You are on page 1of 9

7634

J. Phys. Chem. B 2002, 106, 7634-7642

Alternative Methods for the Preparation of Gold Nanoparticles Supported on TiO2


Rodolfo Zanella, Suzanne Giorgio, Claude R. Henry, and Catherine Louis*,
Laboratoire de Re actiVite de Surface, UMR 7609 CNRS, UniVersite Pierre et Marie Curie,
4 place Jussieu, 75252 Paris Cedex 05, France, and CRMC2 CNRS, Campus de Luminy, case 913,
13288 Marseille Cedex, France
ReceiVed: December 10, 2001; In Final Form: May 23, 2002

The best current way to prepare Au/TiO2 catalysts is the method of deposition-precipitation with NaOH (DP
NaOH) developed by Haruta and co-workers. With this method, it is possible to obtain small gold metal
particles (2-3 nm), but the corresponding gold loading remains rather low (3 wt %). The main goal of this
work is to investigate other methods of preparation of Au/TiO2 catalysts to obtain small gold metal particles
(2-3 nm) and a higher Au loading. It is shown that anion adsorption with AuCl4- (AA) does not produce Au
loading higher than 1.5 wt % and the average particle size is not very small (4 nm). Cation adsorption with
Au(en)23+ (CA) leads to small particles (2 nm) when the solution/support contact time is moderate (1 h), but
the Au loading does not exceed 2 wt %. The most promising method of preparation appears to be depositionprecipitation with urea (DP urea). Indeed, samples with gold particles as small as those obtained with DP
NaOH (2 nm) can be prepared, and all gold in solution is deposited on TiO2 in contrast to DP NaOH. The
DP urea samples reported in this paper can reach a Au loading as high as 8 wt % using a TiO2 support with
a surface area of 45 m2 g-1. The possible mechanisms of deposition of gold on the TiO2 support by the
different methods of preparation are discussed.

I. Introduction
Gold metal becomes catalytically active in several chemical
reactions when it is finely divided and supported on metal
oxides.1-3 The most remarkable catalytic properties of supported
gold were observed by Haruta et al. in 19874-6 in CO oxidation
at subambient temperature. Since then, the most studied catalyst
has been gold supported on TiO2. Taken separately, Au and
the TiO2 support are catalytically inactive for this reaction, but
Au/TiO2 shows a drastic synergetic effect for the reaction of
CO oxidation. The optimum size of gold particles is smaller
than 5 nm for catalytic applications and about 2-3 nm for CO
oxidation.2,7,8 Such particle sizes can be achieved by a careful
control of the conditions of preparation.
The catalytic activity of Au/TiO2 for CO oxidation depends
on the preparation method; for example, Au/TiO2 catalysts
prepared by coprecipitation are less active than catalysts
prepared by deposition-precipitation.9,10 The parameters used
in the preparations are also important. For instance, for the
deposition-precipitation method, Haruta et al.3 described how
the catalytic activity is sensitive to gold concentration, pH and
temperature of the solution, calcination temperature, and addition
of magnesium citrate.
Haruta developed a preparation method of Au/TiO2 catalysts
by deposition-precipitation with NaOH as precipitating agent
(DP NaOH).11,12 With a nominal amount of 13 wt % of Au in
solution and within a pH range between 7 and 10, this method
permits the deposition of up to 3 wt % of Au and the formation
of small metal particles with an average size of about 3 nm.
Higher Au loading of 8 wt % could be achieved at pH 5.5, but
* To whom correspondence should be addressed. E-mail: louisc@
ccr.jussieu.fr.
Universite
Pierre et Marie Curie.
CRMC2 CNRS.
Associated with the Universities of Aix-Marseille II and III.

much larger particles were obtained (10 nm). It may be noted


that the amount of Au deposited on TiO2 by DP NaOH is always
lower than the amount of Au contained in solution, that is, the
yield of DP is lower than 100%.
It may be also noted that this method of preparation does
not exactly correspond to the principle of the method of DP,
largely developed by Geus13,14 and then extensively studied by
our group for the preparation of Ni/SiO2 samples.15-18 In the
DP method, the metal precursor is added to an aqueous
suspension of the support and subsequently precipitated as a
hydroxide by raising the pH. The surface of the support acts as
a nucleating agent, and this method, if it is properly performed,
leads to the greater part of the active precursor being attached
to the support. The key factor of this preparation is the
prevention of precipitation away from the support surface. The
method of deposition-precipitation developed by Geus et al.13,14
using urea (CO(NH2)2) as the precipitating base permits the
gradual and homogeneous addition of hydroxide ions throughout
the whole solution, CO(NH2)2 + 3H2O f 2NH4+ + CO2 +
2OH-, and avoids local increase in pH and the precipitation of
metal hydroxide in solution.
In the preparation of Au/TiO2 catalysts, one can also take
into account the fact that TiO2 is an amphoteric oxide (isoelectric
point, IEPTiO2 ) 6 19). Therefore, this oxide can be used for
preparation (i) by cation adsorption when the solution pH is
higher than IEPTiO2 (the main surface species is O-, so the TiO2
surface is negatively charged) and (ii) by anion adsorption when
the pH is lower than IEPTiO2 (the main surface species is OH2+,
so the TiO2 surface is positively charged).
Following these principles, we decided to prepare Au/TiO2
catalysts by three methods: (1) deposition-precipitation with
urea (DP urea) (Bond and Thompson,1 had suggested this
method, and Dekkers et al.20 had prepared Au/TiO2 samples
by this method. However, they obtained rather large gold

