Precipitation of nanostructured copper oxalate: Substructure and growth mechanism

Article abstract of DOI:10.1021/jp0606816

The possibility of controlling materials properties by tailoring their substructure at the nanometer scale is a current topic of great interest. To do so, a fundamental understanding of the growth mechanism is of key importance and an analytical challenge as nanostructured materials are often produced by precipitation methods at high supersaturations where formation kinetics are fast. The current study focuses on the precipitation of copper oxalate, which has been previously shown to produce self-assembled ordered nanostructered particles with the promise of being able to tailor this nanometer substructure. In the current study we investigate in detail the growth mechanism and kinetics of precipitation by using in-situ particle size measurement or by stopping the reaction at various stages and using ex-situ methods. Combining the ex-situ methods of high-resolution scanning electron microscopy, transmission electron microscopy, and X-ray powder diffraction along with the in-situ methods, we were able to follow the growth process from 2 min to 2 weeks. The results in the 2-30 min period lead to the proposal of a core-shell growth model with a poorly ordered core and a well-structured shell of nanosized crystallites (50-70 nm), adding support to the brick-by-brick model previously proposed for this phase of particle growth. Particle evolution over long periods up to 2 weeks show a ripening which produces lens-shaped particles that eliminate the high surface energy faces observed in the earlier stages of growth. A more complete growth mechanism for copper oxalate precipitation at moderate supersaturations is proposed similar to recent findings for other self-assembled nanostructured particles.

Full text of DOI:10.1021/jp0606816

J. Phys. Chem. B 2006, 110, 17763-17771
17763
Precipitation of Nanostructured Copper Oxalate: Substructure and Growth Mechanism
L. Cristina Soare, Paul Bowen,* Jacques Lemaitre, and Heinrich Hofmann
Powder Technology Laboratory, Institute of Materials, Ecole Polytechnique F e´ d e´ rale de Lausanne (EPFL),
CH-1015 Lausanne, Switzerland
ReceiVed: February 1, 2006; In Final Form: July 13, 2006
The possibility of controlling materials properties by tailoring their substructure at the nanometer scale is a
current topic of great interest. To do so, a fundamental understanding of the growth mechanism is of key
importance and an analytical challenge as nanostructured materials are often produced by precipitation methods
at high supersaturations where formation kinetics are fast. The current study focuses on the precipitation of
copper oxalate, which has been previously shown to produce self-assembled ordered nanostructered particles
with the promise of being able to tailor this nanometer substructure. In the current study we investigate in
detail the growth mechanism and kinetics of precipitation by using in-situ particle size measurement or by
stopping the reaction at various stages and using ex-situ methods. Combining the ex-situ methods of high-
resolution scanning electron microscopy, transmission electron microscopy, and X-ray powder diffraction
along with the in-situ methods, we were able to follow the growth process from 2 min to 2 weeks. The
results in the 2-30 min period lead to the proposal of a core-shell growth model with a poorly ordered core
and a well-structured shell of nanosized crystallites (50-70 nm), adding support to the brick-by-brick model
previously proposed for this phase of particle growth. Particle evolution over long periods up to 2 weeks
show a ripening which produces lens-shaped particles that eliminate the “high” surface energy faces observed
in the earlier stages of growth. A more complete growth mechanism for copper oxalate precipitation at moderate
supersaturations is proposed similar to recent findings for other self-assembled nanostructured particles.
Introduction
ionic strength solution may re-align themselves into a lower
free energy state of the system before being fixed into place.14
The synthesis of tailored particles of specific size, structure,
and morphology is a key factor in the further development of
new materials in fields as diverse as catalysis, medicine, elec-
tronics, ceramics, pigments, and cosmetics. Although micrometer-
sized particles with various morphologies, such as spheres, rods,
and cubes, can be obtained by the use of controlled precipitation
conditions in the presence of additives, the synthesis of such
particles using the self-assembly of nanosized building blocks
is still a challenge. This bottom-up approach provides a way
Penn and Banfield have shown that some materials grow by an
oriented attachment which is directed by the crystallographic
structure to minimize the total free energy of the system,15
leading to either single crystals or well-organized nanostructures.
Such growth mechanisms via aggregation or self-assembly are
not often understood in detail, and to fully exploit nanostructured
materials and to tailor their structures for novel applications, a
better understanding of the growth mechanisms is needed.
1
2
3
For precipitation routes S plays an important role in the
nucleus formation and the growth mechanism. For batch
precipitation, used in most laboratory and exploratory work
including the present study, S varies with time as the reactants
create a solid phase. Different growth mechanisms may result
from this changing state of S, as illustrated in Figure 1 for
classical crystal growth, where different mechanisms dominate
depending on S and the crystal size, r.
Point Z in Figure 1 is located in a region corresponding to
an arbitrary initial S. The dotted line represents the “life history”
of the crystals during their growth, the crystals moving along
this path in the direction of the arrow. Very small crystals are
formed in the beginning of precipitation and start growing.
