Solid-state chemistry of ammonium and cesium 1-vanado-11-molybdophosphate and ammonium 12-molybdosilicate: Application to oxidation catalysis

Article abstract of DOI:10.1016/S0021-9517(03)00242-2

The solid-state behavior of (NH₄)₄[PMo ₁₁V^VO₄₀], crystallized from a water/dioxane solution, has been studied. Decomposition of one of the four ammonium cations occurs easily and a thermal treatment in air at 220°C leads to (NH ₄)₃H[PMo₁₁VO₄₀]. Further ammonium decomposition was observed between 220 and 300°C up to (NH₄) _0.8H_3.2[PMo₁₁VO₄₀]. Only one cubic phase was evident and the lattice parameter increases as ammonia is eliminated. Rietveld refinement of X-ray patterns of all samples is better with an occupancy of 34 for anionic sites, as previously proposed for Cs ₄[PMo₁₁VO₄₀]. Solid-state substitution of cesium for ammonium cations has been performed by thermal treatments at 300°C of the ammonium salt impregnated by cesium nitrate. Only three ammonium cations can be substituted. The catalytic behavior of these salts for the oxidative dehydrogenation of isobutyric acid reveals high catalytic activity and selectivity to methacrylic acid for all the samples up to three cesium cations but fall down hereafter.

Full text of DOI:10.1016/S0021-9517(03)00242-2

Journal of Catalysis 220 (2003) 172–181
www.elsevier.com/locate/jcat
Solid-state chemistry of ammonium and cesium
-vanado-11-molybdophosphate and ammonium 12-molybdosilicate:
application to oxidation catalysis
1
∗
N. Laronze, C. Marchal-Roch, N. Guillou, F.X. Liu, and G. Hervé
Institut de Réactivité, Electrochimie et Microporosités, UMR 8637, Université de Versailles, 45 Avenue des Etats-Unis, 78035 Versailles cedex, France
Received 24 February 2003; revised 16 May 2003; accepted 16 May 2003
Abstract
The solid-state behavior of (NH ) [PMo V O ], crystallized from a water/dioxane solution, has been studied. Decomposition of one
V
4
4
11
40
◦
of the four ammonium cations occurs easily and a thermal treatment in air at 220 C leads to (NH ) H[PMo VO ]. Further ammonium
4
3
11
40
◦
decomposition was observed between 220 and 300 C up to (NH )
H
[PMo VO ]. Only one cubic phase was evident and the lattice
4
0.8 3.2 11 40
3
4
parameter increases as ammonia is eliminated. Rietveld refinement of X-ray patterns of all samples is better with an occupancy of for
anionic sites, as previously proposed for Cs [PMo VO ]. Solid-state substitution of cesium for ammonium cations has been performed
by thermal treatments at 300 C of the ammonium salt impregnated by cesium nitrate. Only three ammonium cations can be substituted.
The catalytic behavior of these salts for the oxidative dehydrogenation of isobutyric acid reveals high catalytic activity and selectivity to
methacrylic acid for all the samples up to three cesium cations but fall down hereafter.
4
11
40
◦

