New preparation method of gold nanoparticles on SiO2

Article abstract of DOI:10.1021/jp060601y

It is shown that adsorption of the [Au(en)₂]³⁺ cationic complex can be successfully employed for the deposition of gold nanoparticles (1.5 to 3 nm) onto SiO₂ with high metal loading, good dispersion, and small Au particle size. When the solution pH increases (from 3.8 to 10.5), the Au loading in the Au/SiO₂ samples increases proportionally (from 0.2 to 5.5 wt %), and the average gold particle size also increases (from 1.5 to 2.4 nm). These effects are explained by the increase in the amount of negatively charged sites present on the SiO₂ surface, namely, when the solution pH increases, a higher number of [Au(en) ₂]³⁺ species can be adsorbed. Extending the adsorption time from 2 to 16 h gives rise to an increase in the gold loading from 3.3 to 4.0 wt % and in the average particle size from 1.8 to 2.9 nm. Different morphologies of gold nanoparticles are present as a function of the particle size. Particles with a size of 3-5 nm show defective structure, some of them having a multiple twinning particle (MTP) structure. At the same time, nanoparticles with an average size of ca. 2 nm exhibit defect-free structure with well-distinguishable {111} family planes. TEM and HAADF observations revealed that Au particles do not agglomerate on the SiO₂ support: gold is present on the surface of SiO₂ only as small particles. Density functional theory calculations were employed to study the mechanisms of [Au(en)₂]³⁺ adsorption, where neutral and negatively charged silica surfaces were simulated by neutral cluster Si₄O ₁₀H₄ and negatively charged cluster Si₄O ₁₀H₃, respectively. The calculation results are totally consistent with the suggestion that the deposition of gold takes place according to a cationic adsorption mechanism.

Full text of DOI:10.1021/jp060601y

J. Phys. Chem. B 2006, 110, 8559-8565
8559
New Preparation Method of Gold Nanoparticles on SiO2
,‡
Rodolfo Zanella,*,†,‡ Alberto Sandoval,† Patricia Santiago,§ Vladimir A. Basiuk, and
Jose´ M. Saniger†,‡
Centro de Ciencias Aplicadas y Desarrollo Tecnolo´gico, UniVersidad Nacional Auto´noma de Me´xico (UNAM),
Circuito Exterior S/N, A. P. 70-186, C. P. 04510, Ciudad UniVersitaria, Me´xico D. F., Mexico, Instituto de
F´ısica, UNAM, Circuito de la InVestigacio´n Cient´ıfica S/N, C. P. 04510, Ciudad UniVersitaria, Me´xico D. F.,
Mexico, and Instituto de Ciencias Nucleares, UNAM, Circuito Exterior, A. P. 70-543, C. P. 04510,
Ciudad UniVersitaria, Me´xico D. F., Mexico
ReceiVed: January 27, 2006; In Final Form: March 9, 2006
It is shown that adsorption of the [Au(en)2]3+ cationic complex can be successfully employed for the deposition
of gold nanoparticles (1.5 to 3 nm) onto SiO2 with high metal loading, good dispersion, and small Au particle
size. When the solution pH increases (from 3.8 to 10.5), the Au loading in the Au/SiO2 samples increases
proportionally (from 0.2 to 5.5 wt %), and the average gold particle size also increases (from 1.5 to 2.4 nm).
These effects are explained by the increase in the amount of negatively charged sites present on the SiO2
surface, namely, when the solution pH increases, a higher number of [Au(en)2]3+ species can be adsorbed.
Extending the adsorption time from 2 to 16 h gives rise to an increase in the gold loading from 3.3 to 4.0 wt
% and in the average particle size from 1.8 to 2.9 nm. Different morphologies of gold nanoparticles are
present as a function of the particle size. Particles with a size of 3-5 nm show defective structure, some of
them having a multiple twinning particle (MTP) structure. At the same time, nanoparticles with an average
size of ca. 2 nm exhibit defect-free structure with well-distinguishable {111} family planes. TEM and HAADF
observations revealed that Au particles do not agglomerate on the SiO2 support: gold is present on the surface
of SiO2 only as small particles. Density functional theory calculations were employed to study the mechanisms
of [Au(en)2]3+ adsorption, where neutral and negatively charged silica surfaces were simulated by neutral
cluster Si4O10H4 and negatively charged cluster Si4O10H3, respectively. The calculation results are totally
consistent with the suggestion that the deposition of gold takes place according to a cationic adsorption
mechanism.
Introduction
oxygen.6,8,16-21 To elucidate this question, the electronic struc-
ture and catalytic activity of Au nanoparticles has to be
At the end of the 80’s an exceptionally high catalytic activity
of gold nanoparticles supported on metal oxides was discovered
for the CO oxidation,1-6 which was interpreted as the formation
of an active Au/support interface along the perimeter of gold
particles.7,8 The Au/support interface is a model of a metal-
semiconductor or metal-insulator junction bearing unique size-
dependent electronic properties, which could play an essential
role in the origin of the high catalytic activity.
Following this assumption, the specific support materials
should play an important role in the catalytic behavior of gold.