Although their size rapidly increases, S remains almost constant
because the mass of solute deposited on the crystals is very
small. A line almost parallel to the size axis therefore represents
the initial portion of the life history. With increasing crystal
size, the mass deposited on the crystals increases and S begins
to fall. In this region, the precipitation process is represented
by a curve. In the last stage of precipitation S decreases to its
final value but the crystal size changes little, as here a large
amount of mass is required to produce even a small increase in
to use the unique physical properties of the nanoscale primary
particles to tune the physical and chemical properties of
materials. By controlling the distance between the assembled
4
particles, additional effects such as magnetic coupling can be
achieved.4
In many cases presented in the literature, spherical particles
are produced by chemical precipitation routes and the poly-
crystalline particles consist of randomly oriented nanocrystal-
lites.5,6 Besides these spherical particles several examples of
ordered assemblies of the primary particles exist. Particles
7
8
showing such behavior are (NH4)3PW12O40, R-Fe2O3, barium
9
10
11,12
sulfate, barium chromate, and copper and cobalt oxalates.
The presence of additives seems to contribute greatly to the
self-assembly,13 and hydrophobic/hydrophilic interactions are
9
often cited as being the driving forces. In precipitation routes
supersaturation, S, during the early stages is usually high as
the electrical double layer is compressed and hence electrostatic
repulsive barriers are low. This is expected to lead to the
randomly oriented polycrystalline particles. Particles in a lower
*
To whom correspondence should be addressed. Telephone: 0041 21
6
93 4907. Fax:. 0041 21 693 3089, E-mail: paul.bowen@epfl.ch.
1
0.1021/jp0606816 CCC: $33.50 © 2006 American Chemical Society
Published on Web 08/18/2006

17764 J. Phys. Chem. B, Vol. 110, No. 36, 2006
Soare et al.
detail. We used high-resolution scanning electron microscopy
(HRSEM) to investigate precipitates after the reaction was
stopped for different aging times. Also in-situ particle size
distributions measured by laser diffraction (PSD) were used
follow the later stages of the particle growth. Complementary
techniques were used to investigate the internal structure in detail
such as cross-section transmission electron microscopy (TEM)
and X-ray diffraction. Also the solution chemistry has been
considered using a previously developed thermodynamic model.20
With the information from these measurements a modified
growth mechanism is proposed.
Experimental Section
Synthesis. All chemicals used were of analytical grade or
highest purity available. Copper oxalate precipitates investigated
throughout the present work was prepared as follows:
2
+
2-
+
2-
Cu + 2NO3 + 2Na + C O S CuC O V +
2
4
2
4
+
-
2
(Na + NO ) (1)
3
Figure 1. Precipitation map for BaSO
in the individual regions are as follow: (A) mononuclear or convection;
B) polynuclear or convection; (C) convection; (D) mononuclear or
screw dislocation; (E) polynuclear or screw dislocation; (F) polynuclear;
4
. Controlling growth mechanism
The reactants, copper nitrate trihydrate (Merck 1.02753, p.a.)
and sodium oxalate (Merck 1.06557, p.a.) were used without
purification. The precipitates were prepared using 0.01 M solu-
tions; both solutions were filtered through a 0.2 µm membrane
(RCN) before carrying out the precipitation experiments to avoid
possible dust. A more detailed description of particle synthesis
(
1
6
(G) diffusion; (H) mononuclear.
crystal size. Therefore in this final stage the life history curve
is almost parallel to the S axis. Thus during the course of their
growth, crystals can pass through a number of regions repre-
senting different controlling mechanisms. Thus growth strongly
depends on the structure of the material, the properties of the
solution, the nature of the interface between crystals, and the
surrounding solution and will be just as important for self-
assembled or aggregation dominated growth of nanosized pri-
1
1,18
can be found elsewhere.
The value of the initial supersatu-
2
0
1/x
ration is of the order of S ) log(P /K ) ) 0.70, where P is
r
s
s
s
the ionic product, K is the solubility constant at 25 °C, and x
s
is the number of the ionic species involved for the dissolution
process (here assumed to be 2). For the reaction described in
eq 1 cushionlike particles are formed. Cube shape particles were
also investigatedsprepared by adding 0.005 g/L HPMC 100 to
1
7
mary particles as for crystal growth by molecular attachment.
1
1
Metal oxalates are interesting candidates for nanostructured
particle synthesis. Besides the possibility to precipitate non-
spherical nanostructured particles, the metal oxalates can be
transformed to oxides or metals without losing the ordered
structure.18 Copper oxalate has been shown to be ordered self-
assembled particles with nanosized crystallites (20-70 nm), and
a brick-by-brick growth mechanism was proposed.11 The copper
oxalate particle morphology can be modified using hydroxy-
propyl methylcellulose (HPMC) from a cushionlike morphology
without polymer through cubes to elongated particles as the
polymer concentration is increased. Also as the polymer con-
centration increases, the size of the primary particles decreases,
and they become more elongated. The anisotropic shape is
linked with a specific growth in the 001 direction attributed to
a specific adsorption onto certain crystallographic faces. From
the anisotropic nature of the crystal structure,19 the crystallites
and particles are expected to show two distinct types of surface
either hydrophobic (ꢀ) or hydrophilic (R).
