Preparation of a Pt/SiO2 catalyst: I. Interaction between platinum tetrammine hydroxide and the silica surface

Article abstract of DOI:10.1006/jcat.2002.3616

This paper describes a detailed study of exchange, drying, and thermal decomposition processes utilized to prepare Pt/SiO₂ catalysts from Aerosil silica (Degussa) and platinum tetrammine hydroxide. It is shown that exchange is complete after about 1 min. However, more than 1 h is required to achieve a uniform distribution of the platinum complex over the silica surface. The exchanged complex does not lose any NH₃ ligands upon drying at 25°C under vacuum. At 100°C and under vacuum the exchanged complex forms clusters of [Pt(NH₃)₄(OH)₂]_n deposited on the surface. Upon increasing the temperature under reduced pressure, the adsorbed complex looses the NH₃ ligands progressively. At 200°C, about half the ammonia ligands are lost, and at 300°C the complex is fully decomposed, mainly into PtO particles of 1-2 nm. About 10% of PtO is reduced to Pt⁽⁰⁾ by NH₃ decomposition. Further reduction under hydrogen at 400°C does not alter the particle size. Chemisorption of H₂ and O₂, electron microscopy, and EXAFS analyses were used to characterize the various species formed. The results suggest that drying is the crucial step in the preparation of a Pt/SiO₂ catalyst. This result was unexpected, because none of the earlier studies found this step to be crucial. The proposed procedure leads to a metal dispersion of 65 ± 5%, which is close to the maximum value reported in the literature. The second part of this article presents time-resolved mass spectrometric analyses of the species released during thermal treatment in various atmospheres. They support the conclusions given here.

Full text of DOI:10.1006/jcat.2002.3616

Journal of Catalysis 209, 135–144 (2002)
doi:10.1006/jcat.2002.3616
Preparation of a Pt/SiO Catalyst
2
I. Interaction between Platinum Tetrammine Hydroxide and the Silica Surface
∗
∗
,1
A. Goguet, M. Aouine,† F. J. Cadete Santos Aires,† A. De Mallmann,‡ D. Schweich, and J. P. Candy‡
∗
Laboratoire de G e´ nie des Proc e´ d e´ s Catalytiques (UMR CNRS-CPE), ESCPE Lyon, 69616 Villeurbanne, France; †Institut de Recherche sur la
Catalyse (UPR 5401-CNRS), 2 Av. Albert Einstein, 69626 Villeurbanne, France; and ‡Laboratoire de Chimie Organom e´ tallique
de Surface (UMR CNRS-CPE), ESCPE Lyon, 69616 Villeurbanne, France
Received November 21, 2001; revised February 15, 2002; accepted March 22, 2002
given carrier and metal precursor, the following procedure
This paper describes a detailed study of exchange, drying, and can be followed: pretreatment of the carrier, “adsorption”
thermal decomposition processes utilized to prepare Pt/SiO2 cata- of the precursor onto the surface, drying, thermal decompo-
lysts from Aerosil silica (Degussa) and platinum tetrammine hy-
droxide. It is shown that exchange is complete after about 1 min.
However, more than 1 h is required to achieve a uniform distribu-
sition (in various atmospheres or under vacuum), and even-
tually reduction. A Pt/SiO2 catalyst such as EUROPT-1 is
probably the simplest catalyst, which illustrates this prob-
lem (1). Table 1 lists the wide range of operating conditions
tion of the platinum complex over the silica surface. The exchanged
◦
complex does not lose any NH3 ligands upon drying at 25 C under
reported in the literature for preparing this catalyst using
◦
vacuum. At100 Candundervacuumtheexchangedcomplexforms
2
4
+
Pt(NH3) as the precursor. As a result, metal dispersion
clusters of [Pt(NH3)4(OH)2]n deposited on the surface. Upon in-
creasing the temperature under reduced pressure, the adsorbed
varies from 5 to no more than 80%. Finally, this dispersion is
often found to depend on the amount of catalyst prepared.
◦
complex looses the NH3 ligands progressively. At 200 C, about half
◦
the ammonia ligands are lost, and at 300 C the complex is fully Thus, it seems obvious that the variability in these condi-
decomposed, mainly into PtO particles of 1–2 nm. About 10% of tions and results reveals a lack of a clear understanding
PtO is reduced to Pt by NH3 decomposition. Further reduction of the mechanisms involved in the successive preparation
underhydrogenat400 Cdoesnotaltertheparticlesize. Chemisorp-
(
0)
◦
steps.
tion of H2 and O2, electron microscopy, and EXAFS analyses were
used to characterize the various species formed. The results suggest
that drying is the crucial step in the preparation of a Pt/SiO2 cata-
lyst. This result was unexpected, because none of the earlier studies
lyst is a relevant problem. This paper describes a detailed
found this step to be crucial. The proposed procedure leads to a
Although Pt/SiO2 is not a typical industrial catalyst, it
is widely used as a model catalyst in basic research. Thus,
understanding and controlling the preparation of this cata-
study of the exchange, drying, and thermal decomposition
processes utilized to prepare a Pt/SiO2 catalyst from aerosil
silica (Degussa) and platinum tetrammine hydroxide. This
metal dispersion of 65 ± 5%, which is close to the maximum value
reported in the literature. The second part of this article presents
time-resolved mass spectrometric analyses of the species released
during thermal treatment in various atmospheres. They support the first paper (part I) is devoted to the analysis of (i) the kinet-
conclusions given here.
ics of the exchange reaction between the silanol groups of
ꢀc 2002 Elsevier Science (USA)
Key Words: Pt/SiO2; platinum particles preparation; platinum the silica surface and the platinum tetrammine hydroxide
tetrammine hydroxide.
in the aqueous solution, (ii) the consequences of the drying
process, and (iii) the various species formed during the de-
composition of the exchanged species at increasing temper-
1
. INTRODUCTION
ature under reduced pressure. The second paper (part II)
will describe more realistic preparation conditions using
time-resolved mass spectrometric analysis of the gaseous
products that evolve during the drying and thermal decom-
position steps in various atmospheres.
