Kinetics of nucleation, growth, and stabilization of cobalt oxide nanoclusters

Article abstract of DOI:10.1021/jp0355004

Metallic fragments, or metallomers (similar to the mer -based nomenclature for polymers), formed via the decomposition of organometallic complexes, are highly reactive, generating a nucleation and growth process that culminates in the formation of nanocrystals. In the absence of stabilizing molecules, the aggregation process is self-restricting mainly because of the decreasing mobility of the particles and their declining diffusional rates as a function of their increasing size. On the other hand, in the presence of a polymer in the reaction medium, the growing metallic particles are stabilized by the adsorption of the polymer chains onto their surfaces, thus lowering their surface energy and creating a barrier to further aggregation. Studies of the nucleation and aggregation kinetics of metallic particles formed from the decomposition of organometallic precursors have been used to shed light on the mechanism of their formation. In these studies, the rate of decomposition of the precursor organometallic complexes used has been deemed representative of the overall rate of the process. Moreover, it has implicitly been assumed that the formation kinetics of the metal nanoclusters directly mirrors the decomposition kinetics of the precursors. In this study, we attempt to decouple the kinetic characteristics of the various steps that comprise the overall nucleation and aggregation process for cobalt oxide nanoclusters. A combination of infrared and X-ray photoelectron spectroscopies, transmission electron microscopy, and dynamic light scattering is used to identify the individual contribution of each step to the overall mechanism of metal nanocluster formation.

Full text of DOI:10.1021/jp0355004

J. Phys. Chem. B 2003, 107, 12097-12104
12097
Kinetics of Nucleation, Growth, and Stabilization of Cobalt Oxide Nanoclusters
†
‡
,†
S. King, K. Hyunh, and R. Tannenbaum*
School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332-0245,
and Department of Chemical Engineering and Materials Science, UniVersity of Minnesota,
Minneapolis, Minnesota 55455
ReceiVed: May 29, 2003; In Final Form: August 27, 2003
Metallic fragments, or metallomers (similar to the “mer”-based nomenclature for polymers), formed via the
decomposition of organometallic complexes, are highly reactive, generating a nucleation and growth process
that culminates in the formation of nanocrystals. In the absence of stabilizing molecules, the aggregation
process is self-restricting mainly because of the decreasing mobility of the particles and their declining
diffusional rates as a function of their increasing size. On the other hand, in the presence of a polymer in the
reaction medium, the growing metallic particles are stabilized by the adsorption of the polymer chains onto
their surfaces, thus lowering their surface energy and creating a barrier to further aggregation. Studies of the
nucleation and aggregation kinetics of metallic particles formed from the decomposition of organometallic
precursors have been used to shed light on the mechanism of their formation. In these studies, the rate of
decomposition of the precursor organometallic complexes used has been deemed representative of the overall
rate of the process. Moreover, it has implicitly been assumed that the formation kinetics of the metal nanoclusters
directly mirrors the decomposition kinetics of the precursors. In this study, we attempt to decouple the kinetic
characteristics of the various steps that comprise the overall nucleation and aggregation process for cobalt
oxide nanoclusters. A combination of infrared and X-ray photoelectron spectroscopies, transmission electron
microscopy, and dynamic light scattering is used to identify the individual contribution of each step to the
overall mechanism of metal nanocluster formation.
1
. Introduction
by a cascade of ligand dissociation and complex rearrangements,
giving rise to coordinatively unsaturated, multinuclear metal
Understanding the molecular-level mechanisms responsible
complexes having a higher metal/ligand ratio than the original
for the nucleation and aggregation of metal nanoclusters is
essential in any process involving the self-assembly of hierarchi-
cal systems formed from metal nanocluster building blocks.
Moreover, the physical and chemical properties of such metal
nanoclusters are determined by their size, shape, and spatial
distribution, and hence, the ability to control these variables
affords a high level of latitude in the design and fabrication of
materials with nanoscale dimensions. This level of control of
metal cluster nanostructure can be achieved through the proper
choice of the synthesis and processing conditions. A synthesis
method that is well suited for these requirements is the
decomposition of organometallic complexes under controlled
conditions in the presence of stabilizing molecules such as
10-12
precursor.
These metallic fragments, or metallomers (similar
to the “mer”-based nomenclature for polymers), created by the
energy-induced decomposition of organometallic complexes are
2
,3
highly reactive, generating a nucleation and growth process
that culminates in the formation of nanocrystals. In the absence
of stabilizing molecules, the aggregation process is self-
restricting mainly because of the decreasing mobility of the
particles as a function of their increasing size and because of
the declining diffusional rates of the larger metallic fragments
achievable in the reaction medium. The reactivity of the
coordinatively unsaturated organometallic species formed during
the decomposition process, even at low precursor concentration,
constitutes the driving force for the formation of the metallomers
via van der Waals attraction forces, and hence, the nucleation
process is not generated by supersaturation, as suggested by
1
-5
polymers to form stable, relatively uniform metal or metal oxide
dispersions of very small particle size.6-9 The overall reaction
may be described as follows:
1
3,14
the LaMer mechanism.
The aggregation process, however,
is strongly limited by the mobility and, consequently, by the
diffusional rates of the growing particles, in accordance with
both the LaMer mechanism and various growth models.1
polymer, solvent
n
M
(
L
)
y
8
x
e
n
e
r
g
y
,
c
o
n
t
r
o
l
l
e
d
a
t
m
o
s
p
h
e
r
e
5-19
(0)
(i)
Mk and/or (M O ) + nyL
j
ij/2 h
On the other hand, in the presence of various reactive
polymers in the reaction medium, that is, polymers having
various functional groups, the growing metallic particles are
stabilized by the adsorption of the polymer chains onto the
surface of the growing metal fragments, lowering their surface
energy and creating a barrier to further aggregation. Moreover,
the increased solution viscosity due to the presence of even low
concentrations of polymer has the effect of decreasing the
(
where M ) metal, L ) organic ligand, k ) nx, and h ) nx/j).