In the final product, the amount of the adsorbed polymer
corresponds to the geometrical external ꢀ surfaces. This indicates
that the particle assembly is governed by a dynamic mechanism
of polymer adsorption and desorption. The transmission electron
microscopy (TEM) results obtained for the nanostructures
material showed an alignment of the nanocrystals parallel to
the 110 direction, parallel to the stacking of the ...Cu(C2O4)-
Cu(C2O4)... ribbons perpendicular to each other19 within the
crystal structure. The above observations led the authors to
propose the brick-by-brick growth mechanism. However all the
measurements were made in ex-situ conditions on the final pre-
cipitate.11 In the current study the kinetics and growth mecha-
nism of copper oxalate precipitation have been studied in more
the copper nitrate solution.
Particle Characterization. Microscopy. The evolution of the
particle morphology from the initial stage of formation toward
the final stage of precipitation was followed by filtration through
20 nm ceramic membrane (Whatmann) for different reaction
times (2, 8, 12, 15, 30, and 60 min and 2 weeks). The samples
on the ceramic membrane were investigated using HRSEM XL
30-FEG Sirion (FEI, Eindhoven, The Netherlands) operating
at 3 kV, with a working distance of 5 mm and a spot size of 3.
The samples were analyzed without any further special treatment
such as gold coating, to avoid visual artifacts.
Copper oxalate particles with both cushionlike and cubic
morphology were analyzed using cross-section TEM. The cross-
section was obtained by ultramicrotomy. The dry powder of
copper oxalate with cubic or cushion morphology was embedded
in an epoxy resin matrix of Agar100, dodecenyl succinic
anhydride (DDSA), methyl nadic anhydride (MNA), and 2,4,6-
tris(dimethylaminomethyl)phenol (DMP 30). The mixture of the
resin and powder was put in a drier at 60 °C during 12 h for
the polymerization. Then the block was cut in a dry environment
at room temperature with a 35° Diatom diamond knife (radius
of 5 nm) mounted on Reichert-Jung Ultracut E ultramicrotom.
The thickness of the cross-section was of the range of 80 nm.
The particle structures were studied on a Philips CM 200 TEM
operating at 200 kV at room temperature.
For atomic force microscopy (AFM), we used a Park Instru-
ment Autoprobe CP (0.6 µm Ultralever SiN; contact mode, force
of 1-2 nN). The samples under investigation were placed on a
ceramic membrane and coated with a thin film of gold.
Particle Size Distribution. Measurements of the particle size
evolution of copper oxalate precipitates without HPMC have

Nanostructured Copper Oxalate
J. Phys. Chem. B, Vol. 110, No. 36, 2006 17765
Figure 2. (a) pH evolution as a function of time with the points A-D corresponding to filtration time and HRSEM analyses; (b) reaction yield
evolution with the time of precipitation from pH measurements and a thermodynamic model.
TABLE 1: Optical Parameters Used for Copper Oxalate for
PSD Evolution Using Laser Diffraction
refractive index refractive index refractive index
devices
(CuC
2
O
4
)
(CuC
2
O
4
)
2
H O
real part
1.77
imaginary part
0.01
Mastersizer S
1.33
been carried out in-situ by laser diffraction (Malvern Mastersizer
S). The total volume of the cell is about 100 mL, and the 300
mm lens used probes in a PSD range from 50 nm to 100 µm.
Before starting the measurements with the two reactant solutions
of copper nitrate and sodium oxalate, a background was
measured. The background measurement was carried out using
decarbonated water that had been filtered through a 20 nm
ceramic membrane to avoid possible dust and bubbles. A 50 mL
aliquot of each reactant solution (at a concentration of 0.01 M)
was injected within 20 s and then circulated at 1400 rpm. The
data were collected every minute up to 25 min and then at 30,
45, and 60 min. To transform these types of measurements from
diffraction patterns into PSDs, the optical data for the copper
oxalate were not found in the literature and were thus determined
by comparing several techniques, namely, imaging analysis,
photo centrifugation, and laser diffraction.18 The optical data
are reported in Table 1.
X-ray Diffraction. The diffraction patterns have been collected
using an X-Pert Philips with copper radiation, wavelength
λ(Cu KR1) ) 0.1506 nm, U ) 50 kV, and I ) 40 mA. Samples
were rotated in order to minimize textural effects, in a range of
2
0-90°. XRD was used in order to determine the phase under
investigation and then to estimate the coherent crystalline
domain or crystallite size, from line broadening using the
following Scherer eq 2
Kλ
B cos(θ)
l )
(2)
where λ is the wavelength of the X-ray source (nm), B is the
line width at half of the maximum intensity (radians), Θ is the
diffraction angle, l is the average diameter of the crystallites
Figure 3. HRSEM micrographs of copper oxalate particles as a
function of precipitation time at high and low magnification after (a)
(
nm), and K is a factor related to the crystallite shape, which is
2
min, (b) 8 min, (c) 12 min, (d) 30 min, and (e) 2 weeks.
