Preparation of a Pt/SiO2 catalyst: II. Temperature-programmed decomposition of the adsorbed platinum tetrammine hydroxide complex under flowing hydrogen, oxygen, and argon

Article abstract of DOI:10.1016/S0021-9517(03)00189-1

The temperature-programmed decomposition of platinum tetrammine hydroxide mixed with or exchanged on silica in a flow of neutral, oxidizing, or reducing gas is studied using mass-spectrometric analysis of the effluent. A unified interpretation mechanism is proposed whatever the nature of the gas. It involves (≡SiO)₂Pt(NH₃)_y with y ≈ 2 as the intermediate species. This species anchors the particles defined at the drying step. Anchoring takes place at 100-150°C whatever the gas. The subsequent decomposition of this species leads either to Pt⁽⁰⁾ (reducing gas), mobile PtO (oxidizing gas), or a mixture of both (neutral gas). The dispersion of the platinum particles ranges between 70 and 80% for metal loading up to 5% when the decomposition is performed under the reducing gas up to 500°C. In the neutral gas, the dispersion ranges from 42 to 72% and depends on the final temperature and on metal loading. In an oxidizing gas, the dispersion can be about 65% provided that the final temperature does not exceed 250°C. Subsequent reduction at a higher temperature reproduces the decomposition under the reducing gas. When the final temperature is raised above 250°C in the oxidizing gas, the dispersion decreases and falls to 20% at 500°C. It is proved that PtO is responsible for low dispersions. Conversely, under reducing conditions, the size of the particles obtained upon drying is kept unchanged. The proposed mechanism explains why a neutral gas gives intermediate results that depend on pressure and metal loading. In the case of high metal loading, decomposition under the reducing gas leads too small particles grouped together, although dispersion remains within 70-80%.

Full text of DOI:10.1016/S0021-9517(03)00189-1

Journal of Catalysis 220 (2003) 280–290
www.elsevier.com/locate/jcat
Preparation of a Pt/SiO catalyst
2
II. Temperature-programmed decomposition of the adsorbed platinum
tetrammine hydroxide complex under flowing hydrogen,
oxygen, and argon
a
a
b,∗
A. Goguet, D. Schweich, and J.-P. Candy
a
Laboratoire de Génie des Procédés Catalytiques (UMR-CNRS), ESCPE Lyon, 69616 Villeurbanne cedex, France
b
Laboratoire de Chimie Organométallique de Surface (UMR-CNRS), ESCPE Lyon, 69616 Villeurbanne cedex, France
Received 16 January 2003; revised 7 April 2003; accepted 8 April 2003
Abstract
The temperature-programmed decomposition of platinum tetrammine hydroxide mixed with or exchanged on silica in a flow of neutral,
oxidizing, or reducing gas is studied using mass-spectrometric analysis of the effluent. A unified interpretation mechanism is proposed
whatever the nature of the gas. It involves (≡SiO) Pt(NH )y with y ≈ 2 as the intermediate species. This species anchors the particles
2
3
◦
defined at the drying step. Anchoring takes place at 100–150 C whatever the gas. The subsequent decomposition of this species leads either
to Pt (reducing gas), mobile PtO (oxidizing gas), or a mixture of both (neutral gas). The dispersion of the platinum particles ranges between
0 and 80% for metal loading up to 5% when the decomposition is performed under the reducing gas up to 500 C. In the neutral gas, the
(0)
◦
7
dispersion ranges from 42 to 72% and depends on the final temperature and on metal loading. In an oxidizing gas, the dispersion can be about
◦
6
5% provided that the final temperature does not exceed 250 C. Subsequent reduction at a higher temperature reproduces the decomposition
◦
under the reducing gas. When the final temperature is raised above 250 C in the oxidizing gas, the dispersion decreases and falls to 20%
at 500 C. It is proved that PtO is responsible for low dispersions. Conversely, under reducing conditions, the size of the particles obtained
◦
upon drying is kept unchanged. The proposed mechanism explains why a neutral gas gives intermediate results that depend on pressure and
metal loading. In the case of high metal loading, decomposition under the reducing gas leads too small particles grouped together, although
dispersion remains within 70–80%.

2003 Elsevier Inc. All rights reserved.
Keywords: Pt/SiO catalyst; Platinum particles; Platinum tetrammine hydroxide
2
◦
◦
1
. Introduction
NH3 ligand upon drying at 25 C under vacuum. At 100 C
still under vacuum the exchanged complex forms clusters of
In the first part of this paper [1], a detailed study of the
[
Pt(NH3)4(OH)2]n deposited on the surface. Increasing the
exchange, drying, and thermal decomposition processes un-
der vacuum when preparing Pt/SiO2 catalysts from Aerosil
silica (Degussa) and platinum tetrammine hydroxide was de-
scribed. It was demonstrated that exchange reaches comple-
tion 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
temperature under vacuum progressively removes the NH3
◦
ligands from the exchanged complex. At 200 C, about half
◦
of the ammonia ligands are lost and at 300 C the complex
is fully decomposed mainly into PtO particles of 1–2 nm
in size. About 10% of PtO is reduced to Pt( due to re-
duction of Pt by NH3. A further reduction under hydrogen
0)
◦
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
obtained suggest that drying could be the crucial step of the
Pt/SiO2 catalyst preparation.
*
Corresponding author. Laboratoire de Chimie Organométallique de
Surface (UMR CNRS-CPE), ESCPE Lyon, 3 avenue Victor Grignard,
9616 Villeurbanne cedex, France
6
E-mail address: candy@cpe.fr (J.-P. Candy).
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00189-1

