Nanocrystal arrested precipitation in supercritical carbon dioxide

Article abstract of DOI:10.1021/jp011815c

FluorocarbOn-coated silver, iridium, and platinum nanocrystals ranging in size from 20 to 120 Aì? in diameter are synthesized in supercritical (sc)-CO₂ by arrested precipitation from soluble organometallic precursors. The synthesis is performed in a single CO₂ phase by reduction with H₂ at elevated temperatures ranging from 60 to 100 ?°C. Precursor degradation and particle nucleation occur in the presence of stabilizing perfluorooctanethiol ligands, which bind to the surface of the metal agglomerates and quench particle growth. The ligands are sufficiently solvated by CO₂ to provide a steric barrier to uncontrollable aggregation during synthesis. The particles redisperse in acetone and fluorinated solvents. The dominant mechanism to particle growth is through cluster agglomeration followed by ligand passivation, leading to self-similar size distributions with a standard deviation of ?±47%. Additionally, the nanocrystal size is tunable with precursor concentration, with higher precursor loadings resulting in larger nanocrystal sizes.

Full text of DOI:10.1021/jp011815c

J. Phys. Chem. B 2001, 105, 9433-9440
9433
Nanocrystal Arrested Precipitation in Supercritical Carbon Dioxide
Parag S. Shah, Shabbir Husain, Keith P. Johnston,* and Brian A. Korgel*
Department of Chemical Engineering and Texas Materials Institute, Center for Nano- and Molecular Science
and Technology, The UniVersity of Texas, Austin, Texas 78712-1062
ReceiVed: May 11, 2001
Fluorocarbon-coated silver, iridium, and platinum nanocrystals ranging in size from 20 to 120 Å in diameter
are synthesized in supercritical (sc)-CO
The synthesis is performed in a single CO
0 to 100 °C. Precursor degradation and particle nucleation occur in the presence of stabilizing perfluoro-
octanethiol ligands, which bind to the surface of the metal agglomerates and quench particle growth. The
ligands are sufficiently solvated by CO to provide a steric barrier to uncontrollable aggregation during synthesis.
2
by arrested precipitation from soluble organometallic precursors.
2
phase by reduction with H at elevated temperatures ranging from
2
6
2
The particles redisperse in acetone and fluorinated solvents. The dominant mechanism to particle growth is
through cluster agglomeration followed by ligand passivation, leading to self-similar size distributions with
a standard deviation of (47%. Additionally, the nanocrystal size is tunable with precursor concentration,
with higher precursor loadings resulting in larger nanocrystal sizes.
Introduction
than those of hydrocarbon solvents, which makes it more like
1
3-16
a fluorocarbon.
colloidal dispersions of organic and inorganic solids, and
Thus, the key to stabilizing CO2-based
Nanocrystals, 20-100 Å in diameter, exhibit unique size-
dependent optical, catalytic, magnetic, and electronic properties
compared to their bulk counterparts and could enhance a variety
of technologies including coating, environmental, chemical
1
7
18
19
water-in-CO2 microemulsions, has been the design of surfac-
tants with suitable “CO2-philic” molecular groups to provide
the steric barrier to aggregation. Surfactants with fluorinated
tails, for example fluoroacrylates, alkanes, and ethers, have
proven to be highly effective, as the weak van der Waals forces
are consistent with those of CO2.
1
-3
processing, medical, electronic, and sensing applications.
Arrested precipitation has proven to be one of the most
successful approaches for synthesizing nanocrystals in this size
range. This wet chemical synthetic method relies on organic
ligands to passivate particle surfaces during growth to provide
size control and colloid stabilizaton. Supercritical fluids (SCF)
offer potential advantages over conventional solvents for arrested
precipitation synthesis and nanocrystal processing: they exhibit
characteristics of both liquids and gases, with sufficient densities
2
0
21
Recently, silver and cadmium sulfide nanocrystals were
synthesized in water-in-CO2 microemulsions. In these systems,
it can be challenging to recover the particles from the micro-
emulsions without irreversible aggregation because the surfactant
is not chemically bound to the nanocrystal surface.22 Shah et
al. showed that nanocrystals capped with partially fluorinated
thiols (synthesized in an acetone/water mixture using a modi-
fication of the biphasic synthesis developed by Brust and co-
(and thus solvent strengths) required for ligand solvation, but
4
with much lower viscosities and higher diffusivities. SCF
density can be altered through modest changes in pressure and
temperature. The solvent density strongly influences the interac-
tions between surfactant tails and solvent molecules that are
directly responsible for steric stabilization,5 making SCF
solvation characteristics highly tunable. Indeed, lattice-fluid
2
workers ) disperse in both liquid and supercritical carbon
2
3
dioxide. Although arrested precipitation has been studied
extensively in conventional organic solvents, we have utilized
this approach to synthesize sterically stabilized monolayer-
protected nanocrystals in CO2 for the first time.