The energy-induced decomposition of these organometallic
precursors is initiated by the loss of at least one ligand, followed
*
Corresponding author. E-mail: rina.tannenbaum@mse.gatech.edu.
Georgia Institute of Technology.
†
‡
University of Minnesota.
1
0.1021/jp0355004 CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/15/2003

12098 J. Phys. Chem. B, Vol. 107, No. 44, 2003
King et al.
polymer layer bonded onto the surface, ultimately determine
the equilibrium cluster size.
Studies of the nucleation and aggregation kinetics of metallic
particles formed from the decomposition of organometallic
precursors have been used to shed light on the mechanism of
their formation. In these studies, the rate of decomposition of
the precursor organometallic complexes used has been deemed
9
,27
representative of the overall rate of the process.
Moreover,
it has implicitly been assumed that the formation kinetics of
the metal nanoclusters directly mirrors the decomposition
kinetics of the precursors. This assumption is due to the lack
of direct experimental data with respect to the dynamics of metal
cluster growth either in the absence or in the presence of
polymers or any other stabilizing molecules. Hence, informa-
tion that may arise from the overlooked mismatch of the kinetics
of the two processes has potentially a direct bearing on the
elucidation of the mechanistic aspects of the nucleation and
aggregation process as a whole.
Figure 1. Schematic description of the formation of metallomers and
the nucleation, growth, and stabilization of cobalt oxide nanoparticles.
diffusional rates of all species in the reaction solution,20,21
thereby reducing the rates of both metallomer formation and
metal nanocluster formation. The ability to control metal
nanocluster size by manipulating their surface interactions with
25
6,22
various polymer types, molecular weights, and concentrations
supports the presence of an autocatalytic surface-growth mech-
anism, such as that suggested by Finke et al.23,24 The nucleation
and growth of metal nanoclusters in such a case may be divided
into three main steps (shown schematically in Figure 1):
formation of stable metallomers under the experimental condi-
tions, aggregation of the metallomers into larger fragments, the
structure of which is defined by the presence of a large fraction
of the metal atoms at the surface of the cluster, and finally,
adsorption of the polymer chains onto the metal cluster surface,
creating a stabilizing polymer layer capable of suppressing the
van der Waals attraction forces between neighboring frag-
In this study, we attempt to decouple the kinetic characteristics
of the various steps that comprise the overall nucleation and
aggregation process and identify the individual contribution of
each step to the overall mechanism of metal nanocluster
formation. The reaction chosen for this purpose is the formation
of cobalt oxide particles by the thermal decomposition of Co2-
(CO)8 in the presence of poly(methyl methacrylate), PMMA,
as the stabilizing agent. The decomposition of Co2(CO)8 in the
presence of various polymers has been studied in detail by us
39-41
and others,
but most of the attention has been concentrated
2
5-27
ments.
on obtaining zerovalent nanoscale metallic particles under inert
conditions. Unfortunately, such a system does not lend itself to
the independent measurement of the kinetics of the metallic
cobalt cluster formation, and hence, it cannot be used as the
model system for kinetic decoupling. When the decomposition
reaction is performed in an oxidative atmosphere, it will be
possible to determine the independent rate of the formation of
cobalt oxides and then compare it to the independent rate of
the decomposition of the carbonyl complex. Moreover, when
the changes that occur in the ester groups of the PMMA chains
are followed, it will be possible to assess the kinetics of the
adsorption of the polymer onto the cobalt oxide surface and
reach some conclusions regarding the interim stages in the
nucleation and growth process and the critical nanocluster sizes
involved in each step of the overall process.
Hence, the steady-state size of the clusters formed is
significantly influenced by the nature and molecular weight of
the polymers. The type of functional groups and chain flexibility
determine the strength of the interfacial interactions between
the growing metal cluster and the adsorbed polymer layer. The
nature of the interfacial bonding between the growing metal
clusters and the polymer chains ranges from weak van der Waals
forces to strong coordination or ionic bonds, depending on the
metal and the functional groups of the polymer.28,29 Because of
their small size, the metal clusters exhibit high surface reactivity
and ensure good mechanical coupling of the particles to the
polymer matrix.30-32
The interactions between reactive metallomers and the
surrounding polymer may occur via two different mechanisms:
formation of a bond between a metallic center in a low-
nuclearity cluster and the functional group (or groups) on the
polymer chain, resulting in a polymer-metal complex, and
adsorption of the polymer chains onto the surface of a high-
nuclearity cluster via surface interactions. The first mechanism
occurs primarily in systems in which the functional groups on
the polymer react vigorously and irreversibly with the coordi-
natively unsaturated metallic species generated in the first stages
of the decomposition reaction, and hence, it directly competes
with the metal-metal aggregation process. In the second
mechanism, the rates of the surface interactions between the
polymer and the metallic cluster are slower than the metal-
metal aggregation process, and they take place upon the
formation of a pseudostable metallic aggregate. In a given
system, both mechanisms occur simultaneously, but the relative
dominance of one or the other is a function of the polymer
reactivity (type of functional groups), polymer molecular weight
2
. Experimental Method
2.1. Kinetic Analysis of the Oxidative Decomposition of
Co (CO) . Five grams of PMMA pellets (Polyscience, Inc., Mh
2
8
w
) 323 000) and 50 mL of chlorobenzene (Fisher-Reagent) were
mixed for 48 h at room temperature in a 250 mL round-bottom
flask. The chlorobenzene was dried using molecular sieve
-
2
pellets. Fifty milliliters of a 5 × 10 M solution of Co (CO)8
2
in chlorobenzene was added dropwise to the PMMA-chlo-
robenzene solution, and the resulting solution was mixed for 5
min. The solution was subsequently heated to 90 °C in an
oxidative atmosphere (Air Products, 20% O , 78% N , and 2%
2
2
H O vapor). The overall kinetics of the reaction was determined
2
by removing aliquots of approximately 1 mL at different time
intervals, quenching them with 4 mL of cold chlorobenzene and
analyzing them by Fourier transform infrared spectroscopy. The
infrared absorption band (measured on Nicolet MX-1 and
Nicolet Nexus 870 Fourier transform infrared spectrophotom-
(
chain size), and type of metal precursor and its concen-
9
,20,27,28,31,33-38
-1
tration.