A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
281
This second part of the paper describes the thermal de-
composition of the exchanged complex under oxidizing,
neutral (or inert), and reducing atmospheres, using time-
resolved mass spectrometric analyses of the gaseous prod-
ucts evolved.
in the SiC bed before entering the catalyst bed. The temper-
ature of the catalyst was measured at the catalyst bed inlet
by a thermocouple. Residence time distribution experiments
showed that plug flow was achieved.
Literature data exhibit large discrepancies concerning the
influence of the nature of the flowing gas on the size of
the platinum particles. Some authors [2–5] claimed that
the smallest particles were obtained under an oxidizing at-
mosphere (calcination), whereas other authors [6–12] ob-
served that they were obtained under a reducing atmosphere.
Recently, Oudenhuijzen et al. [13] observed that starting
with Pt(NH3)4(NO3)2, the smallest particles were obtained
2.1.2. Procedure
Two grams (6 mm layer) of catalyst was introduced into
the reactor.
The TPD experiments were performed under neutral
(10% He, 90% Ar) reducing (10% He, 10% H2, 80% Ar),
or oxidizing (10% He, 10% O2, 80% Ar) atmospheres, with
a constant flow rate of 150 mL min STP.
The temperature program was as follows: dwell 30 min
at room temperature, ramp at 2 C min up to 500 C, and
−1
◦
◦
−1
◦
after heating at 400 C under a flowing mixture of argon and
◦
helium. Finally, the nature of the flowing gas could have an
influence on the shape of the particles. According to Schmidt
and co-workers [12,14] the platinum particles are perfect
cubes after heating in hydrogen, whereas they are spherical
after heating in oxygen.
then dwell at 500 C for 30 min. For some experiments the
final dwell temperature was 200, 300, or 400 C.
◦
At the end of a TPD experiment, the catalyst was treated
◦
at 500 C under the reducing atmosphere.
In order to understand the decomposition mechanism of
the [Pt(NH3)4(OH)2]n clusters deposited on the silica sur-
face, the nature and the amount of the gases evolved during
the temperature-programmed decomposition under various
flowing gases were studied. The influence of the nature of
the feed gas and the metal loading was investigated as well
as the influence of the final temperature. The various cata-
lysts thus obtained were characterized by chemisorption of
probe molecules (H2 and O2) and electron microscopy and
the relation between decomposition conditions and platinum
particle size was clarified.
2.1.3. Preparation of the samples
2.1.3.1. Support. Two grams of silica (Aerosil from De-
2
−1
gussa, 200 m g ) was treated under flowing dry air at
◦
500 C for 5 h and then introduced into a vessel filled with
60 mL of water. After 1 h of stirring (500 rpm), the solid was
filtered and dried under vacuum at room temperature.
2.1.3.2. Mechanical mixture. Two grams of silica was
treated under flowing dry air at 500 C for 5 h and then
◦
mixed at room temperature in a grinder with 160 mg of
Pt(NH3)4(OH)2 (0.5 mmol, metal loading: 5% w/w). The
hydroxide solution was prepared as described elsewhere [1]
and the solid Pt(NH3)4(OH)2 was obtained by drying the
solution at room temperature under vacuum.
2
2
2
. Experimental
.1. Thermodesorption
2.1.3.3. Ion-exchanged silica. Fifteen grams of silica was
treated under flowing dry air at 500 C for 5 h and then intro-
◦
.1.1. Apparatus
duced into a vessel filled with 450 mL of a Pt(NH3)4(OH)2
−1
A home-made temperature-programmed decomposition
solution (0.25 or 2.0 g L , metal loading: 0.5 or 4% w/w).
After 1 h of stirring (500 rpm), the solid was filtered, washed
with 50 mL of water, and dried under vacuum at room tem-
perature (see [1]). TEM picture of the 4% w/w sample
shows small particles (1.5 to 2 nm diameter) with low con-
trast, similar to that previously reported for 4.5% w/w Pt
(
TPD) apparatus was built according to the previous work
of Falconer and Schwarz [15]. Analysis of the exhaust gas
was performed using a mass spectrometer. The quadrupole
mass spectrometer from Vacuum Generator was calibrated
with respect to the following species: H2O, NH3, N2, N2O,
and NO. Eleven m/e were recorded during the TPD exper-
iments. The five molecular concentrations were then calcu-
lated solving the over determined linear system by singular
value decomposition [16]. The various gases (N2, O2, Ar,
He) were fed to the TPD apparatus through Brooks mass
flow controllers. A constant flow rate of helium was super-
imposed to the main gas stream and was used as an internal
calibration standard.
◦
samples after 25 C vacuum drying Pt(NH3)4(OH)2/SiO2 in
part I (Fig. 5a).
2.2. Characterization of the samples
The various samples were characterized by elemental
analysis, H2 or O2 chemisorption, and electron microscopy,
according to the procedure already described elsewhere [1].
Metal dispersion (D = Pts/Pt) was deduced from H2 or
O2 chemisorption, assuming 1.8 H and 1.0 O per Pts [17,18].
The decomposition reactor was made of a Pyrex tube
3
5 mm i.d. and 270 mm long located in an oven. The cat-
◦
alyst was deposited on a sintered glass support. Then SiC
The samples were reduced under flowing hydrogen at 500 C
3
◦
was used to fill the remaining dead space volume (200 cm )
and then out gassed under vacuum at 300 C before H2
◦
of the reactor. The gas flowed downward and was preheated
chemisorption at 25 C. They were then out gassed under