The supports, generally being metal oxides, are divided into
two categories: the reducible (Fe2O3, TiO2, Co2O3, CeO2, ZrO2)
and the nonreducible supports (SiO2, Al2O3, MgO).8 The nature
of the active species is still being discussed. It has been
suggested that the role of the metal oxide support is the
stabilization of the gold nanoparticles, and that the CO oxidation
takes place on the gold surface.9-15 Other authors have proposed
that the reaction takes place at the gold/metal oxide interface,
and that the metal oxide support could act as a source of
characterized in the absence of any supporting material;
however, this is technically difficult. In our opinion, an
alternative approach for an experimental study of this subject
is the deposition of Au nanoparticles onto a noncatalytic, inactive
substrate. Silica satisfies these requirements well.
Silica is used as a support because of its inert character, which
allows investigation of the effect of the metal, minimizing the
effect of the metal-support interaction. The weak interaction
between SiO2 surfaces and Au particles makes silica supports
very convenient for studying the size effect of the gold
nanoparticles on their catalytic activity, minimizing the distur-
bance of the support interaction.22
Gold catalysts on an inert oxide support, such as SiO2, need
to be prepared in a highly dispersed state. Indeed, for these
systems, oxygen activation is not expected to occur on the
support, but on the gold particle surface.8 An enhanced activity
may also arise from geometric effects associated with defect
sites such as kinks, steps, and edges, or from electronic effects
arising from variations in the density of states of small gold
particles.15,23,24
Liquid-phase preparation methods such as deposition-
precipitation, coprecipitation, and impregnation have not been
effective in supporting small gold particles on SiO2.6,25-27 The
difficulty of depositing highly dispersed gold nanoparticles on
SiO2 by liquid-phase methods arises from the fact that the point
* Address correspondence to this author. E-mail: zanella@
aleph.cinstrum.unam.mx. Phone: +(52) 555-622-8602, ext. 1115. Fax.
+(52) 555-622-8651.
† Centro de Ciencias Aplicadas y Desarrollo Tecnolo´gico.
‡ Members of UNAM Nanotechnology Project (PUNTA).
§ Instituto de F´ısica.
Instituto de Ciencias Nucleares.
10.1021/jp060601y CCC: $33.50 © 2006 American Chemical Society
Published on Web 04/11/2006

8560 J. Phys. Chem. B, Vol. 110, No. 17, 2006
Zanella et al.
of zero charge (PZC) for SiO2 surfaces is close to pH 2. This
fact means that the SiO2 surface becomes negatively charged
above pH 2, complicating the deposition of anionic gold species
such as (AuCl4)-. For this reason, when these methods are
applied to deposit gold on SiO2, the average nanoparticle
diameter becomes very large (20 nm),25 and no detectable
catalytic activity for CO oxidation at low temperature is
observed.26
anhydrous diethyl ether (Sigma), ethylenediamine (Sigma), and
crystalline HAuCl4‚3H2O (Aldrich) as gold precursor were used
for the preparation of [Au(en)2]Cl3. The [Au(en)2]3+ complex
was prepared as described by Block and Bailar.33 Before cation
adsorption of [Au(en)2]3+, SiO2 was dried in air at 100 °C for
at least 24 h. All the preparations were performed in the absence
of light, which is known to decompose the gold precursors.
For the preparation of Au/SiO2 samples, 1 g of SiO2 was
dispersed in 100 mL of an aqueous solution of [Au(en)2]Cl3
(4 × 10-3 M) previously heated to 45 °C. The amount of gold
in solution corresponds to a maximum gold loading of 7 wt %
on SiO2. To promote the cationic adsorption of [Au(en)2]3+
complex on SiO2, the solution pH was adjusted to the values
Many attempts have been undertaken to obtain small gold
particles on SiO2 by using different experimental approaches.
Gold particles of 10-50 nm diameter on SiO2/Si(100) were
created by Ar+ ion implantation of a bulklike Au/SiO2/Si(100)
thin film.22 A strategy to immobilize gold nanoparticles (5 nm)
on SiO2 surfaces has been developed by embedding the gold
particles into a layer of octadecylsiloxane or into multiple layers
of SiO2 created by deposition and oxidation of octadecylsiloxane
self-assembled monolayers.28 Okumura et al.29 prepared gold
particles (6.6 nm) on SiO2 by chemical vapor deposition (CVD),
which showed a high catalytic activity in the CO oxidation.
Au/SiO2 prepared by the CVD method exhibits catalytic activity
for CO oxidation as high as Au/Al2O3 and Au/TiO2 do. This
could indicate that the reason for the low catalytic activity,
usually found for the Au/SiO2 systems, cannot be simply
attributed to the nature of the support, but that the procedure of
preparation is also crucial. Au/SiO2 has also been prepared by
CVD of dimethylgold(III) â-diketone,30 but the problem of CVD
methods is that the volatile Au precursors used are very
expensive.
higher than IEPSiO by dropwise addition of 0.1 or 1 M
2
ethanediamine solution. The suspension was vigorously stirred
for 1 to 16 h at 45 °C in a thermostated vessel.