10.1021/jp0144810 CCC: $22.00 2002 American Chemical Society


Published on Web 07/13/2002

Small Metal Particles in Au/TiO2


particles with an average size of 7.5 nm for a gold loading of
4.5 wt %); (2) anion adsorption (AA) with AuCl4- complex;
(3) cation adsorption (CA) with Au(en)23+ complex (en )
ethanediamine) (this method was first successfully developed
by Guillemot et al.21,22 for the introduction of gold into Y
zeolites).
The goal of the study is to investigate whether it is possible
by these three methods to prepare Au/TiO2 catalysts with small
metal particles, in the same range of size as samples prepared
by DP NaOH (2-3 nm),3,11,23 but with a better control of the
Au loading, especially with higher Au loading. In this paper,
we show for the first time that deposition-precipitation with
urea (DP urea) can be successfully applied to the preparation
of Au/TiO2 catalysts with high metal loading and small Au
particle sizes.
For comparison, other Au/TiO2 samples were also prepared
by (i) deposition-precipitation with NaOH (DP NaOH) in the
presence or in the absence of magnesium citrate and (ii) incipient
wetness impregnation (Imp) with HAuCl4 (this is the very first
method reported in the literature for the preparation of supported
gold catalysts.28,29 However, the gold particle sizes are large
even at low metal loading. In addition, the samples contain large
amounts of chlorides, which are known to poison catalysis for
many reactions).
II. Experimental Section
1. Au/TiO2 Preparations. Titania Degussa P25 was used as
the support (BET surface area ) 45 m2 g-1, nonporous, 70%
anatase and 30% rutile, purity > 99.5%) and solid HAuCl4
3H2O (Acros) as the gold precursor. Before preparation, TiO2
was previously dried in air at 100 C for at least 24 h. All of
the preparations were performed in the absence of light, which
is known to decompose the gold precursors. For most of the
preparations, 1 g of TiO2 was added to 100 mL of an aqueous
solution of gold precursor (4.2 10-3 M). The amount of gold
in solution corresponds to a maximum gold loading of 8 wt %
on TiO2.
After deposition of gold onto TiO2 according to the various
methods described below, all the solids were submitted to the
same procedure: (i) separation from the precursor solution by
centrifugation (12 000 rpm for 10 min); (ii) washing (the solids
were suspended in water (100 mL g-1), stirred for 10 min at
RT, and centrifuged again. This washing procedure was repeated
four times to remove residual Cl- and Na+ ions as well as Au
species not interacting with the support); (iii) drying under
vacuum at 100 C for 2 h; (iv) calcination at 300 C (300 mg
of sample was heated in a flow (30 mL min-1) of industrial air
(Air Liquide) from room temperature to 300 C with a rate of
2 C min-1 then maintained at 300 C for 4 h. Calcination
treatment leads to the decomposition of the Au(III) complexes
into gold metal particles); (v) storage of the samples away from
light and under vacuum in a desiccator at RT. Indeed, a strong
increase in the average particle size of the calcined samples
was observed when samples are stored in air even for a short
period (for instance, from 1.8 to 3 nm for a sample left in air
for about 10 days). After several months of storage away from
light in a desiccator, the average particle size also slightly
increases (for instance, from 1.7 to 2.1 nm after 10 months of
storage). To avoid this effect, the samples have been stored after
drying, and calcination is performed when needed.
a. Deposition-Precipitation with NaOH. In the standard
preparation conditions (similar to Harutas preparations24), 100
mL of an aqueous solution of HAuCl4 (4.2 10-3 M) was
heated to 80 C. The gold concentration in solution corresponds

J. Phys. Chem. B, Vol. 106, No. 31, 2002 7635


to a theoretical Au loading of 8 wt % in the case of a complete
deposition-precipitation (DP yield ) 100%). This gold loading
was chosen because Haruta23 reported that the most active
catalyst for CO oxidation contained 8 wt % of Au. The pH was
adjusted to 8 by dropwise addition of NaOH (1 M), and then 1
g of TiO2 was dispersed in the solution, and the pH was
readjusted to 8 with NaOH. The suspension thermostated at 80
C was vigorously stirred for 1 h and then centrifuged, and the
solid was washed, dried, and calcined following the previously
reported procedure. The main parameters studied were the DP
time (1, 2, 4, or 16 h), the addition of magnesium citrate (Mg3(C6H5O7)2, 6.9 10-3 M) in the suspension after the first
adjustment of pH, and the HAuCl4 concentration.
b. Incipient Wetness Impregnation. Titania was impregnated
with aqueous solutions of HAuCl4 (1.3 mL per g of TiO2) of
various concentrations (0.04, 0.08, and 0.16 M) to obtain
samples with 1, 2, and 4 wt % of Au, respectively. In all cases,
the solution pH was less than 1. The samples were aged at room
temperature (RT) for 1 h. Samples (2 and 4 wt %) were then
divided into two parts. One part was directly dried (Imp
samples), whereas the other one was washed before drying
(ImpW samples) to determine whether some Au species interact
with the TiO2 support. All of the samples were calcined at 300
C.
c. Anion Adsorption. A total of 1 g of TiO2 was added to
100 mL of an aqueous solution of HAuCl4 (4.2 10-3 M).
Under such conditions, the solution pH (2) was lower than
the IEPTiO2, that is, in adequate conditions of pH for anion
adsorption. The suspension thermostated at 25 or 80 C was
vigorously stirred for 15 min, 1 h, or 15 h and finally centrifuged. The solids were washed, dried, and calcined at 300 C.
d. Cation Adsorption. Gold was also deposited on TiO2 by
cation adsorption of the Au(en)23+ complex (en ) ethanediamine) the synthesis of which was described by Block and
Bailar.25 Au(en)2Cl3 was dissolved in 100 mL of water (4.2
10-3 M). By dropwise addition of an ethanediamine solution
(1 M), the pH was adjusted to a value of 9.4 or 10.3, that is, at
a pH higher than the IEPTiO2. Hence, the adsorption of the Au(en)23+ complex is theoretically possible. The suspension was
vigorously stirred for 1 or 16 h at 80 C in a thermostated vessel.
After the adsorption, the samples were centrifuged, washed,
dried under vacuum, and calcined at 300 C.
e. Deposition-Precipitation with Urea. In the so-called
standard preparation conditions, 1 g of TiO2 was added to 100
mL of an aqueous solution of HAuCl4 (4.2 10-3 M) and of
urea (0.42 M). The initial pH was 2. The suspension
thermostated at 80 C was vigorously stirred for 4 h (pH
increases) and then centrifuged, washed, dried, and calcined at
300 C. The following parameters were studied: the DP time
(1, 2, 4, 16, and 90 h), the temperature of DP (80 and 90 C),
the gold concentration (1.1 10-3, 1.6 10-3, and 4.2 10-3
M), the urea concentration (0.42 and 0.84 M), and the addition
of magnesium citrate (6.9 10-3 M).
2. Techniques of Characterization. Chemical analysis of
Au, Cl, Mg, Na, C, and N in the samples was performed by
inductively coupled plasma atom emission spectroscopy at the
CNRS Center of Chemical Analysis (Vernaison, France). The
detection limit is 300 ppm for Cl and 1000 ppm for C and N.
Chemical analysis was performed after sample calcination. The
Au weight loading of the samples is expressed in grams of Au
per grams of sample calcined at 1000 C: wt % Au ) [mAu/
(mAu + mTiO2)] 100.
Calcined Au/TiO2 samples were examined by transmission
electron microscopy (TEM) with a JEOL 2000FX electron