282
A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
◦
vacuum at 300 C and O2 chemisorption was performed at
2
Only water was detected all along the spectrum and its
◦
−4
−1
5 C. We checked that the difference between dispersions
concentration never exceeded 10
mol L (STP). This
measured with either H2 or O2 was never greater than 10%
and the dispersion reported in this article is the average
value.
means that the water signal shown in subsequent thermo-
decomposition spectra is only significant if it is higher than
10 mol L .
−4
−1
Assuming that the metallic particles are of spherical
shape [19], the platinum particle diameter (d) was deduced
from electron microscopy observation. In the case of plat-
inum, the dispersion is given by D = 1.017/d if d is ex-
pressed in nanometers (nm).
3
.1.2. Thermodecomposition of Pt(NH3)4(OH)2 · xH2O–
SiO2 under neutral atmosphere
The thermodecomposition at increasing temperature un-
der flowing Ar of a mechanical mixture of Pt(NH3)4(OH)2 ·
xH2O and SiO2, and Pt(NH3)4(OH)2 · xH2O exchanged on
silica are reported in Fig. 1.
3
3
3
. Results
After thermal decomposition, the elemental analysis of
samples gave the Pt loading indicated in Table 1. The curves
for the exchange samples were similar whatever the metal
loading; however, they are very noisy at 0.5% w/w and only
the curves corresponding to the greater Pt loading were re-
ported in Fig. 1. For any sample, water was mostly evolved
.1. Thermodecomposition
.1.1. Silica
A blank experiment with 2 g of pure silica was first per-
formed under neutral, reducing, and oxidizing atmospheres.
◦ ◦
as a broad peak from 25 to 180 C. NH3 appeared at 90 C
Fig. 1. Thermodecomposition of a thin layer (2 g) of a mechanical mixture of Pt(NH ) (OH) · xH O and SiO , and of an exchanged sample. Neutral gas
3
4
2
2
2
−1
◦
−1
◦
flow (10% He, 90% Ar), 150 mL min , 2 C min , from 25 to 500 C.

A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
283
◦
Table 1
be guessed (Tb about 120–130 or 170–180 C), although
they are less well defined than under the neutral atmosphere.
Here again, the exchanged catalyst is more reluctant to de-
composition than the mechanical mixture. Conversely, no N2
was detected.
Amounts of gas species evolved per Pt atom under neutral atmosphere
Species (x)
H O
NH
NH
N
2
2
3
3
Mechanical mixture, 5% w/w Pt
◦
T range ( C)
25–150
100
3.4
90–180
126 and 153
1.5
180–320
209 and 281
2.0
180–320
209 and 281
0.3
The quantities of gases evolved in the various domains
are reported in Table 2.
For both the mechanical mixture and the exchange cat-
alyst, the amount of water evolved between 25 and 180 C
is equivalent and slightly above the amount observed under
the neutral atmosphere. The number of nitrogen atoms per
Pt atom was found to be 4.6 for the mechanical mixture and
◦
Tmax ( C)
X/Pt
◦
Exchanged samples: 3.8 and 0.5% w/w Pt
◦
T range ( C)
25–150
100
4.5
90–280
180
2.3
280–500
340
1.4
280–500
340
0.4
◦
Tmax ( C)
X/Pt (3.8% Pt)
X/Pt (0.5% Pt)
–
2.0
1.2
0.2
5
.1 and 3.7 for the exchanged catalyst (3.8%) and (0.5%).
The deviation from 4 will be discussed later.
for both systems. The spectra can be divided into two main
temperature domains depending on whether N2 was evolved
or not. The low temperature domain is made of the first two
3.1.4. Thermodecomposition of Pt(NH ) (OH) · xH O–
3
4
2
2
SiO under oxidizing atmosphere
2
◦
NH3 peaks between 90 and 180 C for the mechanical mix-
Fig. 3 illustrates the same experiment as Fig. 1 using an
◦
ture and of a single broad peak between 90 and 280 C for
oxidizing atmosphere (10% O , 10% He, 80% Ar). After
2
the exchanged catalyst. The second temperature domain is
made of two peaks of NH3 and N2 between 180 and 320 C
thermal decomposition, the elemental analysis of the sam-
◦
ples gave the Pt loading indicated in Table 3. As in Figs. 1
and 2, the effect of the Pt loading is essentially reflected in
the signal/noise ratio.
for the mechanical mixture, and of one peak between 280
◦
and 450–500 C for the exchanged catalyst. Later on, Tb will
represent the temperature at the boundary between the two
domains. These domains have two important features: (a) the
second domain is shifted by more than 100 C toward higher
For both cases, water evolved first as a broad peak from
◦
25 to 150 C and then as a peak with a maximum at 166 or
◦
◦
267 C for the mechanical mixture and the exchanged cata-
◦
temperatures for the exchanged catalyst; (b) the amount of
nitrogen atoms recovered (as NH3 and N2) in each domains
is approximately 50 ± 10% of the amount introduced (as N
in Pt(NH3)4(OH)2 · xH2O) for both systems.
lyst, respectively. As usual, NH appeared at 90 C as a wide
peak with a maximum around 116 or 180 C. The most strik-
3
◦
ing features are: (a) the synchronous narrow peaks of water,
◦
N , and N O at either 166 or 267 C; (b) the trace level of
2
2
The quantities of gases evolved in the various domains
are reported in Table 1. They are calculated by integration
of the curves versus time within the specified temperature
range.
free ammonia after these temperatures.
The quantities of gases evolved in the various domains
are reported in Table 3.
The agreement between the amount of nitrogen atoms re-
covered as NH , N , or N O and the amount of platinum is
For both solids, the amount of water evolved between 25
3
2
2
◦
and 150 C is equivalent and the whole quantity of NH3 ini-
2 2
excellent (4.0 N per Pt). The production of N and N O co-
tially present on the samples was evolved as NH3 mainly and
a small amount of N2. The number of nitrogen atoms per Pt
atom was found to be 4.1 for the mechanical mixture and 4.5
and 3.6 for the exchanged catalysts (3.8%) and (0.5%). The
deviation from 4 will be discussed later.
incides with the production of about 3.9 H O molecules per
Pt atom. This means that there are from 5.5 to 5.9 other wa-
2
ter molecules per Pt. This is in agreement with the 5 H O/Pt
2
found under the reducing atmosphere.
3
.2. Metal dispersion
3
.1.3. Thermodecomposition of Pt(NH3)4(OH)2 · xH2O–
SiO2 under reducing atmosphere
3.2.1. Influence of metal loading and of the atmosphere
Fig. 2 illustrates the same experiment as Fig. 1 except that
the gas contained 10% of hydrogen. After thermal decom-
position, the elemental analysis of the samples gave the Pt
loading indicated in Table 2. As in Fig. 1, peak locations are
independent of Pt loading of the exchanged samples. Con-
versely, the lower the Pt loading, the more noisy the curves,
and the less accurate the mass balances.
The dispersion of the metallic Pt particles obtained after
◦
reduction at 500 C at two metal loading was measured by
H and O chemisorption and electron microscopy. The re-
2
2
sults are reported in Table 4 and illustrated in Fig. 4.
In both cases, the dispersion of the metallic particles de-
pends drastically on the nature of the gas. It is obvious that
decomposition under oxidizing conditions gives very low
dispersions whereas decomposition under reducing condi-
tions gives high dispersions. Curiously, under neutral at-
mosphere, the dispersion of the metallic particles depends
on metal loading. However, for high metal loading (3.8% Pt),
the particles are grouped together as islands or strings. This
Here again, water was mostly evolved as a broad peak
◦
◦
from 25 to 180 C, and NH3 appeared at 90 C for both
systems. Clearly, the presence of H2 in the flowing gas
promoted complex decomposition in a narrower and lower
range of temperatures. Two temperature domains can again