The exact preparation method of many catalysts is still
a matter of know-how, even in the simple case of a
monometallic catalyst supported on alumina or silica. For
this type of simple catalyst, activity and selectivity are
mainly governed by the size of the metal particles. It is still
unknown which step(s) in the preparation method is (are)
predominantly responsible for the size of the particles. For a
2
. BRIEF LITERATURE REVIEW
2
+
It is widely recognized that dissolved Pt(NH3) cations
4
1
interact with the silica surface via ion exchange provided
that the pH of the solution is high enough. According to
To whom correspondence should be addressed. E-mail: candy@
mulliken.cpe.fr.
135
0
021-9517/02 $35.00
c 2002 Elsevier Science (USA)
All rights reserved.
ꢀ

1
36
GOGUET ET AL.
TABLE 1
After exchange, the solid is recovered, generally by filtra-
tion, and dried, commonly at about 100 C in an oven. This
preparation step is usually considered to be unimportant
13, 14) and, to the best of our knowledge, a fundamental
study of the processes that take place during drying has not
been made.
Following the drying step, the supported precursor is
thermally decomposed in order to obtain platinum or plat-
inum oxide particles on the surface. Previous studies report
decomposition performed in flowing oxidizing (7, 8, 15–21),
neutral (15, 16), reducing atmospheres (7–9, 15, 17–20, 22,
◦
Typical Experimental Conditions Reported in the Literature
for the Preparation of Pt/SiO2 Catalysts
(
Step
Conditions
References
Impregnation
Drying
From 3 min to 17 days
8, 9, 11, 15, 17–19
7–9, 11, 15, 17–19, 23
From 2 to 20 h; from
◦
2
5 to 120 C
Thermal
decomposition
O2/inert or inert gas;
7, 8, 15, 17–21, 23
from 0.5 to 12 h; from
◦
5
0 to 700 C (various
temperature programs)
H2/inert; from 0.5 to 5 h;
2
3)orundervacuum(9). Eveniftheconditionsunderwhich
Reduction
7–9, 15, 17–20, 22, 23
◦
from 300 to 800 C (various
those thermal treatments are performed are usually speci-
fied (gas flow rate, temperature program. . . ), no unambigu-
ous fundamental studies of the behavior of the precursor
during those treatments have been carried out.
temperature programs)
Brunelle (2), the charge in a silica surface is close to 0 from
pH 1–7, and surface ionization becomes significant above
pH 7 according to
3. EXPERIMENTAL
3.1. Materials
Si–OH + OH− ≡Si–O + H O.
←
−
≡
→
2
The silica Aerosil 200 support was purchased from
Degussa. Its surface area, measured by nitrogen ad-
sorption (77 K, N2, B.E.T., Micromeretics ASAP 2010)
However, at pH greater than 10 a significant amount of
silica dissolves (2). It is, thus, obvious that a cationic plat-
inum precursor may interact with the silica surface within
a rather narrow range of basic pH. Platinum tetrammine
hydroxide, referred to as Reiset base (3), is a white powder
at room temperature, and it melts at 110 C (4). In water so-
lution, it is a strong base and appears to be a good precursor
for Pt/SiO2 catalyst (2).
2
−1
is 194.2 m · g . The adsorption–desorption isotherms
showed that this silica is nonporous (Type II isotherms with
no hysteresis loop). Although the individual nonporous sil-
ica particles are about 20 nm in diameter, the bulk powder
is made up of aggregates of about 200 elementary particles
with diameters of about 100 nm.
◦
Platinum tetrammine hydroxide (Pt(NH3)4(OH)2) was
obtained from platinum tetrammine chloride (Strem
Most authors (2, 5–8) suggest that adsorption of dissolved
Pt(NH3)4(OH)2 is described by the cationic exchange re-
action
−
Chemicals; purity, 99%), by exchange of the Cl ions with
−
OH using an ion-exchange resin (Aldrich, IRA 400), ac-
−
+
−
2+
cording to the procedure described by Benesis et al. (22).
2
(≡Si–O , H ) + (OH )2, Pt(NH3)
4
H2O
−
2+
−
+
←
− (≡Si–O )2, Pt(NH3)4 + 2(OH , H ).
−
→
3.2. Analysis
3.2.1. Platinum concentration and pH of the aqueous
This reaction can be very fast and numerous authors sug-
gest that a reaction time as short as 3 min may be enough
to reach equilibrium. However, although no detailed study
was undertaken, other authors (2, 9) observed that after
solution. The concentration of Pt in the suspension was
determined using ICP (Jobin Yvon) analysis after filtering
through a Millipore SLSR025NB.
ThepHofthesuspensionwasmeasuredusingaSCHOTT
CG 840 pH meter.
2
days of solid–liquid contact, the homogeneity of the plat-
inum distribution over the silica surface clearly improves.
+
Elemental analysis of dried samples was performed in
the C.N.R.S. laboratory SCA (Service Central d’Analyses).
It was also proposed (5, 10) that the addition of NH ions
4
to the solution slows down the reaction as a result of the
following exchange reaction:
3
.2.2. Hydrogen and oxygen chemisorption studies. H2
◦
or O2 chemisorption experiments were carried out at 25 C
−
+
−
2+
2
(≡Si–O , NH ) + (OH )2, Pt(NH3)
4
4
using conventional Pyrex volumetric adsorption equipment
H2O
−6
−
2+
−
+
4
←
− (≡Si–O )2, Pt(NH3)4 + 2(OH , NH ).
(
24). The vacuum (10 mbar) was achieved with a liquid-
−
→
nitrogen-trapped mercury diffusion pump. The equilibrium
Some authors (11, 12) suggested that the exchange of pressure was measured with a Texas Instrument gauge
the platinum complex on the silica surface may occur with (pressure range, 0–1000 mbar with an accuracy of 0.1 mbar).
the loss of two NH3 ligands and the formation of the The catalyst sample was placed in a Pyrex cell and out-
◦
(
≡Si–O)2Pt(NH3)2 surface complex.
gassed at 25 and then at 300 C for 2 h under vacuum before

PREPARATION OF Pt/SiO2, PART I
137
the chemisorption measurements. The oxidation state of rameters were adjusted using the simplex and least-squares
platinum was deduced from the amounts of H2 and O2 methods: the number, N, of atomic neighbors of definite
adsorbed.
type (O, Pt. . .), the distance R between the Pt atom and
its neighbor, and the Debye–Waller factor, σ, character-
izing the thermal and structural disorder. The adjustment
residue ρ was computed using the following formula:
ꢀ
3
.2.3. Electron microscopy studies. The dry catalysts
were suspended in ethanol. A drop of the suspension was
deposited onto a carbon film supported on a copper grid,
and the ethanol evaporated quickly at room temperature.