7636 J. Phys. Chem. B, Vol. 106, No. 31, 2002

Zanella et al.

TABLE 1: Au/TiO2 Samples Prepared by


Deposition-Precipitation with NaOH at 80 C
preparation

results

DP
theor Au Au
Cl
average standard particle size
time
loading loading loading particle deviation distribution
sample (h) pH (wt %) (wt %) (wt %) size (nm) (nm)
(nm)
DPN1
1
DPN2
2
DPN3
4
DPN4 16
DPN5a 1
DPN6
1
DPN7
1
a

8
8
8
8
8
7
8

8
8
8
8
8
8
16

1.8
2.4
2.1
2.4
3.1
3.3
3.6

0.026
0.025
0.035
0.077
0.09
0.031
0.030

1.8
1.5
1.6
1.5
1.4
1.4
3.3

0.62
0.40
0.36
0.30
0.36
0.34
1.92

0.7-4.3
0.7-3.5
1.0-3.1
1.0-2.4
0.7-2.7
0.7-2.4
0.7-16.0

Addition of magnesium citrate (6.9 10-3 M).

microscope. Except when especially mentioned, the histograms


of the metal particle sizes were established from the measurement of 300 to 1000 particles. The size limit for the detection
of gold particles on TiO2 is 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 formula )
[((di - dh)2)/ni]1/2. In some samples, the presence of Au was
also determined by energy-dispersive X-ray spectroscopy coupled
to TEM observations.
III. Results
1. Deposition-Precipitation with NaOH. Table 1 shows that
when the pH of the DP NaOH solution is 8 (DPN1 sample),
the average Au particle size is 1.8 nm and the Au loading is
1.8 wt % (Figure 1). When the pH is lowered to 7 (DPN6
sample), the average particle size is smaller (1.4 nm) and the
Au loading is higher (3.3 wt %). Another parameter studied is
the DP time, from 1 to 16 h (samples DPN1 to DPN4). The
gold loading slightly increases from 1.8 to 2.4 wt %, while the
average particle size seems to slightly decrease from 1.8 to 1.5
nm (Table 1). The particle size distribution becomes narrower.
It may be noted that our preparations can provide smaller
gold particles than those obtained by Haruta: at pH 8, 1.8 nm
for 1.8 wt % of Au (DPN1, Table 1) to be compared to 2.9 nm
for 2.2 wt %,12 and at pH 7, 1.4 nm for 3.4 wt % (DPN6) to be
compared to 3.3 or 3.6 nm for 3.6 wt %.12 As for Harutas
samples, the Au loading in our samples is lower than the
nominal amount of gold in solution, indicating that the yield of
DP is less than 100% and that part of the gold is not deposited
on TiO2. When the concentration of gold in solution is higher
(8.4 10-3 M), the gold loading on TiO2 is higher (3.6 wt %
in DPN7) but the average particle size is much larger (3.3 nm)
and the size distribution much broader.
A sample was also prepared in the presence of magnesium
citrate. According to Haruta et al.,3,11,23 magnesium citrate leads
to higher Au loadings and small metal particles because it sticks
on the TiO2 surface and prevents gold particles from sintering
during calcination. Magnesium citrate is also known to reduce
the gold(III) precursors into metallic gold.11,26 Table 1 shows
that when magnesium citrate is added (DPN5), the Au loading
is indeed higher (3.1 compared to 1.8 wt % for DPN1), and the
average particle size is smaller (1.4 instead of 1.8 nm) (Figure
2).
The gold particles in DPN5 are also smaller than those
obtained by Haruta et al. under close experimental conditions:
1.4 nm to be compared to 2.8-3.8 nm.27 However, DPN5 does
not reach gold loadings as high as 6 or 8 wt % as reported by
Haruta in two papers.23,27 However, in the first one,27 it is not
clear whether the Au loadings reported are those in solution or

Figure 1. (a) TEM image of calcined sample DPN1 prepared by


deposition precipitation with NaOH (DP time ) 1 h, pH ) 8, 1.8 wt
% Au); (b) size histogram of gold particles.

on TiO2, and in the second one,23 the amount of gold in solution


is not reported.
2. Incipient Wetness Impregnation. Table 2 summarizes
the results obtained for the Au/TiO2 samples prepared by
impregnation and by impregnation followed by washing. All
of the Imp samples contain only a few large Au particles (>10
nm) after calcination (Table 2). These results are consistent with
published results. Indeed, Haruta et al. 28 obtained gold particles
between 10 and 30 nm for 1 wt % Au/TiO2 catalysts, and
Vannice et al. 29 obtained gold particles between 25 and 35 nm
for 2 wt % Au/TiO2 sample. Although X-ray fluorescence and
chemical analyses indicate the presence of Au, only a small
number of Au particles could be observed by TEM in the Imp
samples, so the values for the particle size distribution are not
reliable. This problem had already been reported by Haruta et
al.7
Chemical analysis shows that about 0.9 wt % of gold remains
on TiO2 after washing whatever the initial Au loading, and the
average particle size is between 3 and 5 nm after calcination
(Table 2). Hence, part of the gold species deposited during
impregnation is in strong interaction with TiO2 because it
remains on the support after washing. Because of this strong
interaction, sintering of gold particles is prevented during
calcination, so the metal particles are smaller than those obtained
after impregnation.
It may be noted that although the Au loading of Imp3 is twice
that of Imp2, both the Au loading and the average particle size
of ImpW3 are smaller than those of ImpW2. We did not attempt