22
assumed to be 0.89. Some powder diffractograms have been
fitted using TOPAS 2000 software that takes into consideration
the crystallographic parameters. JCPDS files used for the copper
oxalate identification is 21-0297, and the theoretical density of
Results
HRSEM Analysis. The particle evolution was followed as a
3
copper oxalate is 3500 kg/m .
function of reaction time by stopping the precipitation process
by filtration through 20 nm ceramic membranes. Figure 2a
shows the pH evolution with time for the copper oxalate pre-
cipitation, and it is indicated where the reaction was stopped
for HRSEM analysis. Figure 2b shows the precipitation yield
calculated from the pH evolution using a thermodynamic
Particles Density. The particle density measurement has been
carried out by helium pycnometry (Microline 380 Micrometrics).
Before the measurement the powder under investigation was
thermally treated at 85 °C, for 4 h under air to eliminate possible
adsorbed water on particles surfaces. For each sample 10
measurements were carried out.
2
0
model.

17766 J. Phys. Chem. B, Vol. 110, No. 36, 2006
Soare et al.
Figure 4. (a) Topography of a typical particle with an aging time of 2 min measured with atomic force microscopy; (b) 3D view of typical particle
under construction.
Samples collected after 2, 5, 8, 12, 30, 60, and 40 230 min
2 weeks) were analyzed using HRSEM and are shown in Figure
blocks have been cemented together with the remaining solute
as the reaction approaches the equilibrium state. After 2 weeks
the morphology has continued to evolve and the smaller surfaces
have more or less disappeared and show only the two large
surfaces, suggesting a ripening or slow growth process (Fig-
ure 2e). These last stages (60 min to 2 weeks) could be expected
from the residual copper oxalate supersaturation predicted by
the solubility simulation (S ) 0.40 after 30 min).
Although we have presented here more evidence for the self-
assembly of nanostructure particles via the brick-by-brick
mechanism, to further confirm the proposed mechanism, a direct
link between the kinetics and observed particle size distribution
and substructure using a population balance approach is still
needed. This is currently under investigation using a recently
developed approach (by Testino et al.), which was successful
in describing the growth of nanostructured barium titanate.17
AFM Analysis. Particles under construction with an aging
time of 2 min have been analyzed in detail by AFM in order to
determine the height and width of the growth steps and compare
them to the XRD crystallite size measurement. Figure 3 shows
a typical particle measured by AFM with an aging time of 2 min.
The steps of particles under construction are clearly visible. The
height and the width of the steps measured for several particles
are summarized in Table 2. The measurements were made with
an experimental error of 10 nm, which corresponds to the radius
of the instrument pointer.
(
3
at low and high magnification.
The pH evolution of copper oxalate precipitation was
unexpected from the thermodynamic calculations if the only
2
0
solid-phase assumed to precipitate was copper oxalate. The
precipitation analysis was consequently split into two parts,
before and after 2 min, the point at which the pH stops increas-
ing and starts to decrease. From the thermodynamic simulation
the increasing part corresponds to copper oxalate precipitation.
The second part of the pH evolution, a slow decrease, could
only be explained by the precipitation of a minor (0.4 wt %)
secondary-phase malachite, the copper hydroxy carbonate. The
carbonate comes from CO2 in ambient air.
After 2 min the HRSEM micrographs show particles around
500 nm under “construction” with steps and kinks the same
size as the crystallite size measured by XRD. These particles
under construction are surrounded by nanoparticles with a size
under 100 nm, supporting the brick-by-brick growth mechanism
previously proposed for copper oxalate and also seen for other
2
3
systems. The micrographs with aging times of 12 min show
a broad range of particle sizes; some around 1 µm, some around
200 nm, and some nearer 50 nm. After 12 min the small particles
were much reduced in frequency and rarely observed, again
coherent with the hypothesis that growth by aggregation
dominates in this period. The particles then grow to sizes of
1
-3 µm between 15 and 60 min. These cushionlike particles
For copper oxalate precipitated under similar conditions, with
30 min aging time, the crystallite size was found to be 70 nm
in the 110 direction and 50 nm in the 001 direction.11 The height
of the step measured by AFM has the same order of magnitude
as, or is close to an integer multiple of, the XRD crystallite
size (70 nm). These steps and their correspondence with the
XRD line broadening crystallite size add significant weight to
the brick-by-brick growth mechanism.
show two large surfaces (R) and four small surfaces (ꢀ) (Fig-
ure 3c), which were shown previously to be linked to the
anisotropic nature of the crystallographic structure of copper
oxalate.11 After 12 min the larger surfaces have a rough appear-
ance, while the four lateral ꢀ surfaces appear smoother. This is
even more accentuated after 30 min with a very smooth lateral
surface (Figure 2d), suggesting the nanocrystallite building
TABLE 2: Dimension of the Particles and of Their Corresponding Step Size Determined by Atomic Force Microscopya
particle
step
width (nm)
height (nm)
width (nm)
height (nm)
particle no.
length (nm)
w ) width d001
h ) height d110
length (nm)
w ) width d001
h ) height d110
1
2
3
4
5
1000
1300
1100
1530
1000
700
780
927
1225
875
280
450
505
275
280
420
800
710
760
500
360
380
650
900
448
73 (1)b
214 (3)
287 (4)
210 (3)
145 (2)
a
Measurements were made with a precision of (10 nm. b Integer value no. of 70 nm units making up the step height.