After the deposition of gold onto SiO2, all the samples were
submitted to the following procedure: solid separationsthe
suspended solids were separated from the precursor solution
by centrifugation (5000 rpm for 10 min); washingsthe separated
solids were resuspended in water (100 mL g-1), stirred for 10
min at 40 °C, and centrifuged again with this washing procedure
being repeated four times to remove residual Cl- ions as well
as Au species not interacting with the support; dryingsthe
washed solids were dried under vacuum at 25 °C for 12 h;
thermal treatments100 mg of sample were heated in a flow
(100 mL min-1) of industrial-grade air or ultrapure hydrogen
(both from Praxair) from room temperature to the final tem-
perature at a heating rate of 2 °C min-1, followed by a 4-h
isothermal step at the final temperature (this thermal treatment
leads to the decomposition of Au(III) complexes into gold metal
particles; it must be remembered that the thermal treatment used
to reduce AuIII into Au0 can be performed with any gas (reducing
gases such as H2, or oxidizing gases such as air); when the
supported gold precursor thermally decomposes in air, it forms
Au0 because of the instability of AuIII species (∆Hf(Au2O3) )
19.3 kJ mol-1)37); storagesall the samples were kept away from
light and under vacuum in a desiccator at room temperature (a
careful storage of the samples under the conditions specified is
crucial; in fact, a strong increase in the average particle size in
the calcined samples was observed when the samples were
exposed to air even for a short period38).
Guillemot with co-workers35,39 showed in a work on the
preparation of gold catalysts supported on Y zeolite that the
[Au(en)2]3+ complex tends to decompose when it is heated to
temperatures above 60-70 °C. As a consequence, they sug-
gested that gold is reduced by ethanediamine ligands so that
the preparation has to be performed at temperatures below 60
°C to avoid the complex decomposition and the reduction of
gold. Indeed, the XANES study of Au/TiO2 prepared by cationic
adsorption of [Au(en)2]3+ performed by one of us25 revealed
the presence of metallic gold in the sample prepared at 80 °C.
The low stability of [Au(en)2]3+ complex is certainly due to
the metallic gold, which originated from the decomposition of
[Au(en)2]3+ in solution during the adsorption and subsequent
deposition of metal gold colloids. In the same work, it was
shown that the samples prepared at 25 °C do not contain metallic
gold, but at the same time the maximum gold loading is low
(∼1 wt %). Therefore, in this work we decided to work at 45
°C, to avoid the decomposition of [Au(en)2]3+ and to favor the
cationic adsorption.
Taking into account the fact that SiO2 is an amphoteric oxide
(isoelectric point IEPSiO ) 2),31 this oxide can be used for the
2
adsorption of cationic metallic species when the solution pH is
higher than IEPSiO . Iler32 has proposed a mechanism of silica
2
dissolution in basic medium, which involves OH- ions as
catalysts of silica depolymerization; indeed, for pH lower than
11, the Si(OH)5- species released is hydrolyzed into silicic acid
Si(OH)4, and the OH- ions are regenerated.
In basic medium, Si(OH)4 units can condense, be ionized
[SiO(OH)3]-, and form different types of monomeric, oligo-
meric, and negatively charged polymeric chains.
In the case of gold, [Au(en)2]3+ complex (en ) ethanedi-
amine), whose synthesis was described by Block and Bailar,33
has been successfully used by Guillemot et al.34,35 for the
introduction of gold into Y zeolites by cationic exchange and
25,36
by one of us to deposit Au/TiO2
by cationic adsorption.
Until now (to the best of our knowledge) nobody has tried the
adsorption of cationic [Au(en)2]3+ complex on SiO2.
The goal of this study is therefore to investigate whether it
is possible to prepare Au/SiO2 catalysts with small metal
particles based on the adsorption of cationic [Au(en)2]3+
complex, in the same size range as for the samples prepared on
TiO2 by the same method.25,36 In this way, a new and simple
liquid-phase method for the preparation of Au/SiO2 would be
available, allowing for a more reliable comparison of the
activities of gold nanoparticles supported on reducible and inert
supports, both prepared by the same experimental approach.
In this paper, we show for the first time that adsorption of
cationic [Au(en)2]3+ complex (CA) can be successfully applied
for the preparation of Au/SiO2 catalysts with high metal loading,
good dispersion, and small Au particle sizes.
Experimental Section
1. Au/SiO2 Preparations. SiO2 Degussa Aerosil 200 was
used as support (BET surface area: 198 m2 g-1, average particle
size of 12 nm, purity >99.5%). Absolute ethyl alcohol (Aldrich),
2. Characterization Techniques. Thermally treated Au/SiO2
samples were examined by transmission electron microscopy
(TEM) with a 2010 FasTem analytical microscope equipped

Small Metal Particles of Au/SiO2
J. Phys. Chem. B, Vol. 110, No. 17, 2006 8561
TABLE 1: Au/SiO2 Samples Prepared by Cationic
Adsorption of [Au(en)2]3+ at 45 °C and Reduced in H2 at
200 °C
preparation
results
time of
Au loading av particle
standard
sample contact (h) pH
(wt %)
size (nm) deviation (nm)
CA1
CA2
CA3
CA4
CA5
CA6
2
2
2
2
4
3.8a
7
9
10.5
9
9
0.2
2.1
3.3
5.5
3.8
4.0
1.5
1.5
1.8
2.4
2.5
2.9
0.42
0.51
0.44
0.52
0.48
0.82
Figure 1. Optimized geometries of the neutral cluster Si4O10H4 and
the negatively charged cluster Si4O10H3 used to model neutral and
negatively charged silica surfaces, respectively.