Small Metal Particles in Au/TiO2

J. Phys. Chem. B, Vol. 106, No. 31, 2002 7637


TABLE 3: Au/TiO2 Samples Prepared by Anion Adsorption
with AuCl4results
preparation

Au
Cl
average standard
time of T
loading loading particle deviation
sample contact (C) pH (wt %) (wt %) size (nm) (nm)
AA1 15 h

25 2.5

1.0

0.16

5.7

1.38

AA2 15 min 80 2

1.3

0.16

3.7

1.30

AA3 1 h

80 2

1.0

0.07

5.6

1.50

AA4 15 h

80 2

1.5

0.15

4.4

1.29

particle size
distribution
(nm)
5.0-7.0
(4 particles)
1.7-6.8
(67 particles)
2.0-8.3
(23 particles)
2.0-7.5
(60 particles)

TABLE 4: Au/TiO2 Samples Prepared by Cation


Adsorption with Au(en)23+ at 80 C
preparation

results

time of
Au
Cl
average standard
contact
loading loading particle deviation
sample (h)
pH (wt %) (wt %) size (nm) (nm)
CA1

9.4

1.1

2.1

0.66

CA2

16

9.4

6.1

0.03

4.1

1.01

CA3

10.3

1.7

0.04

1.8

0.33

CA4

16

10.3

6.4

0.08

4.6a

2.74

particle size
distribution
(nm)
1.0-3.5
(134 particles)
1.7-6.4
(140 particles)
1.0-2.5
(88 particles)
1.7-15.8
(157 particles)

a Plus some very large particles (80-150 nm) not taken into account
in the calculation of the average size.

Figure 2. (a) TEM image of calcined sample DPN5 prepared by


deposition precipitation with NaOH (DP time ) 1 h, pH ) 8, citrate
addition, 3.1 wt % Au); (b) size histogram of gold particles.

TABLE 2: Au/TiO2 Samples Prepared by Incipient Wetness


Impregnation at RT
preparation

sample
Imp1
Imp2
Imp3
ImpW2a
ImpW3a
a

results

theor Au
Au
Cl
average standard
loading loading loading particle deviation
(wt %) (wt %) (wt %) size (nm) (nm)
1
2
4
2
4

0.9
0.6

0.09
0.08

>10
>10
>10
4.9
2.7

1.36
1.52

particle size
distribution
(nm)

1.9-8.5 (165 particles)


1.4-5.8 (78 particles)

The ImpW samples are washed after impregnation.

to repeat these experiments to check these unexpected results


because the main goal of these experiments was to determine
whether some gold remained in interaction with the support after
washing.
3. Anion Adsorption. Because some gold species (e0.9 wt
%, Table 2) remain adsorbed on the TiO2 surface after washing
of the impregnated samples, preparations by anion adsorption
with AuCl4- have been attempted. Table 3 shows that under
our experimental conditions the Au loading does not exceed
1.5 wt % and the average particle size is about 4-6 nm whatever
the samples. As in the case of Imp samples, only a few particles
are observed by TEM, so the average size measurements are
not very accurate. However, it can be mentioned that (i) the
gold loading is higher when anion adsorption is performed at
80 C (AA4) rather than at RT (AA1) and (ii) the equilibrium
adsorption seems to be reached fast (<15 min) because beyond
this time the Au loading does not further increase.

As for the ImpW samples, the solution pH is ,IEPTiO2 and


the samples are washed. The particle sizes (4-6 nm) obtained
in the AA samples are not very different from those obtained
in the ImpW samples (Table 2), but it is possible to reach
slightly higher Au loading.
4. Cation Adsorption. Table 4 shows that for the samples
prepared at pH 9.4, when the adsorption time is short (1 h,
sample CA1), the gold particles are small (2 nm) and the Au
loading is small as well (1.1 wt %) (Figure 3). The increase of
the adsorption time (from 1 to 16 h) leads to a significant
increase in the Au loading because almost all of the Au in
solution is deposited (6.1 wt %, sample CA2), but the average
particle size is also much larger (4 nm).
When the pH of the preparation is 10.3, (i) for the short
adsorption time the Au loading is higher (1.7 wt %, CA3) than
at pH 9.4 (1.1 wt %, CA1), the average particle size is smaller
(1.8 instead of 2.1 nm), and the size distribution is narrower
(Table 4) and (ii) for the long adsorption time the gold loading
is the same as for pH 9.4 but the average particle size is larger
and the size distribution is much broader (CA4). It is worth
noting that en is fully decomposed during calcination at 300
C (<0.1 wt % of C, that is, below the detection limit).
5. Deposition-Precipitation with Urea. The results reported
in Table 5 first show that all of the gold in solution has
precipitated on TiO2 within the first hour of DP (DPU1). Once
all of the gold has precipitated, the samples mature. Indeed,
the increase in the DP time induces a narrowing in the particle
size distribution and a decrease in the average particle size, rather
drastic between 2 and 4 h (5.2 to 2.7 nm, Figure 4) and then
moderate between 4 and 90 h (2.7 to 2.4 nm) (DPU3 to DPU5).
It may be noted that the final pH of the solution increases with
the DP time (Table 5).
Slightly smaller average particle size, narrower size distribution, and higher solution pH are also obtained (Table 5) when
(i) the DP temperature is set at 90 instead of 80 C (DPU6 and
DPU3, Figures 4 and 5), (ii) the urea solution is twice more
concentrated (DPU7) than that of DPU3, and (iii) magnesium

7638 J. Phys. Chem. B, Vol. 106, No. 31, 2002

Zanella et al.