284
A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
Fig. 2. Same as Fig. 1, reducing gas flow (10% He, 10% H , 80% Ar).
2
Table 2
dizing atmosphere. Conversely, the dispersion was found to
be nearly independent of the final temperature under the re-
ducing atmosphere. As expected, an intermediate behavior
was observed with the neutral atmosphere.
Amounts of gas species evolved per Pt atom, under reducing atmosphere
Species (x)
H O
2
NH
3
Mass-spectrometric analyses of the effluent were per-
formed during the reduction step. It was found that: (a) no
nitrogen-containing species were detected when the sample
Mechanical mixture, 5% w/w Pt
◦
T range ( C)
25–150
106
5.0
90–250
106
4.6
◦
Tmax ( C)
X/Pt
◦
was previously decomposed up to 500 C under a neutral,
Exchanged samples: 3.8 and 0.5% w/w Pt
reducing, or oxidizing atmosphere; (b) a broad NH3 peak ex-
◦
◦
◦
T range ( C)
25–150
90–350
189
5.1
tending from 150 to 300 C (maximum about 190–200 C)
was detected for samples previously decomposed under the
◦
Tmax ( C)
X/Pt (3.8% Pt)
X/Pt (0.5% Pt)
90
5.3
–
◦
neutral atmosphere up to 300 C or under the oxidizing at-
3.7
◦
mosphere up to 250 C. This peak was similar to the NH3
peak obtained above Tb under the reducing atmosphere.
structure was not observed when the catalyst (4.5% Pt) was
decomposed under vacuum (see part I) [1].
4
. Discussion
3
.2.2. Influence of the final decomposition temperature
Fig. 5 shows the drastic decrease of the dispersion with
The mass spectrograms shown in Figs. 1 to 3 clearly il-
increasing final decomposition temperature under the oxi-
lustrate the differences in behavior of the samples according

A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
285
Fig. 3. Same as Fig. 1, oxidizing gas flow (10% He, 10% O , 80% Ar).
2
Table 3
Table 4
Amounts of gas species evolved per Pt atom, under oxidizing atmosphere
Pt dispersions according to metal loading and gas nature
Species (x)
Metal loading
Atmosphere
Reducing
H O
2
H O
2
NH
N
N O
2
(w/w)
Neutral
Oxidizing
3
2
(
a) Mechanical mixture, 5% w/w Pt
0.5%
3.8%
42%
72%
80%
70%
26%
26%
◦
T range ( C)
25–150 150–350 80–150 150–300 150–300
◦
Tmax ( C)
100
3.6
166
6.2
116
1.4
166
0.9
166
0.4
X/Pt
mass balances reported in Tables 1 to 3 is about 20% (from
the amount of total nitrogen recovered per platinum atom).
This can be attributed to baseline drift, calibration uncer-
tainty, interference between the various m/e of the products,
and determination of metal loading by elemental analysis.
As a consequence, the mass balances can only be used for
comparing orders of magnitude. Nevertheless, we can ob-
serve that the amounts of NH , N , and N O evolved are
Exchanged samples: (b) 3.8 and (c) 0.5% w/w Pt
◦
T range ( C)
25–150 150–500 80–260 180–500 180–500
◦
Tmax ( C)
100
4.1
–
267
5.3
–
180
1.4
1.6
267
0.9
0.8
267
0.4
0.4
X/Pt (3.8% Pt)
X/Pt (0.5% Pt)
to the method of preparation (exchanged complex or me-
chanical mixture) and/or the atmosphere under which they
were treated. The quantitative analyses of the spectra are
more difficult. One may observe that the uncertainty in the
3
2
2
proportional to the Pt loading.
After drying at room temperature under vacuum, the plat-
inum tetrammine hydroxidecomplex forms [Pt(NH3)4(OH)2