16
a Natural pH.
with a Z-contrast annular detector. The histograms of the metal
particle sizes were established from the measurement of 500 to
1000 particles. The size limit for the detection of gold particles
on SiO2 was about 1 nm. The average particle diameter dh was
calculated from the following formula: dh ) ∑nidi/∑ni, where
ni is the number of particles of diameter di. The standard
deviation was calculated from the following formula: σ )
respectively. Therefore, the increase of pH (from 3.8 to 10.5)
leads to a significant increase in the Au loading (from 0.2 to
5.5 wt %). On the other hand, the gold particle size (1.5 nm)
did not increase when pH was changed from 3.8 to 7, and
slightly increased at pH 9 and 10.5 (to 1.8 and 2.4 nm,
respectively).
Another parameter studied was the adsorption time, varied
from 2 to 16 h (samples CA3, CA5, and CA6). The gold loading
slightly increased from 3.3 to 4.0 wt %, while the average
particle size increased from 1.8 to 2.9 nm (Table 1). The particle
size distribution became broader when the adsorption time
increased (Table 1).
[(∑(di - dh)2)/∑ni]1/2
.
Chemical analysis of Au in the samples was performed by
energy-dispersive X-ray spectroscopy (EDS) with a detector
Oxford-ISIS coupled to a scanning electron microscope (JEOL
JSM-5900-LV). Chemical analyses were performed after thermal
treatment of the samples. The Au weight loading is expressed
In all cases, the TEM images showed highly dispersed gold
nanoparticles on the surface of SiO2 as is exemplified in Figure
2a. Conventional HRTEM images are shown in Figure 2b. The
images were obtained at the optimum defocus condition to
observe Au nanoparticles. Different morphologies are present
as a function of the particle size. The particles with a size of 3
to 5 nm show defective structure, some of them exhibiting a
multiple twinning particle (MTP) pattern, whereas the nano-
particles with an average size of ca. 2 nm have a defect-free
structure with readily visible {111} family planes (Figure 2c,d).
Additional information on the spatial distribution of Au nano-
particles over SiO2 was obtained from the Z-contrast imaging
of Au/SiO2 (Figure 3). The High Angle Annular Dark Field
technique (HAADF) can provide highly detailed images of
nanocrystal surfaces, 3D information, and mass contrast simul-
taneously. The contrast in this type of images depends basically
on the atomic number (Z); in Figure 3, the brighter zones
correspond to Au nanoparticles. Figure 3a demonstrates the
surface distribution of Au nanoparticles, where the differences
in contrast indicate that gold is not agglomerated. Gold is present
on the surface of SiO2 only as small particles, which is evident
from Figure 3a,b, where the Z contrast technique generates
images with the high angle scattered electrons.
In all the tested experimental conditions, the solution pH
(from 3.8 to 10.5) was higher than IEPSiO2 (∼2),31 so that the
adsorption of [Au(en)2]3+ was always possible. However, as
was mentioned above, the gold loading is very dependent on
the solution pH. This can be explained by the increase in the
amount of negatively charged species present on the SiO2
surface. The more the solution pH is above the pH of IEP, the
more sites at the SiO2 surface become negatively charged, and
correspondingly the higher number of [Au(en)2]3+ species is
adsorbed. This indicates that the deposition of gold probably
takes place according to a cationic adsorption mechanism, which
leads to a high dispersion of the gold precursors and to small
metal particle sizes after thermal treatment.
in grams of Au per gram of sample wt % Au ) [mAu/(mAu
+
m
SiO )] × 100.
2
Diffuse reflectance UV-visible spectra were obtained by
using an Ocean Optics USB2000 miniature fiber optic spec-
trometer. Thermal analyses were carried out on a Jupiter Netzch
simultaneous TGA-DSC analyzer, under air (50 mL/min) at a
heating rate of 10 °C min-1. The Raman spectra were obtained
with a Nicolet Almega micro-Raman dispersive equipment,
using an excitation source of 532 nm.
3. Theoretical Calculations. The electronic structure calcula-
tions were performed with DMol3 numerically based density-
functional theory (DFT) computer software implemented in
Materials Studio Modeling 3.2 package from Accelrys Inc. We
employed the Perdew-Wang PW91 general gradient ap-
proximation (GGA) in conjunction with double numerical basis
set DND having a polarization d-function on all non-hydrogen
atoms. Due to the presence of gold atoms in the model systems,
DFT semilocal pseudopotentials were used instead of all-electron
core treatment. Apparently for the same reason, the default
convergence settings did not work well for geometry optimiza-
tion, and convergence problems were found; a way to override
them was to increase the maximum number of iterations and
SCF cycles to 1000. The global orbital cutoff was 4.000 Å.
Neutral and negatively charged silica surfaces were simulated
by neutral cluster Si4O10H4 and negatively charged cluster
Si4O10H3, respectively (Figure 1).