-7
theory,8 Monte Carlo simulation, light scattering and
neutron scattering1 have confirmed that colloid steric stabiliza-
tion in supercritical fluids varies with solvent density. A tunable
SCF solvent could provide reversible stabilization and desta-
bilization of colloidal dispersions, which could improve many
aspects of nanocrystal processing, such as size-selective separa-
tions, synthesis and self-assembly.
,9
10
11
Here, we report a single-phase sc-CO2-based arrested pre-
cipitation synthesis of silver, platinum, and iridium nanocrystals,
2
2
0-120 Å in diameter. Hydrogen is used to reduce an
organometallic precursor (silver acetylacetonate (Ag(acac)) for
example) in the presence of perfluorooctanethiol, a CO2-philic
capping ligand. The thiolate binds to the nanocrystal surface
and carbon dioxide solvates the partially fluorinated hydrocarbon
chains. The monolayer of adsorbed ligands surrounding each
particle provides steric stabilization of the nanocrystals. The
particles can be collected and redispersed in acetone or
fluorinated solvents without significant aggregation or increase
in particle size. Recently, Watkins and co-workers showed that
reduction of organometallic precursors with hydrogen in carbon
Compared to other supercritical fluids, carbon dioxide (CO2)
offers a variety of attractive features. CO2 is nonflammable,
essentially nontoxic, and environmentally benign, and its low
critical temperature and pressure of 31 °C and 71 bar,
respectively, are easily accessible. There are, however, signifi-
cant challenges in using sc-CO2 for arrested precipitation of
nanocrystals. CO2 has a low polarizability per volume (and
refractive index) resulting in far weaker van der Waals forces
2
4-26
dioxide leads to highly uniform thin metal films.
Experimental Section
*
To whom correspondence should be addressed. Phone: (512) 471-
633. Fax: (512) 471-7060. E-mail: korgel@mail.che.utexas.edu; johnston@
che.utexas.edu.
Synthesis. Silver nanocrystals were synthesized in sc-CO2
by reducing a soluble precursor, silver acetylacetonate
5
1
0.1021/jp011815c CCC: $20.00 © 2001 American Chemical Society
Published on Web 08/31/2001

9434 J. Phys. Chem. B, Vol. 105, No. 39, 2001
Shah et al.
TABLE 1
precursor thiol/precursor
av nanocrystal
diameter
(Å)
concn
exp (mM)
ratio
(mol/mol)
temp
(°C)
std dev
%
(Å) std dev
A
B
C
D
E
F
G
H
I
1.8
2.2
2.9
3.4
3.4
3.4
3.6
3.9
4.9
14.9
10.3
8.2
6.0
5.7
6.0
5.8
6.4
11.0
60
90
70
80
90
100
70
70
39
26
50
57
55
59
56
69
107
18
11
23
33
27
31
22
34
40
46
45
45
57
49
52
40
49
38
70
point-to-point resolution operating with a 200 kV accelerating
voltage and GATAN digital photography system. TEM images
were obtained by depositing the nanocrystals on 200 mesh
carbon-coated copper grids. Particle size and size distributions
were determined using Scion Image for Windows software. At
least 400 nanocrystals were measured for each sample. Energy-
dispersive X-ray spectroscopy (EDS) was performed in situ on
a JEOL 2010 TEM equipped with an Oxford Link ISIS EDS
instrument.
Figure 1. Schematic of reaction apparatus.
Ag(acac)), Aldrich Chemical Co.), with hydrogen at slightly
elevated temperatures ranging from 60 to 100 °C. Iridium and
platinum nanocrystals were synthesized using (methylcyclo-
pentadienyl)(1,5-cyclooctadiene)iridium(I) and dimethyl(1,5-
cyclopentadiene)platinum(II), respectively (Strem Chemicals,
Inc.). A thiolated fluorocarbon molecule, 1H,1H,2H,2H-
perfluorooctanethiol (Oakwood Products Inc.) (C6F13C2H4SH),
was used as the stabilizing ligand. Carbon dioxide (purity >
(
Results
9
9
9.99%) (Matheson Gas Products) and hydrogen (purity >
9.999%) (Praxair, Inc.) were used as received. All reactions
Silver nanocrystals were synthesized in sc-CO2 by reducing
Ag(acac) at elevated temperatures in the presence of hydrogen
and perfluorooctanethiol (see Table 1). The reaction proceeds
with hydrogen aiding Ag(acac) reduction to silver, followed by
the formation of silver-silver bonds in nucleated clusters of
silver atoms. The reaction temperatures listed in Table 1 (60-
were carried out in a temperature-controlled stainless steel, high-
pressure variable volume view cell equipped with a piston and
a sapphire window in the front and an inside diameter of
1
.75 cm.
In a typical experiment, 5.9-16.0 mg of Ag(acac), was loaded
1
00 °C) are significantly lower than those typically used (∼140
into the reaction cell with a working volume (volume with piston
in cell) of 27 mL. The cell was sealed and filled with 14-18
mL of CO2 at 138 bar and 20 °C. The cell was then pressurized
to 276 bar with carbon dioxide on the backside of the piston,
using a Dionex model 501 Computer Controlled Syringe Pump,
and heated to the reaction temperature with heating tape attached
to an Omega CN76000 temperature controller. The precursor
does not decompose under these conditions. Fluorinated thiol
°C) for chemical vapor deposition of metal films using similar
precursors. The reactivity at low temperatures is a manifesta-
tion of the very high solubility of hydrogen and organometallic
27
2
8,29
precursors in CO2.