286
A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
Fig. 4. TEM pictures of samples after treatment under neutral, reducing, or oxidizing atmosphere.
exchanged complex is stabilized by the support which means
that it exhibits chemical interaction with the surface ≡SiOH
ꢀ
groups, as it will be explained below [Eq. (2 )].
◦
In the range 25–180 C, dehydration of the silica surface
occurs together with dehydration of the platinum complex
to give [Pt(NH3)4(OH)2]n clusters for both the mechanical
mixture and the exchanged complex. Referring to a single
Pt atom or ≡SiOH group, this dehydration process obeys
ꢀ
Eqs. (1) and (1 ) which explain the broad water peak in
the low-temperature domain whatever the sample and feed
gas:
Pt(NH3)4(OH)2 · xH2O → Pt(NH3)4(OH)2 + xH2O,
(1)
ꢀ
ꢀ
ꢀ
≡
SiOH · x H2O → ≡SiOH + x H2O.
(1 )
Fig. 5. Dispersion of the samples obtained after decomposition under neu-
tral (Ar), oxidizing (O ), and reducing (H ) atmospheres at increasing tem-
perature, followed by reduction under flowing hydrogen at 500 C. Metal
loading: 1.5%.
Under both neutral or oxidizing atmospheres, about half
2
2
◦
of the NH3 ligands are lost in the low-temperature do-
main whereas the remaining ligands are lost at a much
higher temperature together with N2 (and N2O). The tem-
perature boundary, Tb, between these domains depends
on the solid and on the feed gas composition. Compar-
ing the mechanical mixtures and the exchanged samples
shows that the low-temperature domains are more similar
than the second domains. We may thus interpret the re-
b
·
xH2O]n clusters, regardless of whether they are supported
on silica or unsupported (see part I) [1].
Figs. 1 to 3 show that the decomposition profiles of the
[
Pt(NH3)4(OH)2 · xH2O]n clusters are similar to those ob-
served with the mechanical mixture, but a clear shift of
the peaks toward high temperature is observed for the ex-
changed samples. It is obvious that above ca. 150 C, the
sults as follows: Between 90 and T , under argon or oxygen
(Figs. 1 and 3), the [Pt(NH ) (OH) ] clusters of the me-
3
4
2 n
◦
chanical mixture lose about half of the NH3 ligands and

A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
287
the remaining water to form [OPt(NH3)y]n species, follow-
Combining (3) and [4] and assuming complete reduction
ing Eq. (2):
of PtO and y = 2, one obtains the following balance:
Pt(NH3)4(OH)2 → OPt(NH3) + (4 − y)NH3
(≡SiO)2Pt(NH3)2
y
4
1
+
H2O, y ≈ 2.
(2)
ꢀ
(4 )
→
Pt + 3NH3 + N2 + H2O + ≡Si–O–Si≡.
3
In this low temperature domain, the [Pt(NH3)4(OH)2]n
From Table 1 we observe 1.4 NH3 and 0.4 N2 per Pt in-
clusters resulting from the exchange process could melt on
ꢀ
stead of 4/3 and 1/3, respectively following (4 ). It is thus
◦
the surface above 100 C and then could become “rafts”
concluded that most of PtO is reduced to Pt( in the exper-
iment of Fig. 1. As a consequence the dispersion is almost
as high as under a reducing atmosphere as shown by Table 4
0)
(
See [1]), where Pt atoms are close to the oxygen atoms of
the silanol groups, and form (≡SiO)2Pt(NH3)y species, fol-
ꢀ
lowing Eq. (2 ):
(
metal loading 3.8%). The effect of NH3 partial pressure on
the extent of partial reduction is further demonstrated by the
decomposition under vacuum. In the latter case, the partial
pressure was very low and only about 10% of the platinum
was reduced (Part I) [1]. Similarly, at low metal loading, the
partial pressure of NH3 is low when the sample is reduced
under an inert atmosphere, then, the extent of partial reduc-
tion and thus the dispersion are low (Table 4, metal loading
0.5%). All these quantitative observations give support to the
proposed description.
Pt(NH3)4(OH) + 2 ≡SiOH → (≡SiO)2Pt(NH3)
2
y
ꢀ
(2 )
+
(4 − y)NH3 + 2H2O, y ≈ 2.
This was already observed when the tetrammine complex
decomposes under vacuum (see part I, EXAFS studies) [1].
Later on, we will speak of “reactions” like (2) and (2 ). How-
ever, one must keep in mind that these “reactions” are no
more than atom balances with no mechanistic meaning. In
other words, the chemical formulas cited in the reactions do
not reflect necessarily the structure of the species.
ꢀ
◦
Under oxygen (Fig. 3), at 166 C (mechanical mixture)
◦
It would be hazardous to claim that reactions (2) and
2 ) occur alone in the mechanical mixture and on the ex-
or at 267 C (exchanged sample), there is a sudden oxida-
ꢀ
(
tion of the remaining NH3 ligands with formation of H2O,
N2, and N2O. These reactions are strongly exothermic and
the decomposition is very fast; the thermocouple at the bed
inlet recorded an overshoot of several degrees centigrade
at the same time. The overall mechanism can be described
mainly by Eqs. (5) and (6) for the mechanical mixture and
changed catalyst, respectively. The mechanical mixture was
obtained in a grinder under ambient conditions and traces of
exchanged complex cannot be excluded owing to ambient
humidity; furthermore a fraction of the complex in mechan-
ꢀ
ical contact with silica may react according to (2 ). It is
ꢀ
ꢀ ꢀ
by Eqs. (5 ) and (6 ) for the exchanged sample:
evident that reactions (2) and (2 ) could also compete on the
ꢀ
exchanged catalyst. For reactions (2) and (2 ), one or two
3
y
3y
y
H2O + N2,
2
water molecule(s) per Pt atom should be evolved in the first
domain, respectively. Tables 1 and 3 shows that more water
is released from the exchanged catalyst than from the me-
OPt(NH3) +
O2 → PtO +
(5)
y
4
2
3
y
(
≡SiO)2Pt(NH3) +
O2
y
4
◦
chanical mixture in the range 25–150 C. Unfortunately, the
3
y
y
ꢀ
→
PtO + ≡Si–O–Si≡ + 2 H2O + N2,
(5 )
difference falls within the uncertainty of the water mass bal-
ances.
2
3
y
y
OPt(NH3) + yO2 → PtO +
H2O + N2O,
(6)
Above Tb, under argon (Fig. 1), the remaining NH3 lig-
y
2
2
ands are progressively removed from the complex, leading
(
≡SiO)2Pt(NH3) + yO2
to Pt(0) and PtO particles or particles made of a mixture of
y
3
y
y
metal and oxide (part I) [1]. PtO is produced either by re-
ꢀ
→
PtO + ≡Si–O–Si≡ + 2 H2O + N2O.
(6 )
ꢀ
2
action (3) (mechanical mixture) or reaction (3 ) (exchanged
catalyst):
These reactions are only indicative of the mechanism.
Nothing excludes that the complex does release gas-phase
NH3 which is then rapidly oxidized to N2 and N2O. The last
water and nitrogen peaks of the mechanical mixture are ob-
OPt(NH3) → PtO + yNH3,
(3)
y
ꢀ
(
≡SiO)2Pt(NH3) → PtO + yNH3 + ≡Si–O–Si≡.
(3 )
y
◦
served at T ≈ 270 C, i.e., close to the sharp peak of the
◦
The presence of small amounts of dinitrogen and water
exchanged catalyst at T = 267 C (see Fig. 3). This can be
indicates that a part of PtO is reduced by NH3, to form Pt(
H2O, and N2 [Eq. (4)] as was previously proposed (part I)
1,13,20].
0)
,
ꢀ
explained by exchange during grinding or reaction (2 ) at the
mechanical contact between the complex and silica.
[
Under hydrogen (Fig. 2), the complex fully decomposes
at temperatures lower than under neutral or oxidizing at-
mospheres. The two temperature domains still exist, al-
though they are overlapping. The interpretation used for
neutral and oxidizing atmospheres can thus be extended. At
3
PtO + 2NH3 → 3Pt + 3H2O + N2.
(4)
Comparing the heights of the H2O and N2 peaks at
◦
3
40 C (Fig. 1) gives a ratio H2O/N2 of 2.4 which agrees
◦ ꢀ
about 100–120 C reactions (2) and (2 ) take place leading to
with reaction (4) within 20%.