The samples were observed using a JEOL 2010 electron
microscope with energy-dispersive X-ray (EDX) analysis.
For EDX analysis, the electronic beam was focused on
a single silica particle, and the Pt/Si ratio was determined
after calibration. About 100 silica particles of each sample
were observed to obtain the Pt/Si frequency distribution.
The unsupported precursor was also examined. It was
prepared using a drop of platinum tetrammine hydroxide
previously dissolved in ethanol.
3
3
(k)]2
[
k χ (k) − k χ
k
exp
calc
ρ =
ꢀ
.
3
2
[
k χ (k)]
k
exp
3.3. Adsorption of the Pt Precursor
Prior to the studies, 3 ± 0.02 g of silica were treated
◦
under flowing dry air at 520 C for 5 h in order to remove any
pollutants that may have adsorbed. Moreover, treatment
◦
at temperatures above 500 C leads to an almost fully de-
hydroxylated silica surface. Rehydroxylation of the treated
The mean metallic particle size was determined from the
examination of at least 500 particles.
≡
sample leads to Si–OH silanol groups only. There is no for-
mation of the Si(–OH)2 or –Si(–OH)3 silanol groups ini-
tially present (27). After cooling to room temperature, the
treated silica was introduced into a “Pyrex” reactor filled
with a known volume of water and stirred well with a mag-
netic agitator to generate a suspension. The pH of the silica
suspension was then increased to pH 9 using an aqueous
solution of NH3. In some experiments, NH3 was not added.
For kinetic studies, a known volume of solution con-
taining the desired amount Pt(NH3)4(OH)2 was added to
the silica suspension. Analyzing samples of the suspension
taken at regular time intervals enabled the study of ex-
==
3
.2.4. EXAFS studies. All the samples were taken
under dry nitrogen atmosphere and placed into air-proof
cellsequippedwithkaptonwindows. SpectraofPtO2 andan
aqueous solution of Pt(NH3)4(OH)2 (0.17 mol · L ) were
also recorded as a reference. Moreover, the spectrum of a
EUROPT-1 reference catalyst was recorded after reduction
−1
◦
for 1 h at 400 C under hydrogen.
The X-ray absorption spectra were recorded at the Lab-
oratoire pour l’Utilisation du Rayonnement Electromag-
netique (L.U.R.E.) at Orsay, France, on the DCI ring using
2
+
change kinetics. The concentration of Pt(NH3) fixed on
1
2
.85-GeV positrons with an average current intensity of
50 mA. The acquisitions were carried out on the D44
4
the silica surface was deduced from the difference between
2
4
+
the amount of Pt(NH3) introduced and the amount of
beam line at the LIII platinum transition edge using a trans-
mission mode between 11,400 and 12,600 eV and an incre-
ment step of 2 eV. The counting time was 2 s. A Si(111)
double crystal was used as a monochromator, and the γ
photons flux was measured at the entrance and the exit of
the cell using ionization chambers.
The EXAFS analyses were performed according to the
single scattering theory using Michalowicz’s software (25).
The support absorption (µ ) was computed using the theo-
2
4
+
Pt(NH3) remaining in the solution.
For studies of platinum distribution over the silica sur-
face, samples were prepared using the method described
for kinetic studies (see above) and contact times of 5 min,
1
h, and 24 h. Then, the samples were filtered and washed
with 50 ml of distilled water. The subsequent heat treat-
ment included the following steps: (i) drying under vacuum
◦
at T = 25 C for 24 h, and (ii) temperature increase from
0
◦
◦
◦
T = 25 to 300 C (2 C/min), 1 h at 300 C under vacuum,
retical expression developed by Lengeler and Eisenberger
◦
and cooling to 25 C. Finally, some of the samples were re-
(
26), and the platinum atomic absorption (µ ) was mod-
1
◦
duced at 400 C for 2 h under hydrogen. The samples were
eled by a fifth- or sixth-order polynomial between 11,600
and 12,400 eV. Each EXAFS contribution was carefully
extracted in order to minimize the low-frequency noise.
The edge energy E0 (around 11,565 eV) was measured
at half height of the absorption threshold. The pseudo ra-
dial distribution function around the platinum |F(R)| was
observed by electron microscopy with energy-dispersive
X-ray (E.D.X.) analysis (JEOL 2010).
2
4
+
3.4. Drying and Decomposition of Pt(NH3)
Adsorbed on the Silica Surface
obtained by a Fourier transform of the weighted spec-
3.4.1. In situ measurement of ammonia released during
3
trum ω(k) · k · χ(k), where ω(k) is a Kaiser-type function the drying and thermal decomposition steps. The experi-
˚
−1
(
τ = 2.5) defined between 3 and 13.5 A
.
mental setup for exchange, drying, and thermal decompo-
The spectra simulations were performed using theoreti- sition consisted of two flasks connected from above. They
cal files from FEFF software for phases and amplitudes and could be evacuated singly or simultaneously or fed with an
the results obtained from the reference compound simula- inert gas. The “main” flask (200 ml) was used to prepare the
tions were taken into account. The following structural pa- catalyst. The “trap” flask (50 ml) was used to recover the

1
38
GOGUET ET AL.
gaseous products released during the successive steps de-
2
,5
1
scribed below. Silica (3 ± 0.02 g) was treated under flowing
◦
dry air at 520 C for 5 h. After cooling to room temperature,
0
-1
1
[Pt] : 16 mmol.L + NH
3
the solid sample was introduced into the main flask together
−
4
with 30 ml of water and 2.3 × 10 mol of Pt(NH3)4(OH)2
*
*
−4
or 6.4 × 10 mol of NH3. The suspension was stirred for
h (magnetic stirrer), the main flask was cooled to liquid
nitrogen temperature, and both flasks were evacuated to
*
1
0
-1
[
Pt] : 10 mmol.L + NH3
−
4
about 10 mbar (mercury diffusion pump).