Perfluorooctanethiol provides steric
stabilization and particle size control by binding to the surface
of the growing silver particles. Nanocrystals ranged from 26 to
107 Å in average diameter, depending on the precursor
concentration (see discussion below). Figure 2a shows a rep-
resentative TEM image of the silver nanocrystals synthesized
(162-324 mg) and hydrogen (6.22 mg) were simultaneously
introduced into the reaction vessel with a high-pressure injection
pump through two six-port rotary valves positioned in series
in sc-CO . Although the particle size distribution was relatively
broad with standard deviations about the mean diameter of
2
(Valco) with attached sample loops. A schematic of the
(47%, extended regions of hexagonal close packing could still
be found in the sample without size fractionation, as seen in
Figure 2b. As revealed in the high-resolution TEM image in
Figure 2c, the silver nanocrystal cores are crystalline, with the
{111} lattice planes visible.
apparatus is shown in Figure 1. The reaction temperature was
verified with a thermocouple inserted into the reaction cell, and
the pressure was held constant throughout the reaction. The
reaction proceeded for 3 h to ensure completion, although visual
observation indicated that the inside of the cell reached its
darkest after approximately 1 h. After the reaction, the cell was
cooled to room temperature and depressurized. CO2 was then
vented as a vapor from the top of the cell, leaving most of the
other components other than H2 in the cell. The nanocrystals
were collected from the cell using Omnisolv spectrophotometry
grade acetone (EM Science). After collection, the nanocrystals
were precipitated using an antisolvent, heptane, to separate
unreacted precursor, unbound capping ligands, and undesired
reaction byproducts that remained suspended in the supernatant,
from the nanocrystals. Nanocrystals were then redispersed in
acetone for further analysis. The nanocrystals were also dispers-
ible in fluorinated solvents, such as Fluorinert brand electronic
liquid and 1,1,2 trichlorotrifluoroethane (Freon).
Figure 3 shows the average particle size and polydispersity
plotted as a function of initial precursor concentration. Under
the conditions explored for the particle growth experiments, the
average particle diameter depended strongly on the precursor
concentration. As shown in Table 1, the average particle
diameter increases from 25 to 107 Å with only a 2-fold increase
in Ag(acac) concentration, thus allowing the nanocrystal size
to be methodically tuned with changes in the precursor
concentration. The polydispersity, however, remained relatively
independent of the precursor concentration and average nano-
crystal diameter with a standard deviation about the mean
diameter of (47%. Figure 4 shows the average particle size
and polydispersity as a function of temperature for experiments
D-G. These experiments were done at approximately the same
precursor concentration and thiol/precursor ratio. On the basis
of these results, it can be seen that neither the final particle
size nor the polydispersity is greatly affected by the reaction
Characterization. Transmission electron microscopy (TEM)
images of the nanocrystals were obtained using a JEOL 2010
high-resolution transmission electron microscope with a 1.7 Å

Nanocrystal Arrested Precipitation in sc-CO2
J. Phys. Chem. B, Vol. 105, No. 39, 2001 9435
Figure 3. Average nanocrystal diameter vs initial precursor concentra-
tion for silver nanocrystals. Error bars correspond to the sample
polydispersity, not the uncertainty of the size measurement tech-
nique. Nanocrystals synthesized at 60 °C (9), 70 °C (b), 80 °C (1),
9
0 °C ([), and 100 °C (2).
Figure 4. Average nanocrystal diameter vs reaction temperature for
experiments D-G. Error bars correspond to the sample polydispersity,
not the uncertainty of the size measurement technique.
fluorinated ligands on the nanocrystal surface.23 The organic
molecules maintain a separation between the nanocrystals of
approximately 20 Å in the TEM images (Figure 2a for example).
The EDS spectra of the silver nanocrystals (Figure 6) confirms
the presence of carbon, fluorine, sulfur, and silver in the sample.
The interparticle spacing of the perfluorocarbon-coated nano-
crystals is significantly larger than the interparticle edge-to-edge
separation found for silver nanocrystals coated with octanethiol
Figure 2. (a) TEM image of silver nanocrystals coated with C
6 13 2 4
F C H -
SH ligands synthesized in CO . (b) TEM image of hexagonally close
2
packed array of silver nanocrystals. (c) HRTEM image of silver
nanocrystal with visible lattice fringes.
temperature. Figure 5 shows the particle size histograms for
reactions A-I.
