An investigation of the thermal decomposition of gold acetate

Article abstract of DOI:10.1007/s10973-008-9173-1

The thermal decomposition characteristics of gold acetate to produce gold nanoparticles were investigated. A rapid and violent fragmentation of the gold acetate particles was observed at approximately 103±20°C when a rapid heating rate of 25°C min^-1<

Full text of DOI:10.1007/s10973-008-9173-1

Journal of Thermal Analysis and Calorimetry, Vol. 95 (2009) 1, 117–122
AN INVESTIGATION OF THE THERMAL DECOMPOSITION OF
GOLD ACETATE
S. D. Bakrania*, G. K. Rathore and Margaret S. Wooldridge
Department of Mechanical Engineering, University of Michigan, 2350 Hayward Street, Ann Arbor, MI 48109, USA
The thermal decomposition characteristics of gold acetate to produce gold nanoparticles were investigated. A rapid and violent
fragmentation of the gold acetate particles was observed at approximately 103±20°C when a rapid heating rate of 25°C min–1 was
used, leading to formation of nanosized spherical and partially coalesced gold particles. Particle size analysis was used to
investigate possible relationships between the gold acetate crystallite size and the gold nanoparticles produced by thermal
decomposition. The results indicate rapid (<0.14 ms) coalescence of the gold particles occurs for fragments in close proximity.
Keywords: gold acetate, nanoparticles, thermal decomposition
Introduction
ics, medical imaging, drug targeting and sensing applica-
tions [10–12]. Combustion synthesis provides a direct pro-
duction route to such revolutionary particles and the range
of materials that can be produced using combustion syn-
thesis has been dramatically increased by diversifying
the precursor compounds to include solid-phase reactant
precursors such as metal acetates [2, 13].
Metal acetates are used as metal precursors in many
nanoparticle synthesis methods, including oxidation [1],
combustion [2, 3], anaerobic solution [4] and thermal
precipitation synthesis methods [5], to name a few. The
many recent advances in these and other synthesis meth-
ods have facilitated exploring broader applications of
metal and metal oxide nanoparticles by utilizing the
synthesis methods to augment the electrical, optical and
chemical properties of the materials. The crystalline
phase, degree of oxidation and particle morphology and
size are strongly dependent on the synthesis process
and, further, the reactant precursor used for the metal
nanoparticles [6]. For example, earlier studies con-
ducted in our laboratory have indicated there is a strong
relationship between the decomposition properties of
the solid phase metal acetate precursors and the product
metal/metal oxide nanocomposite materials created dur-
ing combustion synthesis [2]. However, there is little
understanding of the fundamental physical and chemical
mechanisms that govern the decomposition of metal
acetates in these combustion and other synthesis
systems which rely on thermal decomposition and
oxidation of metal acetates [6].
The current work focuses on identifying the
mechanisms important during thermal decomposition
of metal acetates in an oxidizing environment. The
results of this study have relevance to combustion,
pyrolysis and oxidation synthesis methods which use
metal acetates as reactant precursors and to synthesis
methods which use solid-phase organometallic precursors
for generating nanoparticles [14, 15]. One of the major
objectives of the investigation was to evaluate if there
is a direct relationship between the average crystallite
size of the metal acetate and the resulting metal
nanoparticles produced. Gold acetate was selected for
the study due to the importance of gold nanoparticles
in many advanced optical, electrical and catalytic
applications and the strong dependence of material
performance on the size of the gold nanoparticles [16–19].
In particular, the experiments were designed to elucidate
the early steps in the synthesis process, when metal
acetate particle decomposition occurs. The initial
steps in this process are critical in determining the
characteristics of the final product nanoparticles.
