissolved oxygen than a solution in contact with air.
Pd0-N complexes could not be found in the literature (stable
Pd0 complexes can be formed with, e.g., As-, Sb-, or P-
containing ligands58). The reason can be explained by the
ligand-field theory. The electron configuration of Pd0 is d10
and shows very unfavorable energetics compared to the Pd2+
configuration which is d8 (the stabilization of -24.56 Dq0 with
a ligand field stabilization energy Dq0 > 0.15 eV61 is larger
than a typical chemical bond energy). Therefore the Pd0-N
bond must be much weaker than the Pd2+-N bond, and thus
the Pd0 species can diffuse over the surface.
For this reason we bubbled clean nitrogen through plating
solutions for 15 min before use. After a sample was covered
by the plating solution, the cell was kept in a nitrogen box for
an hour before it was mounted in the STM. However, STM
imaging was carried out in the electrolyte in contact to air.
Although a dispersion of the results was noticed, there was a
clear trend toward islands with a bigger diameter, usually
between 3 and 15 nm. A typical image at these conditions is
shown in Figure 7. In one case we even found islands of 30-
60 nm diameter with 10-20 monolayer height (steps still could
be resolved). Generally areas were often completely covered
by three-dimensional Co deposits. The poor reproducibility is
linked to the not completely controllable contact to air during
the sample preparation (e.g. time for the transfer to the nitrogen
box).
Calculating, on the other hand, the Pd-Pd bond energy in
bulk metal, we found 3.9 eV for a macroscopic Pd solid.60
Calculations with an effective medium code62 estimated the bond
energy for a Pd atom in a Pd(111) island containing 16 atoms
to be ∼3 eV. In both cases (microscopic and macroscopic) the
Pd-Pd bond energy is twice as high as the Pd2+-N binding
energy. This favorable situation allows the formation of stable
Pd islands.
We propose here a model where the Pd2+ entities are
distributed homogeneously over the surface. They are then first
reduced by the reductant to Pd atoms. During the reduction,
the strength of the Pd-N bond is lowered and Pd atoms can
diffuse over the surface. Nucleation and growth of the Pd atoms
results in stable islands.
On the other hand, we repeated the same procedure with a
plating solution saturated with pure oxygen and kept under an
oxygen atmosphere. For these conditions we noticed that the
average size of the island is between 1 and 3 nm, which
corresponds well to the values estimated for the pure Pd islands.
We conclude that the growth strongly depends on the oxygen
concentration during the ELD. Hence, oxygen addition is an
additional control parameter. High oxygen concentrations can
prevent Co deposition, while decreased oxygen concentrations
Let us now concentrate on the discussion of the Co growth
stop. It is necessary to consider the oxidation of the reductant

7588 J. Phys. Chem. B, Vol. 102, No. 39, 1998
Kind et al.
(eq 1), cathodic reduction of the metal ion (eq 2), and oxygen
reduction (eq 3):
influence the ELD. It is known that the H+ reduction is faster
on Pd than on Co. That means that the Pd islands may be partly
covered by H (or even transformed to Pd hydride), which can
influence further metal deposition.
Concentrating on the described effects, we have to conclude
that it is very difficult to reach a final conclusion on the stop of
the growth. Nevertheless, we want to stress that the oxygen in
the deposition bath is the main parameter to control the island
size and thus likely to be responsible for the growth stop. We
are certain that in our case nonlinear oxygen diffusion controls
the deposition at the molecular level.
j(Red): Red T Ox + me-
j(Mn+): n+ + ne- T M
j(O2): O2 + 2H2O +4e- T 4OH-
(1)
(2)
(3)
M
The influence of H+ reduction will be discussed later. The
competition between reactions 1 and 3 determines whether the
metal will be deposited.63 For our system all partial reactions
can take place first at Pd and later at Co. Thus, the overall
reaction is highly complex.64,65 (In fact, the partial reactions
of ELD can occur at different places. One can envisage that
the electrons move from the adsorbed reductant through Pd and
Co islands to another site where Co2+ is reduced. In our case,
electrons can even move from one metal island to another metal
island via the Au substrate since electrons are rapidly transferred
5. Conclusions
Pd can be grown by electroless deposition (ELD) in nanom-
eter-sized islands on self-assembled aminothiolate layers (AT-
SAMs) on Au(111) electrodes. They can be investigated by in
situ STM. The most likely place of the Pd is on top of the
AT-SAM as proven by normal and grazing emission angle X-
ray photoelectron spectroscopy, by the fact that they can be
moved by slightly increasing the tunneling current, and finally
by the strong correlation with the amine surface concentration
in mixed monolayers. The Pd island density can be varied over
a large range by changing the amine concentration in a
aminothiolate/alkanethiolate SAM.