2003 Elsevier Inc. All rights reserved.
Keywords: Catalysis; Heteropolyanion; Ammonium vanadomolybdophosphate; Isobutyric acid oxidation
1
. Introduction
oxidative dehydrogenation, Lee et al. determined an opti-
mum of methacrylic acid (MAA) formation for x = 2.75 in
◦
the series of CsxH4−xPMo11VO40 salts at 350 C [5]. Above
Heteropolymolybdates have been widely used for oxi-
x = 3.5, no more methacrylic acid was formed. In the case
of isobutane oxidation, Mizuno et al. also observed a max-
imum yield in MAA and methacrolein for x = 2.5 in the
series CsxH3−xPMo12O40 or CsxH3.84−xNi0.08PMo11VO40
dation of organic molecules such as alcohols, aldehydes,
isobutyric acid, and alkanes. By a control of the molecular
composition of the heteropolyanion, it is possible to fine tune
its acid and oxidizing properties. The nature of the cation
influences also the crystallographic lattice, the texture, the
thermal stability, and also the acid and redox behaviors of the
heteropolysalts [1]. Therefore, complicated formulas which
combine alkaline cations (K, Cs), transition metals (V, Cu,
Fe, Ni, Mn), and main group metals (Sn, Sb, As) have been
often proposed in many patents or more fundamental studies
2,3].
Acidic cesium salts of the 1-vanado-11-molybdophos-
phoric acid have attracted much attention since the presence
of vanadium increases the catalytic activity and cesium has
been reported to increase both the stability and the reox-
idation rate of the catalyst [4]. For isobutyric acid (IBA)
◦
at 340 C [6,7]. These results have been interpreted by the
enhancement of surface acidity and surface-oxidizing abil-
ity of cesium salts with less than 3 Cs per heteropolyanion.
Nevertheless, in some cases, the true molecular composition
and the structure of the catalysts have been poorly charac-
terized both initially and at the steady state [1,5,6,8,9]. In
situ, as well as ex situ characterizations of acids and partially
salified heteropolyacids have been performed for the oxida-
tion of methacrolein and isobutyric acid. It was shown that
vanadium [10–13] and molybdenum [14,15] migrate from
the heteropolyanion to the crystal lattice under the reaction
conditions and are probably present as cationic oxospecies.
In previous studies, we have demonstrated that solid so-
[
3
−
lutions of mixed ammonium–cesium salts of [PMo12O40]
*
can be obtained by solid-state exchange [16]. The study
has been extended to more active compounds which contain
Corresponding author.
E-mail address: laronze@chimie.uvsq.fr (N. Laronze).
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00242-2

