Transformation of platinum supported on silicon-doped alumina during the catalytic decomposition of energetic ionic liquid

Article abstract of DOI:10.1016/j.jcat.2005.02.005

The stability of 10 wt% platinum supported on Si-doped alumina (Pt/Si-Al₂O₃) during the catalytic decomposition of energetic ionic liquid (or propellant) at 40 °C was studied. The reaction was performed by successive injections of 79 wt% HAN (hydroxylammonium nitrate NH₃OHNO₃) aqueous solution onto the catalysts in a constant volume batch reactor. The four studied catalysts remain active with a fast reaction rate (up to 450 mbar s^-1), even in the presence of an excess of residual aqueous solution. For all catalysts, the structural studies reveal the presence of theta alumina associated with silica and platinum particles. After the propellant decomposition, alteration in metallic particle size is observed as is the formation of nanosized platinum agglomerates in the reaction medium, which are probably responsible for the excellent catalytic activity. A model is suggested, explaining the part played by platinum and silica particles at the surface of the materials during the propellant decomposition reaction.

Full text of DOI:10.1016/j.jcat.2005.02.005

Journal of Catalysis 232 (2005) 10–18
www.elsevier.com/locate/jcat
Transformation of platinum supported on silicon-doped alumina during
the catalytic decomposition of energetic ionic liquid
L. Courtheoux, E. Gautron, S. Rossignol ∗, C. Kappenstein
University of Poitiers, LACCO UMR 6503, Laboratoire de Catalyse par les Métaux, 40 Avenue du Recteur Pineau, F-86022 Poitiers cedex, France
Received 8 November 2004; revised 2 February 2005; accepted 3 February 2005
Available online 7 April 2005
Abstract
The stability of 10 wt% platinum supported on Si-doped alumina (Pt/Si–Al O ) during the catalytic decomposition of energetic ionic
2
3
◦
liquid (or propellant) at 40 C was studied. The reaction was performed by successive injections of 79 wt% HAN (hydroxylammonium
nitrate NH OHNO ) aqueous solution onto the catalysts in a constant volume batch reactor. The four studied catalysts remain active with a
3
3
−1
fast reaction rate (up to 450 mbar s ), even in the presence of an excess of residual aqueous solution. For all catalysts, the structural studies
reveal the presence of theta alumina associated with silica and platinum particles. After the propellant decomposition, alteration in metallic
particle size is observed as is the formation of nanosized platinum agglomerates in the reaction medium, which are probably responsible for
the excellent catalytic activity. A model is suggested, explaining the part played by platinum and silica particles at the surface of the materials
during the propellant decomposition reaction.
2005 Elsevier Inc. All rights reserved.
Keywords: Sol–gel; Platinum-based catalysts; Ionic liquid; Doped alumina; Characterizations; XRD; TEM
1. Introduction
abatic temperature up to 1800 ◦C) and the need for frequent
restarts that involve preheating of the catalyst (300–400 ◦C),
which is currently a drawback for the development of the
engine. Therefore, a high catalytic activity at lower temper-
ature (20–200 ◦C) associated with a high thermal stability of
the catalysts is a critical parameter for the future develop-
ment of new engines.
Previous studies performed by our group have shown that
platinum supported on thermally stable silicon-doped alu-
mina displays a good activity at low temperatures [11]. The
Si-doped alumina support is obtained by the sol–gel pro-
cedure, and the introduction of the silicon precursor makes
possible better thermal stabilization of the transition alumi-
na [12]. The catalysts are then evaluated for the decomposi-
tion of propellants with the use of a lab-made constant vol-
ume batch reactor [7]. The decomposition of a HAN–water
mixture on these catalysts begins at a very low temperature,
40 ◦C or less [13], whereas the thermal decomposition tem-
perature is in the range of 115–120 ◦C [9]. Another key point
concerns the long-term stability of the catalysts, particularly
in the presence of a large amount of propellant. Most of the
Energetic liquid compounds known as monopropellants
are used for propulsion and gas generation. For example, the
orbit and attitude control of satellites is achieved with small
thruster engines, with the use of the catalytic decomposition
of liquid hydrazine (N2H4) on supported iridium catalysts
(Ir/Al2O3) [1]. The high toxicity of this monopropellant cre-
ates high storage and handling costs due to more severe reg-
ulation rules, and its replacement by less toxic propellants
is of current interest [2–5]. The most currently proposed
and studied hydrazine substitutes are energetic ionic liquids
and a representative mixture containing hydroxylammonium
nitrate (HAN, NH3OH+NO3−) as an oxidizer, water, and
an ionic or molecular fuel as a reducer [6–10]. However,
such mixtures involve more drastic conditions because of
the high temperature reached during the decomposition (adi-
*
Corresponding author. Fax: 33 (0)5 49 45 40 20.