3
0
It is worth noting that the thiol/precursor ratios, which ranged
from 6:1 to 15:1, had little effect on the particle size. The thiol/
precursor ratios used here are much higher than what is typically
used in conventional solvents.2,30-32 These experiments were
performed in the presence of a large excess of thiol. The large
excess of thiol was used to ensure complete surface coverage
of the synthesized silver nanocrystals due to the significant
challenges in stabilizing particles in sc-CO2 (see discussion
section). A more intensive study would be required to determine
the effect of the thiol/precursor ratio on particle size, which is
beyond the scope of this work.
(∼10 Å). Self-assembled fluorocarbon monolayers (FSAMs)
have been found to be more rigid than hydrocarbon monolayers
3
3
(SAMs) due to the bulkiness of the fluorine groups. This
rigidity decreases the ability of the capping molecules to flex
and bend as two nanocrystal cores come together, which leads
to larger interparticle separations for fluorocarbon-coated nano-
crystals.
The absorbance spectrum of perfluorooctanethiol-capped
silver nanocrystals formed in sc-CO shows the characteristic
2
silver surface plasmon resonance (Figure 7) with a maximum
3
4
absorbance peak at 400 nm. This spectrum is similar to the
absorbance spectrum of perfluorodecanethiol-capped silver
nanocrystals synthesized by reducing silver nitrate in an acetone/
The nanocrystals could be repeatedly redispersed in acetone
and fluorinated solvents as a result of tight binding of the

9436 J. Phys. Chem. B, Vol. 105, No. 39, 2001
Shah et al.
Figure 5. (a)-(i) Silver nanocrystal size distributions for experiments A-I.
Figure 6. EDS spectra of silver nanocrystals coated with perfluoro-
octanethiol ligand.
water solution.23 The difference in peak width between the two
spectra results from a difference in particle size between the
Figure 7. UV-vis absorbance spectra of silver nanocrystals. (i)
2
Nanocrystals formed by Ag(acac) decomposition in sc-CO with
6 13 2 4 3
C F C H SH, dispersed in acetone. (ii) Nanocrystals formed by AgNO
3
4
two samples; the nanocrystals formed in acetone are slightly
smaller, d ) 55 Å, compared to the nanocrystals made in CO2,
d ) 59 Å. More distinctly, however, the fluorocarbon-coated
nanocrystals exhibit a much broader absorbance peak than
dodecanethiol-capped silver nanocrystals approximately equal
in size. Most likely, the extremely high electronegativity of the
fluorine groups surrounding the nanocrystals gives rise to a
shorter electron mean free path relative to the hydrocarbon-
reduction in water/acetone solution with C
acetone. (iii) Nanocrystals formed by AgNO
8
F
17
C
3
2
H
4
SH, dispersed in
reduction in water/
chloroform biphasic solution with C H SH, dispersed in hexane.
12
25
Discussion
Steric Stabilization in CO2. Sterically stabilized nanocrystals
must exhibit a repulsive force sufficient to overcome the long-
range attractive van der Waals force ΦvdW, between particle
2
3
coated particles.
3
5,36
cores to maintain stability:
Iridium and platinum nanocrystals were synthesized using
methylcyclopentadienyl)(1,5-cyclooctadiene)iridium(I) and di-
(
ΦvdW
)
methyl(1,5-cyclopentadiene)platinum(II), respectively. The reac-
tions in both cases were carried out at 80 °C and 276 bar. For
iridium, the precursor concentration was 3.3 mM and the thiol/
precursor ratio was 7.6:1. In the case of platinum, the precursor
concentration was 3.6 mM and the thiol/precursor ratio was 6.6:
2
2
2
A
2R
2R
d + 4Rd
-
+
+ ln
2
[
2
2
(
2
2
)
]
6 d + 4Rd d + 4Rd + 4R
d + 4Rd + 4R
(1)
1
. Parts a and b of Figure 8 show TEM images of the iridium
Equation 1 is valid for two particles of equal radius R, interacting
across a medium with an edge-to-edge separation d. The most
important parameter in this equation is the Hamaker constant
and platinum nanocrystals, respectively. Like the silver nano-
crystals, these particles exhibit crystalline cores.