Results and Discussion
Table 1 shows the gold loading and average particle size of
Au/SiO2 samples prepared by cationic adsorption. All the
samples presented in Table 1 were reduced in H2 at 200 °C.
The natural pH of the original [Au(en)2]Cl3 solution with SiO2
in suspension is 3.8, and under these conditions (CA1 sample)
the Au loading is low (0.2 wt %) and the gold particle size is
1.5 nm. When the pH was increased (7, 9, and 10.5) the Au
loading in Au/SiO2 samples also increased in a proportional
way. For the samples prepared at pH 7 (CA2), 9 (CA3), and
10.5 (CA4) the gold loading was 2.1, 3.3, and 5.5 wt %,
To give an insight into the adsorption mechanisms of
[Au(en)2]3+ ions on the SiO2 surface, we performed a series of
geometry optimizations using the PW91 GGA method in

8562 J. Phys. Chem. B, Vol. 110, No. 17, 2006
Zanella et al.
Figure 2. (a) Low-magnification TEM micrograph showing the particle distribution on the SiO2 surface. (b) HRTEM image of gold nano-
particles. (c) HRTEM image of gold nanoparticles with different morphologies (A corresponds to a MTP and B corresponds to a FCC nanoparticle).
(d) Gold nanoparticle of 2-nm size exhibiting the {111} family planes.
Figure 3. (a) HAADF image showing surface distribution of the gold nanoparticles. (b) A higher magnification image where it is possible to
observe the contrast differences between the matrix and the nanoparticles. No gold agglomerates are present.
conjunction with the DND basis set. As shown in Figure 4a,
all the Au-N coordination bonds in free [Au(en)2]3+ ion are
almost equal (2.114-2.116 Å). Both trans-N-Au-N angles
are 180°, so that the coordination sphere is planar. These
parameters notably change in the adsorption complexes
[Au(en)2Si4O10H4]3+ and [Au(en)2Si4O10H3]2+. In the former
case, which models [Au(en)2]3+ adsorption on a neutral silica
surface, the length of Au-N bonds remains relatively uniform,
of 2.076 to 2.116 Å (Figure 4b). At the same time, the
coordination sphere undergoes considerable distortion, where
trans-N-Au-N angles reduce to 174.3° and 178.5°. This
distortion becomes even more manifested in the complex
[Au(en)2Si4O10H3]2+, which corresponds to [Au(en)2]3+ adsorp-
tion on a negatively charged silica surface. The trans-N-Au-N
angles further reduce to 173.1° and 178.0°, and Au-N bond
lengths vary from 2.028 to 2.184 Å (Figure 4c). Here one should

Small Metal Particles of Au/SiO2
J. Phys. Chem. B, Vol. 110, No. 17, 2006 8563
Figure 4. Optimized geometries for the gold complexes (a) [Au(en)2]3+, (b) [Au(en)2Si4O10H4]3+, and (c) [Au(en)2Si4O10H3]2+. The values specified
are formation energies of the adsorption complexes (in kcal mol-1) (italics), trans-N-Au-N angles in degrees (green), Au-N bond distances (red)
and some other selected interatomic distances (black) in angstroms. Element colors: C, gray; H, white; N, blue; O, red; and Au and Si, yellow.
Figure 6. Diffuse reflectance UV-visible spectra of the CA4 sample
Figure 5. Diffuse reflectance UV-visible spectra of sample CA4 as
as a function of the temperatures of drying (25 °C) and calcination
a function of the temperatures of drying (25 °C) and reduction under
under air (100, 150, 200, 300, and 400 °C).
hydrogen (100, 150, and 200 °C).
the incident light far exceeds the particle diameter. The plasmon
mention a somewhat unexpected result, namely, that upon
geometry optimization, one hydrogen atom of the (upper rear)
NH2 group transferred completely to the initially deprotonated
Si-O group of the Si4O10H3 cluster. It is the deprotonated amino
group that forms the shortest Au-N bond of 2.028 Å. In none
of the model surface complexes have we found direct interaction
of the Au atom with the cluster; the closest approach was the
one to the oxygen atom of the Si-O-H group, which remained
as long as 3.051 Å in [Au(en)2Si4O10H4]3+ and 2.935 Å in
[Au(en)2Si4O10H3]2+. On the other hand, hydrogen bonding can
be clearly observed between the N-H groups of the (left)
ethylenediamine ligand and siloxane bridges of the silica
clusters. In [Au(en)2Si4O10H4]3+, this bonding involves two
N-H groups, with the NH‚‚‚O distances of 1.648 and 1.831
Å. In [Au(en)2Si4O10H3]2+, only one NH‚‚‚O separation (1.765
Å) can be qualified as an H-bond, whereas the second one
increases to 2.542 Å. The opposite ethylenediamine ligand
remains separated from the Si-O-H group by more than 2.5
Å in both adsorption complexes. The calculated formation
energies (Figure 4) exhibit even larger differences than the
geometric parameters do. While the value of -63.8 kcal mol-1
was obtained for [Au(en)2Si4O10H4]3+, it is -310.2 kcal mol-1
for [Au(en)2Si4O10H3]2+. Thus, the theoretical calculations are
totally consistent with our suggestion that the deposition of gold
takes place according to a cationic adsorption mechanism.
band is clearly visible after the treatment in hydrogen at 100
°C (Figure 5) and does not change at higher temperature (150
and 200 °C), suggesting that most of the gold is in the form of
metallic particles at 100 °C.