On these Pd islands nanometer-sized Co islands can be
deposited via ELD. The size of such islands can be controlled
by the oxygen concentration in the Co plating bath. Processes
causing the stop of the island growth were discussed.
Our concept to fabricate nanometer-sized islands of Co and
Pd should be applicable to many magnetic and nonmagnetic
metals. In combination with microcontact printing to pattern
the AT-SAM, it should be possible to fabricate diverse metal
cluster arrays.
through thiolate layers.39,66
)
As mentioned above, the islands always reached their final
size before the first in situ STM measurement was performed.
This means that the growth had stopped within a few minutes.
To our knowledge a growth extinction has been found before
by two other groups. Jacobs et al. found a lower limit to the
size of nuclei below which ELD does not occur or is stopped
after a short time.63 They explained the growth stop with an
enhanced nonlinear oxygen diffusion under the assumption that
the diffusionally limited rate of oxygen reduction j(O2) is larger
than that of the kinetically controlled oxidation of the reducing
agent j(Red). A further decrease in the electrode dimensions
down to some 10 nm is equivalent to an increase in the mass
transport of reactants relative to the rate of electron transfer.
Thus, reactions that are limited by diffusion at mesoscopic and
macroscopic electrodes can become limited by the rate of the
electron transfer as the electrode size is reduced,63 but j(O2) >
j(Red) will still be assured. For completeness it has to be
mentioned that j(O2) at electrodes below 10 nm is no longer
controlled by diffusion but by kinetics.67 This can reduce the
expected current (calculated for a purely diffusion controlled
mechanism) by a factor of 3-5. Nevertheless, this will not
change that j(O2) > j(Red).
Acknowledgment. We thank H. Vogel for providing the
aminothiol and H. Brune for his help with the effective medium
calculation. A.M.B. would like to express his gratitude to the
Alexander von Humboldt-Stiftung for a Feodor-Lynen-Stipen-
dium.
References and Notes
Van der Putten et al.21,68 showed thatsdepending on the exact
conditionssthe nonlinear diffusion can lead both to an enhanced
local growth or to no growth at all. The latter is possible if a
compound is added that can be reduced at more anodic potentials
than the metal ion reduction, e.g., dissolved oxygen in the plating
bath. Although Van der Putten’s model was only confirmed
for electrodes in the micrometer range, the similarity of the shape
of our islands with their pyramid-like structures is striking. We
always found very few islands with three monatomic high
terraces, some more with two, and a great majority had only
one terrace.
(1) Weber, C. J.; Pickering, H. W.; Weil, K. G. In preparation.
(2) Paunovic, M. Proc. Symp. Electroless Deposition Met. Alloys 1988,
88, 3.
(3) Shipley, C. R. Plat. Surf. Finish. 1984, 71, 92.
(4) Molenaar, A. Proc. Symp. Electroless Deposition Met. Alloys 1988,
88, 37.
(5) Okinaka, Y.; Osaka, T. Electroless Deposition Processes: Fun-
damentals and Applications; VCH: Weinheim, 1990.
(6) Weber, C. J.; Pickering, H. W.; Weil, K. G. J. Electrochem. Soc.
1997, 144, 2364.
(7) Osaka, T.; Takematsu, H. J. Electrochem. Soc. 1980, 127, 1021.
(8) Besenhard, J. O.; Krebber, U.; Hoerber, J. K.; Kanani, N.; Meyer,
H. J. Electrochem. Soc. 1989, 136, 3608.
(9) Mallory, G. O. Electroless Plating: Fundamentals and Applica-
tions; American Electroplaters and Surface Finishers Society: Orlando, FL,
1990.
(10) Sander, D.; Skomski, R.; Schmidthals, C.; Enders, A.; Kirschner,
J. Phys. ReV. Lett. 1996, 77, 2566.
(11) Ewing, A. G.; Dayton, M. A.; Wightman, R. M. Anal. Chem. 1981,
53, 1842.