Nanocrystal Arrested Precipitation in sc-CO2
J. Phys. Chem. B, Vol. 105, No. 39, 2001 9437
A, which represents the core-core interaction strength. The
Hamaker constant depends on the dielectric constant (ꢀ) and
refractive index (n) of the particles and the nature of the
solvating medium. The Hamaker constant can be estimated using
a simplification of Lifshitz theory.37
removing the solvent to enable redispersion. Hexanethiol-capped
silver nanocrystals, for example, are not easily dispersed in
conventional organic solvents; therefore, it is not surprising that
perfluorooctanethiol-capped particles with only six fluorocarbons
are not easily dispersed in CO , which induces even stronger
2
core-core interactions. In polar solvents, however, the steric
2
1
2 2
2
3
A121 ) kT
4
ꢀ1
- ꢀ2 2
repulsion provided by the chains appropriately reflects the
complete chain length. In acetone and ethanol for example, the
large dipole-dipole interactions between the solvent and the
CH2-CF2 group, which carries a substantial dipole moment,
provides the necessary solvation properties for redispersion.39
3hνe (n - n )
+
(2)
(ꢀ1 + ꢀ
)
2
2 3/2
2
16
x
2 (n + n )
1
2
The subscript 1 represents the particles, while 2 represents the
solvent medium; h is Planck’s constant and νe is the maximum
electronic ultraviolet adsorption frequency typically assumed
to be 3 × 10 s . Using eq 2, CO2 interactions across a vacuum
can be determined by setting ꢀ2 and n2 to 1. The dielectric
constant of silver nanocrystals is size dependent; however, a
reasonable estimate can be found by substituting a known value
for bulk silver interacting across a vacuum of 2.2 eV. Once
these two values are determined, the following relationship can
be used to determine the Hamaker constant for silver interaction
in CO2:37
Growth Mechanism. Hydrogen decomposes the organo-
metallic precursor to metal atoms in the presence of perfluoro-
octanethiol. Chen et al. describe the mechanism for monolayer-
protected nanocrystal formation in solution as a three-step
15 -1
40
process involving nucleation, growth, and passivation. For
example, silver nanocrystal nucleation occurs as the solution
reaches supersaturation and continues until the concentration
of silver atoms falls below the saturation limit. Nucleation
relieves the excess free energy in the supersaturated solution.
The nucleation and growth steps must be temporally separated
in order to achieve narrow particle size distributions, and in
batch systems, rapid nucleation followed by particle growth
3
7
2
A131 ≈ ( A - A )
(3)
x
11
x
33
3
,40-42
typically occurs.
Once nucleated, the particles may grow
A131 is for the interaction of two silver cores across CO2, A11 is
for silver in a vacuum, and A33 is for CO2 across a vacuum.
Silver interactions in CO2 are very strong due to the large
polarizability of the metal and the extremely low polarizability
of the solvent. The Hamaker constant for silver in sc-CO2 at
by a combination of essentially two different mechanisms: (1)
reactant diffusion-limited growth (silver atom condensation on
existing particles) or (2) core fusion (coagulation). The organic
ligands bind to the particle surfaces and halt particle growth at
long times to stabilize nanometer-size particles. At long times
relative to the nucleation event, the particle size distribution
reaches a self-preserving form that depends on the growth
mechanism, independent of the initial size distribution of the
2
5 °C is 1.4 eV, compared to 0.93 eV in acetone. Therefore,
silver nanocrystals experience nearly 50% stronger attraction
in CO2 than they do in acetone. The increased core-core
attraction makes nanocrystal stabilization more difficult in CO2
than in conventional solvents.
43
particle nuclei.
Figure 3 shows that the average nanocrystal size increases
with increasing precursor concentration. To determine if the
precursor concentration affects the particle growth mechanism,
the size distributions (shown in Figure 5) can be plotted in a
Ligand tails must be well solvated to provide steric stabiliza-
tion. The chains must protrude into the solvent to enable the
molecular fluctuations that give rise to steric repulsion between
particles. Fluorination of the hydrocarbon chain induces ligand
solvation in CO2. The repulsive force must overcome ΦvdW to
eliminate aggregation; thus, the chain length of the capping
ligand is very important for stabilization. In ref 23, C10
perfluorocarbon (C8F17C2H4SH) stabilizing ligands were used,
with eight fluorinated carbons and two hydrocarbons adjacent
to the thiol. These chains were long enough to enable dispersion
of nanocrystals in pure CO2. In the present study, C8 perfluoro-
carbon molecules (C6F13C2H4SH) were used to stabilize the
nanocrystals. These chains were sufficient to control particle
growth, and flocculation did not occur during the synthesis.
However, after collection and washing, the nanocrystals would
not redisperse in pure CO2, despite the fact that they could be
redispersed in acetone and fluorocarbon solvents and were well-
capped as observed by TEM. It would appear that the perfluoro-
octanethiol (eight carbons) is simply not long enough to provide
redispersibility in CO2.
reduced form to determine whether they exhibit self-preserving
43-45
behavior.
Friedlander et al. showed that a convenient
similarity transformation using a nondimensional distribution
2
function, ψ1(η) ) n(V)φ/N∞ (where n(V) is the concentration
of particles with volume V, and φ/N∞ is the average particle
volume), can be plotted to compare size distributions between
different systems. The total number of particles N∞, and total
volume of particles φ, are calculated using the relationships
∞
0
∞
0
N∞ ) ∫ n(V) dV and φ ) ∫ Vn(V) dV. Figure 9 shows ψ1(η)
plotted as a function of η ) VN∞/φ, the dimensionless particle
volume fraction. The particle diameter data determined experi-
mentally from TEM were converted to volumes by assuming
spherical particles. For all of the samples studied, ψ1(η) exhibits
self-similar behavior, which indicates that the precursor con-
centration does not affect the particle growth mechanism.
For certain limiting cases of particle growth mechanism, ψ1(η)
has been calculated and can be compared directly with
experimental results. Essentially, two growth mechanisms,
coagulation and condensation, compete to broaden or sharpen
the size distribution. Growth by condensation, also known as
diffusion-limited growth, narrows the particle size distribution
The limited stability with C6F13C2H4SH is somewhat unex-
pected given that octanethiol has been used successfully to
38
sterically stabilize silver nanocrystals in conventional solvents.