[Pt]0 : 3 mmol.L-1
0
,5
0
The flasks were then isolated from the vacuum line, and
◦
[Pt]0 : 3 mmol.L-1 + NH3
the main flask was heated to 25 C for 24 h while the trap
flask was cooled to liquid nitrogen temperature to condense
the vapor of the bulk liquid phase in the main flask into
the trap flask. After this step the pressure in the flasks was
lower than 0.01 mbar. Then both flasks were filled with ar-
gon up to atmospheric pressure, and the main flask was
isolated. The trap flask was heated to room temperature,
and a known amount of HCl in aqueous solution was intro-
duced to neutralize the ammonia that evolved during the
heat treatment. The trap flask was then disconnected, and
the remaining amount of hydrochloric acid was determined
using a titrated NaOH aqueous solution. The difference be-
tween the initial and final amounts of HCl gave the amount
of NH3 evolved from the main flask.
10
20
40
0
Time/min
FIG. 1. Evolution of Pt concentration during Pt(NH3)2 exchange on
silica for various initial Pt concentrations ([Pt] ), corresponding to various
+
4
0
Pt loadings. (∗) 5.9%; (ꢀ) 3.7%; (×) 1.0% Pt with addition of NH3 (pH 9);
and (+) 1.0% Pt without addition of NH3.
contact with 30 ml of Pt(NH3)4(OH)2 aqueous solution (see
Section 3.3).
The interaction between the platinum complex and the
silica surface is clearly very fast: in all cases, the platinum
concentration and pH reach almost constant values after
lessthan1min. FollowingtheadditionofNH3 tothesuspen-
sion, before the interaction (see Section 3.3), the exchange
process does not slow down significantly but affects the final
equilibrium concentration. The platinum residual concen-
tration decreases upon addition of NH3 (Fig. 1) and, con-
versely, the final pH increases (Fig. 2). The initial pH is only
an indication, since pH varies quickly upon addition of the
complex into the suspension. The exchange rate does not
depend significantly on the initial concentration of the plat-
The trap flask was then washed, dried, and reconnected
(
under argon). The same sequence of operations (evacua-
◦
tion under vacuum at 25 C, heating of the main flask, NH3
analysis) was repeated successively at 100, 200, and 300 C
◦
for 30 min using the same sample of solid. The tempera-
◦
−1
◦
ture was increased by 2 C · min from 25 C to the desired
◦
temperature. During the steps at 25, 100, 200, and 300 C
the residual pressure remained constant and lower than
−2
1
0
mbar while the flasks were isolated from the vacuum
line.
−
1
3
.4.2. Other catalyst samples. The catalyst samples re-
inum precursor C0, which varied from 3 to 16 mmol · L
,
quired for EXAFS studies (see below) were prepared as
follows: the exchange reaction was performed according
to the procedure described in Section 3.3 with no addition
of NH3 and with a contact time of 1 h. The wet solid was
placed in the chemisorption apparatus (see Section 3.2.2)
as can be seen from Figs. 1 and 2.
1
1
1
3
2
1
*
◦
−5
and dried at 25 C for 24 h under vacuum (P < 10 mbar
◦
−1
along the vacuum line). It was then heated (2 C · min
)
and maintained for 30 min at a desired temperature (100,
0
-1
[
[
[
Pt] : 16 mmol.L + NH
3
◦
2
00, or 300 C) still under vacuum. When necessary, it was
*
◦
0
-1
*
Pt] : 10 mmol.L + NH
*
reduced under hydrogen at 400 C for 1 h. With the excep-
tion of reduction, these steps were as similar as possible to
those described in Section 3.4.1.
10
9
3
Pt]0 : 3 mmol.L-1 + NH
Pt]0 : 3 mmol.L-1
3
[
8
4
. RESULTS
4
.1. Kinetics of the Exchange Reaction
7
0
10
20
40
Figures 1 and 2 show the variation in the Pt concentration
Time/min
and pH of the liquid phase with time when 3 ± 0.02 g of
silica in suspension in 30 ml of water (pH 9) come into
FIG. 2. Evolution pH corresponding to the conditions given in Fig. 1.

PREPARATION OF Pt/SiO2, PART I
139
0
0
0
0
0
.5
.4
.3
.2
.1
0
70
5
1
2
min
h
8
6
4
2
0
6
5
4
3
2
1
0
0
0
0
0
0
0
4
0
5
10
15
1
4
7
10 13 15 >18
-1
3
[Pt]eq/mmol.lL
10 Pt/Si atomic ratio
FIG. 3. Pt(NH3)2 exchange isotherm on silica.
+
FIG. 4. Pt distribution over the silica support after 5 min, 1 h and 24 h
of contact time.
4
4
.2. Exchange Isotherm of Pt(NH3)4(OH)2
on the Silica Surface
+
It is clear that more than 90% of the exchanged NH can
be recovered from the sample after 30 min at 100 C.
4
◦
From the above observation, it can be assumed that the
exchange equilibrium is reached within 5 min. This leads
4
.4.2. Decomposition of platinum tetrammine exchanged
on silica. The experiment described in Section 4.4.1
was performed using 30 ml of aqueous Pt(NH3)(OH)2
to the “exchange isotherm” shown in Fig. 3. For a residual
−1
platinum concentration close to 7 mmol · L , the amount
of the exchanged platinum complex corresponds to a metal
loading of 7.5% (w/w) and does not significantly increase
with the concentration of the residual platinum.
−
1
(
7.67 mmol · L ) instead of NH3. The results are also listed
in Table 2. Note that silica was in contact with the solution
for 1 h to obtain a rather uniform distribution of the plat-
inum complex over the surface, as suggested by the results
in Fig. 4.
All these observations are in agreement with the ex-
change process described in Section 2, although a quan-
titative estimation of the equilibrium constants was not
attempted. It is, however, certain that the exchange equi-
librium is very favorable for the Pt complex.