288
A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
OPt(NH3)y and (≡SiO)2Pt(NH3)y. At higher temperatures
to either metallic or oxide particles. Dispersion as high as
80% can be obtained under a reducing atmosphere at a tem-
perature as high as 500 C while, under the same conditions,
an oxidizing atmosphere leads to a low dispersion. These
results are in agreement with the works of Benesis et al.
◦ ◦
Tb about 120 C, mechanical mixture; Tb about 180 C,
(
◦
exchanged sample) these species decomposes according to
ꢀ
Eqs. (7) and (7 ) where adsorption of H on Pt is ignored.
OPt(NH3) + H2 → Pt + H2O + yNH3,
(7)
y
[
10], Brunelle et al. [7], or Dorling et al. [6] who found
(
≡SiO)2Pt(NH3) + 2H2
that smaller particles were obtained by decomposing the
exchanged tetrammine complex under reducing conditions.
Conversely, they contradict the observations of Gonzalez
and co-workers [2–5], and they partially agree with those
of Oudenhuijzen et al. [13].
y
ꢀ
→
Pt + ≡Si–O–Si≡ + yNH3 + H2O.
(7 )
The overlap between the temperature domains may be ex-
plained by a decomposition catalyzed by reduced platinum
ꢀ
or activation energies of reactions (7) and (7 ) smaller than
Gonzalez and co-workers [2–5] showed that small par-
ticles can be obtained under oxidizing conditions whereas
a reducing atmosphere leads to larger particles. These au-
thors suggested that under an oxidizing atmosphere, there
ꢀ
ꢀ
ꢀ
those of (3), (3 ), (5), (5 ), (6) and (6 ). The heights of the
water and ammonia peaks give support to this interpreta-
tion. For the mechanical mixture, the heights of the peaks
◦
at 106 C give a ratio H2O/NH3 of 0.9. According to Eq. (2)
was formation of Pt2 species in strong interaction with the
support and that under a reducing atmosphere, the platinum
hydride diammine formed was very mobile on the surface,
leading to large metallic particles. Oudenhuijzen et al. [13]
used the same explanation for comparing decomposition un-
der Ar/He and H2. The work presented here does not al-
low proposing the formation of platinum hydride diammine
since no water would be released when it decomposes (see
+
this would imply y = 2.9. Conversely, the heights of the
◦
peaks at 189 C for the exchanged catalyst give a ratio of 0.4
ꢀ
which leads to y = 2.5 according to (7 ). It must be noted
that because of peak overlap there may be some error in the
interpretation of peak heights. However, considering reac-
ꢀ
ꢀ
tions (2) and (2 ) (any atmosphere), (3), (3 ), and (4) (neutral
atmosphere), (5), (5 ), (6), and (6 ) (oxidizing atmosphere),
7) and (7 ) (reducing atmosphere) and inserting y is close to
ꢀ
ꢀ
ꢀ
(
◦
water peak at 189 C, Fig. 2). Conversely, formation of mo-
2
lead to a unified and coherent interpretation summarized
bile PtO under an oxidizing atmosphere is known [21,22].
by Scheme 1.
◦
Bulk PtO does not melt but decomposes at 550 C and it
According to our preparation method, two key steps gov-
ern the size of the resulting metallic particles. The first one
is difficult to propose that large PtO particles can migrate
quickly on the silica surface. If we consider that the heat of
◦
takes place between 100 and 150–200 C (i.e., Tb) and it
ꢀ
ꢀ
−1
combustion of NH3 is about 293 kJ mol (for comparison,
involves reactions (1), (1 ), (2 ), and to a lesser extent re-
action (2). Reaction (2 ) is responsible for anchoring of the
1
−1
ꢀ
H2 + O2 = H2O is 286 kJ mol ), the exothermicity of the
2
complex and thus determines the starting particle size. Since
y is found close to 2, one may assume that the anchoring
species is (≡SiO)2Pt(NH3)2. The second key step is the de-
ligand combustion could trigger the sintering of the newly
formed oxide. Fig. 6 shows the dispersion of the metallic
particles reported in the literature for oxidizing atmospheres,
as a function of the final decomposition temperature. All the
◦
composition of this species above 150–200 C which leads
Scheme 1. Simplified overview of the changes on the surface during treatments of the exchanged sample with the three gasses.

