Cluster coagulation and growth limited by surface interactions with polymers

Article abstract of DOI:10.1021/jp012784o

The physical and chemical properties of metal nanoparticles differ significantly from those of free metal atoms as well as from the properties of bulk metals, and therefore they may be viewed as a transition regime between the two physical states. Within this nanosize regime, there is a wide fluctuation of properties, particularly chemical reactivity, as a function of the size, geometry, and electronic state of the metal nanoparticles. In recent years, great advancements have been made in the attempts to control and manipulate the growth of metal particles to prespecified dimensions. One of the main synthetic methods utilized in this endeavor, is the capping of the growing clusters with a variety of molecules, e.g., polymers. In this paper we attempt to model such a process and show the relationship between the concentration of the polymer present in the system and the final metal particle size obtained. The theoretical behavior, which we obtained, is compared with experimental results for the cobalt-polystyrene system.

Full text of DOI:10.1021/jp012784o

146
J. Phys. Chem. B 2002, 106, 146-151
Cluster Coagulation and Growth Limited by Surface Interactions with Polymers
†
,‡
Horacio G. Rotstein, and Rina Tannenbaum*
Department of Chemistry and Volen Center for Complex Systems, Brandeis UniVersity,
MS 015, Waltham, Massachusetts 02454-9110, and School of Materials Science and Engineering,
Georgia Institute of Technology, Atlanta, Georgia 30332-0245
ReceiVed: July 18, 2001
The physical and chemical properties of metal nanoparticles differ significantly from those of free metal
atoms as well as from the properties of bulk metals, and therefore they may be viewed as a transition regime
between the two physical states. Within this nanosize regime, there is a wide fluctuation of properties,
particularly chemical reactivity, as a function of the size, geometry, and electronic state of the metal
nanoparticles. In recent years, great advancements have been made in the attempts to control and manipulate
the growth of metal particles to prespecified dimensions. One of the main synthetic methods utilized in this
endeavor, is the capping of the growing clusters with a variety of molecules, e.g., polymers. In this paper we
attempt to model such a process and show the relationship between the concentration of the polymer present
in the system and the final metal particle size obtained. The theoretical behavior, which we obtained, is
compared with experimental results for the cobalt-polystyrene system.
1
. Introduction
tion of metal carbonyl complexes in solutions containing either
pure solvent or, in addition, polymers, which will result in a
The physical and chemical properties of metal nanoclusters
more viscous reaction environment. The main factors governing
the nucleation and growth of the metal particles during the
decomposition process are the metal cluster size, i.e., the number
of reactive sites available for chemical reactions, and the
mobility of the reactive metal clusters and their ability to diffuse
through the solution and collide with each other. Clearly, as
particle size grows, the mobility of the particles decreases and
a preferred average pseudo-equilibrium size distribution is
reached. If the medium in which the thermal decompositions
take place is more viscous than a pure solvent, then it is expected
that the upper bound on particle mobility will be reached faster,
and hence the final average particle size will be smaller.
The particular system studied in this work consists of cobalt
carbonyl complexes thermally decomposed in the presence of
polystyrene (PS). It is important to note that the solvent chosen
for the decomposition reactions was toluene, which is a good
solvent for polystyrene, and therefore it promoted the solvation
of the polymer over a large composition range. Based on this
experimental system, we have developed a mathematical model
that aims at elucidating the effect of the initial concentration of
the polymer present in the reaction solution on the final cobalt
particle size obtained via the thermal decomposition mechanism,
and correlates the model to our experimental results.
differ significantly from those of individual metal atoms, as well
as from the properties of bulk metal,1-5 and hence the size
regime of nanoclusters may be viewed as a transition regime
between the two physical states. Within the cluster regime, there
is a wide fluctuation of properties, particularly chemical reac-
tivity, depending on the size, geometry, electronic state, and
packing of the cluster.6
-13
Uniformity of size and spatial
distribution of the metal nanoclusters is essential in the study
of their properties and can be achieved primarily by conducting
their synthesis in the presence of stabilizers. These materials,
such as surfactants or polymers, adsorb onto the surfaces of
growing clusters and create a “shielding effect”, a chemical bar-
rier that prevents the effects of van der Waals interactions be-
tween particles, thus inhibiting the particle aggregation pro-
cess.14-23
Polymers are frequently used as stabilizers for metal
clusters because they are transparent, permeable, and noncon-
ductive, and as such, do not interfere with and/or mask the
potential optical, electrical, and catalytic properties of these
clusters.