Figure 3. (a) TEM image of calcined sample CA1 prepared by cation


adsorption with Au(en)23+ complex (CA time ) 1 h, pH ) 9.4, 1.1 wt
% Au); (b) size histogram of gold particles.

citrate is added to the suspension (DPU8). When the gold


concentration is lower (DPU10 and DPU9), 100% of the gold
is still deposited on TiO2 and the average particle size is
smaller: 2 nm for 2 wt % Au (DPU9) and 2.3 nm for 3 wt %
Au (DPU10) instead of 2.7 nm for 8 wt % Au (DPU3).
It may be noted that our preparations by DP urea provide
much better results than the very first preparations performed
by Dekkers et al.20 They report particles with an average size
of 7.5 nm for a gold loading of 4.5 wt % in a Au/TiO2 sample
prepared by DP urea at 80 C (pH ) 8.5) and then washed,

Figure 4. (a) TEM image of calcined sample DPU3 (DP time ) 4 h,


T ) 80 C, 7.7 wt % Au); (b) size histogram of gold particles.

dried at 80 C, and calcined at 400 C. The reason for the


differences in the particle size is not straightforward because
some preparation parameters are missing and it is well-known
that every step of preparation may have an influence on the
final state of the catalysts. One relevant point is that the pH
reached by Dekker et al. is higher than ours.
If one compares the DP urea samples (Table 5) to our own
DP NaOH samples (Table 1), one can note that the average
particle sizes are slightly larger and the size distributions slightly
broader, except for the preparation performed at 90 instead of
80 C (DPU6 in Table 5). Samples of each type of preparation,

TABLE 5: Au/TiO2 Samples Prepared by Deposition-Precipitation with Urea


preparation

sample

DP
time
(h)

T
(C)

DPU1
DPU2
DPU3
DPU4
DPU5
DPU6
DPU7c
DPU8d
DPU9
DPU10

1
2
4
16
90
4
4
4
4
4

80
80
80
80
80
90
80
80
80
80

results

pHa

theor Au
loading
(wt %)

Au
loading
(wt %)

Cl
loading
(wt %)

average
particle
size (nm)

standard
deviation
(nm)

particle size
distribution
(nm)

2.99
6.26
7.04
7.35
7.84
7.27
7.21
6.3
7.1
7.0

8b
8
8
8
8
8
8
8
2
3

7.8
6.5
7.7
6.8
7.4
7.1
7.1
7.7
2.0
3.4

0.041
0.122
<0.03
<0.03
<0.03
0.045
<0.03
0.04
<0.03
0.079

5.6
5.2
2.7
2.5
2.4
1.7
2.0
1.5
2.0
2.3

1.66
1.10
0.90
0.74
0.70
0.35
0.60
0.29
0.52
0.61

2.3-10.2
2.0-7.5 (135 particles)
0.7-5.5
1.0-5.1
0.7-5.0
0.7-3.7
0.7-4.1
0.7-2.4
1.0-4.0 (85 particles)
1.0-4.4

a pH at the end of the preparation. b The theoretical Au loading corresponds to the Au loading that could be reached if all of the gold in solution
was deposited on TiO2 (100% yield of DP); it is in fact comprised between 7.4 and 7.7 wt %. c Twice higher [urea] ) 0.84 M. d Addition of
magnesium citrate (6.9 10-3 M).

Small Metal Particles in Au/TiO2

J. Phys. Chem. B, Vol. 106, No. 31, 2002 7639


2. Anion Adsorption. Under our conditions of preparation
(RT or 80 C, 4.2 10-3 M of Au, and pH 2), the main
species in solution is AuCl3(OH)- at RT34 and remains the same
up to 150 C.37 Because the solution pH is lower than IEPTiO2
(6), AuCl3(OH)- could electrostatically interact with the TiO2
surface. However, according to the literature data related to the
adsorption of gold hydroxy chlorides on various oxides,38-43
the mechanism of adsorption of gold hydroxy chlorides is not
an electrostatic interaction but the formation of a surface
complex by reaction with surface OH, that is, the formation of
an inner-sphere complex such as

TiOH + AuCl3(OH)- T TiOAuCl2 + H2O + Cl- (I)

Figure 5. (a) TEM image of calcined sample DPU6 (DP time ) 4 h,


T ) 90 C, 7.1 wt % Au); (b) size histogram of gold particles.

DP NaOH (DPN1) and DP urea (DPU6), with the same average


small particle size were recently tested in the reaction of CO
oxidation (25 mg of catalyst, T ) 5 C, 1% CO, and 4% O2 in
N2, 99.3 cm3 min-1).30 The results show that the activity
expressed in molCOconverted gcat-1 s-1 is much higher for DPU6
than for DPN1. This result is fully consistent with the higher
Au loading in the DP urea samples. More interesting is the fact
that DPN1 and DPU6 show the same activity when it is
expressed in molCOconverted molAu-1 s-1. This result confirms that
the DP urea method leads to the same dispersion of gold onto
titania as DP NaOH and that it is an outstanding method of
preparation of Au/TiO2.
IV. Discussion
Although this is not the main goal of the present paper, a
discussion on the mechanism of deposition of gold on TiO2
during the various preparations described above is attempted,
on the basis of literature data only, because no characterization
studies of the Au/TiO2 samples before calcination have been
performed up to now.
1. Impregnation and Impregnation-Washing. This preparation was performed at RT with gold solutions at pH < 1 and
concentrations between 0.04 and 0.16 M. Under these conditions, the main gold species in solution is AuCl4-.31-36 Because
at this low pH the TiO2 surface is positively charged, some
AuCl4- can electrostatically interact with the TiO2 surface. This
would explain that part of the Au remains on the support after
washing (1 wt %, Table 2).