E-mail address: sylvie.rossignol@univ-poitiers.fr (S. Rossignol).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.02.005
2005 Elsevier Inc. All rights reserved.

L. Courtheoux et al. / Journal of Catalysis 232 (2005) 10–18
11
papers concerning this topic are available as AIAA confer-
ence proceedings [14].
electron microscope (Philips CM 120) coupled with an en-
ergy dispersive X-ray spectrometer (EDX) made it possible
to determine the particle size and the distribution of Pt, Al,
and Si atoms by X-ray mapping at the corresponding en-
ergies (Pt-Lα at 9.441 keV; Al-Kα at 1.486 keV; Si-Kα at
1.739 keV).
The objectives of this work were to evaluate the catalytic
activity during several successive injections of propellant, in
order to simulate the pulse mode of the satellite thrusters,
and to characterize the catalysts before and after the re-
action, by X-ray diffraction (XRD), transmission electron
microscopy (MET), and specific surface area measurements
(SBET).
We performed the catalyst evaluation by decomposing
an aqueous solution of HAN (79 wt% hydroxylammonium
nitrate; NH3OH+NO3−) in a constant volume batch reac-
tor (167 cm3) [7], in the increase temperature mode or the
isothermal mode at 40 ◦C. This last temperature was spec-
ified in a previous work, where we could show that the
prepared 10 wt% Pt catalysts permitted the decomposition
of the HAN solution at about 40 ◦C [13]. When the cata-
lyst is heated from the bottom, the reactor displays a tem-
perature gradient, in which the top is the cold part; thus
the gas temperature is lower than the catalyst tempera-
ture. In the temperature increase mode, 100 µl of propellant
(1.23 mmol; density 1.50 g cm3) is injected at room temper-
ature on 160 mg of catalyst, and the reactor is heated with a
ramp of 10 K min−1. For the isothermal mode, we proceeded
to make successive manual injections of 100 µl of propel-
lant (i.e., 150 mg, 1.23 mmol HAN) on 160 mg of catalyst.
Each injection led to fast increases in pressure and tempera-
ture, which returned to equilibrium values within 3 to 4 min;
therefore, each injection was carried out after an interval of
approximately 4 min. Pressure and temperature variations
against time made it possible to determine the evolution of
the catalyst performances in the presence of a large cumula-
tive amount of propellant (HAN/Pt molar ratio in the range
of 225–323 for 15 injections).
2. Experimental
All catalysts are based on 10 wt% platinum supported
on silicon-doped alumina. The support was prepared in our
laboratory by the sol–gel method and has already been de-
scribed [15,16]. We have prepared samples of nominal com-
position (Al2O3)0.88–(SiO2)0.12 (atomic Si/Al ratio 6:94)
according to the Yoldas procedure [17] under an excess of
water (H2O/Al molar ratio 100:1). The synthesis was per-
formed with Al(O–secC4H9)3 as a molecular precursor. After
1 h at 60 ◦C, the corresponding amount of Si(OC2H5)4 was
slowly added, followed by a small quantity of HCl (Al +
Si/HCl molar ratio 1:0.07). The temperature was raised to
80 ◦C and maintained for 2 h in a covered beaker to avoid
solvent evaporation. Then the beaker was left open for sev-
eral hours at the same temperature, and gelling occurred.
After gelling and cooling, the sample was dried at 120 ◦C
for 12 h in an oven, leading to the xerogel form, and was
then fired for 5 h at 1200 ◦C.