(12) Morris, R. B.; Franta, D. J.; White, H. S. J. Phys. Chem. 1987, 91,
3559.
Moreover, we estimated j(O2) on Pd and Co.63,69 Both values
are of the same order of magnitude and do not explain the
growth stop of Co (a reasonable explanation would be j(O2)Co
. j(O2)Pd: After the Pd island is covered with Co, j(O2)Co would
consume all electrons delivered by the reductant and the growth
would cease).
Another problem is the hydrogen evolution during the plating.
Van der Meerakker stated that hydrogen evolves during almost
all electroless reactions and proposed a general mechanism based
on the formation of atomic hydrogen during the oxidation of
the reducing agents.65 During our experiments we observed
small hydrogen bubbles at the liquid/solid interface. This will
(13) Penner, R. M.; Heben, M. J.; Longin, T. L.; Lewis, N. S. Science
1990, 250, 1118.
(14) Ewing, A. G.; Wightman, R. M.; Dayton, M. A. Brain Res. 1982,
249, 361.
(15) Tanaka, K.; Kashiwagi, N. Bioelectrochem. Bioenerg. 1987, 17,
519.

Electroless Deposition of Metal Nanoislands
J. Phys. Chem. B, Vol. 102, No. 39, 1998 7589
(16) Tanaka, K.; Kashiwagi, N. Bioelectrochem. Bioenerg. 1989, 21,
95.
(17) Conti, F.; Neher, E. Nature 1980, 285, 140.
(18) Craston, D. H.; Lin, S. W.; Bard, A. J. J. Electrochem. Soc. 1988,
135, 785.
(19) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Funda-
mentals and Applications; John Wiley & Sons: New York, 1980.
(20) Fleischmann, M.; Lasserre, F.; Robinson, J.; Swan, D. J. Electroa-
nal. Chem. 1984, 177, 97.
(21) Van der Putten, A. M.; de Bakker, J. W. J. Electrochem. Soc. 1993,
140, 2229.
(22) Dressick, W. J.; Dulcey, C. S.; Georger, J. H.; Calvert, J. M. Chem.
Mater. 1993, 5, 148.
(23) Vargo, T. G.; Gardella, J. A.; Calvert, J. M.; Chen, M. S. Science
1993, 262, 1711.
(24) Potochnik, S. J.; Pehrsson, P. E.; Hsu, D. S.; Calvert, J. M. Langmuir
1995, 11, 1841.
(41) Wyatt, I. R. Chem. Weekbl. 1966, 62, 310.
(42) Kazakova, V. I.; Ptitsyn, B. V. Russ. J. Inorg. Chem. 1967, 12,
323.
(43) Nabivanets, B. I.; Kalabina, L. V. SoV. Prog. Chem. 1975, 41, 81.
(44) Nabivanets, B. I.; Kalabina, L. V. Russ. J. Inorg. Chem. 1970, 15,
818.
(45) Elding, L. I. Inorg. Chim. Acta 1972, 6, 683.
(46) Elding, L. I. Olssen, L. F. J. Phys. Chem. 1978, 82, 69.
(47) Bekker, P. Z.; Robb, W. J. Inorg. Nucl. Chem. 1975, 37, 829.
(48) Vargaftik, M. N.; Igoshin, V. A.; Syrkin, Y. K. IzV. Akad. Nauk
SSSR, Ser. Khim. 1972, 6, 1426.
(49) Rund, J. V. Inorg. Chem. 1970, 9, 1211.
(50) Maitlis, P. M. The Organic Chemistry of Palladium-I. Metal
Complexes; Academic Press: New York, 1971.
(51) Haruyama, S.; Ohno, I. Proc. Electrochem. Depos. Thin Films 1993,
93, 70.
(52) Sondag-Huethorst, J. A.; Fokkink, L. G. Langmuir 1995, 11, 4823.
(53) Camillone, N.; Leung, T. Y.; Scoles, G. Surf. Sci. 1997, 373, 333.
(54) Cavalleri, O. Ordering and electrodeposition at the alkanethiol/Au-
(111) interface. Ph.D. Thesis, EPF Lausanne, 1997.
(55) Bain, C. D.; Evall, J.; Whitesides, G. M. J. Am. Chem. Soc. 1989,
111, 7155.
(25) Kumar, A.; Whitesides, G. M. Appl. Phys. Lett. 1993, 63, 2002.