N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
173
IV
◦
◦
−1
vanadium, derived from (NH4)5[PMo11V O40] [17]. The
to 250 C in 5 h (0.77 C min ) and maintained for 20 h
at this temperature; (b) the temperature was raised again to
thermal behavior of this salt showed that a phase, isostruc-
tural of the cubic phase of (NH4)3[PMo12O40], was ob-
tained after a thermal treatment at 210 C and was stable
◦
◦
−1
300 C at 0.21 C min and maintained for 3 h at this tem-
perature; (c) the sample was cooled to room temperature.
The sample 0 Cs is obtained by the same thermal treatment
of the initial ammonium salt not impregnated by cesium ni-
trate.
For Cs4[PMo11VO40], two solutions were prepared: (A)
The amount of 5.5 g of H4[PMo11V O40] · 29H2O was dis-
◦
◦
+
up to 410 C. Moreover, solid solutions containing NH4
+
and Cs cations can be obtained up to only three cations for
an heteropolyanion [18]. The same limitation to three ce-
sium cations has been observed by Lee et al. when cesium
is substituted for protons in H4[PMo11VO40]. On the con-
trary, cubic salts CsxH4−x[PMo11VO40], with 0 ꢀ x ꢀ 4,
have been described in a recent study of Berndt et al. [19],
which propose a general explanation of their cubic structure
according to the cationic composition.
The aim of this work was to prepare well-defined stoi-
chiometric ammonium and cesium 1-vanado(V)-11-molyb-
dophosphates, in order to study their structure and their
thermal behavior. They were compared to those of (NH4)4-
V
solved in 20 mL of dioxane. Then the heteropolyacid was
neutralized by the addition of a lithium hydrogenocarbonate
solution 0.5 M until pH ≈ 4.5 (19.2 mL). The lithium hy-
drogenocarbonate solution was obtained by bubbling CO2
in a lithium hydroxide solution 0.5 M and it was titrated
by HCl. (B) The amount of 1.863 g of cesium nitrate was
dissolved in 20 mL of the water/dioxane mixture (1/1). So-
lution B was added dropwise to Solution A under stirring.
An orange precipitate of cesium molybdovanadophosphate
appeared after 1 min. The solid was collected by filtration
through a fine frit, washed with cold water, and dried at
room temperature for 2 days. Analysis, % exp (calcd): Mo
41.8 (42.1), Cs 21.46 (21.19), P 1.41 (1.23), V 1.91 (2.03),
C < 0.3.
[
SiMo12O40], an isostructural and isocharge compound
without vanadium.
Moreover, mixed ammonium–cesiumsalts obtained by so-
+
+
V
lid-state Cs /NH4 exchange with (NH4)4[PMo11V O40]
have been prepared and characterized in order to determine
the influence of the initial oxidation state of vanadium on
the reactivity for the catalytic oxidative dehydrogenation of
isobutyric acid.
(NH4)4[SiMo12O40] was obtained in two steps, first syn-
thesis of the 12-molybdosilicic acid H4[SiMo12O40] as pre-
viously described [21], and then neutralization of this acid.
The heteropolyacid was neutralized by an ammonium hy-
drogenocarbonate solution, 1.1 M, in water up to pH ≈ 4.
The solution was then precipitated by NH4Cl. The solid
was collected by filtration through a fine frit, washed with
cold water, and dried at room temperature. Analysis, % exp
(calcd): Mo 55.34 (56.56), N 2.52 (2.75).
2
. Experimental
2
.1. Preparations
V
(
NH4)4[PMo11V O40] was obtained in two steps, first
synthesis of the 1-vanado(V)-11-molybdophosphoric acid
V
H4[PMo11V O40] as previously described [20], and then
2.2. Characterizations of the solids
neutralization of this acid. The heteropolyacid H4[PMo11-
V
−1
V O40] was dissolved in water/dioxane v/v (0.36 mol L ),
and then solid ammonium hydrogenocarbonate was added
up to a rapid increase of pH until pH ≈ 4 (pH meter previ-
ously calibrated in water). Orange crystals of (NH4)4[PMo11-
Nitrogen adsorption–desorption isotherms were obtained
using a Micromeritics ASAP 2010 apparatus. The samples
were evacuated at 150 C for 2 h. Results were analyzed by
the ASAP 2010 V3.00 program.
Density measurements were performed using a Micro-
meritics AccuPyc 1330 gas pycnometer (He).
◦
V
◦
V O40] were obtained from the solution placed at 4 C. This
◦
ammonium salt was heated at 120 C for 24 h in air flow in
order to eliminate dioxane. The solid was rehydrated at room
temperature. Analysis, % exp (calcd): Mo 54.27 (52.25), N
Morphology of the solids was observed by scanning elec-
tron microscopy using a JEOL JSM-5800LV microscope.
X-ray diffraction patterns (XRD) were recorded at room
temperature with a Siemens D 5000 diffractometerusing Cu-
Kα radiation (λ = 1.5418 Å) and were fitted by means of
the Socabim program PROFILE. Unit cell parameters were
determined with the program DICVOL91 [22] and were
refined by means of the program NBS*AIDS83 [23]. Ri-
etveld refinements were undertaken by using the FullProf
[24] program integrated in WinPLOTR [25] software. X-ray
thermodiffractometry (XRTD) was performed in air in an
Anton Paar HTK16 high-temperature device of a Siemens
D 5000 diffractometer (θ–θ mode) using Co-Kα radiation
(λ = 1.7903 Å), and equipped with a M Braun linear po-
sition sensitive detector (PSD). Samples were deposited di-
2
.69 (2.77), P 1.63 (1.53), V 2.53 (2.52), C 0.5.
Mixed ammonium–cesium salts containing x cesium
cations per heteropolyanion, noted as xCs (x = 1, 2, 2.7, 3,
3
1
.2, 3.5), were prepared from the ammonium 1-vanado(V)-
1-molybdophosphate. Exchange between ammonium and
cesium cations was carried out by thermal treatment of
V
(
NH4)4[PMo11V O40] impregnated by CsNO3. A sample
V
−3
of (NH4)4[PMo11V O40] (5 g, 2.47 × 10 mol) was im-
pregnated at incipient wetness by a solution of CsNO3
−
3
(
x · 2.47 × 10 mol in 2 mL of hot water). It was dried
◦
at 100 C for 30 min and was then ground with a pestle and
−1
a mortar. This solid was heated in air flow (20 mL min ) us-
ing the following procedure: (a) the temperature was raised

