High Yield, Large Scale Synthesis of Thiolate-Protected Ag
7
Clusters
Zhikun Wu, Eric Lanni, Wenqian Chen, Mark E. Bier, Danith Ly, and Rongchao Jin*
Carnegie Mellon UniVersity, Department of Chemistry, 4400 Fifth AVenue, Pittsburgh, PennsylVania 15213
Received September 8, 2009; E-mail: rongchao@andrew.cmu.edu
Noble metal clusters have stimulated major research interest in
recent years. Scientific interest in these clusters is due to their unique
material properties that bridge the gap between those of small
molecules (e.g., organometallic compounds) and of nanocrystals
(typically >2 nm).
1-4
Such nanoclusters hold potential in a wide
range of applications including nanoelectronics, optics, sensing,
biomedicine, and catalysis.
4-8
For both fundamental studies and
practical applications of this new type of material, the first major
task is to develop synthetic methods (especially wet chemical
approaches) that permit the synthesis of robust, atomically mono-
disperse clusters with precise control over the number of metal
atoms in the cluster.
A well-established route to prepare solution phase metal clusters
is to use strong ligands (such as thiols) to protect clusters and effect
size control.
9-12
Among the noble metals, gold clusters are being
extensively studied due to their chemical stability and relative ease
of preparation (e.g., under ambient conditions). Some well-defined
monodisperse Au
n
nanoclusters have been reported, and their exact
formulas have been determined by mass spectrometry analysis.
9-12
Bimetal Au/Ag clusters, phosphine, or iron-carbonyl protected Au
n
clusters have also been reported.
13,14
Small Ag
n
clusters (n < 10
atoms) stabilized by DNA have received particular attention due
to their strong fluorescence.
15,16
For thiolate-capped Ag clusters,
no truly monodisperse Ag
n
(SR)
m
clusters have been reported.
Nevertheless, there have been several notable recent studies of Ag
n
thiolate clusters (n > 10 atoms) and of chiroptical (e.g., circular
dichroism) properties.
17-21
Very recently, Cathcart et al. reported
an interesting cyclic reduction in an oxidative condition (CTOC)
method to prepare silver nanoclusters; a well-defined optical
spectrum was obtained for the as-prepared silver cluster.
19
Bakr et
al. synthesized a silver thiolate species with eight absorption
bands.
20
In another study, a 7 kDa silver cluster species was
reported.
21
But the exact composition of these silver clusters was
not attained in MS analysis,
19,20
although the clusters were
estimated to be ∼25 silver atoms. Overall, the synthesis, isolation,
and precise composition determination still remain a big challenge
in Ag cluster research.
Herein we report a rationally designed wet chemical method for
synthesizing monodisperse Ag
7
clusters stabilized by four meso
2,3-dimercaptosuccinic acid (DMSA) ligands (donoted as
Ag
7
(DMSA)
4
). This is the first report that achieves a precise MS
determination of the composition of silver thiolate clusters. The
DMSA ligand was found to be important for the synthesis of
relatively stable silver clusters.
In a typical experiment (details in the Supporting Information),
silver salt (AgNO
3
, 34.0 mg) was dissolved in ethanol (Figure 1).
The solution was cooled to ∼0 °C in an ice bath; DMSA was then
added. After the formation of Ag
x
(DMSA)
y
intermediates, NaBH
4
(powders) was slowly added to the solution under vigorous stirring.
The reaction mixture slowly turned from yellowish green to deep
brown, indicating the reduction of Ag
x
(DMSA)
y
and formation of
Ag clusters (Figure 1).
The silver clusters have a low solubility in the chosen reaction
medium (ethanol); hence, they spontaneously precipitated out of
solution. After reaction for ∼12 h, the product suspension was
centrifuged briefly, and the resultant black precipitates were
collected, washed thoroughly with methanol, and then dissolved
in water. The aqueous solution of the as-prepared Ag clusters
showed a pronounced absorption peak at ∼500 nm, Figure 2
(dashed profile). To further improve the purity of the product, the
clusters were precipitated by addition of MeOH. Recrystallization
2-3 times leads to highly pure Ag
n
(DMSA)
m
clusters. This
purification process was evaluated by polyacrylamide gel electro-
phoresis (PAGE) analysis. The crude product (prior to recrystal-
lization) shows at least two diffuse and broad bands, indicating
the existence of impurities (Figure 2, inset a), while the recrystal-
lized clusters show a more well-defined band (Figure 2, inset b),
indicating a higher purity. The as-purified Ag clusters show a strong
absorption peak at ∼500 nm and weak peaks at ∼415 and ∼625
nm (Figure 2, red profile).