(26) Kumar, A.; Biebuyck, H. A.; Whitesides, G. M. Langmuir 1994,
10, 1498.
(27) Xia, Y.; Whitesides, G. M. Angew. Chem., Int. Ed. 1998, 37, 550.
(28) Delamarche, E.; Schmid, H.; Bietsch, A.; Larsen, N. B.; Rothuizen,
H.; Michel, B.; Biebuyck, H. J. Phys. Chem. B 1998, 102, 3324.
(29) Hidber, P. C.; Helbig, W.; Kim, E.; Whitesides, G. M. Langmuir
1996, 12, 1375.
(56) Delamarche, E.; Michel, B.; Biebuyck, H. A.; Gerber, C. AdV.
Mater. 1996, 8, 719.
(30) Brandow, S. L.; Dressick, W. J.; Marrian, C. R.; Chow, G. M.;
Calvert, J. M. J. Electrochem. Soc. 1995, 142, 2233.
(31) Marrian, C. R.; Perkins, F. K.; Brandow, S. L.; Koloski, T. S.;
Dobisz, E. A.; Calvert, J. M. Appl. Phys. Lett. 1994, 64, 390.
(32) Nuzzo, R. G.; Zegarski, B. R.; Dubois, H. J. Am. Chem. Soc. 1987,
109, 733.
(33) Bryant, M. A. J. Am. Chem. Soc. 1991, 113, 8284.
(34) Camillone, N.; Chidsey, C. E.; Liu, G.-Y.; Scoles, G. J. Chem.
Phys. 1993, 98, 4234.
(35) Calvert, J. M.; Calabrese, G. S.; Bohland, J. F.; Chen, M. S.;
Dressick, W. J.; Dulcey, C. S.; Georger, J. H.; Kosakowski, J.; Pavelcheck,
E. K.; Rhee, K. W.; Shirey, L. M. J. Vac. Sci. Technol. B 1994, 12, 3884.
(36) Calvert, J. M. J. Vac. Sci. Technol. B 1993, 11, 2155.
(37) Dressick, W. J.; Dulcey, C. S.; Georger, J. H.; Calabrese, G. S.;
Calvert, J. M. J. Electrochem. Soc. 1994, 141, 210.
(38) Van der Putten, A. M.; de Bakker, J. W.; Fokkink, L. G. J.
Electrochem. Soc. 1992, 12, 3475.
(57) Takami, T.; Delamarche, E.; Michel, B.; Gerber, C. Langmuir 1995,
11, 3876.
(58) Livingstone, S. E. ComprehensiVe Inorganic Chemistry; Bailor, K.,
Ed.; Pergamon Press: Oxford, 1973; Vol. 3.
(59) Baes, C. F.; Mesmer, R. E. The Hydrolysis of Cations; Wiley-
Interscience: New York, 1976.
(60) Mortimer, C. T. ReV. Inorg. Chem. 1984, 6, 233.
(61) Huheey, J. E. Inorganic Chemistry: Principles of Structure and
ReactiVity; Harper Row: New York, 1983.
(62) Stoltze, P. Condens. Mater. 1994, 6, 9495.
(63) Jacobs, J. W.; Rikken, J. M. J. Electrochem. Soc. 1988, 135, 2822.
(64) Wiese, H.; Weil, K. G. J. Electroanal. Chem. 1987, 228, 347.
(65) Van der Meerakker, J. E. J. Appl. Electrochem. 1981, 11, 395.
(66) Smalley, J. F.; Feldberg, S. W.; Chidsey, C. E.; Linford, M. R.;
Newton, M. D.; Liu, Y. P. J. Phys. Chem. 1995, 99, 13141.
(67) Seibold, J. D.; Scott, E. R.; White, H. S. J. Electroanal. Chem.
1989, 264, 281.
(68) Van der Putten, A. M.; de Bakker, J. W. J. Electrochem. Soc. 1993,
140, 2221.
(69) Meruva, R. K.; Meyerhoff, M. E. Anal. Chem. 1996, 68, 2022.
(39) Cavalleri, O.; Bittner, A. M.; Kind, H.; Kern, K.; Greber, T. Z.
Phys. Chem., in press.
(40) Gilbert, S. E.; Cavalleri, O.; Kern, K. J. Phys. Chem. 100, 12123
1996.