Synthetic routes for the formation of nanoclusters include
ultrahigh vacuum techniques (UHV) and “wet” chemical
synthesis. The UHV processes, e.g., chemical vapor deposition,
have traditionally dominated in the preparation of nanoclusters
due to their inherent controllability, but the they lack the
flexibility necessary to manipulate a variety of properties within
the same system. The chemical synthetic methods, e.g., reduction
of metal halides and decomposition of organometallic com-
plexes, are becoming more popular in recent years due to the
ability to introduce stabilizing agents and therefore afford
limitation of aggregate size.
2
. Experimental Procedure
.1. Preparation of Polystyrene Solutions. Polystyrene
2
(
Aldrich Chemicals, Mh w ) 120 000, powder form; 3.24 g) was
dissolved in 250 mL of toluene (Fluka Chemical Corp., bp 110
°
%
-4
C, d ) 0.867) to produce a 1.08 × 10 M solution (or 1.5 wt
). Other polystyrene solutions had initial concentrations
The motivating physical system for this work is the chemical
synthesis of nanoscale metal clusters via the thermal decomposi-
-5
-3
ranging from 3.60 × 10 to 8.16 × 10 M (corresponding to
0
.5 wt % to 85 wt %, respectively). The toluene used in these
solutions was dried by molecular sieve pellets (Matheson,
Coleman, and Bell) and redistilled under an N2 stream. The
solutions were mixed in a 500 mL round-bottom flask for 24 h
*
Corresponding author, e-mail: rina.tannenbaum@mse.gatech.edu.
Brandeis University.
Georgia Institute of Technology.
†
‡
1
0.1021/jp012784o CCC: $22.00 © 2002 American Chemical Society
Published on Web 12/11/2001

Cluster Coagulation and Growth Limitations
J. Phys. Chem. B, Vol. 106, No. 1, 2002 147
3. Results and Discussion
at room temperature. The more concentrated solutions (poly-
styrene concentrations above 40 wt %) were refluxed under N2
at 90-100 °C to ensure homogeneity. All solutions were either
used immediately or stored under a nitrogen atmosphere.
The general chemical synthesis of metal clusters has been
designed to accommodate the manipulation of several process
variables and is best described in three stages: (a) The
preparation of homogeneous solutions of organometallic com-
plexes in a carefully selected solvent, i.e., it readily dissolves
the organometallic complex and is also a good solvent for the
polymer; (b) The mixing of these solutions with a polymer
solution resulting in a viscous medium which contains the
homogeneously dispersed organometallic complexes; (c) The
thermal decomposition of the organometallic complexes to form
uniform dispersions of small particle size. The decomposition
of metal carbonyls in various media has been reported in the
literature for the preparation of fine metallic particles. The
thermolysis of transition metal carbonyls in solution under an
inert atmosphere is a well-known technique for the preparation
2
.2. Decomposition of Cobalt Carbonyl Solutions. The
solution decompositions of either Co2(CO)8 or Co4(CO)12 in
an inert atmosphere were performed in a closed system
according to the classical stability diagram of cobalt carbonyls
as a function of temperature and p(CO).24 Freshly prepared Co4-
(
CO)12 crystals (1.287 g), synthesized by the method of Natta
25
et al., were dissolved in 50 mL of toluene (Fluka Chemical
-
2
Corp., d ) 0.867), corresponding to a 4.51 × 10 M solution.