The metallic phase (10 wt%) is introduced by impregna-
tion of H2PtCl6 precursor solution, followed by reduction
(r) at 400 ◦C (4 h). The sample name is 10rx (10 for 10 wt%
Pt; r for reduced sample; x for xerogel support). Some cata-
lysts have been prepared from two successive impregnations,
each corresponding to 5 wt% platinum, on xerogel supports,
with an intermediate reduction between impregnations (sam-
ple 5r + 5rx) or without this intermediate reduction (sample
(5 + 5r)x). For development applications, a sphere-shaped
sample (1 mm diameter) was made from a xerogel powder
and impregnated with the platinum precursor (three impreg-
nations followed by reduction for a total of 10 wt% Pt); this
sample is called 3i.Pt.MF.
To identify the gaseous reaction products, a dynamic re-
actor was designed [19]. This was a fixed-bed reactor with
online analysis by a mass spectrometer (Omnistar; Pfeiffer,
Quadstarr 422 software). The catalyst (160 mg) was put into
a quartz reactor and was preheated at 40 ◦C. Fifteen succes-
sive injections of HAN solution were performed every 4 min.
3. Results
3.1. Catalytic tests: temperature increase mode
The specific surface area of the samples was determined
by the BET method from the nitrogen adsorption isotherms
at −196 ◦C in an automated Micromeritics Tristar 3000 ap-
paratus after drying for 90 min at 350 ◦C. Particle sizes were
determined by X-ray diffraction experiments performed on
a Siemens D 5005 powder θ–θ diffractometer with Cu-Kα
radiation (λKα = 0.15418 nm) and a secondary graphite
back monochromator. The diffractograms were obtained un-
der the following conditions: dwell time: 2 s; step: 0.04◦ 2θ;
divergence slit: 1◦. The crystalline phases were identified by
comparison with ICDD standards (PDF number: θ alumi-
na: 23-1009; platinum: 04-0802). The crystallite sizes were
determined with the Scherrer equation [18]. A transmission
The experiments in temperature increase mode allowed
us to determine the onset decomposition temperature of the
monopropellant for the different catalysts, the pressure in-
crease due to the formation of gaseous products, and a “re-
action rate” defined as the slope of the pressure increase, as
shown in Fig. 1. We observed a pressure spike followed by
a step and a temperature peak reaching 120 ◦C; the temper-
ature decrease was due to the thermal transfer to the reactor.
The results are presented in Table 1; all prepared catalysts
are very promising because they display very low decompo-
sition temperatures: 62 ◦C for the sphere-shaped catalyst and
22–33 ◦C for the powdered catalysts. The pressure increases
are on the same order of magnitude (about 120 mbar), and

12
L. Courtheoux et al. / Journal of Catalysis 232 (2005) 10–18
Fig. 1. Pressure and temperature (gaseous phase and catalyst) versus time
during the decomposition of 79 wt% HAN (100 µl) in the temperature in-
(a)
−1
crease mode (10 K min ) on the (5 + 5)rx catalyst (160 mg), pressure
increase (ꢀP ) and slope between points 1 and 2.
Table 1
79 wt% HAN onset decomposition temperature, pressure increase and slope
for the reaction in temperature increase mode for various catalysts
Name
T
ꢀP
Slope
(mbar s
◦
−1
)
( C)
(mbar)
10rx
22
33
25
62
126
119
129
123
87
73
120
107
5r + 5rx
(5 + 5)rx
3i.Pt.MF
(b)
the best reaction rate is observed for the (5 + 5)rx catalyst
(120 mbar s−1), measured by at least 5 points. For the cata-
lyst 3i.Pt.MF, we observed, after the reaction, a mechanical
breaking of the spheres, probably due to the strong exother-
micity and pressure increase in the pores during the HAN
decomposition. This mechanical breaking could be avoided
by preheating at higher temperature, as is the case for hy-
drazine decomposition [1]; this preheating helps to accel-
erate the decomposition and thus to avoid accumulation of
liquid inside the pore. Thus the local overpressure and the
mechanical strains inside the pores during the decomposi-
tion were limited.
Fig. 2. (a) Pressure and temperature (gaseous phase and catalyst) versus
time during 15 injections of 79 wt% HAN solution at 40 C on the 10rx
catalyst (160 mg; 15 × 100 µl); (b) enlargement to determine pressure in-
◦
crease and slope.
gradually decreasing (112 to 80 ◦C) because of the increas-
ing amount of water (as reactant in the HAN solution and
product) in the sample holder, up to about 1.1 cm3 after 15
injections. Another important feature is the ability of the cat-
alyst to maintain a good activity in the presence of this water
excess.