Consider, however, that it is the fluorinated segment of the
ligand that CO2 most effectively solvates, not the CH2 groups.
The fluorinated part of the perfluorooctanethiol ligand is 20%
shorter than in the perfluorodecanethiol ligand. In a sense, this
is like capping the nanocrystals with a C6 chain, which simply
does not provide sufficient spatial separation (or large enough
low cohesive energy density domains) between the particles after
4
2,46,47
at long times.
Coagulation, on the other hand, leads to
broad particle size distributions. Essentially, we are interested
in determining the extent to which coagulation and condensation
control particle growth. The high degree of polydispersity seen
in these samples (standard deviation of 47%), along with the

9438 J. Phys. Chem. B, Vol. 105, No. 39, 2001
Shah et al.
Figure 10. ψ
2
(η) versus η. Experiments A (O), B (0), C (]), D (4),
E (b), F (9), G ([), H (2), and I (1). ψ2theory (s) represents the
theoretical coagulation solution.
condensation. The parameter r1 is the arithmetic mean radius
defined as ∑ri/N∞, r3 is the cube-mean radius defined as
3
3
∑
r /N , and rh is the harmonic mean radius defined as
i ∞
x
N∞/∑(1/ri). For a monodisperse sample, r1 ) r3 ) rh, and µ1 )
µ3 ) 1. In our case, the average value for µ1 is 1.48 and µ3 is
0
.82. Values of µ1 > 1.25 and µ3 < 0.905 indicate that
coagulation controls growth.
An effective method of comparing size distributions is to
compare the nondimensional distribution function, ψ2(η) ) NV/
∞
N∞ ) ∫ ψ1(η) dV. NV represents the total number of particles
η
∞
larger than V: NV ) ∫ n(V) dV. In Figure 10, ψ2(η) are plotted
V
as a function of η. Again, the data exhibit a high degree of
self-similarity. The theoretically calculated values for ψ2(η) for
Brownian coagulative growth at long times, assuming a particle
sticking probability of 1, are also plotted in Figure 10. The model
calculations agree well with the data apart from a slight variation
in the middle of the size distribution. The data skew slightly to
a higher average volume. The deviation from the model curve
can be interpreted as an indication that the particle sticking
probability is less than 1. Given that the particles, particularly
at long times, are coated with the organic ligands, a sticking
probability less than 1 is expected. Certainly, the sticking
probability must decrease essentially to values of zero in order
to maintain stable particles.
Figure 8. TEM image of (a) iridium nanocrystals coated with
6 13 2
SH and (b) platinum nanocrystals coated with C F C -
H SH.
4
C
6
F
13
C
2
H
4
Figure 11 shows TEM images of polycrystalline silver
nanocrystals with multiple misaligned lattice fringes. The
internal crystalline domains possibly relate to the initial particle
nuclei prior to coagulation. It is important to note, however,
that the majority of the particles examined appear to be single
crystals, as seen in Figure 2c. Although one might expect all of
the particles to be polycrystalline in a system dominated by
coagulation, Yacam a´ n and co-workers recently observed atomic
reorganization and restructuring in gold particles as large as 70
4
8
Å in diameter after coagulation. In our system, it appears that
crystal planes can realign within the nanocrystal core; however,
the polycrystallinity resulting from the coagulative growth
process persists in some particles.
Figure 9. ψ
1
(η) versus η. Experiments A (O), B (0), C (]), D (4),
E (b), F (9), G ([), H (2), and I (1).
Conclusions
log-normal shape of the size distributions, indicate a coagulation-
controlled growth process. Quantitatively, the moments of the
size distribution function µ1 ) r3/rh and µ3 ) r1/r3 can be used
to determine the significance of coagulative growth relative to
Fluorocarbon-coated silver, iridium, and platinum nanocrys-
tals ranging in size from 20 to 120 Å in diameter were
synthesized in sc-CO2 by arrested precipitation. These fluoro-

Nanocrystal Arrested Precipitation in sc-CO2
J. Phys. Chem. B, Vol. 105, No. 39, 2001 9439
6 13 2 4 2
Figure 11. TEM images of polycrystalline silver nanocrystals coated with C F C H SH ligands synthesized in CO . (a)-(c) Multiple misaligned
lattice fringes visible in nanocrystal core. (d) Coagulated nanocrystal with partial realignment of crystal planes.
carbon-coated nanocrystals represent the first synthesis of
nanocrystals in pure CO2. The nanocrystals can be redispersed
in acetone and fluorinated solvents.