Combustion synthesis has proved to be a versatile
production technique for metal oxides and carbon-based
particles such as silica (SiO2), titania (TiO2), alu-
mina (Al2O3) and carbon black. Recent developments have
expanded the scope of combustion synthesis to the gen-
eration of metal nanoparticles [7], nanocomposites [8] and Experimental
one-dimensional structures [9] – focusing on the morphol-
ogy and size, as much as the composition of the material
itself. This renewed interest is attributed to new discov-
eries on functional nanomaterials for catalysis, electron-
A combination of experimental strategies was employed
to characterize the early stages of the gold acetate
decomposition process. In one set of experiments, gold
*
Author for correspondence: bakrania@rowan.edu
1388–6150/$20.00
Akadémiai Kiadó, Budapest, Hungary
Springer, Dordrecht, The Netherlands
© 2008 Akadémiai Kiadó, Budapest

BAKRANIA et al.
acetate particles were heated on a hot-plate in air.
The as-received gold acetate powders were deposited
on copper tape and sputter coated with Au–Pd (to im-
prove image contrast) and then imaged using a scan-
ning electron microscope (SEM, Philips XL30). The
products of the heated gold acetate particles were col-
lected by direct deposition of the particles ejected
during decomposition onto transmission electron mi-
croscope (TEM) grids. The TEM grids (3 mm diame-
ter, Electron Microscopy Sciences, carbon film,
300 mesh copper) were placed vertically at distances
of 1, 2 and 4 mm away from the initially unreacted
gold acetate particles. Samples of the ejected particles
were obtained in order to monitor the evolution of the
particles after fragmentation. The morphology of the
decomposition products of the gold acetate were recorded
using TEM imaging (Philips CM-12 Transmission
Electron Microscope). Particle size information was
determined from the TEM images using ImageJ, a
Java-based image processing program [20]. For the
image processing, the TEM files were converted to
8-bit grey-scale for auto-thresholding to separate the frag-
mented particles from the background of the images.
Unreacted (as-received) gold acetate powders and
partially decomposed gold acetate powders were analyzed
by XRD. Scans for phase identification and for average
crystallite size of the unreacted powders were obtained
using an automated Scintag Theta–Theta XRD with
increments of 0.02° 2q and CuKa radiation
(l=1.5406 ). The scans were obtained over a 2q
range of 5–70° at a scan rate of 1° 2q min–1. Spectral
scans for average crystallite size were measured over
a 2q range of 11–16° at a scan rate of 0.5° 2q min–1.
The average crystallite size was determined from the
XRD spectra using the Scherrer equation,
High-speed imaging was used to capture the fragmentation
of the metal acetate particles. Electron microscopy of
unreacted and product particles ejected during
decomposition was used to identify characteristic
particle sizes and morphologies. X-ray diffraction (XRD)
of the unreacted gold acetate was used to determine
the average crystallite size of the powders. XRD
spectra were also obtained for partially decomposed
gold acetate to investigate changes in the reactant
composition during the early stage of decomposition
process. Because little is known regarding the thermal
decomposition of gold acetate, another set of
experiments using thermogravimetric analysis (TG)
was used to confirm the general characteristics
observed in the hot-plate studies.
All experiments were conducted using gold(III)
acetate (Au(C2H3O2)3, Alfa Aesar, 99.96% Au). The
as-received powders ranged in color from light brown
to black. The source of the variability in the color
(e.g. differences in particle size, partial decomposition
of the powders, etc.) was not identified by the
manufacturer. Based on the results of this study (presented
and discussed below), the color change is likely due
to partial decomposition of the gold acetate powders.
For the hot-plate experiments, the powders were
heated at a rate of approximately 25°C min–1 (Fisher
Scientific, Thermix Hot Plate Model 300T) in room
air. This heating rate was selected because it was
considered more representative of synthesis processing
conditions for metal acetates than lower heating rates
and lower heating rates did not yield physical
fragmentation of the reactant particles. A high-speed
digital imaging video camera (Vision Research,
Phantom v.7.1, Nicor 50 mm lens f/0.95) was
positioned approximately 100 mm from the hot plate
for side-view imaging of the gold acetate particles. A
15 mm c-mount extension tube was used with the
camera to optimize image quality. A metal halide
lamp (ED 28 Universal bulb, 400W) was used to
illuminate the decomposing particles on the hot plate.