The strength of the polymer interaction
eters) at 1858 cm (bridging carbonyl stretching) was observed
with the surface atoms in a cluster, that is, the permanence of
the adhesive bond formed, and the effective thickness of the
to change during decomposition of the cobalt-containing
polymer solutions and was used to follow the kinetics of the

Kinetics of Nucleation, Growth, and Stabilization
J. Phys. Chem. B, Vol. 107, No. 44, 2003 12099
thermal decomposition reaction of Co2(CO)8. The absorption
angle, ∼80°, and the reflected light from the sample surface is
then collected by a BaF2 lens and focused onto an MCT (cooled
with liquid N2) detector. Also in these experiments, the
-
1
band at 886 cm , corresponding to the stretching vibration of
the Co-O bond in Co2O3, was used to monitor the formation
of the cobalt oxide nanoclusters.
-
1
resolution was 2 cm , and 3000 scans were collected for the
bulk PMMA films and the adsorbed polyethylene samples
containing the PMMA-coated cobalt oxide nanoclusters.
2.5. X-ray Photoelectron Spectroscopy (XPS). The X-ray
photoelectron spectroscopy experiments were carried out using
a Perkin-Elmer model 5400 spectrometer equipped with a UHV
chamber using a double pass cylindrical mirror analyzer with
nonmonochromatized Mg KR (1253.6 eV) radiation and an
analyzer pass energy of 35.75 eV. The source was operated at
2
.2. Preparation of PMMA-Coated Cobalt Oxide Nano-
cluster Samples. To extract measurable amounts of PMMA-
coated cobalt oxide nanoclusters, several experiments, as those
described in section 2.1., were performed. In each case, the
reaction was stopped at a predetermined time (e.g., after 2 h, 5
h, etc.) by immersing the flask in an ice bath. The solution was
divided into small portions (∼5 mL each) and placed in an
Eppendorf Concentrator 5301 centrifuge (14 000 rpm). This
separated excess solvent from the polymer-coated nanoclusters.
The supernatant chlorobenzene solution was removed, and the
remaining particles were washed with both chlorobenzene and
hexane to remove any excess unbound polymer or partially
reacted cobalt carbonyl fragments. The suspension was centri-
fuged again, and the process was repeated three times. The
remaining wet paste was resuspended in a 5 mL, 10 wt % xylene
solution of polyethylene (Aldrich, Mh w ≈ 125 000), and the
suspension was allowed to mix well to obtain reasonable
suspension homogeneity. Then the polymer suspension contain-
ing the PMMA-coated cobalt oxide nanoparticles was cast on
three different substrates: (1) on a glass slide, (2) on a NaCl
circular crystal, and (3) on a glass slide coated with 1000 Å of
Ag. The film formed on the glass slide was analyzed by X-ray
photoelectron spectroscopy (XPS), while the other two samples
were analyzed by transmission FTIR and reflection-absorption
FTIR, respectively.
1
00 W to minimize the damage to the PMMA films in the
samples that were examined. Under these conditions, C(1s)
spectra of bulk PMMA indicate that samples could be safely
exposed to the source for about 15 min. The operating pressure
-10
of the vacuum chamber was 2 × 10 Torr. The measurements
were made at 45° angle of incidence. The data analysis was
carried out on a Perkin-Elmer 7000 computer.
3. Results and Discussion
The nucleation and growth of cobalt oxide nanoclusters
formed by the decomposition of cobalt carbonyl complexes in
the presence of an oxidative atmosphere may be divided into
three main steps: (a) the formation of stable cobalt oxide nuclei
under the experimental conditions, (b) the aggregation of the
cobalt nuclei into larger oxide fragments the structure of which
is defined by the presence of a large fraction of the oxide
molecules at the surface of the cluster, and (c) the adsorption
of the PMMA chains onto the oxide cluster surface, reducing
the surface energy and limiting the growth of the particles. In
previous studies, the rate of decomposition of the precursor
organometallic complexes used in similar reactions has been
deemed representative of the overall rate of the process. In this
study, we attempt to deconvolute the kinetic characteristics of
the various steps that comprise the overall process and identify
the individual contribution of each step to the overall mechanism
of cobalt oxide nanocluster formation.
2
.3. Particle Size Determination. Samples for transmission
electron microscopy (TEM) were prepared by depositing a drop
of the reaction solution onto a Formvar coated copper TEM
grid (Ted Pella). The grid rested on a thin piece of tissue, such
that when the drop of solution was deposited on the grid, the
tissue pulled the liquid through the grid, leaving a very thin
film on the grid itself. Thin films were necessary for TEM
imaging because electrons cannot penetrate thicker films. TEM
analysis was done on both a JEOL 100CX TEM at an operating
voltage of 100 keV and two high-resolution microscopes, a
Hitachi HF-2000 and a JEOL 4000EX, both with operating
voltages of 200 keV.