constant, independent of nanocrystal size. Nondimensional
analysis of the particle size distributions showed a self-similar
solution, indicating consistency in the reaction mechanisms
between the experiments. The log-normal shape of the size
distributions, along with the polydispersity of the samples and
the first and third moments of the size distributions indicate a
growth mechanism that is dominated by coagulation between
the nanocrystal cores. The perfluorooctanethiol capping ligand
used in this study is unable to provide redispersibility in pure
2
Because of the wide variety of available organometallic
precursors soluble in sc-CO2, and the large solubility of
hydrogen, many different types of metal and semiconductor
nanocrystals can feasibly be synthesized. The most effective
precursors dissolve easily in CO2, are stable at moderate
temperatures, and must decompose quickly to the base metal
with the addition of hydrogen. If hydrogen cannot reduce the
precursor at moderate conditions, then higher temperatures or
stronger reducing agents become necessary for particle forma-
tion. The nature of the binding group on the stabilizing ligand
must also be considered when applying these methods to other
materials. The stabilizing ligand must interact with the solvent
to provide protection against irreversible aggregation and exhibit
a binding strength weak enough to allow particle growth, yet
strong enough to arrest particle growth in the nanometer size
range.
sc-CO due to its length. The extremely low polarizability of
sc-CO results in approximately 50% larger van der Waals
2
attractions for silver as compared with the case in conventional
solvents. Partially because of this factor, C F C H SH ligands
8
17
2
4
stabilize silver nanocrystals in pure CO2 much more effectively
than C6F13C2H4SH. In conclusion, the reaction method described
here provides a pathway to the synthesis of metal nanocrystals
with unique dispersibility characteristics in an environmentally
benign solvent.
Under the reaction conditions explored here, the precursor
concentration predominantly determines the nanocrystal size
with higher Ag(acac) concentrations, giving rise to larger
particles. The average diameter ranged from as small as 26 Å
to as large as 107 Å. The polydispersity, however, remained
Acknowledgment. This work is supported in part by the
STC Program of the National Science Foundation under
Agreement No. CHE-9876674. We also acknowledge support
from the Petroleum Research Fund, the Welch Foundation, and

9440 J. Phys. Chem. B, Vol. 105, No. 39, 2001
Shah et al.
the Separations Research Program at UT Austin. B.K. also
thanks DuPont for partial support through a young professor
grant. We also thank John Mendenhall at the Institute for Cell
and Molecular Biology at UT Austin for TEM assistance.
(22) Silicon nanowires have been formed in supercritical hexane at 500
C by utilizing gold seed crystals stabilized by thiol ligands. Holmes, J.
°
D.; Johnston, K. P.; Doty, R. C.; Korgel, B. A. Science 2000, 287, 1471.
23) Shah, P. S.; Holmes, J. D.; Doty, R. C.; Johnston, K. P.; Korgel,
(
B. A. J. Am. Chem. Soc. 2000, 122, 4245-4246.
(
24) Watkins, J. J.; McCarthy, T. J. Chem. Mater. 1995, 7, 1991-1994.
(25) Watkins, J. J.; McCarthy, T. J. Method of Chemically Depositing
References and Notes
a Material onto a Substrate; University of Massachusetts: Amherst, 1998.
(26) Watkins, J. J.; Blackburn, J. M.; McCarthy, T. J. Chem. Mater.
1999, 11, 213-215.
(
1) Alivisatos, A. P. Science 1996, 271, 933-937.
(2) Brust, M.; Walker, M.; Bethell, D.; Schiffrin, D. J.; Whyman, R.
J. Chem. Soc., Chem. Commun. 1994, 801-802.
3) Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc.
993, 115, 8706-8715.
4) Eckert, C. A.; Knutson, B. L.; Debenedetti, P. G. Nature 1996,
83, 313-318.
5) Johnston, K. P. Curr. Opinion Colloid Interface Sci. 2000, 5, 351-
(27) Dryden, N. H.; Vittal, J. J.; Puddephatt, R. J. Chem. Mater. 1993,
5, 765-766.
(28) Poliakoff, M.; George, M. W.; Howdle, S. M. Inorganic and Related
Chemical Reactions in Supercritical Fluids; Eldik, R. v., Hubbard, C. D.,
Eds.; Wiley: New York, 1997.
(
1
3
3
(
(
56.
(29) Darr, J. A.; Poliakoff, M. Chem. ReV. 1999, 99, 495-541.
(30) Korgel, B. A.; Fullam, S.; Connolly, S.; Fitzmaurice, D. J. Phys.
Chem. B 1998, 102, 8379-8388.
(6) Johnston, K. P.; DaRocha, S. R. P.; Holmes, J. D.; Jacobson, G.
B.; Lee, C. T.; Yates, M. Z. Interfacial Phenomena with CO2-Soluble
Surfactants; DeSimone, J., Tumas, W., Eds.; Oxford University Press:
Oxford, 2001.
(
31) Leff, D. V.; Ohara, P. C.; Heath, J. R.; Gelbart, W. M. J. Phys.
Chem. 1995, 99, 7036-7041.
32) Previous studies using the biphasic approach have used thiol/
(
(7) Meredith, J. C.; Johnston, K. P. Theory and Simulation of Colloid
30
precursor ratios less than 1. Additionally, Brust found that well-capped
gold nanocrystals, 10-30 Å in diameter, had a final thiol/gold ratio of 1:3.