On the other hand, Figure 6 presents the diffuse-reflectance
UV-visible spectra of the same sample (CA4), dried under
vacuum at room temperature and calcined in air at 100, 150,
200, 300, and 400 °C. The evolution of the plasmon band with
temperature is different from that observed for the same samples
treated in hydrogen. A weak plasmon band at ∼550 nm is visible
in the spectrum of the sample treated at 100 °C, which further
increases after calcination at 150 °C. This confirms that metallic
gold is present in the sample even when it is treated in air. The
plasmon band intensity first increases with an increase in the
treatment temperature, and then remains constant between 300
and 400 °C. It is known that the gold precursor decomposes in
air with temperature to form Au0 due to the instability of AuIII
species. Guillemot et al.35 also suggested that Au3+ cations can
be reduced by ethylenediamine ligands. As should be expected,
gold is more easily reduced in hydrogen that in air. Furthermore,
the spectra of the samples treated in air at 100, 150, and 200
°C exhibit an additional band in the region of 245-290 nm
with a maximum at ∼275 nm, which is not observed for the
samples treated in hydrogen, even at the lowest temperature of
reduction (100 °C). This band disappears after 300 °C. Even if
the literature data regarding the UV-visible absorption bands
of gold coincide only in one point (i.e., that the absorption in
the range of 500-600 nm results from the plasmon resonance
of metallic gold particles), the band with maxima in the region
Figure 5 shows the diffuse-reflectance UV-visible spectra
between 200 and 800 nm of the CA4 sample dried under
vacuum at room temperature, and thermally treated in hydrogen
at 100, 150, and 200 °C. The spectrum of the sample dried at
room temperature under vacuum only presents an increasing
reflectance signal starting at about 400 nm. The spectra of the
samples treated in hydrogen exhibit a band around 550 nm, the
intensity of which increases with the treatment temperature. This
band is characteristic of the plasmon resonance of metallic gold
particles. Such a resonance appears when the wavelength of
of 240-320 nm was attributed to charge-transfer transitions in
δ+ 40-44
Au(I), Au(III) species, and gold charged clusters Aun
.
From these results it is assumed that for samples treated in air
at low temperature (e200 °C), both metallic and oxidized gold
species are present. It has been shown by CO adsorption
followed by FTIR studies that positive gold clusters are present

8564 J. Phys. Chem. B, Vol. 110, No. 17, 2006
Zanella et al.
(onset point 210 °C) associated with a low-intensity endothermal
peak occurs, which can be related to the evolution of chlorides
probably in the form of Cl2 or even HCl (hydrogen atoms could
be obtained from the amino groups). This assignment is
consistent with the stoichiometric mass loss expected for the
evolution of three chlorine atoms (plus eventually three more
hydrogen atoms) in [Au(en)2]Cl3, and with the occurrence of
an endothermal peak. Finally, another complex mass loss of
25% (onset point 400 °C, end point 580 °C) takes place along
with two exothermal peaks. Stoichiometrically this mass loss
is consistent with the evolution of ethylenediamine species
through a combustion process responsible for the exothermal
peaks. At higher temperature, further changes are observed, and
the residual mass (>40%) agrees with the expected one for
metallic gold.
Figure 7. Raman spectra of the gold precursor [Au(en)2]Cl3 and CA4
sample dried under vacuum at 25 °C.
Conclusions
Liquid-phase methods such as deposition-precipitation, co-
precipitation, and impregnation have not been effective to
deposit small gold particles (<5 nm) onto SiO2. In this work
we developed a new liquid-phase approach for the synthesis of
small gold particles by the adsorption of [Au(en)2]3+ cationic
complex on an SiO2 surface. We proposed and confirmed by
theoretical calculations that the deposition of gold effectively
takes place according to a cationic adsorption mechanism.
When the solution pH increases, the Au loading in Au/SiO2
samples increases in a proportional way and the average gold
particle size slightly increases. This can be explained by the
increase in the number of negatively charged species present
on SiO2 surface: when solution pH increases, a higher number
of [Au(en)2]3+ species can be adsorbed. Extending the adsorption
time also produced a slight increase in gold loading. In all cases
gold is not agglomerated on the SiO2 support. The morphologies
of gold nanoparticles were different depending on the particle
size. The particles with an average size of 2 nm exhibit mainly
fcc structures, whereas the particles of 3-5-nm size exhibit a
multiple twinning particle structure.
To decompose the [Au(en)2]3+ complex and generate gold
nanoparticles, the dried samples were thermally treated in
hydrogen or in air. According to UV-visible absorption
measurements, a major fraction of gold is present in the form
of metallic particles when samples were treated in hydrogen at
100 °C. For samples treated in air at low temperature (e200
°C), both metallic and oxidized gold species are present. During
the cationic complex adsorption and subsequent drying, gold is
not reduced, and the gold species deposited on SiO2 remain as
the [Au(en)2]3+ complex.