Nanostructured Copper Oxalate
J. Phys. Chem. B, Vol. 110, No. 36, 2006 17767
Figure 5. Copper oxalate TEM cross-section-cushion morphology: (a) diffraction patterns for micrograph b; (c) diffraction patterns for micrograph d.
Figure 6. Copper oxalate TEM cross-sections-cubic morphology: (a) diffraction patterns from the particle core; (b) image of particle core;
c) diffraction pattern of outer region; (d) image of outer region showing larger crystallites at the surface of the cubic copper oxalate particles.
(
TEM Cross-Section. The copper oxalate particles were
TEM was carried out on cubic copper oxalate formed with the
1
1,23
investigated by TEM cross-section to observe their internal
structure (Figure 5). It was very difficult to decide which region
of the cushion-shaped particle was being sampled because of
the damage or random orientation of the cross-section during
sample preparation. Previous results on cubic copper oxalate
showed an order in the 110 direction.1 Therefore cross-section
addition of HMPC,
and the results are shown in Figure 6.
The cubic shape of the particles made it much easier to identify
which region of the particle was being imaged in the TEM.
The central region of the particles shows rings in the electron
diffraction pattern which were indexed as copper oxalate,
indicating a random orientation between the crystallites. The
1

17768 J. Phys. Chem. B, Vol. 110, No. 36, 2006
Soare et al.
Figure 8. Particle size evolution of copper oxalate versus time for
d
v10 (µm) for two different runs.
Figure 9. SEM images for copper oxalate precipitation with an aging
time of 23 min: (a) particles with a size less than 200 nm; (b) a large
distribution of “primary” particles, particles under construction, and
well formed cushion-shaped particles.
Before 12-15 min it was not possible to detect the formation
of particles, the scattered light intensity being too low. This is
perhaps due to two contributing factors, a low particle volume
concentration (solid/liquid ) 0.004) and the low scattering
coefficient for particles less than 100 nm which are expected
Figure 7. Particles size distribution of copper oxalate as a function of
time for two repeated experiments.
TABLE 3: Particle Size Distribution Data in Aqueous
Solution Determined by Laser Diffraction as a Function of
Time (Span ) (Dv90 - Dv10)/Dv50
25
to be most numerous in the initial stages of precipitation. After
time (min)
D
v10 (µm)
D
v50 (µm)
D
v90 (µm)
span
15 min, particles larger than 500 nm are detected and a dv50 at
around 1 µm is measured. The particles then increase in size to
a dv50 ) 1.23 µm (Figure 7, 17.5 min). After 20 min the PSD
shows a significant decrease with small particles under 200 nm
being detected (Table 3). This transition takes place during a
short period of about 5 min, when the smaller particles are
detected and then the growth process continues and the dv50
increases again. A plot of dv10 as a function of time shows this
decrease in size very clearly for two different precipitation
experiments (Figure 8). The phenomenon is reproducible, but
the exact time at which the smaller particles appear is not the
same in all cases. This is probably due to difficulties in repro-
ducing mixing in the laser diffraction cell of 100 mL. This
sudden appearance of a second population of fine particles
smaller than 500 nm is indicative of a second nucleation event.
The smallest size that can be measured in these conditions, using
laser diffraction is 50 nm.
To confirm this production of small particles, 250 µL of
suspension was filtered through 20 nm ceramic membranes and
then analyzed using HRSEM. Figure 9 shows particles with an
aging time of 23 min. Three types of particles can be seen, one
with a size less than 200 nm (which look like random aggregates
of primary particles), a second population under construction
(100-500 nm), and a third population of well-formed particles
(1500 nm).
1
1
2
2
3
4
6
5
7.5
0
2
0
5
0
0.59
0.70
0.22
0.26
0.78
0.98
1.05
1.10
1.23
1.14
0.95
1.23
1.76
2.00
2.45
2.74
2.73
2.53
2.05
3.05
3.72
1.68
1.51
2.19
2.38
1.02
1.17
1.32
electron diffraction patterns from the outer regions show sharp
dots with minor broadening indicative of a strong alignment
between the primary particles or crystallites. From indexation
of the diffraction pattern this alignment is in the 110 direction
parallel to the CuC2O4 ribbon axis of the crystal structure. The
earlier brick-by-brick mechanism assumed very good alignment
throughout the interior of the particle; here we see more of a
core-shell structure as seen recently for cobalt oxalate. The
poor order observed from the TEM cross-section suggested a
certain degree of porosity may be present in this zone. This is
confirmed by the density measurements, which showed an
2
3
3
internal porosity of 14%, i.e., a density of 3 g/cm instead of
3
3
.5 g/cm as calculated from the crystallographic structure.