HAN decomposition can lead to thermodynamic products
(water, nitrogen, and oxygen, Eq. (1)) or kinetic products
(nitrogen oxides, Eqs. (2), (3), and (4)) [20–22]. The reaction
enthalpy, taking into account the composition of the solution,
is given in parentheses (ꢀrH◦ in kJ mol−1) [23]
3.2. Catalytic tests: constant-temperature mode
A typical example of temperature and pressure evolu-
tion versus time during 15 successive injections is shown
in Fig. 2(a) for the 10rx catalyst preheated at 40 ◦C. The
samples display a significant activity after 15 propellant in-
jections. After each injection, we observe a pressure spike
due to the high decomposition rate, followed by a step. The
gas phase and catalyst temperatures display peaks due to the
exothermic decomposition reaction followed by heat trans-
fer to the reactor wall and retrieval of the initial temperature
(25 ◦C for the gas phase and 40 ◦C for the catalyst). Dur-
ing HAN decomposition, the amount of water present in the
sample holder increases, because only a fraction is vaporized
and condenses at the top of the reactor (cold part). There-
fore, the maximum temperature reached by the catalyst is
NH3OHNO3(aq) → 2H2O(l) + N2(g)
+ O2(g)
(−227.7),
(−193.5),
(1)
NH3OHNO3(aq) → 2H2O(l) + 0.5N2(g)
+ NO2(g or aq)
(2)
(3)
NH3OHNO3(aq) → 2H2O(l) + 2NO(g) (−45.11),
NH3OHNO3(aq) → 2H2O(l) + N2O(g)
+ 0.5O2(g)
(−146.1).
(4)
Nitric acid was proposed as an intermediate reaction
species [21], but we did not observe its quantitative forma-

L. Courtheoux et al. / Journal of Catalysis 232 (2005) 10–18
13
◦
Fig. 3. ꢀP values as a function of injection number for four experiments with different 79 wt% HAN–water and 10rx catalyst preparations at 40 C.
Table 2
◦
BET surface area, platinum content determined by CNRS analyses and particle size (from X-ray diffraction) before and after reaction at 40 C for different
samples. The last column gives the average of the ꢀP value for the 15 injections
Catalysts
Before reaction
BET
After 15 injections of HAN
ꢀP average
(mbar)
σ (ꢀP)
(mbar)
% Pt
(wt)
Size
(Å)
BET
% Pt
(wt)
Size
(Å)
2
−1
2
−1
)
(m g
)
(m g
10rx
70
58
61
43
8.3
8.4
7.2
220
180
90
58
99
71
44
4.9
5.0
4.6
320
160
80
−
115
121
107
127
18
17
16
20
5r + 5rx
(5 + 5)rx
3i.Pt.MF
10.0
160
10.0
tion in the final products, the pH of residual water being
between 2 and 3, due probably to the dissolution of nitro-
gen dioxide (vide infra).
90 mbar (Eq. (2)) to 360 mbar (Eq. (1)). The experimental
data are in between, in agreement with the formation of ther-
modynamic and kinetic products (vide infra, Fig. 5) and thus
lead to the same selectivity [13].
The pressure increase (ꢀP ) and the pressure slope
(ꢀP/ꢀt) were obtained for each peak from peak enlarge-
ment (Fig. 2b). To check the reproducibility of the results,
the 15-injection test was performed four times with three in-
dependent HAN solution preparations and two 10rx catalyst
preparations. The ꢀP values for the successive injections
are reported in Fig. 3 for four propellant–catalyst associa-
tions. The dispersion of the results is linked to the manual
injection procedure, in which the syringe is filled and intro-
duce through the septum for each injection. An automatic in-
jection apparatus was built, from a remote-controlled micro-
burette, but failed for this pulse injection mode. The pressure
average and the corresponding standard deviation (given in
parentheses; the two very low values were ignored) for each
test were
In Fig. 4, we present the pressure slope as a function of
the injection number. The results are very sensitive to the
injection procedure, displaying a high dispersion that can
take the form of some kind of oscillations. Nevertheless, we
can deduce general trends: the average slope increases up to
the seventh to ninth injection, then reaches a plateau. The
very low-slope data corresponding to injection 11 (sample
3i.Pt.MF) or injections 14 and 15 (sample (5 + 5)rx) could
be explained by a propellant drop reaching the side of the
sample holder before coming into contact with the catalyst
(which cannot be detected, as our reactor has no window).