Figure 8. Thermogravimetric and differential scanning calorimetric
analysis of [Au(en)2]Cl3.
in Au/Y-zeolite samples prepared by cation exchange of
[Au(en)2]3+ after a calcination-reduction process.34
The absence of the plasmon band in the sample dried at room
temperature indicates that after preparation, and therefore during
drying, gold is not reduced and probably remains in the
oxidation state of the precursor, i.e., AuIII. Thus, the reduction
of [Au(en)2]3+ species on SiO2 begins at low temperatures
(<100 °C) even if the sample is treated in air.
Noteworthy is the change of color of the samples during
preparation/treatment, which is also an indication of the oxida-
tion state of gold. In the CA samples, the white color of the
room temperature dried sample changed to dark red-brown after
the reduction in hydrogen at 100 °C or higher temperatures.
This dark red color is typical for metallic gold particles on
SiO2.5,45,46 The Au/SiO2 samples also presented a reddish-brown
color after calcinations (in air) at temperatures above 80 °C.
The [Au(en)2]Cl3 and room tempeature-dried Au/SiO2 samples
were analyzed by Raman spectroscopy (Figure 7). Both spectra
are remarkably coincident when they are superimposed, showing
that the gold species deposited on SiO2 during CA is precisely
the [Au(en)2]3+ complex. This result corroborates the fact that
gold is not reduced during the CA process. A precise assignment
of the vibrational modes of solid amine compounds is frequently
complicated due to the band splitting.47 Given this circumstance,
the medium-intensity band at 1329 cm-1 as well as the weak
and broad band at 1212 cm-1 can be attributed to the stretching
ν(C-N) vibrations, while the relatively weak bands between
1000 and 800 cm-1 are likely related to the NH2 wagging/
bending vibrations. Finally, the high-intensity band at 584 cm-1
together with those at 482, 348, and 297 cm-1 can be attributed
to Au-N vibrations.48
The results presented here suggest a new and simple liquid-
phase method for the deposition of gold nanoparticles onto SiO2.
As we showed previously, the cation adsorption can produce
small gold particles on reducible supports such as TiO2. Now
this preparation method will allow a more reliable comparison
of catalytic activities of gold nanoparticles supported on
reducible and inert supports, prepared through the same
experimental approach. In addition these Au/SiO2 samples could
be used in studies of linear and nonlinear optics. We also expect
that the cation adsorption of [Au(en)2]3+ can be used for the
preparation of small gold particles on mesoporous materials such
as MCMs y SBAs.
Figure 8 shows the thermogravimetric analysis (TGA) data
of the complex [Au(en)2]Cl3 treated in air at a heating rate of
10 °C/min. Initially, a mass loss of 5% (onset point 60 °C) along
with a medium-intensity exothermal peak are observed, which
may be due to the desorption/decomposition of residual chemical
species coming from the synthesis of [Au(en)2]Cl3 (ethanol was
used for washing). After that, a continuous mass loss of 29%
Acknowledgment. We are thankful to R. Sato for Raman
spectra acquisition assistance and L. Rendon for HRTEM
observation assistance. The authors are also thankful to the
Central Microscopy facilities of the Institute of Physics, UNAM

Small Metal Particles of Au/SiO2
J. Phys. Chem. B, Vol. 110, No. 17, 2006 8565
(24) Bar, T.; Visart-de-Bocarme, T.; Nieuwenhuys, B. E.; Kruse, N.
Catal. Lett. 2001, 74, 127.
(25) Zanella, R.; Delannoy, L.; Louis, C. Appl. Catal. A 2005, 291, 62.
(26) Wolf, A.; Schu¨th, F. Appl. Catal. A 2002, 226, 1.
(27) Lin, S. D.; Bollinger, M.; Vannice, M. A. Catal. Lett. 1993, 17,
245.
for providing its microscope tools used in this work and the
CUDI project for financial support of the HRTEM work. V.A.B.
appreciates financial support from UNAM (grant DGAPA-
IN100303).
(28) Resch, R.; Meltzer, S.; Vallant, T.; Hoffmann, H.; Koel, B. E.;
Madhukar, A.; Requicha, A. S. G.; Will, P. Langmuir 2001, 17, 5666.
(29) Okumura, M.; Nakamura, S.; Tsubota, S.; Nakamura, T.; Azuma,
M.; Haruta, M. Catal. Lett. 1998, 51, 53.
(30) Schimpf, S.; Lucas, M.; Mohr, C.; Rodemerck, U.; Bru¨ckner, A.;
Radnik, J.; Hofmeister, H.; Claus, P. Catal. Today 2002, 72, 63.
(31) Parks, G. A. Chem. ReV. 1965, 65, 177.
References and Notes
(1) Haruta, M.; Kageyama, H.; Kamijo, N.; Kobayashi, T.; Delannay,
F. Stud. Surf. Sci. Catal. 1988, 44, 33.
(2) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chem. Lett. 1987,
2, 405.
(3) Haruta, M.; Yamada, N.; Kobayashi, T.; Iijima, S. J. Catal. 1989,
115, 301.