The following procedure was used for the hot
plate experiments. Gold acetate particles ranging from
0.5–1.0 mm in size were placed approximately 30 mm
from the edge of the hot plate. The high-speed imaging
camera was operated at 4800 frames per second
(208 ms) with a spatial resolution of 800×600 pixels.
The exposure time was set to 19 ms. The camera was
manually triggered to record the particle decomposition
process after particle heating was initiated. During the
experiments, a K-type thermocouple was used to
record the temperature of the hot plate approximately
2 mm away from the gold acetate particles.
0.9l
d
=
XRD
b1/ 2cosq
where dXRD is the average crystallite size, l is the
source wavelength, b1/2 is the full width at half
maximum of the peak used for the analysis and q is
the XRD scattering angle of the peak. The partially
decomposed powders were obtained by heating the
powders at 110°C for 1 h (prior to particle fragmentation,
discussed above). The XRD scans of these materials
were used to determine the degree of decomposition
of the gold acetate and the relationship between reactant
decomposition and the acetate color.
The TG experiments were conducted in an inert
(nitrogen) environment using a heating rate of 5°C min–1
(TA Instruments Q50). Samples of approximately
4 mg of gold acetate powders were used. The metal acetate
particles were not size-selected and were representative
of the as-received powder. The gold acetate used in
the TG study was darker in color than the powders
used in the hot-plate experiments. As noted earlier,
Powders before and after the decomposition process
were collected for analysis using electron microscopy.
118
J. Therm. Anal. Cal., 95, 2009

THERMAL DECOMPOSITION OF GOLD ACETATE
the darker color indicated some partial decomposition
as-received powders) to dark brown (for the heated
gold acetate powders). As a consequence, we concluded
that the darker gold acetate powders are more partially
decomposed than the light brown powders.
of the reactant materials had occurred. This was not
considered a concern for the TG experiments as the
intention was to verify the decomposition temperature,
which remained clearly identifiable in the TG
time-histories.
Figure 2 shows SEM images of the as-received
gold acetate powders. The images show the powders
have a highly porous structure formed by finer grains
that are less than 1 mm in size. XRD Scherrer analysis
of the as-received light brown powders yielded an
average crystallite size of 160 nm for the gold acetate
used in this study.
Results and discussion
Characteristic particle size and morphology
Figure 3 presents a typical imaging sequence of
the gold acetate decomposition and fragmentation process
that occurred during each experiment. The first panel
of Fig. 3 shows three gold acetate particles on the hot
plate at the start of the experiment. The second panel
shows the initial asymmetric fragmentation of the
center gold acetate particle. A sharp cracking sound
was also heard at this time. Figure 3c shows the
formation of a nebula of particle fragments that
occurred a fraction of a second (Dt=208 ms) after the
first decomposition event. The decomposition
temperature, defined as the temperature of the hot plate
when the particles fragmented, was 103°C with an
uncertainty of ±20°C (based on the standard deviation
of multiple experiments). Although not apparent in the
images in Fig. 3, the gold acetate particles also
changed in color from light brown to dark
brown/black before fragmentation was observed.
After the experiments had concluded, pink streaks
forming fairly uniform radial patterns of nanosized
metallic gold were identified on the hot plate.