From these data, the catalysts can be ordered according to
the following sequence:
(5 + 5)rx ∼ 3i.Pt.MF < 10rx < 5r + 5rx.
Test 10rx(1) + HAN(1):
Test 10rx(1) + HAN(2):
115(18) mbar;
105(13) mbar;
To identify the gaseous products formed during the
monopropellant decomposition, catalytic tests were per-
formed in the dynamic reactor, with the same injection
procedure. The preliminary results deduced from these ex-
periments (Fig. 5) reveal the very fast formation of N2,
NO, and O2 as primary products. Then the formation of
N2O and traces of NO2 are noted with parallel profiles.
As for the thermal decomposition [13], the catalytic de-
composition leads to the formation of kinetic and ther-
Test 10rx(2) + HAN(2):
Test 10rx(2) + HAN(3):
102(17) mbar;
126(10) mbar.
The results display similar pressure increases, thus in-
dicating both reproducibility and repeatability of the tests.
Table 2 shows similar pressure averages for the different cat-
alysts. The calculated pressure increase is in the range of

14
L. Courtheoux et al. / Journal of Catalysis 232 (2005) 10–18
Fig. 4. Evolution of the slope (ꢀP/ꢀT ) as a function of the injection number for different catalysts during 15 injections.
◦
–water in the dynamic reactor at 40 C on
Fig. 5. Variation of the relative intensity of the main products as a function of time during 15 injections of HAN
the 10rx catalyst.
79%
modynamic products [19]. The different profiles for ni-
trogen oxides, by comparison with nitrogen and oxygen,
can be related to slow adsorption–desorption equilibriums.
Therefore, the experiments performed in two different re-
actors lead to similar results; to quantify these data, cali-
brations of the mass spectrometer are currently in progress,
with the use of representative mixtures in the same condi-
tions.
corresponding support fired at 1200 ◦C (65 m2 g−1) [6] de-
spite the important platinum load.
XRD data (Fig. 6; 10rx catalyst as typical example) con-
firm the presence of θ-alumina in the support without change
after impregnation and the platinum metal phase. The deter-
mination of crystallite size is based on the last two diffrac-
tion peaks (see enlargement in Fig. 6), where the contribu-
tion of the support is very weak. The high platinum contents
(between 7 and 10%) induce a sufficient intensity of plat-
inum diffraction peaks to determine the crystallite size. The
comparison of our results with the literature data is not suit-
able because of the low amount of platinum generally used
(0.1–2%) [24,25]. Three samples display almost the same
average size in the range 160–220 Å; sample (5 + 5)rx, pre-
pared without intermediate reduction, presents a lower crys-
tallite size, in relation to a higher crystallite number, and thus
better platinum precursor-support interactions. The higher
average platinum size for samples 5r + 5rx and 3i.Pt.MF
(i.e., with intermediate reduction at 400 ◦C) can be explained
by the possibility of reaction of the first Pt crystallites (i.e.,
obtained after the first reduction) with the H2PtCl6 precur-
sor.
To supplement these results, different characterizations
have been carried out, before and after the test reaction.
3.3. Structural data and morphology: before reaction test
The platinum loading content of the powdered samples
(7–8.5 wt% Pt, Table 2) is always lower than the expected
value (10 wt%), mainly because of weak platinum–carrier
surface interactions; therefore part of the platinum is lost
from the carrier surface during the thermal treatments. In
contrast, for the sphere-shaped sample, this part remains in-
side the porosity, leading to the expected value.
BET surface area values measured before reaction (Ta-
ble 2) are in accordance with the value determined for the

L. Courtheoux et al. / Journal of Catalysis 232 (2005) 10–18
15
◦
Fig. 6. Diffractograms of the 10rx catalyst before and after 15 injections of HAN
at 40 C (enlargement to determine Pt particles size; ICDD file: 04-0802).