The solution was placed in a 100 mL three-neck round-bottomed
flask flushed with carbon monoxide. The middle neck was
equipped with a reflux condenser, a side neck with a rubber
stopper, and the other neck with a thermometer. The rubber
outlet was also used for sampling. The reflux condenser was
equipped with a gas inlet-outlet glass-fitting device. The outlet
part was connected to a tube whose other side was inserted
securely into the vent. These special precautions had to be taken
in order to flush away carbon monoxide gas formed during the
decomposition reactions. Then the flask was placed in a
controlled heated oil bath, and the decomposition reaction was
carried out for approximately 48 h at 90 °C under a continuous
N2 stream. Stirring was essential throughout the reaction time.
During the reaction period, samples were removed with a
syringe at various time intervals, and their infrared spectra were
recorded in NaCl liquid cells with toluene as a reference. The
process was deemed complete (and pseudo-equilibrium reached)
when the carbonyl absorption bands completely disappeared.
The decomposition reactions performed in solutions containing
polystyrene were identical to the procedure described so far,
1
4-18
of pure metal powders.
Typically, the clusters formed
following such thermal decompositions are 50-250 Å in
diameter, and are classified as the mid-size nano-regime (0-
1
9-21
5
00 Å range).
Some of the characteristic properties and
advantages of these clusters prepared by the above method are
as follows: (a) The growth of the particles, and hence their
final size, is determined mainly by diffusion distances and the
driving forces between the growing clusters. (b) The desired
oxidation state and surface cleanliness of the metal clusters are
achieved by the strict control of the chemical and physical
environments during synthesis, primarily through the variation
of the decomposition atmosphere. (c) The particles may be
22,23
polycrystalline
and hence contain grain boundaries within
the cluster. As a consequence, a large number of the atoms reside
at the interface between these grain boundaries. (d) Another
large fraction of the atoms in the cluster is on, or close to, the
surface, and hence these atoms do not order themselves
identically to atoms within the “bulk” material. (e) There are
various possibilities for the subsequent reactions of these clusters
with a wide range of particles or molecules, and moreover they
are small enough to allow for a homogeneous dispersion within
a polymer matrix.
but instead of using pure toluene as the solvent, various
polystyrene solutions were used,2
6,27
depending on the desired
polymer concentration. For polystyrene solutions of 40 wt %
and above, some heating with vigorous stirring was necessary
to ensure homogeneity of the solutions.
Since highly reactive intermediate species are formed during
2
.3. Particle Size Measurements. Particle size and size
distribution were measured by transmission electron microscopy
TEM) using the JEOL JEM-1210 TEM instrument. Electron
28,29
the decomposition reaction of metal carbonyls,
there are two
major pathways for reaction available for these species. (a) They
can aggregate or fragment to form larger or smaller clusters,
according to the following general mechanism, respectively:
(
diffraction patterns from a TEM field emission gun were used
to study the composition of the metallic particles. Samples were
prepared by dilution of the original solutions (after the decom-
position process had taken place) to obtain a 0.1% solution of
polystyrene, which was applied dropwise to carbon-coated (100
Å thick) TEM copper grids.
(0)
(0)
(0)
j+k
Coj + Cok f Co
(0)
(0)
(0)
Coj f Coh + Coi
(1)
Infrared spectra were recorded on an IBM-IR-44 Fourier
transform spectrometer using a demountable liquid cell pur-
chased from Crystal Laboratories and equipped with a 0.5 mm
spacer and new NaCl windows. The cell chamber was purged
with dry nitrogen for at least 0.5 h before interferrograms were
collected. The resolution was 0.5 cm , and 3000 scans were
taken for both the sample and the reference solutions. The
empty, clean, and dry liquid cell was used as the background.
Upon completion of each interferrogram, the sample in the cell
was removed by vacuum suction and the cell thoroughly flushed
with toluene. When samples contained viscous solutions (over
where Coj is a randomly chosen cobalt cluster with j units (or
fragments, defined as Co1), j, k ) 1, 2,...., where j + k is
conserved, and h + i ) j. (b) They can interact with the polymer,
according to the following mechanism:
-
1
(0)
Coj + PS f Co - PS
(2)
where j ) 1, 2,...., resulting in metal attachment processes which
may, in some cases, lead to polymer degradation and/or cross-
linking. Such cases, which are a result of the presence of reactive
functional groups on the polymer chains that cause extensive
metal-polymer interactions, are not considered in this paper.