For instance, Nechayev et al.41 reported that the maximum of


Au adsorption on alumina occurs at pH close to the IEPAl2O3
) 8, that is, at a pH at which the number of neutral OH is
maximum. Machesky et al.38 reported that the adsorption of gold
hydroxy chloride on goethite (IEPFeOOH ) 8.1) increases as pH
increases from 4 to 7, which is opposite to typical behavior for
anion adsorption on positively charged oxide surfaces. This
retrograde adsorption is attributed to (i) a shift in Au speciation
when pH increases from AuCl4- at pH 4 to AuCl(OH)3- at pH
7 and AuCl3OH- and AuCl2(OH)2- at intermediate pH and (ii)
the fact that gold hydroxy chloride complexes preferentially react
with the OH of the support and all the more easily because the
amount of substituted OH in the gold coordination sphere
increases.
Because the amount of neutral OH surface species on TiO2
is low at low pH, this could explain that the amount of gold
adsorbed is low in the AA samples (e1.5 wt %, Table 3). In
addition, the fact that the Au loading does not depend on the
solution/support contact time and on the temperature (Table 3),
that is, that gold adsorption is fast, is in agreement with a
mechanism of gold adsorption via a surface complex formation.
If this interpretation is correct, this also means that (i) the nature
of the interaction between Au species and TiO2 is different from
that occurring during impregnation and (ii) the term anion
adsorption used to designate this preparation procedure is
improper because it does not reflect the phenomenon of surface
complex formation.
3. Cation Adsorption. In the case of cation adsorption with
Au(en)23+, the solution pH (9.4 or 10.3) is higher than IEPTiO2.
If one assumes that all of the surface OHs are deprotonated
and that each Au(en)23+ interacts with 3 O- of the TiO2 surface
to compensate the surface charge, a maximum gold loading of
3 wt % is expected (TiO2 Degussa support contains 6 OH/
nm2 44). Because a higher Au loading is reached (Table 4), this
indicates that another chemical phenomenon rather than cation
adsorption is probably involved, but more characterization is
needed to elucidate this point.
4. Deposition-Precipitation with NaOH. Haruta et
al.4-7,9-12,23,24,27,28 reported only a few pieces of information
concerning the possible mechanism of deposition-precipitation
of Au on TiO2 with NaOH. In ref 12, they showed that the
amount of gold deposited reaches a maximum at pH 6, at the
pH of IEPTiO2, that is, when the surface charge is neutral. In
that paper, they reported that the number of Au particles is
almost constant within the pH range of 6-8. They concluded
that gold is deposited on specific sites of the TiO2 surface. From
XANES measurements, they concluded that Au is not metallic
but bound to oxygen, most probably as gold hydroxide. They
proposed that this gold hydroxide is deposited from Au(OH)3Clon specific sites of TiO2, which may act as nucleation sites for
Au(OH)3. The increasing amount of Au with increasing pH up

7640 J. Phys. Chem. B, Vol. 106, No. 31, 2002

Figure 6. pH versus time of deposition-precipitation with urea at 80


C.

to 6 can be accounted for by a decrease in the ion exchange


capacity or a decrease in ion interaction between TiO2 and the
Au species due to the decrease in the positive charge or both.
The decreasing trend when pH increases above 6 is explained
by the increasing solubility of Au(OH)3 with pH.
In our samples, the Au loading barely changes with the DP
time nor does particle size (Table 1). This indicates that, as in
the case of anion adsorption, gold adsorption is fast. However,
the solution pH is much higher (7 or 8 instead of 2), so gold
speciation in solution is different, probably AuCl2(OH)2- or
AuCl(OH)3- or both at 80 C, according to Murphy et al.37
The TiO2 surface charge is different as well; it is negatively
charged. It may be noted that electrostatic adsorption of anionic
gold complexes cannot occur. Our interpretation is that a gold
surface complex can form (Ti-O-AuCl2) in the same way as
in eq I:

TiOH + AuCl2(OH)2- T TiOAuCl2 + H2O + OH- (II)


This is consistent with Harutas results:12 (i) the amount of gold
deposited reaches a maximum pH 6, that of the IEPTiO2, that is,
when the surface charge is neutral; (ii) Au is bound to oxygen
atoms (XANES data), provided that these oxygens are those of
the TiO2 surface (Au-O-Ti) and not those of gold hydroxide.
It may be noted that in the literature data on the interaction of
gold chloride with oxide surfaces at various pHs,38-43,45 the
authors never mention the possibility of the formation of gold
hydroxide. In addition, when one considers the chemistry of
HAuCl4 in aqueous solution,46,47 AuCl4- is hydrolyzed, and
hydroxochloro complexes form by replacement of Cl- ligands
by OH- ligands when pH increases without formation of gold
hydroxide.
5. Deposition-Precipitation with Urea. It is proposed that
the first chemical step occurring during DP urea preparation of
Au/TiO2 is surface complex formation, as for anion adsorption
or DP NaOH. When the DP time increases and before the
solution pH reaches IEPTiO2 (Figure 6), the positive surface
charge decreases, there are more and more neutral surface OHs,
and so more and more sites for the formation of gold surface
complex (retrograde adsorption, see section IV.2.).
However, this reasoning does not explain the fact that all of
the gold precipitates during DP urea and not during DP NaOH.
A second chemical phenomenon must be involved. If one refers
to literature data related to adsorption of gold hydroxy chlorides
on reducible oxides, this second chemical phenomenon could
be the formation of gold colloids. For instance, Greffie et al.45