The characterization of the cobalt oxide nanoparticle products,
performed by both XRD and TEM diffraction patterns, shown
in Figure 2, revealed that Co2O3 is the most abundant cobalt
oxide found in the system. CoO is found as well, but even its
highest intensity band, corresponding to the (111) crystal plane,
was much less intense than the highest intensity band of Co2O3,
as shown in Table 1. We are not ruling out the presence of
Co3O4, but the intensity of the bands corresponding to this
compound is relatively low, indicating a low concentration.
Figure 3 shows the kinetics of the decomposition and
oxidation of Co2(CO)8 in a PMMA solution. The reaction is
assumed to follow first-order kinetics. The decrease in the
concentration of Co2(CO)8 is monitored by following the
In parallel, dynamic light scattering (DLS) particle size
determination was performed on 10 mL samples removed from
the solutions prepared similarly to those described in section
2.2. The samples were placed in standard scintillation vials
(
∼27.4 mm diameter) as the sample cells. Measurements were
conducted on a Brookhaven DLS system equipped with a 35
mW, 632.8 nm wavelength HeNe laser and a stepping motor
allowing data collection at angle increments of 0.01°. The results
were analyzed by using CONTIN, NNLS (nonnegative least
squares), and Williams-Watts size distribution programs.
-
1
2
.4. Surface-Sensitive Reflection-Absorption Infrared
decrease of the 1858 and 1867 cm absorption bands, corre-
sponding to the stretching vibrational modes of the bridging
Spectroscopy (IRRAS). The preliminary infrared experiments
on the samples with the Ag substrate were conducted on an
IBM-IR-44 (Bruker) FTIR spectrometer equipped with a Harrick
9
,44-48
-1
carbonyls of the cobalt complex.
The 1858 cm band is
used as the analytic band. Because the intensity of this
absorption band is directly proportional to the concentration of
the cobalt complex in the polymer solution (via the relationship
E ) ꢀdc), it is not necessary to engage in the lengthy process
of calculating the molar extinction coefficient of Co2(CO)8 in
the PMMA matrix, but it is sufficient to express the kinetic
behavior as ln(E1858) vs time and extract the rate coefficient.
The rate coefficient observed for this reaction is consistent with
previous experiments conducted in similar circumstances in the
presence of PMMA and other similar polymeric matrixes.
4
2
variable-angle specular reflection (IRRAS) attachment. The
cell chamber was purged with dry nitrogen for at least half an
hour before collecting interferograms. The resolution was 2
-
1
cm , and 3000 scans were collected for the bulk and the
adsorbed polymer samples. Additional experiments were con-
ducted on a Nicolet Nexus 870 equipped with a PM-IRRAS
attachment. The polarization was achieved by using a 50 kHz
43
ZnSe photoelastic modulator (PEM). The polarization modu-
lated IR light impinges on the sample surface near the grazing

12100 J. Phys. Chem. B, Vol. 107, No. 44, 2003
King et al.
Figure 3. The kinetics of the decomposition and oxidation of Co
2
-
(
CO) in a PMMA solution (as indicated by the changes in the intensity
8
-
1
of the 1858 cm infrared band). The rate coefficient is directly
calculated from the plot of ln(E1858) vs time because the band intensity
is directly proportional to the cobalt precursor concentration (via the
relationship E ) ꢀdc).
compared to the plot of the absorption intensity of the 1858
-
1
cm band as a function of time and shown in Figure 4b. It is
-
1
interesting to note that the appearance of the 883 cm infrared
absorption band has an induction period, suggesting a delay in
the formation of the oxide compared to the decrease in the
concentration of the precursor. This suggests that the cobalt
oxide nuclei are formed as a result of a reaction cascade in which
2
5
the oxidation occurs only in the final step. This, again, is
consistent with various suggested mechanisms of cobalt carbonyl
decompositions, in which the formation of stable nuclei is due
1
1,50
to a systematic decrease in the CO/Co ratio,
generating
coordinatively unsaturated and highly reactive cobalt species,
which in turn undergo oxidation under ambient atmospheric
conditions. An indication as to the critical size of cobalt
fragments that are sufficient to generate the onset of the
oxidation process can be assessed indirectly from the duration
of the induction period. According to Figure 4b, at least in the
first hour of the decomposition reaction, the intensity of the
Figure 2. XRD (top) and TEM (bottom) diffraction patterns of the
cobalt oxide nanocluster products obtained via the thermal decomposi-
tion of metal carbonyl precursors in the presence of PMMA and in an
oxidative atmosphere.