Polystyrene exhibits only a limited degree of interaction with
the cobalt cluster fragments, via the weak coordination between
the π-orbitals of the arene group on the polystyrene and the
d-orbitals of the cobalt clusters. We are currently studying
40 wt % polystyrene), the liquid cell was disassembled, the NaCl
crystals and spacer were washed in toluene, and then the cell
was reassembled using the same spacer. Spectral subtraction
was performed using a subtraction factor of 1.00 in order to
avoid possible spectral distortions.

148 J. Phys. Chem. B, Vol. 106, No. 1, 2002
Rotstein et al.
systems in which polymer degradation and/or cross-linking are
extensive, and those results will be published soon. All the
reactions described here are considered at 90 °C and in the
presence of N2.
The overall chemistry of the system will therefore be a
combination of both pathways described above and will be
determined by the type of the reactive groups of the polymer
in solution and the nature of the metallic species formed. The
overall chemical reaction considered in this work is given by
The process can therefore be modeled by the following system
of nonlinear differential equations:
3
2,33
j-1
∞
1
2
c˘ j )
[Rj-k,k cj-k ck - Q
j-k,k
ck] -
[Rj,k c c - Q cj+k] - δ1,j c1 p
j,k
j
k
∑
∑
k)1
k)1
{
p˘ ) - c p
1
(4)
∞
j)1
for j ) 1, 2, ..., together with initial conditions satisfying ∑
j cj(0) ) 1 and p(0) ) p0. In eq 4, p0 is the initial dimensional
concentration of reactive sites divided by C, Rjk is the
dimensional coefficient of coagulation divided by λ, and Qjk is
the dimensional coefficient of fragmentation divided by λC. All
are nonnegative and dimensionless. Since, as we pointed out
before, a pseudo-equilibrium state is reached as a consequence
of the fragmentation process and the decrease in the mobility
of large clusters, we assume that the concentration of clusters
whose size is larger than some finite N is negligible. This is
also reflected in our simulation, and therefore, the ∞ bound on
the summation in eq 4 can be replaced by N - j. Note that in
that case, the conservation of mass is assured. When the ∞ bound
is left in the summation, conservation of mass may be lost for
t g tc for some nonnegative tc (tc ) gelation time). In that
case, the model will be valid up until t ) tc.
As we pointed out before, and as it is reflected in the model,
the process of metal coagulation and cluster formation (inde-
pendent of the existence of polymer interactions) is a complex
series of reactions, which can be grouped into two categories:
(
m
0)
n[Co (CO) ] + PS f Co + Co - PS*
(3)
x
y
where the asterisk (*) stands for a metal-polymer complex
compound, x ) 1, 2, 4, or 6 and y ) 4, 8, 12, or 16, respectively,
y/x e 4 (according to the coordination chemistry of metal
30
carbonyl complexes ), and m ) j + k in eq 1. In the main
reaction, the clusters are formed from smaller reactions. The
Co-PS* complexes formed during this process are the products
of the secondary reaction described in eq 2, which competes
with the main coagulation-fragmentation process described in
eq 1. The extent of the secondary reaction depends on the
reactivity of the cobalt precursor and on the interaction
coefficient between the reactive cobalt species and the poly-
styrene molecules. Therefore, the rate of the main coagulation
reaction will depend not only on the concentration of the cobalt
precursor and the rate coefficients of coagulation and fragmen-
tation but also on the concentration of the available reactive
sites on the polystyrene in the system that are capable of
interacting with the cobalt precursor.
In this manuscript we present a mathematical model to
qualitatively understand the dynamics of the complex chemical
system under consideration. The model is as simple as possible,
but reflects the main chemical features of the system. To our
knowledge, no attempt to model chemical systems of this kind
has been reported to date.