A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
289
Oudenhuijzen’s spectrograms under O2 and Ar/He are more
reminiscent of our results obtained with the mechanical mix-
tures than those obtained with the exchanged samples. The
authors explained their results by the formation of HNO3
and Pt(NH3)4(OH)2 during impregnation and drying, and
then of NH4NO3 and Pt( (under H2 and Ar/He) during the
early stage of the decomposition process. The subsequent
decomposition of NH4NO3 is then responsible for various
gaseous species according to the atmosphere. If they are
true, the decomposition of their Pt(NH3)4(OH)2 intermedi-
ate should be similar to that observed in our experiments
although they are not. We have no explanation for these dif-
ferences; we can only invoke the nature of the counteranion
0)
− −
(
NO3 versus OH ), the impregnation method (incipient
wetness versus solid-liquid suspension), and the nature of
the support (porous silica versus nonporous aerosil).
Fig. 6. Dispersion of the metallic particles as a function of the final decom-
position temperature under flowing oxidizing atmosphere, from literature
data and from this work.
Let us now compare the results obtained by decomposing
the complex either under a reducing atmosphere or under
vacuum (see Part I) [1]. We can observe that the size of the
metallic particles measured by electron microscopy and the
dispersion of the samples measured by chemisorption of H2
results point toward the negative effect of too a high final
temperature.
It is worth noting that dispersions as high as 70% were
obtained under oxidizing atmospheres provided that the final
temperature does not exceed 200–250 C, i.e., just below the
and O are the same in both cases, respectively, about 1.5 nm
and 70%. However, as demonstrated by electron microscopy,
2
◦
the metallic particles are isolated from each other and regu-
larly distributed on the support when decomposition is done
under vacuum (see Part I) [1] and they are, in some cases, ag-
glomerated when decomposition is performed under flowing
gas (Fig. 4). There is no clear explanation for this observa-
tion. It is only suggested that some gaseous species (e.g.,
water) could be responsible for the metallic particle mobil-
ity under nonvacuum conditions. Conversely, under vacuum,
the partial pressure (of water?) would be to small to promote
mobility.
temperature at which NH3 oxidation is “fired” (see Fig. 3).
This means that samples decomposed under oxidizing at-
◦
mospheres up to 200–250 C are only partially decomposed
into anchored (≡SiO)2Pt(NH3)y. Further reduction above
◦
2
50 C then reproduces the decomposition under a reducing
atmosphere and leads to stable and small Pt particles. This is
supported by the effluent analysis during reduction (see Sec-
tion 2.1.3). It is then clear that the increase in particle size
◦
which occurs above 350 C, under oxidizing atmospheres, is
due to mobile PtO species as proposed above. If this inter-
pretation reconciles the results of Gonzalez and co-workers
[
2–5] with those of other authors, it does not explain why
5. Conclusion
Gonzalez and co-workers obtained low dispersions under a
reducing atmosphere. Under a neutral atmosphere, the ex-
From the results of the two parts of this study, the decom-
position of exchanged platinum tetrammine complex under
flowing neutral, reducing, or oxidizing atmosphere can be
interpreted by a stepwise process.
◦
changed complex decomposes above 300 C and particles of
both PtO and Pt( are formed, depending on metal loading.
0)
It is then logical that the dispersion measured after treat-
◦
ment above 300 C under a neutral atmosphere lies between
◦
that measured under reducing and oxidizing atmospheres
• First, upon drying up to about 100 C, the exchanged
(
Fig. 5).
Oudenhuijzen et al. [13] used mass spectrometry and
complex loses the electrostatic interaction with the
−
≡SiO groups and then loses water to give
Quick EXAFS spectroscopy to study the decomposition of
Pt(NH3)4(NO3)2. These authors observed that complete de-
[Pt(NH ) (OH) ] clusters.
3
4
2 n
◦
• At about 100 C, these clusters melt on the surface to
form “rafts,” where Pt atoms are close to the oxygen
atoms of the silanol. The surface complex loses some
NH ligands and forms (≡SiO) Pt(NH ) species an-
composition of the Pt(NH3)42 complex under hydrogen
+
◦
occurs at a much lower temperature (150–200 C) than un-
der argon/helium (200–250 C). They also observed, as we
◦
3
2
3 y
did, that decomposition under oxygen gives a low disper-
sion. These observations partially agree with our results.
Conversely, the thermo decomposition profiles measured by
Oudenhuijzen et al. under argon/helium and oxygen did not
chored to the surface with y of the order of 2. This
step is independent of the nature of the flowing gas.
The melting/anchoring process freezes the particle size.
Since the size is independent of whether the complex is
supported or not, this is the size of the clusters obtained
◦
show the peaks that we observed in the range 250–350 C
◦
(
see Figs. 1 and 3). If we compare the temperature ranges,
upon drying up to about 100 C.