TABLE 1: Analysis of the Diffraction Patterns of the Cobalt
Oxide Nanoparticle Products Obtained from XRD and TEM
Experiments
-
1
1858 cm band has decreased by ∼15% (that is, about 15%
of the initial Co2(CO)8 has reacted) without generating a similar
cobalt oxide
exptl XRD
d spacing
exptl XRD
relative
-1
increase of the 883 cm band. TEM micrographs and dynamic
exptl E-beam
Co
O
2 3
CoO
light scattering (DLS) measurements taken from samples
generated during the induction period and at the end of the
process (shown in Figure 4c on the left and right, respectively)
show that no measurable particles were formed after 1 h of
reaction, while the particles formed at the end of the process
have an average size of ∼70 nm. Because the sensitivity limit
of the DLS instrument is ∼3 nm, we conclude that the critical
size for the onset of the oxidation of the cobalt reactive nuclei
is e3 nm, that is, e860 cobalt atoms. Finally, the intensity of
(
Å)
d spacing (Å)
d (Å)
d (Å)
intensity (%)
3
2
2
2
2
2
2
2
2
1
1
1
1
1
.326
.901
.799
.484
.405
.315
.105
.075
.017
.712
.662
.572
.492
.450
3.210
2.970
2.817
2.481
2.330
2.324
2.163
2.024
1.960
1.852
1.630
1.576
1.472
1.379
3.321
14.3
14.9
32.9
52.7
100.0
6.8
2.836
2.460
2.130
2.380
2.320
5.2
13.1
15.1
6.1
10.9
20.0
14.8
15.8
-1
the 883 cm infrared absorption band tapers off and reaches a
1.780
1.645
1.570
plateau, suggesting a saturation of the cobalt oxide formation.
FTIR measurements of the washed cobalt oxide particles
formed during the decomposition reaction, shown in Figure 5,
reveal that an extensive reaction between the polymer and the
metal oxide particles has taken place, resulting in the formation
1.506
1.397
-
Figure 4 shows the direct formation of the cobalt oxide
clusters, as evidenced by the changes observed in the infrared
absorption bands of the Co-O bonds in Co2O3 at 883 and 832
of RCOO groups (where R is the PMMA main chain) and,
-
III
hence, possibly a new (RCOO )-Co complex. As the reaction
progresses, three distinct functional groups on the PMMA are
observed, (a) the initial CdO stretching vibration band at 1741
-
1
cm , corresponding to the asymmetric and symmetric vibra-
49
-1
-1
tional modes of the oxide, respectively. The increase of the
83 cm infrared absorption band, corresponding to the
asymmetric Co-O stretch, is shown in Figure 4a. The plot of
the absorption intensity of this band as a function of time is
cm , (b) the COOH stretching vibration at 1702 cm , and
-
1
-
-1
8
the COO stretching vibration at 1674 cm . The presence of
these bands in the PMMA adsorbed onto the cobalt oxide
particles points to the hydrolysis of the ester group of the PMMA

Kinetics of Nucleation, Growth, and Stabilization
J. Phys. Chem. B, Vol. 107, No. 44, 2003 12101
Figure 4. The direct formation of the cobalt oxide cluster as indicated by the appearance of the infrared absorption bands of the Co-O bonds in
-
1
2 3
Co O as a function of the growth of the cobalt oxide nanocluster: (a) the increase of the 852 cm infrared absorption band corresponding to the
Co-O stretch; (b) the plot of the absorption intensity of this band as a function of time compared to the plot of the absorption intensity of the 1858
-
1
cm band as a function of time; (c) TEM micrographs and dynamic light scattering (DLS) measurements taken from samples generated during the
induction period (left image) and at the end of the process (right image).
ester group resides on the side chain of the polymer, the
hydrolysis of the ester bond does not result in chain scission.
The intensities of these three bands corresponding to the
concentration of the three different functional groups on the
adsorbed PMMA chains are shown in Figure 6a. During the
induction period observed for the formation of cobalt oxide,
no particles could be isolated, and hence, the first successful
measurement was performed only after 2 h since the start of
the reaction. Of particular interest is the delay in the appearance
-
of the COO functional group because it is similar to the
induction period observed in the formation of the oxide. Previous
studies involving the reactive adsorption of PMMA on chro-
mium oxide surfaces have shown that the PMMA chains adsorb
onto the metal oxide surface via the direct ionic bonding of the
carboxylate ion that was generated by the cleavage of the ester
Figure 5. FTIR spectra of the washed cobalt oxide particles formed
during the decomposition reaction and stabilized by PMMA.
5
4
bond through hydrolysis, facilitated by the metal oxide surface.
with the subsequent formation of a carboxylic group followed
by the formation of a carboxylate ion,51-54 as shown:
-
As a result, the formation of the COO group is intimately
linked to the formation of the cobalt oxide particles, and hence,
the delay in the formation of the latter is mirrored in the
formation of the former. Moreover, the formation of the
carboxylate ion is initially slower than the formation of the
carboxylic acid (i.e., the rate of the hydrolysis is greater than
the dissociation of the newly formed COOH group), but the
rate increases as the reaction proceeds. The relative intensities
-
The methanol formed by this hydrolysis can diffuse through
the thin PMMA film adsorbed on the surface of the oxide
nanoparticles and slowly evaporate (bp 64.5 °C). Because the
of the COOH band and the COO band expressed as the ratio
of these bands with the CdO band are shown in Figure 6b.
However, the overall rate of formation of the carboxylate ion

12102 J. Phys. Chem. B, Vol. 107, No. 44, 2003
King et al.
TABLE 2: The Characteristic Core-Level C(1s) XPS Peaks
of a 250 Å PMMA Film Deposited by Spin-Coating on a
Polystyrene-on-Glass Substrate and Considered for This
Study as Bulk PMMAa
a
The specific peaks are marked with identifying band numbers that
appear in the text.
Figure 7. The high-resolution C(1s) core-level electron XPS contour
spectra of the PMMA-stabilized cobalt oxide nanoparticles: (a) XPS
spectra obtained at different reaction times and the contour spectrum
of the bulk PMMA; (b) plot of the increase in the height of the 287.94
eV peak as a function of time.