◦
At all temperatures from 25 to 300 C, the residual pres-
−
2
sure in the isolated flasks was close to 10 mbar at each
step. This very low pressure corresponds to 0.04 mmol of
permanent gas in the flasks, which is much less than the
amount of NH3 recovered above 100 C (see Table 2). This
proves that most of the released gaseous products are con-
◦
4
.3. Distribution of the Exchanged Platinum
Complex on the Silica Surface
densed in the trap flask, i.e., hardly any NH3 decomposes to
2
An aqueous solution (30 ml) of Pt(NH3)4(OH)2 N and H . These results show that Pt–NH bonds do not
2
3
−1
◦
(
1.71 mmol · L ) came into contact with 1 g of silica. The break during the exchange and drying processes at 25 C
◦
average Pt/Si value was then 0.003. After 5 min, 1 h, and 24 h under reduced pressure. Up to 100 C, the complex does
of contact, electron microscopy with X-ray emission analy- not decompose significantly, whereas more than 90% of the
sis (EDX) of the samples gave the results shown in Fig. 4.
After 5 min, Pt/Si values as high as 0.018 were obtained for
about 40% of the individual silica particles. After 1 h, the
average Pt/Si value shifts to lower values, and after 24 h,
there is a rather narrow distribution close to the expected
Pt/Si ratio of 0.003.
TABLE 2
+
Ammonia Evolved after Drying and Heat Treatment of NH
4
2+
(
Blank Experiment) and Pt(NH3)4 Exchanged on Silica
◦
◦
◦
◦
Temperature
Time (h)
NH /SiO2
25 C
100 C
200 C
300 C
0.5
0.63 mmol
(99%)
4
.4. Stoichiometry of the Exchanged Platinum
Complex on the Silica Surface
24
0.5
0.5
0.63 mmol
(98%)
+
0.27 mmol 0.59 mmol
42%) (92%)
4
(
.4.1. Blank experiment. Silica was put into contact Pt(NH3)2 /SiO2 0.01mmol 0.025 mmol 0.345 mmol 0.835 mmol
+
4
4
−
1
(1%)
(3%)
(38%)
(91%)
with30mlofanaqueoussolutionofNH3 (21.33mmol · L
)
as described in Section 3.4.1. After drying and heating, the
amount of ammonia evolved was measured. The results are
reported in Table 2.
Note. Cumulative amount of ammonia evolved in millimoles and corre-
sponding percentage of introduced ammonia recovered (in parentheses).
Time: Duration of the temperature step in hours.

1
40
GOGUET ET AL.
◦
FIG. 5. TEM pictures of samples after 25 C vacuum drying. (a) Pt(NH3)4(OH)2/SiO2 (4.5% Pt); (b) unsupported Pt(NH3)4(OH)2.
◦
ammonia was recovered at that temperature in the blank
experiment (Table 2).
Treatment under reduced pressure at 200 and 300 C
• after drying and 30 min at 100 C under vacuum (sam-
ple Pt100),
◦
◦
• after drying and 30 min at 200 C under vacuum (sam-
causes the exchanged complex to progressively lose the ple Pt200),
◦
◦
NH3 ligands. At 300 C, 91% of the ammonia ligands are
• after drying and 30 min at 300 C under vacuum (sam-
recovered; the remaining 9% are decomposed into N2 and ple Pt300),
H2 (see Discussion).
◦
• after drying, 30 min at 300 C under vacuum, and re-
◦
duction at 400 C under hydrogen (sample PtR400).
4
.5. Transmission Electron Microscopy Study
of the Adsorbed Pt(NH3)2+ Complex
The corresponding EXAFS spectra and Fourier trans-
form curves are reported in Fig. 7. The structural parame-
ters obtained for the samples from the simulations (N, R, σ)
are reported in Table 3 together with the known structural
parameters ofbulk platinum oxide(PtO2), 8-µm-thickplat-
inum foil, and Pt(NH3)4(OH)2 in aqueous solution.
4
Figure 5 shows typical TEM pictures obtained with
vacuum-dried supported (Fig. 5a) and unsupported
◦
(
Fig. 5b) silica samples at 25 C.
Both samples exhibit similar small particles (1.5- to
2
-nm diameter) with low contrast. It must be emphasized
that additional microanalysis, carried out on supported and
unsupported samples of Pt(NH3)4(OH)2, revealed a ratio
of 4 nitrogen atoms per platinum atom.
After treatment under vacuum at 300 C, the samples dis-
play similarly low-contrast particles with roughly the same
1
00
◦
8
6
4
2
0
0
0
0
0
diameter (i.e., 1.5–2 nm).
◦
For the sample treated under vacuum at 300 C and then
◦
reduced at 400 C, well-contrasted particles with an average
diameter of 1.5 nm were distributed homogeneously on the
silica surface (Fig. 6).
4
.6. EXAFS Studies
A 4.5% (w/w) silica catalyst was prepared as explained in
0
1
2
3
4
5
Section 3.4.2. The spectra of the solid were recorded after
each step of catalyst preparation:
Particle diameter/nm
FIG. 6. Particle size distribution of a 4.5% Pt catalyst after drying at
◦
•
after drying 24 h under vacuum at 25 C only (sample
Pt025),
◦
◦
2
5 C, vacuum heat treatment at 300 C, and reduction under hydrogen at
◦
400 C.

PREPARATION OF Pt/SiO2, PART I
141
8
4
0
A
B
C
A’
B’
C’
A’’
B’’
4
0
4
-
4
-
4
4
0
8
4
0
0
4
-
4
-4
0
8
4
0
C’’
D’’
4
0
4
-4
-4
0
8
4
D
D’
4
0
0
4
-4
-
4
0
8
4
Experimental
Simulated
E
E’’
E’
4
0
0
4
4
0
-4
-
4
6
8
10
12
0
2
4
0
2
4
-
1
k (Å )
R (Å)
R (Å)
FIG. 7. EXAFS spectra of the various samples. (A) Pt025; (B) Pt100; (C) Pt200; (D) Pt300; and (E) PtR400 and corresponding Fourier transform.