Representative XRD patterns of the as-received gold
acetate and heated sample are presented in Fig. 1. The
JCPDS database does not have a reference pattern for
gold acetate. Consequently, the XRD spectra for gold
acetate reported by Kristl and Drofenik [21] is provided
in the figure as a basis for comparison. As seen in the
figure, the XRD pattern for the gold acetate used in
this work is consistent with the previous XRD pattern
for gold acetate documented in the literature; however,
each spectra indicate some gold impurities (e.g. the
features at 2q=38.17 and 44.37°). The heated gold ac-
etate spectra were used to determine if the gold impurity
was from partial decomposition of gold acetate. The
degree of decomposition can be assessed by evaluat-
ing the gold <111> peak intensity relative to the gold
acetate peak intensity of the feature at approximately
2q=14.5° for the as-received and heated gold acetate
powders. A comparison of the difference in peak in-
tensities normalized by the gold acetate peak height
indicates greater presence of gold in the heated sam-
ple (Igold acetate – Igold/Igold acetate=0.11) compared to the
As noted earlier, samples of the particles ejected
during decomposition were acquired at various
locations near the hot plate by direct deposition onto
as-received powders (Igold
– Igold/Igold acetate=0.43);
acetate
suggesting greater degree of decomposition has oc-
curred in the heated sample. This experiment also
yielded a change in color from light brown (for the
Fig. 2 SEM images of as-received gold acetate powder
Fig. 1 Typical X-ray diffraction patterns of as-received (light
brown in color) and heated (dark brown) gold acetate
powders. A reference XRD pattern presented in Kristl
and Drofenik [19] for gold acetate and the JCPDS
reference for metallic gold are provided for comparison
Fig. 3 Typical imaging sequence showing the progression of
gold acetate particle decomposition and fragmentation
after heating at 25°C min–1 was initiated at t=0 min:
a – t=0 min, b – t=~4 min, c – t=4 min 0.2 ms
J. Therm. Anal. Cal., 95, 2009
119

BAKRANIA et al.
TEM grids. In order to estimate the residence time of
examples of two such clusters captured from two
separate grids located 1 mm from the initial gold acetate
particles. Grid samples collected at 4 mm did not produce
such cluster patterns. Instead much larger (with diameters
ranging from 50 to 100 nm) and highly dispersed
spherical gold particles were observed, such as shown
in Fig. 4d. Fast sintering rates are clearly present in
this system, as samples obtained at 4 mm were dominated
almost exclusively by large coalesced particles.
The TEM images were analyzed to determine the
characteristic features of the gold fragment particles.
In particular, the area of individual particles and cluster
projected areas (defined as the total area of co-located
and clustered fragmented particles) were measured
from the images. The areas of the particles were used
to determine the characteristic diameters of the discrete
spherical particles, the diameters of the particles co-located
within a cluster and the equivalent diameter that would
be obtained if all the particles located in a cluster were
coalesced. All image analysis assumed spherical
particles. Analysis conducted on two grids sampled at
1 and 2 mm (obtained from two experiments) yielded
coalesced sphere diameters of 140±65 and 171±54 nm,
respectively. Here, the uncertainties represent the
standard deviation in the particles analyzed. Individual
particles in the clusters had diameters in the range of
5–10 nm. TEM images of the discrete spherical particles
that were not part of the clusters yielded average
diameters of 12.8 4.9 nm.
the particles sampled, a radial particle ejection
velocity was calculated from the high-speed imaging
sequence based on the evolution of the particle plume
relative to the center of the gold acetate particle. The
average minimum initial particle ejection velocity
was 14 m s–1. This average ejection velocity yields a
residence time of 0.14 ms for a grid located 2 mm
from the center of the source particle.
TEM images of the particle fragments collected
during decomposition revealed generally spherical
nanoparticles dispersed non-uniformly on the grids.
All the particles present on the grid were identified as
metallic gold using X-ray energy dispersive spectroscopy
(XEDS). A typical TEM image of particles sampled at
1 mm is provided in Fig. 4a. There were two categories
of particles observed in the TEM imaging as seen in
Fig. 4a: discrete spherical particles and more densely
co-located particles which often appear partially
coalesced (circled features). Figures 4b and c show
The TEM data can be compared to the characteristic
dimensions of the unreacted gold acetate powders,
where dXRD=160 nm, to develop an understanding of
the effects of the decomposition process on the gold
nanoparticles produced. Our hypothesis, initially
proposed in Bakrania et al. [2], is that the gold acetate
crystallite grains are direct precursors to the initial gold
particles formed and the crystallite grain dimensions
therefore dictate the minimum dimension of the gold
nanoparticles. The results of this work provide further
data to consider this theory. Accounting for the molecular
weight and density of gold acetate and gold, assuming
complete conversion of gold acetate to metallic gold
and assuming the gold acetate fragments along crystallite
grains, the average crystallite size of 160 nm would
yield spherical gold particles with an average diameter
of 63 nm. This average is between the characteristic
sizes determined for the larger coalesced particles
(140±65 nm) and the smaller discrete particles
(12.8±4.9 nm). While the results do not isolate a
unique decomposition pathway, the results do point
toward a relationship between the average crystallite
size of the metal acetate reactant and the resulting
metal nanoparticles.