To translate the physical system considered here into math-
ematical terms, we first scale the time variable by multiplying
it by the coefficient of reaction between cobalt fragments and
polystyrene, λ, a positive constant, and the initial cluster mass
C, also a positive constant. λ in this case is analogous to the
Flory-Huggins interaction parameter,31 and it represents the
strength of the interaction between the metal clusters and the
functional groups of the polymer. Then we define cj as the
concentration of Co clusters of size j divided by C, and similarly
p as the concentration of the reactive sites on the polystyrene,
whose initial value is assumed to be a linear function of the
polymer concentration, divided by C. For the model developed
in this paper, we assumed that the only significant irreversible
interaction between cobalt fragments and polystyrene will occur
for c1 with a coefficient of reaction λ, while for other clusters,
j g 2, the interactions with the polystyrene will be reversible
and weaker.30 Therefore, each c1 particle, which is involved in
this interaction, will be excluded from further participation in
the main coagulation reaction. Moreover, the chemical bonding
between the c1 particles and the polystyrene reduces the number
of reactive sites on the polymer and hence, limits the extent
and rate of the cobalt-styrene interaction. It is important to stress
that the overall chemistry of the system and the final particle
size distribution of the cobalt clusters are dependent on the
relative importance and contribution of the two competing
reactions. These considerations, as we will see later, will be
reflected in the results obtained from our model, which will
illustrate the main features of the dynamics of the system, while
disregarding a more exact and detailed description.
3
2
(a) The growth of clusters either by the formation of atom-
atom bonds between two metal fragments (metal fragment )
c1) and between a small cluster and a metal fragment, or by the
development of surface-surface interactions between two large
clusters; and (b) The decomposition of large metal clusters into
smaller fragments. To properly model these processes, it is
important to construct the appropriate reaction rate expressions
both for the cluster formation step and the cluster dissociation
step (with or without polymer interaction).
The simplest coagulation expressions found in the literature
R
R
that are pertinent for our model are Rjk ) j k (predominantly
34-40
R
R
chemical bond),
interactions),
and Rjk ) j + k (predominantly surface
where in both cases 0 e R e 1. In this model,
34,35
we recognize the fact that there is a decline in particle mobility
through the medium as a function of increased particle size.
Moreover, the presence of the polystyrene molecules in the
reaction solution contributes to the formation of a viscous
41,42
medium,
which will impose an upper bound on the ability
of larger clusters to interact via surface-surface interactions,
and hence these types of interactions will have a negligible
contribution to the overall coagulation process. These assump-
tions are consistent with the fact that in large clusters the fraction
of atoms on the surface is small compared to the total number
of atoms in the cluster, and therefore only this small fraction is
essentially coordinatively unsaturated. In effect, we can view
large clusters as being thermodynamically stable in real time.
This is not the case for small clusters whose surface atoms
represent a considerable fraction of the total metal mass. In this
case, the model has to promote the growth of particles formed
from smaller fragments, because this growth will minimize their
free energy by reducing the number of coordinatively unsatur-
4
3,44
ated surface atoms.
Therefore, the expression that we propose for the coefficients
of coagulation of the cobalt fragments in our model (with or

Cluster Coagulation and Growth Limitations
without polymer interactions) is
J. Phys. Chem. B, Vol. 106, No. 1, 2002 149
σµ
j + k
R
R
j + k 1
R )
j k +
(5)
j,k
â
â
σ
j k
for j,k ) 1, 2, ..., where 0 e R, â e 1, 0 < µ e 1 is a weighting
parameter that denotes the fraction of cobalt clusters that are
permitted to enter into the coagulation reaction (µ ) 1 if no
polymer is present), and σ > 0 denotes the critical size of the
cobalt clusters beyond which the coagulation rates will drop as
a function of size increase. The first part of this equation
describes the mechanism of cluster formation due a combination
of chemical bonding (for small clusters) and surface interactions
(for larger clusters), without specifying the critical crossover
cluster size. The second part of this equation incorporates
mobility and diffusion considerations into the rate expression,
by reducing the contribution of the larger clusters, due to their
lower mobility and slower diffusion through the reaction
medium. The dissociation of cobalt clusters (i.e., fragmentation
of cobalt cluster flocculates) into smaller particles has a direct
impact on the final equilibrium cluster-size distribution of the
system. We assume that large clusters dissociate preferentially
into smaller clusters of similar size rather than into clusters with
a large variation in size. Therefore, the expression that we
propose to describe the dissociation of large cobalt clusters into
smaller clusters is
Figure 1. The theoretical dependence of cobalt cluster size on the
initial concentration of reactive sites of polystyrene in the reaction
solution, p . Note that the number of reactive sites is directly
0
proportional to the polymer concentration for a particular polymer
molecular weight. The cluster size is expressed as the average number
of precursor fragments, kavg. Each data point in the plot is the result of
a full simulation cycle.