Figure 6. Comparison of the kinetics of the three functional groups
formed during the reaction of PMMA with reactive cobalt oxide
clusters: (a) plot of the three bands of interest, (9) the initial CdO
-
1
stretching vibration band at 1741 cm , (2) the COOH stretching
-
1
-
vibration at 1702 cm , and (b) the COO stretching vibration at 1674
-
1
cm ; (b) plot of the relative intensities of (2) the COOH band and
performed but are not shown): (1) the 284.57 eV peak
corresponds to the hydrocarbons in the main chain, (2) the
(
b) the COO- band expressed as the ratio of these bands with the
CdO band; (c) plot of the overall rate of formation of the carboxylate
287.06 eV peak corresponds to the methyl group singly bonded
ion compared to that of the formation of the COOH group.
to an oxygen atom, and (3) the 288.69 eV peak corresponds to
55,56
is similar to that of the formation of the COOH group, as shown
in Figure 6c. First-order formation rates are assumed, but
because the overall reaction is composed of at least two
consecutive reactions, this assumption may be only partially
correct.
the ester carbon.
The experimental ratios among these three
peaks are 67:17:16, in reasonable agreement with the theoretical
ratios of 3:1:1. Because the correct value for the binding energy
of the C(1s) core electrons of a uncharged aliphatic hydrocarbon
5
4-56
compound is 284.80 eV,
we conclude that there was no
Further support for the hydrolysis mechanism of PMMA and
the delayed formation of COO- groups concurrent with the
delayed formation of the cobalt oxide nanoparticles was obtained
from XPS experiments. As a first step, the XPS spectra of a
charging effect, which may take place when analyzing polymer
films.
57
The high-resolution C(1s) core-level electron contour XPS
spectra of the PMMA-stabilized cobalt oxide nanoparticles
obtained at different reaction times are shown in Figure 7a,
together with the contour spectrum of the bulk PMMA. A new
peak is observed at 287.94 eV, corresponding to a carbon-
oxygen bond the bond order of which is greater than 1 and
smaller than 2 and hence the bond character is between that of
250 Å PMMA film deposited by spin-coating on a polyethylene-
coated glass substrate was considered for this study as bulk
PMMA, and the precise position of the characteristic core C(1s)
peaks was established, as shown in Table 2. Bulk PMMA gives
rise to three C(1s) peaks (peak deconvolutions have been

Kinetics of Nucleation, Growth, and Stabilization
J. Phys. Chem. B, Vol. 107, No. 44, 2003 12103
TABLE 3: A Summary of the C(1s) Core-Level Electron
Peaks Obtained at Different Times during the
obtained during the synthesis of cobalt oxide nanoclusters
stabilized by PMMA chains. Unlike traditional kinetic studies,
in which the determination of the precursor decomposition
reaction rate is assumed to be representative of the overall cobalt
oxide formation rate, we have shown here that there are several
crucial steps in the nanoparticle aggregation process and that
the traditional assumptions regarding the dynamics of the
process are not correct. The formation of cobalt oxide nano-
particles with an average size above the detection threshold of
dynamic light scattering (∼3 nm) and TEM (∼2-3 nm) does
not mirror the decomposition of the precursor during the same
time frame. Moreover, the stabilization process, that is, the
adsorption of the polymer chains onto the nanocluster surface,
is shown to occur in the later stages of the aggregation, as is
evidenced by the common and parallel induction period
observed in the formation of the cobalt oxide nanoclusters and
the transformations that occur in the ester functional groups of
PMMA. It is therefore clear that the aggregation and stabiliza-
tions processes occur only after the reactive metallic fragments,
that is, metallomers, have reached a critical size and a distinct
surface structure.
Decomposition Reaction of Co
Containing PMMA
2
(CO)
8
in a Solution
reaction
time (h)
peak
position (eV)
peak
height (N /E)
E
2
5
0
5
287.94
287.94
287.94
287.94
0.46
1.71
2.23
3.07
1
1
a carbonyl and a carbon-oxygen single bond.54,58-61 A summary
of the C(1s) core level peaks obtained at different reaction times
is given in Table 3, and the increase in the height of the 287.94
eV peak is plotted as a function of time and shown in Figure
7
b. The intensity of the new peak increases sharply after the
end of the induction period discussed earlier, but stabilizes after
10 h. This is in good agreement with the slowing rate of
growth of the cobalt oxide nanoclusters, as evidenced by the
∼
-1
slower increase in the intensity of the 883 cm absorption band
shown in Figure 4b. The most striking features of the XPS
spectra are the dramatic decrease in the relative intensity of the
carbonyl carbon atoms and a considerable increase in the
carbons associated with the new peak. The peak generated by
carbon atoms singly bonded to oxygen does not seem to be
affected by the interaction of PMMA with the cobalt oxide
nanoclusters. These observations imply two main conclusions:
Acknowledgment. We thank the Surface Science Center at
the University of Minnesota and its director, Dr. Raul Carreta,
for assistance with the XPS measurements. This work is
supported by a University of Minnesota Faculty Research Award
and by the Georgia Tech Research Institute Equipment Grant
(a) there has been an extensive reaction of the polymer with
the surface of the cobalt oxide nanocluster and (b) there may
be more than one single mode of bonding between the PMMA
and the surface. The bonding may occur either via the interaction
of the carboxylate ion with the cobalt oxide, in which case only
one cobalt center per functional group will be involved, or via
the separate interaction of each oxygen atom in the carboxylate
ion, in which case two cobalt centers per functional group will
(
both to R. Tannenbaum).
References and Notes
(
(
(
1) Martin, T. P. Solid State Ionics 2000, 131(1), 3-12.
2) Klaubunde, K. J.; Tanaka, Y. J. Mol. Catal. 1983, 21, 57.