It is not possible to distinguish O from N as Pt neighbors quired by the coordination sphere of Pt were assumed to
by EXAFS. Thus, based on the results of Section 4.4.2, we be oxygen.
assume that there were about two nitrogen neighbors per
Pt atom at 200 C and none at 300 C. The other atoms re- Pt(NH3)4(OH)2 in aqueous solution and of Pt025 (vacuum
From these results, it appears that the EXAFS spectra of
◦
◦

1
42
GOGUET ET AL.
TABLE 3
Structural Parameters Obtained for the Studied Samples
˚
2
˚ 2
Sample
Scattering atom
N
R (A)
2ꢀ (A )
ꢁ
Pt025
Pt100
Pt200
Pt300
N
O
N
O
4.0
2.2
4.1
2.04
3.71
2.02
3.56–3.74
2.03 + 1.99
3.60–3.80
2.00
0.0060
0.012
0.0048
14%
19%
16%
35%
2.5–3.5
2.0 + 2.0
2.3–3.2
2.93
∼1.2
5.8
0.0098–0.0104
0.0051 + 0.0075
0.0056
N + O
O
O
0.0095
0.013
0.0172
0.0169
Pt
Pt
Pt
Pt
Pt
O
∼2.57 ± 0.1
PtR400
2.69
15%
14%
EUROPT-1
EUROPT-1 (29, 30)
PtO2(39)
5.6
5.5
8
2
2.68
2.76
3.350
1.984
O
4
2.004
Pt (foil)(40, 41)
Pt
Pt
N
12
6
4
2.772
3.920
2.040
Pt(NH3)4(OH)2(40)
Note. ꢀ: Debye–Waller factor. First six rows, our results; next four rows, literature data (reference
in parantheses).
◦
dried at 25 C only) are very similar. The platinum environ- and hydroxyl anions according to
ment (four light atoms in the first shell) is only slightly mod-
ified by heat treatment at 100 C. The distances obtained
Pt(NH3)4(OH)2 → Pt(NH3) + 2OH−.
←
2+
[2]
◦
4
from simulations are fully compatible with Pt–N bonds in
When the platinum tetrammine solution is added to the
silica suspension, there is a strong increase in pH that is
responsible for silica ionization according to Eq. [1]. Then
platinum tetrammine cations are exchanged for the surface
H cations. WhenanaqueoussolutionofNH3 isusedtocre-
ate NH as competing cations, the residual platinum con-
centration is lower (Fig. 1). This is explained easily by the
higher pH, which promotes ionization of the silica surface.
Although NH4 competes with platinum there are many
more sites available for exchange. Thus, it is not recom-
mended that an aqueous solution of NH3 be used as a com-
peting agent.
Depending on the mixing intensity, the surface ionization
process and, thus, the distribution of the adsorbed complex
may be more or less heterogeneous. The driving force for
redistribution is mixing in the bulk solution and diffusion
across the external boundary layer or inside the silica ag-
gregates (31). The diffusional flux is proportional to the dif-
ference between the concentrations of the actual and the
equilibrium platinum in the liquid phase. Although this dif-
ference is not known, the concentration is almost certainly
lower than the residual concentration. Consequently, the
lower the residual platinum concentration, the smaller the
diffusional flux and the slower the redistribution process.
This explains why about 24 h are required to obtain a homo-
geneous distribution for the lower Pt loading (Section 4.3,
Fig. 4). According to this explanation, with a higher residual
2
4
+
◦
Pt(NH3) (40). At 200 C, about half the NH3 ligands are
lost, and there are still four light atoms (O or N) as neigh-
bors. The distances obtained from simulations are in the
domain of the Pt–O bonds in PtO2 (39). This would imply
stronger bonding of the complex to the ≡SiO surface. For
+
+
◦
4
the sample heated to 300 C, the spectrum was more diffi-
cult to simulate. The results obtained for this sample did
not show clearly that platinum was a second neighbor (like
the Pt–O–Pt structure observed in bulk PtO2). This may be
compatible with partially reduced platinum mixed with ox-
ide within the same particle. Joyner (28) did not observe Pt
second neighbors when studying the EUROPT-1 catalyst
+
◦
previously reduced at 300 C under hydrogen and then ex-
posed to air. Finally, as expected, the spectra of the reduced
sample (PtR400) and the EUROPT-1 reference catalyst are
similar. Note that the structural parameters obtained (5.8 Pt
˚
˚
at 2.69 A for PtR400 and 5.6 Pt at 2.68 A for EURPOT-1)
are in good agreement with the results obtained by Wells
˚
and co-workers (29, 30) (5.5 Pt at 2.76 A for EUROPT-1).
5
. DISCUSSION
In aqueous solution and at pH above the point of zero
charge, the silica surface ionizes according to
←
−
+
≡
SiOH ≡SiO + H .
[1]
→
The platinum tetrammine hydroxide in aqueous solution concentration or in the presence of a pH-neutral competing
is completely dissociated into platinum tetrammine cations cation, redistribution is probably faster.

PREPARATION OF Pt/SiO2, PART I
143
From our kinetic measurements (Fig. 1), it is clear that particles after thermal decomposition and reduction. The
ammonium ions do not significantly lower the rate of the ex- process that governs the cluster size is still unknown.
◦
change processes, as was proposed by Ribeiro and Marcilly
Heat treatment at 100 C does not lead to NH3 release
(
10). This suggests that the exchange rates and selectivity (EXAFS and NH3 release measurements). This result is
+
factors of the platinum complex cation compared to NH
or H are similar.
in agreement with the work of Kinoshita et al. (32), who
4
+
found that thermal decomposition of Pt(NH3)4(OH)2 does
◦
We did not perform further measurements of the Pt/Si not occur below 150 C.