79%
For two representative catalysts (5r + 5rx and (5 + 5)rx)
the transmission electron microscopy images and the parti-
cle size distributions are indicated in Figs. 7a and 7b. The
catalyst with intermediate reduction, 5r + 5rx, displays a
very large distribution, whereas the second sample shows
more numerous, smaller crystallites, in agreement with the
XRD data. These extended particle size distributions are re-
lated to the high platinum percentage and low metal–support
interactions. The volume-averaged size taken from these dis-
tributions was, respectively, 760 Å for 5r + 5rx and 170 Å
for (5 + 5)rx samples. These average data are much higher
than the values obtained from XRD, leading to the suspi-
cious single crystal character of the largest particles. This
point was corroborated by conical dark-field TEM images,
with the use of (111) reflections (Fig. 7c), which obviously
show that these particles are made of aggregates of smaller
crystallites. Therefore the calculated volume-averaged value
is overestimated. This concerns mainly the particles with a
size greater than 200–300 Å.
aggregates and an increase in the medium-sized crystallites;
whereas sample (5 + 5)rx (Fig. 7b) displays a shift in the
distribution of the nanosized particles. This is can be due
to the loss of the smallest crystallites or to the formation of
agglomerates; a sintering effect seems to be less probable
at this low temperature, in agreement with the same XRD
average size. The characteristics of the 10rx and 3i.Pt.MF
catalysts are similar to those of the (5 + 5)rx sample. Nev-
ertheless, for the 3i.Pt.MF sample, it was difficult to com-
pare the sample before and after the reaction because of the
breaking of the spheres.
All of the powdered samples displayed an important loss
of platinum, about 40% detected after reaction (Table 2).
This loss can be correlated with the presence of very small
crystallites (Fig. 8a) on the carbon membrane of the cop-
per grid used for the TEM experiment. To clarify this loss,
we performed X-ray mapping on some catalysts (Figs. 9a
and 9b). The 10rx catalyst shows the heterogeneity of a xe-
rogel support with the presence of silica particles, as estab-
lished in a previous work [11]. However, before the catalytic
reaction, the platinum atom distribution appears to be almost
homogeneous, and the content of platinum next to silica par-
ticles remains low (Fig. 9a). After the decomposition reac-
tion, the presence of disconnected silica and platinum parti-
cles (Fig. 9b) is indicated. The heterogeneous distribution
of platinum particles displays either the sintering of plat-
inum crystallites or more probably the formation of nano-
sized platinum agglomerates (Fig. 8b). This effect confirms
the possibility of platinum crystallite migration for the par-
ticles linked to silica groups. Moreover, these silica species
3.4. Structural data and morphology: after reaction test
After the 15 HAN solution injections, the BET surface
area remains of the same order of magnitude, except for
the 5r + 5rx sample, which presents an area increase with-
out structural change (same diffractograms for all samples;
see Fig. 6). The XRD average size increases for one sample
(10rx, Table 2).
After the reaction, the size distribution of sample 5r +
5rx (Fig. 7a) is explained by a dislocation of the platinum

16
L. Courtheoux et al. / Journal of Catalysis 232 (2005) 10–18
(a)
(b)
(c)
Fig. 7. Distributions of particles sizes deduced by transmission electron microscopy measurements (about 120 particles counted) and corresponding images
◦
before and after 15 injections of HAN
at 40 C for the (a) and (5 + 5)rx (b) catalysts, (c) conical dark-field TEM images using 111 reflections of 5r + 5rx
79%
catalyst (1 bright field and 2 dark field).
(a)
(b)
at 40 C: (a) platinum crystallites on membrane, (b) plat-
◦
Fig. 8. Transmission electron microscopy images of the 10rx catalyst after 15 injections of HAN
inum agglomerates on catalyst.
79%

L. Courtheoux et al. / Journal of Catalysis 232 (2005) 10–18
17
(a)
(b)
◦
Fig. 9. X-ray mapping of 10rx catalyst (a) before and (b) after 15 injections of HAN
at 40 C (Pt-L at 9.441 keV; Al-K at 1.486 keV; Si-K at 1.739 keV).
α
α
α
79%
can be responsible for the primary loss of platinum in rela-
tion to the calculated values (i.e., sample 10rx: 8.3 instead of
10), which occurs during the catalyst preparation. Indeed, in
this case the metal–support interactions are weaker for sil-
ica than for alumina [26,27]. This explanation provides an
idea of the mechanism that occurs at the surface but does
not explain the significant loss of platinum atoms. The plat-
inum loss estimated from XRD data for sample 10rx (Fig. 6;
intensity ratio of Pt diffraction peaks related to the diffrac-
tion peaks of alumina) is in the range of 10–20%; this value
takes into account the largest crystallites, which remain pref-
erentially on the support surface. This is in agreement with
the observed slight increase in the XRD crystallite size (Ta-
ble 2).