3) Kanai, H.; Tan, B. J.; Klaubunde, K. J. Langmuir 1986, 2 (6), 760.
(4) Bockstaller, M.; Kolb, R.; Thomas, E. L. AdV. Mater. 2001, 13
be involved, as shown:2
7,54,62,63
(23), 1783-1786.
(
5) Zhao, S.; Wang, S.; Ye, H. J. Phys. Soc. Jpn. 2001, 70 (10), 2953-
957.
6) Tannenbaum, R.; Reich, S.; Flenniken, C. L.; Goldberg, E. P. AdV.
Mater. 2002, 14 (19), 1402-1405.
7) Tannenbaum, R.; Flenniken, C. L.; Goldberg, E. P. J. Polym. Sci.,
Polym. Phys. Ed. 1987, 25, 1341.
8) Tannenbaum, R.; Flenniken, C. L.; Goldberg, E. P. J. Polym. Sci.,
Polym. Phys. Ed. 1990, 28, 2421.
9) Tannenbaum, R. Langmuir 1997, 13 (19), 5056 and pertinent
2
(
(
(
(
The XPS spectra do not provide sufficient information to
distinguish between the two bonding possibilities. The XPS
spectra of the corresponding O(1s) core electrons are not shown
because they could not be used for the analysis of the reaction
mechanism. The bulk PMMA samples generated two O(1s) core
electron peaks, corresponding to the carbonyl oxygen and the
methoxy oxygen. However, the O(1s) core electron peaks for
bare cobalt oxide nanoclusters could not be used as the
appropriate baseline for the oxygen peaks originating from the
surface because the in situ synthesis of cobalt oxide clusters in
the absence of PMMA produces large cobalt oxide aggregates
and hence their surface cannot be reliably compared to the bare
surface of smaller cobalt oxides, such as those formed in the
presence of PMMA. Moreover, even if one would assume
similar surface properties for large and small cobalt oxide
particles, there is still an insurmountable difficulty in applying
surface-sensitive techniques to these particulate systems that will
allow the identification of surface-layer cobalt oxide molecules
versus inner-layer cobalt oxide molecules.
references therein.
(10) Heck, R. F.; Breslow, D. S. J. Am. Chem. Soc. 1961, 83, 4023.
(
(
11) Ungv a´ ry, F.; Mark o´ , L. J. Organomet. Chem. 1969, 20, 205.
12) Werner, P.; Ault, B. S.; Orchin, M. J. Organomet. Chem. 1978,
162, 189-194.
(13) LaMer, V. K.; Dinegar, R. H. J. Am. Chem. Soc. 1950, 72, 4847.
(14) LaMer, V. K. Ind. Eng. Chem. 1952, 44, 1270.
(
(
(
(
15) Smoluchowsky, M. V. Z. Phys. Chem. 1917, 92, 129.
16) McLeod, J. B. Q. J. Math. Oxford (2) 1962, 13, 119.
17) Ziff, R. M. J. Stat. Phys. 1980, 23, 241.
18) Leyvraz, F.; Tschudi, H. R. J. Phys. A: Math. Gen. 1982, 15, 1951.
(19) Rotstein, H. G.; Novick-Cohen, A.; Tannenbaum, R. J. Stat. Phys.
998, 90, 119.
1
(
20) Bird, R. B.; Curtis, C. F.; Hassager, O.; Armstrong, R. C. Dynamics
of Polymeric Liquids: Vol. 2, Kinetic Theory; Wiley: NewYork, 1987.
21) Tang, J. G.; Hu, K.; Liu, H. Y.; Guo, D.; Wu, R. J. J. Appl. Polym.
Sci. 2000, 76, 1857-1864.
22) For example: (a) Zanchet, D.; Hall, B. D.; Ugarte, D. J. Phys.
(
(
Chem. B 2000, 104 (47), 11013-11018. (b) Whaley, S. R.; English, D. S.;
Hu, E. L.; Barbara, P. F.; Belcher, A. M. Nature 2000, 405, 665-668. (c)
Ahmadi, T. S.; Wang, Z. L.; Green, T. C.; Henglein, A.; El-Sayed, M. A.
Science 1996, 272, 1924-1925. (e) Petroski, J. M.; Wang, Z. L.; Green, T.
C.; El-Sayed, M. A. J. Phys. Chem. B 1998, 102, 3316-3320. (f) Wang,
Z. L.; Mohamed, M. B.; Link, S.; El-Sayed, M. A. Surf. Sci. 1999, 440,
L809-L814. (g) Bradley, J. S.; Tesche, B.; Busser, W.; Maase, M.; Reetz,
M. T. J. Am. Chem. Soc. 2000, 122, 4631-4636. (h) Wang, Z. L.; Gao, R.
P.; Nikoobakht, B.; El-Sayed, M. A. J. Phys. Chem. B 2000, 104, 5417-
4. Summary
The intent of the work presented in this paper was to decouple
5420. (i) Link, S.; Burda, C.; Nikoobakht, B.; El-Sayed, M. A. J. Phys.
the various contributions embedded in the kinetic information
Chem. B 2000, 104, 6152-6163.

12104 J. Phys. Chem. B, Vol. 107, No. 44, 2003
King et al.
(
23) Watzky, M. A.; Finke, R. G. J. Am. Chem Soc. 1997, 119, 10382-
(43) Wang, B. Spectroscopy 1997, 12 (1), 30.
1
5
0400.
(44) Ungv a´ ry, F.; Mark o´ , L. J. Organomet. Chem. 1974, 71, 283.
(45) Bor, G.; Dietler, U. K. J. Organomet. Chem. 1980, 191, 295.