◦
distribution. Consequently, the tentative explanation of ion
After heating under vacuum at 200 C, the surface com-
exchange given above should be checked by a detailed and plex loses 38% of the NH3 ligands (Table 2) and it must
specific analysis of the exchange process together with the be concluded from the EXAFS analysis that there are still
˚
surface ionization process.
four light atoms (either O or N) about 2 A away from
Our results definitely show that the adsorbed complex Pt. It seems that the surface complex is usually bonded to
◦
does not lose NH3 ligands up to 100 C, and that the the silica surface by two Pt–O bonds to form the surface
formation of the (≡SiO)2Pt(NH3)2 species, proposed by species (≡SiO)2Pt(NH3)2. It is noteworthy that the melt-
Sermon and Silvaningam (11) and by Mang et al. (12), ing point of bulk platinum tetrammine hydroxide is close
◦
◦
must be excluded. It is thus proposed that, at 25 C, the to 110 C (4). It may be that the complex clusters melt on
◦
platinum tetrammine cations in a water solution are sim- the surface above 100 C and then become “rafts,” where Pt
ply trapped in the electrical double layer at the silica atoms are close to the oxygen atoms of the silanol groups.
interface.
A raftlike structure was proposed by Jackson et al. (29, 30)
◦
After drying at 25 C under vacuum, the sizes of the par- for crystallites of Pt on silica obtained after reduction un-
ticles of the complex measured by TEM are similar regard- der moderate conditions. Our results present evidence that
less of whether or not they are supported by silica. It is the precursor can adopt a raftlike structure prior to reduc-
◦
surprising to observe small particles of the precursor on the tion. Since the amount of NH3 released at 200 C differs
silica surface; instead atomic dispersion of platinum ow- significantly from 50%, the possible explanation above is
ing to the exchange mechanism may have been expected. tentative at best. It will be considered again in the sec-
However, it can be argued that when the solid comes into ond part of this series, when time-resolved mass spectro-
contact with ethanol during the preparation of the sam- metric analyses during T.P.D. experiments will be dealt
ples for TEM, the analysis may yield biased results if the with.
◦
previously exchanged precursor “dissolves.” Several obser-
Heating to 300 C leads to the release of the remaining
vations, however, have shown that this is not so. (i) When NH3 ligands and to the formation of poorly contrasted
a freshly prepared sample is washed with a small amount particles with diameters of 1–2 nm. These particles chemi-
of water, desorption is not observed, in agreement with sorbed either 0.1 oxygen atom or 3.1 hydrogen atoms per
the strong affinity of the Pt cations for the silanol groups. platinum atom. This shows that the platinum tetrammine
(
ii) Even if a small amount dissolves in ethanol, the subse- complex decomposes mainly into PtO (90%) and a small
quent drying of the sample reproduces the previous drying amount of Pt( (10%). The reactions with hydrogen or
0)
step of the catalyst. (iii) EXAFS spectra show similar re- oxygen may be
sults, both for Pt(NH3)4(OH)2 in aqueous solution and for
◦
0.1Pt + 0.9PtO + 0.05O2 → Pt−O (O/Pt = 0.1),
a supported sample dried at 25 C. (iv) The supported sam-
◦
ple treated at 300 C consists mainly of PtO particles, which
0.1Pt + 0.9PtO + 1.55H2
cannot dissolve; nevertheless, the same particle size is ob-
served. This suggests that the exchanged platinum complex
hardly continues to interact electrostaticly with the surface.
The system behaves as if the following “transformation”
had taken place upon drying:
→ Pt−H1.3 + 0.9H O (H/Pt = 3.1).
2
The self-reduction of Pt(NH3)4(OH)2 adsorbed on zeo-
lite (33) or alumina (34) during treatment under He at a
◦
temperature higher than 300 C was proposed elsewhere,
ꢁ
ꢂ
although these authors did not give the extent of self-
reduction.
−
2+
n(≡Si–O )2 Pt(NH3)
+ (x + 2)H2Oadsorbed
4
→
2n(≡Si–OH) + (Pt(NH3)4(OH)2, xH2O) .
n
According to our interpretation, the size of these PtO and
Pt( particles reflects the size of the dry complex clusters
0)
The resulting “free” hydroxide forms [Pt(NH3)4(OH)2, due to the drying step.
◦
xH2O]n clusters (or colloids), regardless of whether they
Reduction at 400 C weakly modifies the size of the parti-
are supported on silica or unsupported. This explains why cles obtained after thermal decomposition under vacuum.
the dry silica-complex system exhibits TEM patterns sim- The resulting solid is very similar to EUROPT-1 (1, 35–37).
ilar to the unsupported dry complex (Fig. 5). The size of This step of the preparation does not appear to be critical
the resulting clusters (or colloids) determine the size of Pt with respect to the size of the particles.

1
44
GOGUET ET AL.
6
. CONCLUSION
pens when the exchanged catalyst decomposes in various
atmospheres and is eventually reduced under hydrogen.
Preparation of Pt/SiO2 catalyst is still based on recipes.
Although many of them lead to well-dispersed platinum,
some of them differ considerably from each other, and
the chemical and engineering processes are not well un-
derstood or controlled. Our study shows the following:
REFERENCES
1
2
3
4
. Bond, G. C., and Wells, P. B., Appl. Catal. 18, 225 (1985).
. Brunelle, J. P., Pure Appl. Chem. 50, 1211 (1978).
. Reizet, J., Ann. Chem. Phys. XI, 417 (1844).
. Duval, C., in “Nouveau Trait e´ de Chimie Min e´ rale” (Masson et Cie,
Ed.), Vol. XIX, p. 860. Libraries de l’Acad e´ mie de M e´ decine, Paris,
1958.
. Brunelle, J. P., and Sugier, A., C. R. Acad. Sci. Paris 276, 1545 (1973).
. Morrow, B. A., and McFarlan, A. J., Langmuir 7, 1695 (1991).
. Gonzalez, R. D., and Miura, H., Rev. Sci. Eng. 36, 145 (1994).
. Zou, W., and Gonzalez, R. D., J. Catal. 133, 202 (1992).
. Boronin, V. S., Nikulina, V. S., and Poltorak, O. M., Russ. J. Phys. Chem.
37, 626 (1963).
0. Ribeiro, F., and Marcilly, C., Rev. IFP 34, 405 (1979).
11. Sermon, P. A., and Silvaningam, J., Colloids Surf. 63, 59 (1992).
2. Mang, T., Breitscheidel, B., Polanek, P., andKnozinger, H., Appl. Catal.
06, 239 (1993).