290
A. Goguet et al. / Journal of Catalysis 220 (2003) 280–290
•
The final step takes place at a higher temperature and in-
volves the loss of the remaining NH3 ligands. It depends
on the nature of the flowing gas:
Under reducing conditions, the final dispersion is inde-
pendent of the metal loading of the catalyst. However, for
high loading, Pt particles may agglomerate whereas ag-
glomeration is not observed if decomposition is performed
under vacuum (see Part I). Under pressure, the higher the
metal loading, the higher the partial pressure of the released
species in the catalyst bed. This means that agglomera-
tion is due to one of the released species. Since water is
a candidate, further experiments with controlled water va-
por pressure would be necessary. If water is responsible for
agglomeration, higher gas flow rates should be used when
decomposing the catalyst under hydrogen. All the observa-
tions point toward a particle size defined at drying by the
size of [Pt(NH3)4(OH)2]n clusters deposited on the silica
surface. It is not clear which process is responsible for the
size of these clusters.
◦
◦
Above 250 C under oxygen, there is a sudden ox-
idation of the NH3 ligands with formation of PtO
particles. When the temperature further increases, the
small and mobile PtO particles merge which leads to
a low metal dispersion. Alternatively, the sudden and
exothermic NH3 oxidation may promote particle sin-
tering. The higher the final temperature, the higher the
mobility of PtO and the lower the dispersion. Below
◦
2
50 C, the complex is not fully decomposed.
◦
◦
◦
Between 150 and 250 C under a hydrogen flow, NH3
(0)
and H2O are released while Pt is formed. The par-
ticle size remains that defined at the previous step.
H2O release in the high temperature domain proves
that platinum hydride diammine cannot be a signifi-
cant intermediate species.
◦
References
Above 300 C under argon, NH3 is released while
PtO is formed. However, a fraction of NH3 reduces
[1] A. Goguet, M.A. Aouine, F.J. Cadete Santos Aires, A. De Mallmann,
PtO to Pt( . Since both Pt and PtO are coexist-
ing, the particle size is intermediate between those
obtained under reducing and oxidizing atmospheres.
For a given gas flow rate, the higher the metal load-
ing, the higher the partial pressure of released NH3.
It is thus logical that the extent of self-reduction,
and thus the dispersion increase with metal load-
ing.
0)
(0)
D. Schweich, J.P. Candy, J. Catal. 209 (2002) 135.
[2] S. Alerasool, R.D. Gonzalez, J. Catal. 124 (1990) 204.
[3] R.D. Gonzalez, H. Miura, Rev. Sci. Eng. 36 (1994) 145.
[
[
[
[
4] W. Zou, R.D. Gonzalez, J. Catal. 133 (1992) 202.
5] W. Zou, R.D. Gonzalez, Catal. Today 15 (1992) 443.
6] T.A. Dorling, B.W. Lynch, R.L. Moss, J. Catal. 20 (1971) 190.
7] J.P. Brunelle, A. Sugier, J.F. Lepage, J. Catal. 43 (1976) 273.
[8] G.R. Wilson, W.K. Hall, J. Catal. 24 (1972) 306.
[9] K. Tsurimi, K. Inoue, Kagaku Kogaku Ronbunshu 22 (1996) 542.
10] A.H. Benesis, R.M. Curtis, H.P. Studer, J. Catal. 10 (1968) 328.
11] T.A. Dorling, M.J. Eastlake, R.L. Moss, J. Catal. 14 (1969) 23.
12] L.D. Schmidt, K.R. Krause, Catal. Today 12 (1992) 269.
13] M.K. Oudenhuijzen, P.J. Kooyman, B. Tappel, J.A. van Bokhoven,
D.C. Koningsberger, J. Catal. 205 (2002) 135.
[
[
[
[
According to this interpretation, decomposition of
Pt(NH3)4(OH)2 under a reducing atmosphere is the best
preparation method which leads to dispersions of 70–80%
for metal loadings up to 5% w/w. Decomposition under a
neutral atmosphere is efficient only when significant self-
reduction takes place. In that case, decomposition under
reduced pressure is not recommended. Decomposition un-
[14] T. Wang, L.D. Schmidt, Surf. Sci. 163 (1985) 115.
[
[
15] J.L. Falconer, J.A. Schwarz, Catal. Rev.-Sci. Eng. 31 (1983) 141.
16] A. Goguet, Ph.D. thesis, Claude Bernard University, UCB-Lyon I,
No. 34-2001 (2001).
[17] J.P. Candy, P. Fouilloux, A.J. Renouprez, J. Chem. Soc., Faraday
◦
der an oxidizing atmosphere above 250 C gives the least
Trans. I 76 (1980) 616.
dispersed catalysts. Oudenhuijzen et al. [13] who studied
the decomposition of Pt(NH3)4(NO3)2 found that the best
dispersion was obtained under Ar/He and suggested that
Pt(NH3)4(OH)2 be used to avoid the “multistep reduction”
during autoreduction. Our results show that autoreduction is
not sufficient especially at low metal loading.
[18] A. Frennet, P.B. Wells, Appl. Catal. 18 (1985) 243.
[
19] A.S. Ramachandran, S.L. Anderson, A.K. Datye, Ultramicroscopy 51
1993) 282.
(
[20] A.C.M. Van den Broek, J. Van Grondelle, R.A. Van Santen, J. Ca-
tal. 167 (1997) 417.
[21] M. Chen, L.D. Schmidt, J. Catal. 55 (1978) 348.
[22] H.H. Lee, E. Ruckenstein, Catal. Rev.-Sci. Eng. 25 (1983) 475.