(32) Iler, R. The Chemistry of Silica; John Wiley and Sons: New York,
1979.
(4) Haruta, M. Catal. Today 1997, 36, 153.
(5) Bond, G. C.; Thompson, D. T. Catal. ReV. Sci. Eng. 1999, 41, 319.
(6) Haruta, M. Cattech 2002, 6, 102.
(33) Block, B. P.; Bailar, J. C., Jr. J. Am. Chem. Soc. 1951, 73, 4722.
(34) Guillemot, D.; Borovskov, V. Y.; Kazansky, V. B.; Polisset-Thfoin,
M.; Fraissard, J. J. Chem. Soc., Faraday Trans. 1997, 93, 3587.
(35) Guillemot, D.; Polisset-Thfoin, M.; Fraissard, J. Catal. Lett. 1996,
41, 143.
(36) Zanella, R.; Giorgio, S.; Henry, C. R.; Louis, C. J. Phys. Chem. B
2002, 106, 7634.
(37) Bond, G. C. Gold Bull. 2001, 34, 117.
(7) Horvath, D.; Toth, L.; Guczi, L. Catal. Lett. 2000, 67, 117.
(8) Schubert, M. M.; Hackenberg, S.; Veen, A. C. v.; Muhler, M.; Plzak,
V.; Behm, R. J. J. Catal. 2001, 197, 113.
(9) Valden, M.; Pak, S.; Lai, X.; Goodman, D. W. Catal. Lett. 1998,
56, 7.
(10) Zanella, R.; Giorgio, S.; Shin, C. H.; Henry, C. R.; Louis, C. J.
Catal. 2004, 222, 357.
(11) Zanella, R.; Louis, C.; Giorgio, S.; Touroude, R. J. Catal. 2004,
223, 328.
(12) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647.
(13) Valden, M.; Goodman, W. Isr. J. Chem. 1998, 38, 285.
(14) Boccuzzi, F.; Chiorino, A.; Manzoli, M.; Andreeva, D.; Tabakova,
T. J. Catal. 1999, 188, 176.
(15) Haruta, M.; Date´, M. Appl. Catal. A 2001, 222, 427.
(16) Bond, C. G.; Thompson, D. T. Gold Bull. 2000, 33, 41.
(17) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M.
J.; Delmon, B. J. Catal. 1993, 144, 175.
(18) Bollinger, M. A.; Vannice, M. A. Appl. Catal. B 1996, 8, 417.
(19) Tsubota, S.; Nakamura, T.; Tanaka, K.; Haruta, M. Catal. Lett.
1998, 56, 131.
(20) Kozlov, A. I.; Kozlova, A. P.; Liu, H.; Iwasawa, Y. Appl. Catal. A
1999, 182, 9.
(21) Liu, H.; Kozlov, A. I.; Kozlova, A. P.; Shido, T.; Iwasawa, Y. Phys.
Chem. Chem. Phys. 1999, 1, 2851.
(22) Guczi, L.; Peto¨, G.; Beck, A.; Frey, K.; Geszti, O.; Molna´r, G.;
Daeo´czi, C. J. Am. Chem. Soc. 2003, 125, 4332.
(23) Peto¨, G.; Molna´r, G. L.; Pa´szti, Z.; Geszti, O.; Beck, A.; Guczi, L.
Mater. Sci. Eng., C 2002, 19, 95.
(38) Zanella, R.; Louis, C. Catal. Today 2005, 107-108, 768.
(39) Guillemot, D. Ph.D. Dissertation, Universite´ Paris VI, 1997.
(40) Salama, T.; Shido, T.; Ohnishi, R.; Ichikawa, M. J. Phys Chem.
1996, 100, 3688.
(41) Pestryakov, A.; Tuzovskaya, I.; Smolentseva, E.; Bogdanchikova,
N.; Jentoft, F. C.; Knop-Gericke, A. Int. J. Mod. Phys. 2005, 19, 2321.
(42) Pestryakov, A. N.; Lunin, V. V.; Kharlanov, A. N.; Kochubey, D.
I.; Bogdanchikova, N.; Stakheev, A. Y. J. Mol. Struct. 2002, 642, 129.
(43) Smolentseva, E.; Pestryakov, P.; Bogdanchikova, N.; Simakov, A.;
Avalos, M.; Farias, M.; Diaz, J.; Gurin, V.; Tompos, A. Int. J. Mod. Phys.
2005, 19, 2496.
(44) Chen, W.; Zhang, J.; Cai, W. Scr. Mater. 2003, 48, 1061.
(45) Bond, G. C.; Sermon, P. A.; Webb, G.; Buchnan, D. A.; Wells, P.
B. Chem. Commun. 1973, 444.
(46) Sermon, P. A.; Bond, G. C.; Wells, P. B. J. Chem. Soc., Faraday
Trans. 1 1979, 75, 385.
(47) Pouchert, C. J. The Aldrich Library of Infrared Spectra, 3rd ed.;
Aldrich Chemical Co.: Milwaukee, WI, 1981.
(48) Socrates, G. Infrared Characteristic Group Frequencies, Tables
and Charts, 2nd ed.; Wiley: New York, 1994.