γ
γ
j k
γ
are allowed to react with the polymer, i.e., will be a factor in
the determination of µ. Due to the complexity of this system of
equations (eq 4), and the lack of extensive experimental data,
we have opted to consider only the temperature of the
experimental reaction.
Q ) η
(6)
j,k
γ
j + k
for j,k ) 1, 2, ..., where 0 e γ e 1 and 0 < η e 1 is a weighting
parameter that denotes the fraction of cobalt clusters that
undergo dissociation.
To qualitatively test our model, eq 4 has been solved
numerically using a Runge-Kutta method of order four with
.01 as time-step and for N ) 400, R ) 0.1, â ) 0.5, γ ) 1,
η ) 0.0001, σ ) 200, and µ ) 0.1. These values of the
parameters have been chosen such that the solution reflects a
pseudo-equilibrium situation in which there is a preferred cluster
average size or, in some cases, preferred average sizes. In Figure
An interesting observation that arises from our simulation
results is the fact that the fragmentation of large cobalt floc-
culates has negligible contribution on the average particle size
obtained for the system. Hence, a similar inverse dependence
of average cluster size on initial polymer concentration is
obtained also when eq 4 does not contain the fragmentation
term (and hence it is assumed that Qjk ) 0). However, the
mobility term included in the expression for Rjk (eq 5) is crucial
for the achievement of a preferred cluster size. In the absence
of this term, the system proceeds toward gelation, independent
of the presence of the polymer in the reaction medium.
Experimental results support the qualitative relationship
between initial concentration of polymer in the reaction solution
0
1
we can see a graph of these preferred average sizes as a
function of p0. The value of η ) 0.0001 in eq 6 shows that for
a system with N ) 400, the fraction of clusters that undergo
fragmentation is low, and so the system is governed primarily
by the coagulation processes. Also, µ ) 0.1 in eq 5 indicates
that only a fraction of the c1 clusters is allowed to enter the
coagulation reaction and the remainder will react with the
polymer. However, the second term in the equation, which
represents the contribution of the larger clusters whose mobility
has been diminished, is not affected by the interaction with the
polymer (since according to this model, only c1 clusters have
an irreversible bonding with the polymer), and hence, equal
weight has been given to both coagulation mechanisms. This
is further supported by the fact that the crossover cluster size is
a cluster containing half the number of c1 fragments in the
system. Other values for these parameters may be chosen for a
different number of initial precursor fragments (N) and a
different operating temperature. The first condition will dictate
the optimal value for the crossover cluster size, the fraction of
clusters undergoing fragmentation and the relative weight of
the two coagulation mechanisms. The second condition, tem-
perature, will directly influence the mobility and diffusional rates
of the larger clusters and therefore will also dictate the relative
weight of the two coagulation mechanisms. Moreover, the
temperature will also influence the fraction of c1 clusters that
4
5
and the final particle average size. Cobalt carbonyl decom-
position reactions were performed at 90 °C in toluene solutions
2
6,45
with different polystyrene concentrations.
Figure 2 shows
the average particle size and particle size distribution obtained
for several initial concentrations of polystyrene in solution. As
the concentration of the polymer increases, a shift to lower
average particle sizes and a compression of size distributions
are observed. Two main effects contribute to the inverse
dependence of particle size on polymer concentration. (a) As
the polymer concentration increases, the solution becomes more
viscous and hence the diffusion of metallic fragments through
the reaction medium to form aggregates decreases. Moreover,
as fragments aggregate, their mobility decreases, and hence the
impact of an increased solution viscosity is considerable. (b)
As the polymer concentration increases, there are more polymer
molecules capable of adsorbing onto the surface of the forming
clusters (the total number of polymer reactive sites increases),
2
6,46
thus creating a more compact “cap”.