Work by Leff showed that thermodynamic control over particle size could
and Interface Science in Supercritical Fluids; Kiran, E., Debenedetti, P.,
Peters, C., Eds.; Kluwer Academic Publishers: Dordrecht, The Netherlands,
2
2
000; pp 211-228.
8) Meredith, J. C.; Johnston, K. P. Macromolecules 1998, 31, 5518-
528.
31
be obtained at thiol/precursor ratios between 1:1 and 6:1. This indicates
(
that even at our lowest thiol/precursor ratio, there is an obvious molar excess
of thiol.
5
(
9) Peck, D. G.; Johnston, K. P. Macromolecules 1993, 26, 1537.
(
33) Chidsey, C. E. D.; Loiacono, D. N. Langmuir 1990, 6, 682-691.
(
10) Meredith, J. C.; Sanchez, I. C.; Johnston, K. P.; de Pablo, J. J. J.
(34) Bohren, C. F.; Huffman, D. R. Absorption and Scattering of Light
Chem. Phys. 1998, 109, 6424-6434.
11) Yates, M. Z.; O’Neill, M. L.; Johnston, K. P.; Webber, S.; Canelas,
D. A.; Betts, D. E.; DeSimone, J. M. Macromolecules 1997, 30, 5060-
067.
12) Lee, C. T.; Johnston, K. P.; Dai, H. J.; Cochran, H. D.; Melnichenko,
Y. B.; Wignall, G. D. J. Phys. Chem. B 2001, 105, 3540-3548.
13) Hoefling, T. A.; Beitle, R. R.; Enick, R. M.; Beckman, E. J. Fluid
Phase Equilib. 1993, 83, 203.
14) McClain, J. B.; Londono, D.; Combes, J. R.; Romack, T. J.; Canelas,
by Small Particles; John Wiley & Sons: New York, 1983.
35) Hiemenz, P. C.; Rajagopalan, R. Principles of Colloid and Surface
Chemistry, 3rd ed.; Marcel Dekker: New York, 1997.
(
(
5
(
36) Hamaker, H. C. Physica 1937, 4, 1058-1072.
(
(37) Israelachvili, J. Intermolecular & Surface Forces, 2nd ed.; Academic
Press: San Diego, 1992.
38) Korgel, B. A.; Fitzmaurice, D. Phys. ReV. B 1999, 59, 14191-
4201.
39) Fluorine Chemistry: A ComprehensiVe Treatment; John Wiley &
Sons: New York, 1995.
40) Chen, S.; Templeton, A. C.; Murray, R. W. Langmuir 2000, 16,
543-3548.
41) Peng, X.; Wickham, J.; Alivisatos, A. P. J. Am. Chem. Soc. 1998,
20, 5343-5344.
42) Watzky, M. A.; Finke, R. G. J. Am. Chem. Soc. 1997, 119, 10382-
0400.
(
(
1
(
(
D. A.; Betts, D. E.; Wignal, G. D.; Samulski, E. T.; DeSimone, J. M. J.
Am. Chem. Soc. 1996, 118, 917-918.
(
(
15) O’Shea, K.; Kirmse, K.; Fox, M. A.; Johnston, K. P. J. Phys. Chem.
3
1
1
1
1
1
991, 95, 7863.
16) O’Neill, M. L.; Yates, M. Z.; Johnston, K. P.; Smith, C. D.;
Wilkinson, S. P. Macromolecules 1998, 31, 2838.
17) DeSimone, J. M.; Maury, E. E.; Menceloglu, Y. Z.; McClain, J.
B.; Romack, T. J.; Combes, J. R. Science 1994, 265, 356.
18) Calvo, L.; Holmes, J. D.; Yates, M. Z.; Johnston, K. P. J. Supercrit.
Fluids 2000, 16, 247-260.
19) Johnston, K. P.; Harrison, K. L.; Clarke, M. J.; Howdle, S. M.;
Heitz, M. P.; Bright, F. V.; Carlier, C.; Randolph, T. W. Science 1996,
(
(
(
(
(43) Friedlander, S. K.; Wang, C. S. J. Colloid Interface Sci. 1966, 22,
(
26-132.
(
26.
44) Pich, J.; Friedlander, S. K.; Lai, F. S. Aerosol Sci. 1970, 1, 115-
(
2
71, 624.
(45) Swift, D. L.; Friedlander, S. K. J. Colloid Science 1964, 19, 621-
647.
(46) Reiss, H. J. Chem. Phys. 1951, 19, 482.
(20) Ji, M.; Chen, X.; Wai, C. M.; Fulton, J. L. J. Am. Chem. Soc. 1999,
1
21, 2631-2632.
(
21) Holmes, J. D.; Bhargava, P. A.; Korgel, B. A.; Johnston, K. P.
(47) LaMer, V. K.; Dinegar, R. H. J. Am. Chem. Soc. 1950, 72, 4847.
Langmuir 1999, 6613-6615.
(48) Yacam a´ n, M. J. Personal Communication.