In our synthetic approach, DMSA was found to be an effective
ligand for stabilization of silver clusters. We found that monothiols
such as phenylethylthiol or dodecanethiol cannot sufficiently
stabilize Ag nanoclusters, albeit these ligands have been extensively
used in making gold nanoclusters.
9-12,22
As to the choice of
reaction medium, ethanol was found to be a good medium for the
Figure 1. Color change during the synthesis of silver clusters. (Left) ethanol
solution of AgNO
3
. (Middle) 4 h after addition of DMSA. (Right) 12 h
after addition of NaBH
4
.
Figure 2. UV-vis spectra of the Ag clusters before and after purification.
Inset: PAGE analysis of the (a) crude Ag clusters and (b) pure clusters
after recrystallization (photograph b shows increasing loading amount of
Ag clusters from left to right). Note that the blue band is from the indicator
dye in the loading buffer. The arrow shows the migration direction.
Published on Web 11/03/2009
10.1021/ja907627f CCC: $40.75  2009 American Chemical Society16672 9 J. AM. CHEM. SOC. 2009, 131, 16672–16674

production of uniform Ag clusters. In our system, ethanol plays
two important roles: (1) it acts as a size-selecting solvent by
terminating or retarding the growth of clusters into larger ones since
larger clusters would be less soluble in ethanol and precipitate out
of solution; this provides an effective way of controlling cluster
size. (2) Ethanol also acts as a reactant, and its slow reaction with
NaBH
4
(in the absence of water) is helpful in slowing the
Ag
x
(DMSA)
y
reduction rate and, hence, producing small and
uniform Ag clusters; otherwise, if water is present, a large amount
of Ag(0) species would be instantly produced from a rapid,
uncontrolled reduction of Ag
x
(DMSA)
y
, and the growth of Ag nuclei
into clusters would be too fast and one would lose control of the
cluster growth kinetics. Using the above strategy, we obtained
highly monodisperse Ag nanoclusters (yield ∼20%); further
optimization of the reaction conditions may improve the yield. The
reaction can be readily scaled up; we have tested a 6X scale-up
and obtained similar yields (see Supporting Information).
To rule out the possibility that the absorption peak might arise
from large Ag nanospheres (which would be >80 nm diameter for
a surface plasmon band at ∼500 nm) or anisotropic silver particles
such as nanoprisms (which would be at least a 30 nm edge length),
23
we performed a high-speed centrifugation test: Ag clusters were
first dissolved in water (10 mg/mL), and the solution was then
centrifuged at 14 000 rpm for 30 min, but no precipitates were
observed, indicating that the prepared Ag species is small clusters
rather than large nanocrystals. The high electrophoretic mobility
of the Ag clusters (Figure 2 inset) also implies small clusters.
Moreover, TEM also confirms that the Ag particles are indeed
subnanometer clusters, and no large nanocrystals were found (see
Figure S1). Note that a diluted solution was used to prepare the
TEM specimen to avoid cluster aggregation in TEM imaging (we
observed that densely distributed Ag clusters quickly agglomerate
upon electron beam irradiation under TEM). The Ag clusters are
barely observable due to their extremely small size (subnanometer).
Taken together, the results from high-speed centrifugation tests,
electrophoresis, and TEM confirm that the Ag species synthesized
in this work is composed of small clusters, and the observed optical
absorption indeed originates from the clusters rather than from larger
nanocrystals.
To determine the exact size of clusters (i.e., the number of Ag
atoms and ligands in the cluster) is a challenging task, particularly
for silver (as opposed to gold), because the isotopic distributions
of silver clusters are complicated significantly by the two abundant
naturally occurring silver isotopes,
107
Ag and
109
Ag, in addition to
carbon and sulfur isotopes. In this work we have succeeded in ESI-
MS analysis of intact silver clusters; this is indeed the first reported
ESI-MS determination of silver thiolate clusters to our knowledge.