Fig. 4 TEM images of gold nanoparticles captured during gold
acetate fragmentation. a – typical gold nanoparticles
sampled at 1 mm; b, c – typical partially coalesced gold
clusters sampled at 1 mm from two experiments;
Figure 5 presents a schematic of a decomposition
mechanism based on the hypothesis described above
d – coalesced gold nanoparticles sampled at 4 mm
120
J. Therm. Anal. Cal., 95, 2009

THERMAL DECOMPOSITION OF GOLD ACETATE
and which is consistent with the experimental data
observed in this study. Initially, heating causes
discoloration of the particles indicating relatively slow
gold acetate decomposition. This is rapidly followed
by the explosive fragmentation stage, where fragments
with a range of sizes (on the order of 160 nm) are
ejected. The decomposition of metal acetates is typically
highly exothermic [22–24]. These fragments can
potentially be at much higher temperatures than the
decomposition temperature of 100°C. The fragments
form clouds of gold particles and vapor phase gold.
Differences in the size of the clouds and the number
density of gold affect the number of initial nucleation
sites present within the cloud. Given similar residence
times (~0.14 ms in the current study), higher number
densities will yield more nucleation sites and higher
interparticle collision frequencies than lower number
densities. Consequently, more dense fragment clouds
will yield more gold particles and larger gold particles
(assuming rapid coalescence rates), such as shown in
the reaction pathway on the left side of Fig. 5. Less
dense clouds of particle fragments will form smaller
and more discrete spherical particles, as shown in the
reaction pathway on the right of Fig. 5. A range of initial
particle fragment sizes is expected as the XRD data
represent an average crystallite dimension. The rates
of interparticle collision and coalescence dictate the
final size of the gold nanoparticles.
Decomposition temperature
Figure 6 presents the thermogravimetric data for gold
acetate mass loss under heating and the derivative of
the percent mass loss with respect to temperature.
Significant mass loss was observed between 120–210°C,
with an average peak mass rate of loss occurring at
T=170°C. Although the decomposition temperature
measured during the hot plate experiments (103±20°C)
was at a significantly different heating rate, the
decomposition temperature is comparable to results
obtained from the TG peak decomposition temperature.
The TG decomposition temperature for gold
acetate agrees relatively well with other metal acetate
thermal analysis such as palladium and copper acetate
[22, 23, 25]. Such studies have suggested acetic acid
is the dominant organic by-product when decomposition
yields a pure metal. Note, the mass loss observed in
these TG experiments was lower than expected;
which is consistent with some partial decomposition
of the powders prior to the analysis.
Fig. 5 Schematic of proposed gold acetate decomposition and
gold particle formation mechanisms under rapid heating
Conclusions
This work provides the first documentation of the
dramatic particle fragmentation process that occurs
Fig. 6 TG curves for gold acetate in nitrogen using a heating
rate of 5°C min–1
J. Therm. Anal. Cal., 95, 2009
121

BAKRANIA et al.
11 O. V. Salata, J. Nanobi., 2 (2004) 3.
during heating of gold acetate in air. Measurements of
the gold acetate decomposition temperature are in
reasonable agreement and are consistent with expectations
for metal acetates. The results of materials analysis
supports that there is a relationship between the gold
acetate crystallite size and the metal nanoparticles
produced. Solid-phase metal-acetate precursors have
tremendous potential to expand the composition and
architecture of nanomaterials. This work provides key
insights into the physical and chemical decomposition
pathways that are important during decomposition of
gold acetate and provides quantitative data that are
critical for refining existing and developing new
synthesis applications.