As a result, the
hydrophobic interactions between the polymer molecules in-
crease and the net effect of van der Waals attraction forces

150 J. Phys. Chem. B, Vol. 106, No. 1, 2002
Rotstein et al.
Figure 2. Cobalt particle size distribution as a result of thermal
decompositions of cobalt carbonyl complexes in solutions containing
Figure 4. The TEM micrograph of zero-valent cobalt particles obtained
by the thermal decomposition of cobalt carbonyl complexes at 90 °C
under an inert atmosphere in a polystyrene matrix (Davg ) 22.9 Å).
polystyrene with different initial polymer concentrations, p
0
. The cluster
size is expressed as the average number of precursor fragments, kavg
.
Figure 5. The experimental relationship between average cobalt cluster
size and the initial concentration of polystyrene in the reaction solution.
The 85 wt % PS concentration is considered a polystyrene matrix, and
is designated as “solid state”.
Figure 3. The TEM micrograph of zero-valent cobalt particles obtained
by the thermal decomposition of cobalt carbonyl complexes at 90 °C
under an inert atmosphere in a toluene solution (Davg ) 210.1 Å). The
dark regions represent the cobalt clusters.
between the metal particles decreased. A transmission electron
microscopy (TEM) micrograph picture of cobalt particles
obtained from a decomposition reaction in the absence of PS
4. Summary
The work presented in this paper is the first attempt to
combine theoretical studies with experimental data in conjunc-
tion with the nucleation and growth mechanism of metal
nanoparticles in the presence of a “reactive wall”. Since the
reactive adsorption of organic molecules, amphiphiles, and in
particular polymers, onto metal clusters, is the method of choice
for the in situ control of particle size, it is imperative to
understand the fundamental mechanisms of the competing
processes that occur in such systems. Unfortunately, not many
experimental data are available in the recent literature regarding
the kinetics, mechanisms, and thermodynamics of the formation
of polymer-capped metal clusters, and there are practically no
theoretical studies in this area, except those dealing with high-
vacuum nucleation and growth of metal clusters in the absence
of polymers. This work shows that a common-sense experi-
mental phenomenon, i.e., the inverse dependence of cluster size
on polymer concentration, can be modeled qualitatively, and
provide the basis for further, more quantitative analysis.
(
solvent only) is shown in Figure 3. The average particle size
obtained in this case is 210.1 Å, and the particle size standard
deviation is 56.26 Å. Another TEM micrograph picture of cobalt
particles obtained from a decomposition reaction in a PS matrix
(85 wt % PS in solution) is shown in Figure 4. The average
particle size obtained in this case is 22.9 Å. Clearly, the presence
of PS in the reaction medium at these extreme concentrations
has a drastic effect on final average particle size, which
decreases by more than an order of magnitude when the medium
consists mainly of PS.
Figure 5 summarizes the experimental data obtained for media
containing a variety of concentrations of PS. Not only does the
average final particle size decrease with increasing initial
concentration of PS, but the average cobalt size distribution
decreases as well. This indicates that the presence of the polymer
in the reaction medium prevents the formation of larger clusters
thus inhibiting the coagulation process.

Cluster Coagulation and Growth Limitations
J. Phys. Chem. B, Vol. 106, No. 1, 2002 151
Acknowledgment. This work was supported in part by the
Fischbach Fellowship to H. G. Rotstein, the Faculty Grant-in-
Aid program at the University of Minnesota, and NSF-ERC
(20) Kernizan, C. F.; Klabunde, K. J.; Sorensen, C. M.; Hadjipanayis,
G. C. Chem. Mater. 1990, 2, 70.
(
21) Brus, L. E. J. Phys. Chem. 1986, 90, 2555.