(46) Bor, G.; Dietler, U. K.; Noack, K. J. Chem. Soc., Chem. Commun.
(
24) O¨ zkar, S.; Finke, R. G. J. Am. Chem. Soc. 2002, 124 (20), 5796-
810.
(
25) Widegren, J. A.; Aiken, J. D., III; O¨ zkar, S.; Finke, R. G. Chem.
Mater. 2001, 13 (2), 312-324.
26) Biffis, A.; D′Archivio, A. A.; Jerabek, K.; Schmid, G.; Corain, B.
AdV. Mater. 2000, 12 (24), 1909-1912.
27) Rotstein, H. G.; Tannenbaum, R. J. Phys. Chem. B 2002, 106 (1),
1976, 914.
(47) Noack, K. HelV. Chim. Acta 1962, 45, 1847 and all pertinent
references therein.
(48) Braterman, P. S. Metal Carbonyl Spectra; Academic Press: New
York, 1975.
(
(
1
46-151.
(49) Uzunova, E. L.; St. Nikolov, G.; Mikosch, H. J. Phys. Chem. A
2002, 106 (16), 4104-4114.
(
28) Tadd, E.; Bradley, J.; Tannenbaum, R. Langmuir 2002, 18 (6),
2
378-2384.
(50) Werner, P.; Ault, B. S.; Orchin, M. J. Organomet. Chem. 1978,
162, 189-194.
(29) Spatz, J. P.; M o¨ ssmer, S.; Hartmann, C.; M o¨ ller, M.; Herzog, T.;
Krieger, M.; Boyen, H.-G.; Ziemann, P.; Kabius, B. Langmuir 2000, 16,
07-415.
30) Tsivline, D.; Stepanyuk, V. S.; Levanov, N.; Hergert, W.; Katsnel-
son, A. A. Comput. Phys. Commun. 1999, 121, 747.
(51) Konstadinidis, K.; Thakkar, B.; Chakraborty, A.; Potts, L. W.;
Tannenbaum, R.; Tirrell, M.; Evans, J. F. Langmuir 1992, 8 (5), 1307-
1317.
(52) Allara, D. L. Polym. Sci. Technol. 1980, 12B, 751.
(53) Mallik, R. R.; Pritchard, R. G.; Horley, C. C. Polymer 1985, 26,
551.
4
(
(
(
31) Dan, N. Langmuir 2000, 16 (8), 4045.
32) Tadd, E.; Zeno, A.; Zubris, M.; Dan, N.; Tannenbaum, R.
Macromolecules 2003, appeared on the Web (ASAP) on 7/29/03.
(54) Tannenbaum, R.; Hakanson, C.; Zeno, A.; Tirrell, M. Langmuir
2002, 18 (14), 5592-5599.
(55) Clark, D. T.; Thomas, H. R. J. Polym. Sci., Polym. Chem. Ed. 1971,
14, 1701.
(
(
33) Flory, P. J.; Fox, T. G. J. Am. Chem. Soc. 1951, 73, 1904.
34) Flory, P. J. Principles of Polymer Chemistry; Cornell University
Press: Ithaca, New York, 1953.
(
(
35) Graessley, W. W. AdV. Polym. Sci. 1974, 16, 1.
(56) Dilks, A. In Electron Spectroscopy: Theory, Techniques and
Applications; Brundle, C. R., Parker, A. D., Eds.; Academic Press: 1981;
Vol. 4, p 277.
(57) Riggs, W. M.; Parker, M. J. In Methods of Surface Analysis;
Czandera, A. W., Ed.; Elsevier Scientific: 1975; p 103.
(58) Bartha, J. W.; Hahn, P. O.; LeGoues, F. K.; Ho, P. S. J. Vac. Sci.
Technol. 1985, A3, 1390.
36) deGennes, P. G. Scaling Concepts in Polymer Physics; Cornell
University Press: Ithaca, New York, 1979.
37) Reetz, M. T. W.; Helbig, S. A.; Quaiser, U.; Stimming, N.; Breuer,
R.; Vogel Science 1995, 267, 367.
(
(
(
38) Hariharan, R.; Russell, W. B. Langmuir 1998, 14 (25), 7104.
39) Reich, S.; Goldberg, E. P. J. Polym. Sci., Polym. Phys. Ed. 1983,
2
1 (6), 869-879.
40) Griffiths, C. H.; O’Horo, M. P.; Smith, T. W. J. Appl. Phys. 1979,
0 (11), 7108-7115.
41) Rotstein, H.; Tannenbaum, R. In Synthesis, Functionalization and
(59) Jordan, J. L.; Sanda, P. N.; Morar, J. F.; Kovac, C. A.; Himpsel, F.
J.; Pollak, R. A. J. Vac. Sci. Technol. 1986, A4, 1046.
(60) Nuzzo, R. G.; Wong, Y.-H.; Schwartz, G. P. Langmuir 1987, 3
(6), 1136-1140.
(
5
(
Surface Treatment of Nanoparticles; Baraton, M.-I., Ed.; Advances in
Nanoscale Materials and Nanotechnology Book Series; American Scientific
Publishers: Stevenson Ranch, CA, 2003; pp 103-126.
(61) Atanasoska, Lj.; Anderson, S. G.; Meyer, H. M., III.; Weaver, J.
H. Vacuum 1990, 40 (1-2), 91-94.
(62) Burkstrand, J. M. J. Appl. Phys. 1981, 52 (7), 4795.
(63) DeKoven, B. M.; Hagans, P. L. J. Appl. Surf. Sci. 1986, 27, 199.
(42) Duevel, R. V.; Corn, R. M. Anal. Chem. 1992, 64, 337.