Zanella et al.
identified two forms of gold species on iron oxides (ferrihydrite
and goethite) after coprecipitation: metallic gold colloids and
Au(III) species, which corresponded to 1-5% of all the
deposited gold). The formation of gold colloids was attributed
to oxidation-reduction reactions between Au(III) and Fe(II)
present as traces: adsorption of gold on oxidizable mineral
surfaces induces the reduction of Au(III) to elemental gold,
which is subsequently sorbed on oxides. Indeed, colloidal gold
particles are negatively surface-charged,48,49 so they can react
with positively charged iron oxides by electrostatic interaction
(IEP ferrihydrite and goethite ) 8-9).
It is highly probable that TiO2 can also reduce gold because
it usually contains defects of Ti3+ ions. Let us then form the
hypothesis that gold colloids can form. The difference in gold
loading between DP NaOH samples and DP urea samples can
be explained by the differences in solution pHs. Because gold
colloids are negatively charged on the surface48,49 and because
the TiO2 surface is also negatively charged during DP NaOH
preparations because the solution pH (8 or 7) is higher than
IEPTiO2, gold colloids could not interact with the TiO2 surface
but would be released into the solution. Thus, only the Au(III)
surface complex would be present on TiO2. This interpretation
is consistent with the low Au loading of the samples and with
Harutas XANES results,12 which show that the gold is not
metallic.
For DP urea, all of the gold is deposited on the TiO2 during
the first hour (Table 5). It can be inferred that gold colloids
can interact with the TiO2 surface, which is positively charged
before the pH reaches IEPTiO2. In consequence, gold species in
the DP urea samples would be a mixture of Au(III) surface
complexes and Au(0) colloids. UV-visible spectroscopy of the
samples gathered after centrifugation and washing (before drying
to avoid possible decomposition of Au(III) into Au(0)) should
allow us to check these hypotheses. Detection of gold colloids
should also be attempted in the solution of DP NaOH gathered
after centrifugation.
Table 5 also shows that as the pH increases, the average
particle size becomes smaller; this is observed when (i) the DP
time increases (DPU1 to DPU5), (ii) the DP temperature
increases from 80 to 90 C (DPU2 and DPU6), and (iii) the
urea concentration is twice higher (DPU2 and DPU7). If we
suppose that the particle size measured after calcination reflects
the size of gold colloids before calcination, it can be inferred
that the increase in pH as the DP time increases and the fact
that urea dissociates faster at 90 than at 80 C and that more
urea dissociates when the urea concentration is higher have an
influence on the kinetics of redispersion of gold colloids
(because all of the gold is deposited within the first hour of
DP). Such a phenomenon of redispersion of gold colloids was
reported in a study on the formation of gold colloids in solution
by reduction of gold chloride by sodium citrate.50 It showed
that the first particles formed were large and had a fluffy
morphology, then they shrink over the course of the reaction.
The authors proposed that particles may nucleate in a short burst,
agglomerate weakly, and continue to grow within the agglomerates. As the reaction proceeds, the large agglomerates fall apart,
giving rise to a continual increase in the number of small
particles. To the light of this interpretation, the mechanism of
DP urea can be revisited. It is proposed that the formation of
gold surface complexes and colloids takes place simultaneously.
At the beginning of the DP, the pH is low (Figure 6) and there
are few OH sites for the formation of gold surface complex
and therefore few nucleation sites for colloid growth. As DP
proceeds and pH increases, more and more surface OH sites

Small Metal Particles in Au/TiO2


are available, so more and more gold surface complexes form,
and there are more and more sites for colloid redispersion.
In conclusion, only on the basis of literature data, we can
infer that the chemical phenomena occurring during DP NaOH
and DP urea are different from each other and also from the
mechanism described by Geus et al.13,14 and Burattin et al.15
(see Introduction) for the preparation of Ni/SiO2 catalysts by
DP urea. In contrast to nickel, no gold hydroxide can precipitate
on oxide supports according to the literature data, and the gold
loading does not depend on the DP time (Tables 1 and 5). In
addition, upon solution basification during DP urea, the TiO2
surface charge switches from positive to negative when the pH
passes the IEPTiO2 at 6 (Figure 6), whereas in the case of silica,
the pH was always higher than its IEPSiO2 (2). In consequence,
the gradual basification of the gold solution upon urea decomposition is not performed to avoid hydroxide precipitation in
solution as in the case of the preparation of Ni/SiO2 by DP urea
but to favor the deposition of all of the gold as Au(III) and
gold colloids and then the redispersion of gold colloids before
the pH reaches IEPTiO2. It may be noted that the term
deposition-precipitation used to designate this preparation
procedure is improper because it does not reflect the chemical
phenomena occurring during preparation.
V. Conclusion
This paper shows that the method of preparation by deposition-precipitation with NaOH developed by Haruta et al.10 gives
slightly smaller metal particles than those obtained in their
previous studies (e2 nm with Au loading of 3 wt %).
However, the yield of DP was always lower than 100%.
In the present work, deposition-precipitation with urea is
successfully used for the preparation of Au/TiO2. Comparison
of the results obtained by DP NaOH and DP urea shows that
the DP urea is a promising method of preparation. Indeed, (i)
the gold particles are as small as those obtained by Haruta, 2
nm average particle size and (ii) the yield of DP is about 100%,
so the Au loading can reach much higher values, as high as 8
wt % for a TiO2 support with a surface area of 45 m2 g-1. It is
even likely that higher Au loading could be reached with more
concentrated gold solutions.
On the other hand, preparation of Au/TiO2 samples by
impregnation with HAuCl4 followed by washings revealed that
part of the gold is in strong interaction with TiO2. This led us
to attempt to prepare Au/TiO2 samples by anion adsorption with
AuCl4-. In fact, the gold complex is AuCl3OH- under the
conditions of preparation. The Au loading in these samples was
always lower than 1.5 wt %, and the average particle size was
not very small (4 nm). The interaction between gold and TiO2
is different between the samples prepared by impregnationwashing and anion adsorption because of the different pHs and
gold concentrations of the solutions, that is, because of the
different gold speciation. AuCl4- can electrostatically interact
with the TiO2 surface in the samples prepared by impregnationwashing, while AuClxOHy- can form a surface complex with
TiO2 in the samples prepared by anion adsorption.
In an effort to exploit to the amphoteric properties of the
TiO2 support (IEP 6), cation adsorption with the Au(en)23+
complex was also attempted. Small particles were obtained (2
nm) when the solution/support contact time was moderate (1
h). However, the Au loading did not exceed 2 wt %.
Acknowledgment. Rodolfo Zanella is indebted to CONACYT (Mexico) and SFERE (France) for his Ph.D. grant, and to
FESC, UNAM. We all thank the Groupement De Recherche