(22) Reich, S.; Goldberg, E. P. J. Polym. Sci., Polym. Phys. Ed. 1983,
(Packaging Research Center at the Georgia Institute of Technol-
21, 869.
(23) Tannenbaum, R.; Goldberg, E. P.; Flenniken, C. L. Metal-containing
ogy) to R. Tannenbaum.
Polymeric Systems; Carraher, C., Pittman, C. U., Sheats, J., Eds.; Plenum
Press: New York, 1985, p 303-340.
References and Notes
(24) Bor, G.; Dietler, U. K.; Pino, P.; Po e¨ , A. J. Organomet. Chem.
1
976, 154, 301.
1) Mason, M. G. Phys. ReV. B 1983, 27, 748.
(25) Natta, G.; Ercoli, R.; Castellano, S. Chim. Ind. (Milan) 1955, 37,
6.
(
(
(
(
(
26) Tannenbaum, R. Langmuir 1997, 13, 5056-5060.
27) Tannenbaum, R. Inorg. Chim. Acta 1994, 227, 233-240.
28) Klabunde, K. J.; Tanaka, Y. J. Mol. Catal. 1983, 47, 843.
29) Imizu, Y.; Klabunde, K. J. Inorg. Chem. 1984, 23, 3602.
30) Cotton, F. A.; Wilkinson, G. AdVanced Inorganic Chemistry, 5th
2
8
ed.; Wiley-Interscience: New York, 1988.
31) Israelaschvilii, J. Intermolecular and Surface Forces, 2nd ed.;
Academic Press: New York, 1991.
32) Rotstein, H. G.; Novick-Cohen, A.; Tannenbaum, R. J. Stat. Phys.
998, 90, 119.
33) Rotstein, H. G. Coagulation Equations with Cluster-wall Interac-
6) Gates, B. C.; Guczi, L.; Knozingen, H. Metal Clusters and
(
(
1
(
tions; M.Sc. Thesis, Department of Mathematics, Technion-Israel Institute
of Technology, 1994.
(
34) Ziff, R. M. J. Stat. Phys. 1980, 23, 241.
(35) Ziff, R. M.; Ernst, M. H.; Hendriks, E. M. J. Stat. Phys. 1983,
31(3), 519.
(
(
(
36) Smoluchowski, M. V. Physik. Z. 1916, 17, 557.
37) Leyvraz, F.; Tschudi, H. R. J. Phys. A.: Math. Gen. 1981, 14, 3389.
(
(38) McLeod, J. B. Q. J. Math. Oxford (2) 1962, 13, 119.
(39) Leyvraz, F.; Tschudi, H. R. J. Phys. A: Math. Gen. 1982, 15, 1951.
(40) Smoluchowski, M. V. Z. Phys. Chem. 1917, 92, 129.
(
(
(41) Bird, R. B.; Curtis, C. F.; Hassager, O.; Armstrong, R. C. Dynamics
1
of Polymeric Liquids: Vol. 2, Kinetic Theory; Wiley: New York, 1987.
(42) deGennes, P. G. Scaling Concepts in Polymer Physics; Cornell
University Press: Ithaca, New York, 1979.
(43) Jarrold, M. F. Clusters of Atoms and Molecules I; Haberland, H.,
Ed.; Springer Series in Chemical Physics; Springer-Verlag: Heidelberg,
1995.
(44) Riley, S. J. Clusters of Atoms and Molecules II; Haberland, H.,
Ed.; Springer Series in Chemical Physics; Springer-Verlag: Heidelberg,
1995.
(
(
1
(
5
(
(
Polym. Phys. Ed. 1987, 25, 1341.
19) Alivasatos, A. P.; Harris, A. L.; Levinos, N. J.; Steigerwald, M.
L.; Brus, L. E. J. Chem. Phys. 1988, 89, 4001.
(45) Tannenbaum, R. Curr. Trends Polym. Sci. 1998, 3, 81-98.
(46) Tang, J. G.; Hu, K.; Liu, H. Y.; Guo, D.; Wu, R. J. J. Appl. Polym.
Sci. 2000, 76, 1857-1864.
(