In the ESI-MS spectrum (acquired in the negative ion mode, Figure
3A), the base peak was found at m/z 1520.40 (labeled 3, see Figure
3A inset) as a member of one isotope cluster amidst a series of
similar, less-intense clusters spaced regularly to either side (labeled
1-6). The unity spacing of the isotopes (Figure 3B) implies that
the ionized clusters bear a -1 charge (z); the m/z values of the
peaks therefore represent true molecule ion mass, which for the
base peak of 1520.40 matches very well with the theoretical exact
mass of [Ag
7
L
4
- 2H + 2Na]
-
(theoretical molecular weight:
1520.12, deviation 0.28), where L ) S
2
C
4
H
4
O
4
, FW: 179.95). This
assignment is also supported by the excellent match of the simulated
and experimental isotopic distributions (Figure 3B and C). The other
ions surrounding peak 3 (Figure 3A inset) correspond to the
following ions: Ag
7
L
4
-
(1476.5, labeled 1), [Ag
7
L
4
- H + Na]
-
(1498.4, labeled 2), [Ag
7
L
4
- 3H + 3Na]
-
(1542.4, labeled 4),
[Ag
7
L
4
- 4H + 4Na]
-
(1564.4, labeled 5), [Ag
7
L
4
- 5H + 5Na]
-
(1586.4, labeled 6). The assignments are supported by their isotopic
distribution patterns (see Figure S2) as well as the 22 Da spacing
(22 ) m
Na
- m
H
).
On the basis of the assigned peaks as well as the negative ion
nature, we conclude that the native cluster is composed of 7 silver
atoms stabilized by 4 DMSA ligands and the Ag
7
core carries a
-1 charge. The emergence of a series of H subtraction/Na addition
(-H + Na) peaks does not affect the cluster’s original charge.
We have also performed matrix-assisted laser desorption ioniza-
tion (MALDI) MS analysis (matrix: sinapic acid). Compared to
ESI-MS, MALDI forms and detects ions over a much higher m/z
range yet fragmentation is often observed. For the Ag
7
(DMSA)
4
clusters, a set of peaks centered at ∼1500 m/z (singly charged,
assigned to Ag
7
L
4
- H + Na) with a spacing of 22 (i.e., m
Na
-
m
H
) were found in MALDI-MS analysis (Figure S3), which is
consistent with the ESI results. The other sets of peaks at lower
m/z values are fragments resulted from the Ag
7
L
4
ions since they
were not found in ESI analysis. No larger Ag
n
clusters were found
in the high mass range (up to m/z 20 000), confirming the
monodispersity of the prepared Ag
7
(DMSA)
4
clusters.
The MS/MS experiments also confirmed that the cluster is
composed of 7 silver atoms. In MS/MS analysis, the dianion
[Ag
7
L
4
- 3H + 2Na]
2-
(m/z 759.57) was chosen as the parent
ion. The ion was subject to further collision at different energies,
and the fragmentation pattern was analyzed. When the collision
voltage was set at 10 V, the parent ion loses a neutral fragment
(SC
4
O
4
H
4
, 73.99 × 2 ) 147.98 Da, SC
4
O
4
H
3
observed at m/z )
146.99) of one of the four ligands but a sulfur atom of the ligand
is still retained on the cluster due to Ag-S bonding, forming
Ag
7
S(S
2
C
4
O
4
H
4
)(S
2
C
4
O
4
H
3
Na)(S
2
C
4
O
4
H
2
Na) (m/z 685.57, z )
-2; note that the Ag
7
core bears a -1 charge and the other
negative charge arises from one ligand bearing COO
-
), Figure
4A. At 25-40 V, the rest of the ligands break up but one S
atom from each ligand is retained on the cluster, forming
fragments (Ag
7
S
4
-
), Figure 4B. Apparently the negative charge
on the Ag
7
core is retained, consistent with the ESI assignment
of Ag
7
L
4
-
. With increasing voltage, smaller fragments were
sequentially observed, including Ag
6
S
4
-
and Ag
5
S
4
-
(Figure
4C-D).
Figure 3. (A) ESI spectra of silver clusters (negative ion mode, inset shows
the zoomed-in spectrum). (B) and (C) show the experimental and simulated
isotopic pattern of Ag
7
L
4
- 2H + 2Na, respectively.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 46, 2009 16673
COMMUNICATIONS