The outermost layer in Figure 6d shows larger crystallites than
the zone just beneath the surface where the diffraction was
made. This may be due the ripening discussed in the HRSEM
section.
Particle Size Distribution. EVolution. All the results pre-
sented above were made on ex-situ samples at the end of
precipitation or for the intermediate times (HRSEM) in static
conditions. These techniques do not give a real time evolution
of the precipitation reaction. Therefore in-situ PSD measure-
ments were made using laser diffraction. The results are shown
in Figure 7 and summarized in Table 3.
Attempts to follow the very early stages of precipitation
nucleation and growth were made using the small-angle X-ray
25
scattering (SAXS) technique. Unfortunately the supersaturation
had to be increased from 0.70 to 1.22 (the concentration was
increased 5 times from 0.005 to 0.025 M after mixing) to
produce a solid content accessible by the SAXS method. For
these conditions of higher supersaturation, the first particles were

Nanostructured Copper Oxalate
J. Phys. Chem. B, Vol. 110, No. 36, 2006 17769
Figure 11. Schematic representation of the variation of system free
energy as the supersaturation and hence ionic strength decrease during
the batch precipitation reaction (high ionic strength at early times
<
2 min, low ionic strength > 2 min).
the particle morphology becomes almost spherical with ran-
domly oriented crystallites throughout the particle as seen in
1
1
earlier studies on copper oxalate. The crystallite sizes also
decrease for precipitates aged for 30 min from 50 nm for the
cushion morphology (0.005M) to 27 nm for the spherical
morphology (0.025M) for the 110 reflection.
Discussion
The mechanism of copper oxalate particles assembly proposed
11
previously starts with a simple fast burst of nucleation, growth
of crystallites by molecular attachment, and then growth by self-
ordered aggregation. The above results suggest that more than
one process may take place at the same time, and the formation
of copper oxalate involves a more complicated mechanism. Each
of the processes of nucleation, growth by molecular attachment,
or growth by aggregation may be influenced by the distribution
and concentration evolution of ions and complexes in the
precipitating system. This will be discussed below, incorporating
information from thermodynamic solubility simulation.20
Proposed Growth Mechanism. Figure 10 shows a schematic
representation of a modified growth mechanism for copper
oxalate precipitation at a low level of supersaturation Sr < 1.00.
Just after the two reactants come in contact, the supersaturation
produces a burst of nucleation in a time of milliseconds, as
2
5
suggested from the SAXS results (point A, Figure 10). The
total system energy decreases quickly represented by the chem-
ical potential change; i.e., supersaturation decreases the nuclei
grow (point B, Figure 10), and then particles begin to agglomer-
ate. The large number density of particles initially induces a
high collision frequency and with a relatively high ionic con-
centration of around 0.005 M, giving a thin electric double layer
and low colloidal stability, leading to a random agglomeration
of the particles (point C, Figure 10). When the particles aggre-
gate in a random fashion by simple contact, little change in the
total system energy would be expected, as they would have an
open porosity, i.e., no reduction of surface area per unit volume.
This open structure for the random organization of the particles
is supported by the presence of an internal porosity of about
Figure 10. Schematic representation of the precipitation mechanism
of copper oxalate formation from the initial stage to final particle as a
function of time and supersaturation: (a) early stages of nucleation
and primary particle growth; (b) evolution between 1 and 5 min showing
the random core and the beginning of the ordered self-asssembly; (c)
evolution from 5 min to 2 weeks with a cross-section of the final
particlesswith a possible increase of supersaturation leading to a second
nucleation event during the ordered assembly period (E).
1
4% from pycnometry analysis and also from the TEM cross-
detected in the time range of milliseconds, they then grew
quickly, and after 25 s the intensity of scattering remains con-
stant. This means that the particles reached a size greater than
section study.
As particle number density and collision frequency decrease,
the particles then start to aggregate in a more ordered behavior,
brick-by-brick eliminating the solution-solid interface to
minimize the system free energy. This more ordered attachment
may be helped by the decrease in the overall ionic strength
(decreasing supersaturation), which would increase the electric
double layer thickness from 3 nm (0.005 M) to 5.7 nm after
100 nm after the first 25 s. These experiments are in general
agreement with the thermodynamic simulation of the pH evolu-
tion (Figure 2), where the majority of the copper oxalate (73%)
precipitates in less than 2 min20 but did not allow us to follow
the PSD evolution at early times. At this higher supersaturation

17770 J. Phys. Chem. B, Vol. 110, No. 36, 2006
Soare et al.
Figure 12. SEM micrographs of partially dissolved copper oxalate cubic particles showing the core-shell structure proposed in the modified
growth model.
2
0
3
0 min (0.001 45 M, from the thermodynamic model ). This
shows very poor order. The dissolution process does not
necessarily mimic the crystallization process, but the regions
are very coherent with the proposed model.