Fig. 10. Scheme displaying the transformation of platinum particles during
the reaction of HAN decomposition on catalysts.
79%
For more detailed explanations, experiments in water and
in diluted nitric acid are being carried out to reproduce the
conditions that occur in the batch reactor. These experiments
consist of stirring a catalyst for 4 h in the presence of a
neutral or nitric acid solution (pH ∼ 2.5) followed by filtra-
tion, to determine the platinum amount retrieved in solution.
Only a very small amount of platinum (0.1%) is dissolved in
the acidic solution, whereas no platinum could be detected
in water. By comparison, after the catalytic decomposition,
atomic absorption and ICP measurements revealed the pres-
ence of 6% initial platinum in solution. No aluminum or
silicon could be detected, meaning that the support is not af-
fected by the strong oxidizing feature of the reactant. Thus,
the main loss of the platinum active phase is due to a disper-
sion of the nanosized platinum particles into the remaining
solution or on the walls of the sample holder.
particles. The platinum active phase is divided into three
groups: (i) small particles fixed on alumina; (ii) small par-
ticles bonded to silica; and (iii) large aggregates on alumina.
During the catalytic decomposition of the ionic liquid, the
metallic species can react according to three possibilities
(Fig. 10): (1) migration of platinum from silica to alumina,
leading to the formation of nanosized platinum agglomerates
deposited on the alumina surface; (2) migration of platinum
from silica to another platinum particle, leading to growth;
and (3) redispersion of the large platinum aggregates. These
mechanisms are the results of strong reaction exothermic-
ity, acidic and drastic oxidizing conditions, and energetic
decomposition.
These mechanisms enable us to understand the catalytic
results. Furthermore, the presence of platinum agglomerates
may be related to the good activity of the catalysts. This ar-
gument can be correlated with the slope results. Indeed, we
have shown that, up to the sixth or seventh injections, the cat-
alyst is able to achieve an excellent performance. During this
step, the formation of nanosized platinum agglomerates may
occur. The metal species weakly linked to silica migrate into
the aqueous solution, and the agglomerates are responsible
for the reaction rate being maintained. In the case of metal-
3.5. Proposed model
To clarify these results, we suggest a scheme consider-
ing the different possibilities for platinum species mobility
under strong oxidizing conditions (Fig. 10). The surface of
supported platinum on silicon-doped alumina catalysts con-
sist of a theta alumina structure and silica and platinum

18
L. Courtheoux et al. / Journal of Catalysis 232 (2005) 10–18
lic aggregate dislocation (catalyst 5r + 5rx), the amount of
small particles increases with the reaction rate and confirms
that the agglomerates of very small crystallites are respon-
sible for the good activity. It is important to remember that
the best catalyst (5r + 5rx; Fig. 4) is a material that presents
a value for the pressure slope higher than 500 mbar s−1. But
regarding the homogeneity, the reproducibility, the presence
of nanometric platinum crystallites, and, more important,
the decomposition behavior of HAN on these catalysts, as
shown in Table 1 (25 ◦C, 129 mbar, 120 mbar s−1 for (5 +
5)rx), the best catalyst to consider is then the (5 + 5)rx sam-
ple.
References
[1] E.W. Schmidt, Hydrazine and Its Derivatives. Preparation, Properties,
Applications, second ed., Wiley, New York, 2001, p. 2121.
[2] H.F.R. Schoyer, W.H.M. Welland-Veltmans, J. Louwers, P.A.O.G. Ko-
rting, A.E.D.M. van der Heijden, H.L.J. Keizers, R.P. van den Berg,
J. Propulsion Power 18 (2002) 131.
[3] M. Farshchi, V. Vaezi, B.D. Shaw, Combustion Sci. Technol. 174
(2002) 71.
[4] Y.-P. Chang, K.K. Kuo, J. Propulsion Power 18 (2002) 1076.
[5] K. Anflo, T.A. Grönland, G. Bergman, M. Johansson, R. Nedar, Proc.
38th Joint Propulsion Conference, Indianapolis, IN, July 2002, AIAA
Paper 2002-3847.