12 B. Delmon, J. Therm. Anal. Cal., 90 (2007) 49.
13 S. D. Bakrania, C. Perez and M. S. Wooldridge, Process
Combust. Inst., 31 (2007) 1797.
14 D. Wostek-Wojciechowska, J. K. Jeszka, P. Uznanski,
C. Amiens, B. Chaudret and P. Lecante, Mater.
Sci.-Poland, 22 (2004) 407.
15 B. E. Hamaoui, L. Zhi, J. Wu, J. Li, N. T. Lucas,
¦. Tomoviº, U. Kolb and K. Müllen, Adv. Funct. Mater.,
17 (2007) 1179.
16 M.-S. Hu, H.-L. Chen, C.-H. Shen, L.-S. Hong,
B.-R. Huang, K.-H. Chen and L.-C. Chen, Nature Mater.,
5 (2006) 102.
17 V. K. Pustovalov and V. A. Babenko, Laser Phys. Lett.,
1 (2004) 516.
18 L. Pasquato, P. Pengo and P. Scrimin, J. Mater. Chem.,
14 (2004) 3481.
19 M. Haruta, Nature, 437 (2005) 1098.
20 M. D. Abramoff, P. J. Magelhaes and S. J. Ram,
Biophot. Intl., 11 (2004) 36.
References
1 M. H. Khedr, A. A. Farghali and A. A. Abdel-Khalek,
J. Anal. Appl. Pyrolysis., 78 (2007) 1.
21 M. Kristl and M. Drofenik, Inorg. Chem. Commun.,
6 (2003) 1419.
2 S. D. Bakrania, T. A. Miller, C. Perez and
M. S. Wooldridge, Combust. Flame, 148 (2007) 76.
3 T. A. Miller, S. D. Bakrania, C. Perez and
M. S. Wooldridge, J. Mater. Sci., 20 (2005) 2977.
4 Z. Jia, S. Kang, D. E. Nikles and J. W. Harrell,
IEEE Trans. Magnetics, 41 (2005) 3385.
5 M. Yin, Z. Chen, B. Deegan and S. O’Brien,
J. Mater. Sci., 22 (2007) 1987.
22 P. K. Ghallagher and M. E. Gross, J. Thermal Anal.,
31 (1986) 1231.
23 A. Obaid, A. Alyoubi, A. Samarkandy, S. Al-Thabaiti,
S. Al-Juaid, A. El-Bellihi and El-H Deifallah, J. Therm.
Anal. Cal., 61 (2000) 985.
24 R. M. Mahfouz, S. M. Alshehri, M. A. S. Monshi and
N. M. Abd el-Salam, Radiat. Effects. Defects Solids,
159 (2004) 345.
6 M. Afzal, P. K. Butt and H. Ahmad, J. Thermal Anal.,
37 (1991) 1015.
25 J. Y. Zhang and I. W. Boyd, Appl. Phys. A: Mater.
Sci. Proc., 65 (1997) 379.
7 R. N. Grass and W. J. Stark, J. Nanoparticle Res.,
8 (2006) 729.
Received: April 11, 2008
8 A. Vital, A. Angermann, R. Dittmann, T. Graule and
J. Töpfer, Acta Mater, 55 (2007) 1955.
Accepted: August 12, 2008
OnlineFirst: November 12, 2008
9 R. L. Vander Wal, L. J. Hall and G. M. Berger,
J. Phys. Chem. B, 106 (2002) 13122.
10 T. A. Miller, S. D. Bakrania, C. Perez and
M. S. Wooldridge, Functional Nanomaterials, (2006).
DOI: 10.1007/s10973-008-9173-1
122
J. Therm. Anal. Cal., 95, 2009