The proposed model can also qualitatively accommodate the
change in the particle morphology from cushions to spheres
could produce a local potential minimum, E1, Figure 11, which
attaches the primary particle to the larger particles but allows
rotation to find the global energy minimum E3, i.e., crystallo-
graphically well-aligned. In the earlier random agglomeration
the local attractive potential is too deep, E2, to allow rotation
and alignment. When primary particles attach in an ordered
fashion, the ions that were in the electric double layer will be
released into solution. Interparticle potential calculations have
not been made because no Hamaker constant for copper oxalate
could be found.
When a significant quantity of “ordered” attachment has taken
place, the “released” ions may be numerous enough to increase
the supersaturation and induce a second nucleation eventsas
modeled and proposed for nanoparticle aggregation by Nomura
et al. This behavior corresponds to point E in Figure 10. In
the in-situ PSD measurements an apparent second nucleation
event between 12 and 20 min was also observed. This is coher-
ent with the above discussion of ordered aggregation between
for higher levels of supersaturation (S > 1.00) used in the SAXS
r
experiments. The number of primary particles per unit volume
increases significantly if we assume that each crystallite comes
from a single nucleation event. For the cushion morphology
1
6
we would have an initial number density around 6 × 10
3
(particles/dm ) and for the spherical morphology around 8 ×
2
0
3
10 (particles/dm .) The mean separation between particles
decrease from around 1000 nm to around 90 nm, and thus the
collision frequency increases. Also with the higher concentra-
tions (0.025 M) the electrical double layer would also be highly
2
6
compressed (around 1.3 nm). So at the higher S the period of
r
the random aggregation, seen at the core of the low-Sr
precipitates, would be expected to last for much longer times
and the local potential minimum E is too deep to allow particle
2
2
and 20 min and the electron microscopy observations. Finally
reorientation before more particles are attached. Thus particles
to move toward the equilibrium state of the system, the particles
continue to aggregate and the free particle surfaces is further
eliminated.
aggregate as spheres to minimize the volume consistent with
the proposed growth model.
After 30 min of precipitation the particles reach a size of
around 2 µm (Figure 2d). The TEM cross-section for the parti-
cles with an aging time of 60 min showed larger crystallites at
the surface than in the core (Figure 6d are the cubes after 60 min
aging). This outer layer, formed from 2 to 3 layers of crystallites,
is assumed to be due to a ripening process or continued crystal
growth by molecular attachment from the residual super-
saturation (Figure 2). During the development of the particles,
once they have taken on the cushion shape, the four small lateral
Conclusions
In the present paper we have investigated, using both in-situ
and ex-situ methods, copper oxalate precipitation from the early
stages of precipitation to the final nanostructured particles. The
growth mechanism has been shown to be more complex than
the simple nucleation, growth, and brick-by-brick self-assembly
mechanism previously proposed. The mechanism can be
described in four stages: (i) initial nucleation and growth of
primary particles; (ii) rapid aggregation to give a core of ran-
domly oriented primary particles or crystallites; (iii) a period
of ordered attachment of crystallites giving an inner shell of
well-aligned nanocrystals; (iv) a slow approach toward equi-
librium with the solution where high-energy surfaces are
eliminated and an outer shell of larger well-cemented crystallites
envelopes the particles. A qualitative mechanism describing the
four stages has been described using knowledge of the super-
saturation and solution evolution from a previously developed
thermodynamic model, adding weight to the direct microscopic
observations.
ꢀ
surfaces decrease in relative importance compared to the two
large basal surfaces. Particles with an aging time of 2 weeks,
point F (Figure 10), are larger in size (5 µm), and the mor-
phology has developed from cushions to lenses (Figure 2 d,e).
The four small ꢀ surfaces of the cushionlike particles have
vanished, and the particles are made up from the two large
curved surfaces. The elimination of the smaller edge faces is
both coherent with a slow desaturation or ripening and their
“
high energy”-hydrophobic nature. Further supporting evidence
for the core-shell structure is illustrated in Figure 12, where
SEM micrographs of partially dissolved particles (under con-
trolled conditions ) are presented. The substructure shows the
dense outer layer (more difficult to dissolve) and an ordered
layer which has been significantly dissolved. The inner core
1
8
The two different regimes of aggregation, a random organiza-
tion followed by well-aligned crystallites, were supported by
both HRSEM and TEM cross-section imaging combined with

Nanostructured Copper Oxalate
J. Phys. Chem. B, Vol. 110, No. 36, 2006 17771
electron diffraction. The specific organization evoked a second
nucleation event, supported by in-situ PSD measurement and
ex-situ microscopy, both showing bimodal particle distributions.
For intermediate times (2-30 min) the measurements generally
confirm the previously proposed brick-by-brick agglomeration
mechanism producing a cushionlike morphology. After 2 weeks
a disk- or lenslike morphology with only two curved surfaces
indicates an equilibrium form has been achieved.
The experimental approach of using a small well-mixed batch
reactor, following PSD evolution in-situ and using high-
resolution SEM and TEM cross-section, ex-situ for specific time
intervals has been successful in further elucidating the growth
mechanism of copper oxalate. With knowledge of this growth
mechanism better control over such nanostructured particles can
be envisaged, although further work on the early stages and
population balance modeling to further confirm the kinetic
aspects should prove fruitful. Such detailed approaches should
allow us to approach the ultimate goal of being able to design
and tailor structures on the nanometer scalesa key issue in
current materials science.