[6] B. Reed, S. Harasim, Proc. 37th Joint Propulsion Conference, Salt
Lake City, UT, July 2001, AIAA Paper 2001-3696.
[7] R. Eloirdi, S. Rossignol, C. Kappenstein, D. Duprez, N. Pillet,
J. Propulsion Power 19 (2003) 213.
[8] L. Courthéoux, R. Eloirdi, S. Rossignol, C. Kappenstein, D. Duprez,
N. Pillet, Proc. 38th Joint Propulsion Conference, Indianapolis, IN,
July 2002, AIAA Paper 2002-4027.
4. Conclusions
This work shows that supported platinum on silicon-
doped alumina catalyst is active and stable after 15 injections
of a HAN-water mixture at 40 ◦C. The X-ray diffraction re-
sults confirm the presence of theta alumina structure without
change after the catalytic reaction. Transmission electron
microscopy images and size distribution confirm the het-
erogeneous distribution of silica and platinum particles. The
activity is mainly due to the formation of nanosized platinum
agglomerates responsible for the efficient decomposition of
ionic liquid. During the catalytic decomposition, the low
metal–silica interactions lead to a migration of the metallic
particles, which are responsible for sintering of the smallest
crystallites and dislocation of the aggregates. The suggested
model relates to the surface change in the catalyst before and
after the reaction. Two possibilities are discussed, one con-
cerning the loss of platinum linked to silica particles and the
other concerning a redistribution of platinum species.
[9] H. Lee, T.A. Litzinger, Combution Flame 135 (2003) 151.
[10] E.W. Schmidt, D.F. Gavin, US Patent 5485722 (1996).
[11] L. Courthéoux, F. Popa, E. Gautron, S. Rossignol, C. Kappenstein,
J. Non-Cryst. Solids 350 (2004) 113.
[12] S. Rossignol, C. Kappenstein, Int. J. Inorg. Mater. 3 (2001) 51.
[13] F. Popa, L. Courthéoux, E. Gautron, S. Rossignol, C. Kappenstein,
Eur. J. Inorg. Chem. (2005) 543.
[14] AIAA papers are available from the American Institute for
Aeronautics and Astronautics Website: http://www.aiaa.org/content.
cfm?pageid=413.
[15] A.F. Popa, S. Rossignol, C. Kappenstein, J. Mater. Chem. 12 (2002)
2866.
[16] M. Nguefack, A.F. Popa, S. Rossignol, C. Kappenstein, Phys. Chem.
Chem. Phys. 5 (2003) 4279.
[17] B.E. Yoldas, J. Mater. Sci. 10 (1975) 1856.
[18] B.E. Warren, X-Ray Diffraction, Dover, New York, 1969, p. 251.
[19] Y. Batonneau, L. Courthéoux, P. Esteves, L. Pirault-Roy, S. Rossignol,
C. Kappenstein, N. Pillet, Proc. 40th Joint Propulsion Conference, Fort
Lauderdale, FL, USA, July 2004, AIAA Paper 2004-3835.
[20] J. Schoppelrei, T. Brill, J. Phys. Chem. A 101 (1977) 8593.
[21] J. Cronin, T. Brill, J. Phys. Chem. A 90 (1986) 178.
[22] H. Lee, T. Litzinger, Combustion Flame 127 (2001) 2205.
[23] A. Roine, HSC Chemistry® for Windows, Chemical Reaction and
Equilibrium Software with Extensive Thermochemical Database, Ver-
sion 5.1, Outokumpu Research Oy, Pori, Finland, 2002, ISBN 952-
9507-08-9.
[24] C. Wang, C. Yeh, Appl. Catal. A 209 (2001) 1.
[25] U. Olsbye, R. Wendelbo, D. Akporiaye, Appl. Catal. A 152 (1997)
127.
[26] A. Douidah, P. Marécot, S. Szabo, J. Barbier, Appl. Catal. A 225
(2002) 21.
Acknowledgments
We thank the European Space Agency (ESA-ESTEC,
Noordwijk, the Netherlands), the French Space Agency
(Centre National d’Etudes Spatiales, Toulouse, France) for
funding this study as well as the 12th CPER (Contrat de
plan Etat Region Poitou Charente, France) for funding the
dynamic reactor.
[27] S. Jackson, G. Kelly, G. Webb, J. Catal. 176 (1998) 225.