3. Marcilly, C., and Le Peltier, F., Rev. IFP 39, 337 (1984).
•
As already shown, the interaction between the dis-
solved complex and the silica surface is governed by ion
exchange. However, this exchange process is extremely fast
and can be affected by the mixing process. A uniform dis-
tribution of platinum over the silica surface may require
5
6
7
8
9
about 24 h at low Pt loading (<1% (w/w)Pt/SiO2). Adding
+
NH3 to create NH counterions within the usual concentra-
4
1
tion range does not slow the exchange process. The possible
+
2+
competition between NH and Pt(NH3) for the exchange
4
4
1
−
+
reaction with SiO H is masked by the increase in the ion-
ization of the silica surface owing to the pH increase. Fur-
1
1
ther studies of silica ionization versus pH, of the exchange 14. Foger, K., in “Catalysis: Science and Technology” (J. R. Anderson and
+
+
M. Boudart, Eds.), Vol. 6, p. 247. Springer-Verlag, Berlin/New York,
984.
5. Brunelle, J. P., Sugier, A., and Lepage, J. F., J. Catal. 43, 273 (1976).
6. Zou, W., and Gonzalez, R. D., Catal. Today 15, 443 (1992).
7. Dorling, T. A., Eastlake, M. J., and Moss, R. L., J. Catal. 14, 23 (1969).
18. Dorling, T. A., Lynch, B. W., and Moss, R. L., J. Catal. 20, 190 (1971).
9. Alerasool, S., and Gonzalez, R. D., J. Catal. 124, 204 (1990).
20. Tsurimi, K., and Inoue, K., Kagaku Kogaku Ronbunshu 22, 542 (1986).
equilibrium between Pt-containing cations and H or NH ,
4
1
and of diffusion-limited redistribution are necessary to bet-
ter define the solid–liquid contact conditions. Nevertheless,
this exchange step does not seem to be crucial provided that
platinum redistribution is achieved.
1
1
1
1
•
The platinum tetrammine complex does not decom-
◦
pose from the exchange step to the heat treatment at 100 C
2
1. Tsurimi, K., and Inoue, K., Kagaku Kogaku Ronbunshu 22, 1294
1986).
2. Benesis, A. H., Curtis, R. M., and Studer, H. P., J. Catal. 10, 328 (1968).
◦
under vacuum. However, between 25 and 100 C the com-
plex consists of clusters of platinum tetrammine hydroxide
(
2
deposited on the silica surface. These clusters define the 23. Wilson, G. R., and Hall, W. K., J. Catal. 24, 306 (1972).
◦
2
2
2
4. Candy, J. P., El Mansour, A., Ferretti, O. A., Mabilon, G., Bournonville,
J. P., Basset, J. M., and Martino, G., J. Catal. 112, 201 (1988).
5. Michalowicz, A., “Logiciel pour la Chimie.” Soci e´ t e´ Fran c¸ aise de
Chimie, Paris, 1991.
final particle size of Pt. At about 100 C, these clusters may
melt, which would enable spreading of the platinum com-
plex over the silica surface in the immediate vicinity of the
original cluster. This would mean that drying is the crucial
6. Lengeler, B., and Eisenberger, P., Phys. Rev. B 21, 4507 (1980).
step. This is an unexpected result since no previous studies 27. Iller, R. K., “The Chemistry of Silica.” Willey, New York, 1979.
2
8. Joyner, R. W., J. Chem. Soc. Faraday Trans. 76, 357 (1980).
reported this step to be crucial.
◦
29. Jackson, S. D., Keegan, M. B. T., McLellan, G. D., Meheux, P. A.,
Moyes, R. B., Webb, G., Wells, P. B., Whyman, R., and Willis, J., Stud.
Surf. Sci. Catal. 63, 135 (1991).
•
At 200 C, about half the NH3 ligands are lost. This
would imply a stronger bond to the ≡SiO surface groups,
which would prevent the melted complex from spreading
over the surface. This could explain why the size of the
crystal, which was frozen at drying, is reflected in the size
30. Jackson, S. D., Willis, J., McLellan, G. D., Webb, G., Keegan, M. B. T.,
Moyes, R. B., Simpson, S., Wells, P. B., and Whyman, R., J. Catal. 139,
191 (1993).
3
3
1. Goguet, A., Aouine, M., Cadete Santos Aires, F. J., Schweich, D., and
of the metal particle.
Candy, J. P., Chem. Commun. 1417 (2000).
◦
•
At 300 C under vacuum the complex essentially fully
2. Kinoshita, K., Routsis, K., and Bett, J. A. S., Thermochim. Acta 10, 109
(1974).
decomposes to PtO. Although PtO is known to be a mo-
bile species, the temperature treatment was probably too 33. Van den Broek, A. C. M., van Grondelle, J., and van Santen, R. A.,
J. Catal. 167, 417 (1997).
short to promote particle sintering (38). A further reduc-
34. Munoz-Paez, A., and Koningsberger, D. C., J. Phys. Chem. 99, 4193
◦
tion under hydrogen at 400 C does not alter the particle
(
1995).
size.
35. Frennet, A., and Wells, P. B., Appl. Catal. 18, 243 (1985).
36. Geus, J. W., and Wells, P. B., Appl. Catal. 18, 231 (1985).
Allthesamplestreatedasdescribedabove(seeSection3) 37. Wells, P. B., Appl. Catal. 18, 259 (1985).
3
3
8. Chen, M., and Schmidt, L. D., J. Catal. 55, 348 (1978).
9. Range, R. J., Rau, F., Klement, U., and Heynes, A. M., Mater. Res. Bull.
2, 1541 (1972).
0. Wyckoff, R. W. G., “Crystal Structures.” Interscience, New York, 1974.
1. Schroder, R. H., Schmitz-Pranghe, N., and Kohlaas, R., Z. Metallkd.
63, 12 (1972).
lead to a metal dispersion of 65 ± 5%, which is close to the
maximum value reported in the literature. Although this
result is encouraging, drying and heating under vacuum is
costly from an industrial point of view. Using time-resolved
mass spectrometry, the next paper will describe what hap-
2
4
4