(5) Privmann, V.; Goia, D. V.; Park, J.; Matijevic, E. J. Colloid
Interface Sci. 1999, 213, 36.
(6) Vacassy, R.; Scholz, St.; Dutta, J.; Plummer, Ch. J.; Houriet, R.;
Hofmann, H. J. Am. Ceram. Soc. 1998, 81, 2699-2705.
(7) Ito, T.; Inumaru, K.; Risono, M. J. Phys. Chem. B 1997, 101,
9
958-9963.
8) Park, G. S.; Scindo, D.; Waseda, Y.; Sugimoto, T. J. Colloid
Interface Sci. 1996, 177, 198-207.
9) Qi, L.; C o¨ lfen, H.; Antonietti, M. Chem Mater. 2000, 12, 2392-
2403.
(
(
(
10) Li, M.; Schnablegger, H.; Mann, S. Nature 1999, 402, 393-395.
(11) Jongen, N.; Bowen, P.; Lemaitre, J.; Valmalette, J. C.; Hofmann,
H. J. Colloid Interface Sci. 2000, 226, 189-198.
12) Pujol, O.; Bowen, P.; Stadelmann, P. A.; Hofmann, H. J. Phys.
Chem. B 2004, 108, 13128-13136.
13) Zemb, T.; Dubois, M.; Debe, B.; Gulik Krzywicki, T. Science
999, 283.
14) Collier, A. P.; Hetherington, C. J. D.; Hounslow, M. J. Cryst.
(
(
1
(
Growth 2000, 208, 513-519.
(15) Penn, R. L.; Banfield, J. F. Geochim. Cosmochim. Acta 1999, 63,
1
549-1557. Penn, R. L.; Oskam, G.; Strathmann, T. J.; Searson, P. C.; et
al. J. Phys. Chem. B 2001, 105, 2177-2182. Penn, R. L.; Banfield, J. F.
Science 2000, 281, 969-971. Banfield, J. F.; et al. Science 1998, 289, 751-
754.
(16) S o¨ hnel, O.; Garside, J. Precipitation Basic Principles and Industrial
Applications; Butterworth-Heinemann: Woburn, MA, 1992.
(
17) Testino, A.; Buscaglia, M. T.; Buscaglia, V. Chem. Mater. 2004,
6, 1563-1543.
18) Soare, L. C. Lausanne Thesis no. 3083, EPFL, 2004 (pdf available
from http://library.epfl.ch/theses/).
Acknowledgment. The authors thank the European Project
COST 523 Nanostructures Materials for the financial support,
Prof. Philippe Buffat, M. Brain Senior, and Mme. Daniele Laub
for the TEM work, Dr. Nathalie Jongen and Dr. Ollivier Pujol
for fruitful discussions, Dr. Heinz Amenitsch from IBR Graz
Austria for the SAXS measurements, and Dr. Christopher
Plummer for the AFM measurements.
1
(
(
(
19) Schmittler, H. Cryst. Res. Technol. 1984, 19, 1225-1230.
20) Soare, L. C.; Bowen, P.; Lemaitre, J.; Hofmann, H. J. Cryst.
Growth 2005, 289, 278-285.
21) Rubattel, S.; Lemaitre, J.; Bowen, P.; Ring, T. A. J. Cryst. Growth
1994, 135, 135-144.
22) Klug, H. P.; Alexander, A. L. X-ray Diffraction Procedures for
Polycrystalline and Amorphous Materials; Wiley: New York, 1954.
23) C o¨ lfen, H.; Mann, S. Angew. Chem., Int. Ed. 2003, 42 (21), 2350-
2365.
(24) Bowen, P.; Dirksen, J. A.; Baker, R. H.; Jelinek, L. Powder
(
(
References and Notes
(
(
(
(
1) Matijevic, E. Curr. Opin. Colloid Interface Sci. 1996, 1, 176.
2) Matijevic, E. Chem. Mater. 1993, 5, 412.
3) C o¨ lfen, H.; Antonietti, M. Angew. Chem., Int. Ed. 2005, 44, 5576-
Technol. 1993, 74, 67-71.
5
591.
(25) Soare, L. C.; Jongen, N.; Bowen, P.; Buscalia, V.; Amenitsch, H.
Annual Report, IBR/ELETTRA, 2002, pp 65-67.
(26) Nomura, T.; Kousaka, Y.; Alonso, M.; Fukunaga, M.; Satoh, T.
J. Colloid Interface Sci. 2000, 221, 195-199.
(4) Baibich, M. N.; Broto, J. M.; Fert, A.; Nguyen Van Dau, F.; Perroff,
F. Phys. ReV. Lett. 1988, 61, 2472-2475. Wecker, J.; von Helmolt, R. Appl.
Phys. Lett. 1993, 62, 1985-1987.