J. Phys. Chem. B, Vol. 106, No. 31, 2002 7641


CNRS Interfaces et Surfaces Sensibles a` la Structure for
supporting Zanellas travel expenses between Paris and Marseille.
References and Notes
(1) Bond, G. C.; Thompson, D. T. Catal. ReV.-Sci. Eng. 1999, 41,
319.
(2) Cosandey, F.; Madey, T. E. Surf. ReV. Lett. 2001, 8, 73.
(3) Haruta, M. The Abilities and Potential of Gold as a Catalyst; Report
No. 393; The Osaka National Research Institute: Osaka, Japan, 1999; pp
1-93.
(4) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chem. Lett. 1987,
2, 405.
(5) Haruta, M.; Kobayashi, T.; Iijima, S.; Delannay, F. Proc. 9th Int.
Congr. Catal., Calgary 1988, 1206.
(6) Haruta, M.; Saika, K.; Kobayashi, T.; Tsubota, S.; Nakahara, Y.
Chem. Express 1988, 3, 159.
(7) Bamwenda, G. R.; Tsubota, S.; Nakamura, T.; Haruta, M. Catal.
Lett. 1997, 44, 83.
(8) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647.
(9) Haruta, M.; Kageyama, H.; Kamijo, N.; Kobayashi, T.; Delannay,
F. Stud. Surf. Sci. Catal. 1988, 44, 33.
(10) Haruta, M. Catal. SurV. Jpn. 1997, 61.
(11) Tsubota, S.; Haruta, M.; Kobayashi, T.; Ueda, A.; Nakahara, Y.
Stud. Surf. Sci. Catal. 1991, 72, 695.
(12) Tsubota, S.; Cunningham, D. A. H.; Bando, Y.; Haruta, M. Stud.
Surf. Sci. Catal. 1995, 91, 227.
(13) Hermans, L. A. M.; Geus, J. W. Stud. Surf. Sci. Catal. 1979, 4,
113.
(14) van Dillen, J. A.; Geus, J. W.; Hermans, L. A.; van der Meijden,
J. In Proceedings of the 6th International Congress on Catalysis, London,
1976; Bond, G. C., Wells, P. B., Tompkins, F. C., Eds.; The Chemical
Society: London, 1977; p 677.
(15) Burattin, P.; Che, M.; Louis, C. J. Phys Chem. B 1997, 101, 7060.
(16) Burattin, P.; Che, M.; Louis, C. J. Phys Chem. B 1998, 102, 2722.
(17) Burattin, P.; Che, M.; Louis, C. J. Phys Chem. B 1999, 103, 6171.
(18) Burattin, P.; Che, M.; Louis, C. J. Phys Chem. B 2000, 104, 10482.
(19) Boehm, H. P. Angew. Chem. 1966, 78, 617.
(20) Dekkers, M. A. P.; Lippits, M. J.; Nieuwenhuys, B. E. Catal. Lett.
1998, 56, 195.
(21) Guillemot, D.; Polisset-Thfoin, M.; Fraissard, J. Catal. Lett. 1996,
41, 143.
(22) Guillemot, D.; Borovskov, V. Y.; Kazansky, V. B.; Polisset-Thfoin,
M.; Fraissard, J. J. Chem. Soc., Faraday Trans. 1997, 93, 3587.
(23) Haruta, M.; Ueda, A.; Tsubota, S.; Torres Sanchez, R. M. Catal.
Today 1996, 29, 443.
(24) Haruta, M. Catal. Today 1997, 36, 153.
(25) Block, B. P.; Bailar, J. C. J. Am. Chem. Soc. 1951, 73, 4722.
(26) Turkevich, J.; Stevenson, P. C.; Hillier, J. Trans. Faraday Soc. 1951,
85, 55.
(27) Torres Sanchez, R. M.; Haruta, M. Proc. 15th Iberoam. Symp. Catal.
1996, 1395.
(28) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M.
J.; Delmon, B. J. Catal. 1993, 144, 175.
(29) Lin, S.; Vannice, M. A. Catal. Lett. 1991, 10, 47.
(30) Zanella, R.; Shin, C. H.; Louis, C. Manuscript in preparation.
(31) Farges, F.; Sharps, J. A.; Brown, G. E. Geochim. Cosmochim. Acta
1993, 57, 1243.
(32) Gammons, C. H.; Williams-Jones, A. E. Geochim. Cosmochim. Acta
1995, 59, 3453.
(33) Gammons, C. H.; Yu, Y.; Williams-Jones, A. E. Geochim.
Cosmochim. Acta 1997, 61, 1971.
(34) Murphy, P. J.; LaGrange, M. S. Geochim. Cosmochim. Acta 1998,
62, 3515.
(35) Pan, P.; Wood, S. A. Geochim. Cosmochim. Acta 1991, 55, 2356.
(36) Peck, J. A.; Brown, G. E. Geochim. Cosmochim. Acta 1991, 55,
671.
(37) Murphy, P. L.; Stevens, G.; LaGrange, M. S. Geochim. Cosmochim.
Acta 2000, 64, 479.
(38) Machesky, M. L.; Andrade, W. O.; Rose, A. W. Geochim.
Cosmochim. Acta 1991, 5, 769.
(39) Mitsyuk, B. M.; Mironov, A. G.; Plyusnin, A. M.; Belomestrova,
N. V. Geochem. Int. 1991, 28, 86.
(40) Nechayev, Y. A.; Nikolenko, N. V. Geochem. Int. 1985, 11, 1656.

7642 J. Phys. Chem. B, Vol. 106, No. 31, 2002


(41) Nechayev, Y. A. Geochem. Int. 1986, 23, 32.
(42) Schoonen, M. A. A.; Fisher, N. S.; Wente, M. Geochim. Cosmochim. Acta 1992, 56, 1801.
(43) Yokoyama, T.; Matsukado, Y.; Uchida, A.; Motomura, Y.; Watanabe, K.; Izawa, E. J. Colloid Interface Sci. 2001, 233, 112.
(44) Degussa-France. Personal communication.
(45) Greffie, C.; Benedetti, M. F.; Parron, C.; Amouric, M. Geochim.
Cosmochim. Acta 1996, 60, 1531.

Zanella et al.
(46) Puddephatt, R. J. The chemistry of gold; Elsevier Scientific
Publishing Company: Amsterdam, 1978.
(47) Baes, C. F., Jr.; Mesmer, R. R. The hydrolysis of cations; John
Wiley & Sons: New York, 1986.
(48) Enzweiler, J.; Joekes, I. J. Geochem. Explor. 1991, 40, 133.
(49) van Olphen, H. An Introduction to Clay Colloid Chemistry;
Interscience Publishers-John Wiley: New York, 1977.
(50) Chow, M. K.; Zukoski, C. F. J. Colloid Interface Sci. 1994, 165,
97.