174
N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
rectly on the heating platinum support connected to a ther-
◦
mocouple. Patterns were recorded from 30 to 610 C every
2
◦
◦
−1
0 C, with a temperature ramp of 1.8 C min . The tem-
perature was stabilized 2 min before measurements; each
data collection lasted 570 s.
Thermogravimetry measurement (TGA) was carried out
−
1
in air flow (60 mL min ) with a Perkin-Elmer electrobal-
◦
−1
◦
ance TGA-7 at a heating rate of 5 C min up to 600 C.
Differential Scanning Calorimetry (DSC) was carried out in
air flow (60 mL min ) with a TA Instruments DSC 2010
−1
◦
−1
◦
at a heating rate of 5 C min up to 600 C. Infrared spec-
tra of KBr pellets were recorded on a Fourier Transform IR
Nicolet 550 apparatus.
Solution 31P NMR spectra were recorded on a Bruker AC
3
00 apparatus at 121.5 MHz. Chemical shifts are referenced
to 85% H3PO4. Initial concentration in heteropolyanion is
−
2
−1
.
6
.5 × 10 mol L
2
.3. Catalytic reaction
Oxidative dehydrogenation of IBA was studied under
standard conditions in order to obtain structure–reactivity
◦
relationships. Catalytic reaction was carried out at 320 C
using a continuous flow pyrex microcatalytic reactor at at-
mospheric pressure. The feed gas consisted of 2% IBA, 5.2%
O2, 3.5% H2O, and N2 balance; total flow rate was 180
−1
mL min . The catalyst (0.3 g) was introduced in the reactor
and the temperature of reaction was reached in 30 min in air
−
1
flow (180 mL min ). All reaction products were analyzed
by on-line gas-phase chromatography with molecular sieves
5
V
Fig. 1. NMR spectra following the neutralization of H [PMo V O
]
4
11
40
V
−2
by NH HCO in water/dioxane (v/v): H [PMo V O ] (6.5 × 10
4
3
4
11
40
Å, Porapak Q, and AT 1200 with 1% H3PO4 columns.
Carbon, oxygen, and hydrogen balances were in the range
0–100%. The catalytic performances were determined af-
ter 18 h of reaction.
−1
mol L ) (a), 4 eq of base/HPA (b), 5 eq of base/HPA (c), 6 eq of
3−
base/HPA (d) (the asterisk indicates the signal of trace of [PMo
O ]
2 40
).
1
9
water/organic solvents such as water/dioxane since hy-
droorganic solvents are known to stabilize the heteropolyan-
ions. Hydrogenocarbonate ion has been chosen to neutral-
3
. Results
V
4−
ize acidity. The stability of the anion [PMo11V O40] in
water/dioxane v/v has been checked by 31P NMR spec-
troscopy during the neutralization by (NH4)HCO3 (Fig. 1).
3
.1. Synthesis and characterization of
(
NH4)4[PMo11VO40]
V
4−
The peak of [PMo11V O40] at −3.31 ppm does not vary
until the neutralization (4 eq/HPA). It is only after the
neutralization that other weak peaks appear (−0.26 ppm
Ammonium or cesium 12-molybdophosphatesor 1-vana-
do-11-molybdophosphates can be obtained by addition of
ammonium or cesium chloride (nitrate) to the solution of
the heteropolyacid and filtration of the solid. An alterna-
tive method is the addition of ammonium (cesium) carbonate
in a stoichiometric amount and evaporation of water. In the
first case, it is difficult to eliminate quantitatively the pro-
tons; salts of composition M3−xHx[PMo12O40] (x ∼ 0.3)
or (NH4)3H[PMo11VO40] are currently obtained [26]. In
the second method, we experienced that the basicity of
(
(
8.1%), 0.58 ppm (7.3%), 2.07 ppm (11.1%), and 2.45 ppm
7.4%) for 6 eq of base/HPA). Therefore it is possible
to neutralize quantitatively the heteropolyacid in this sol-
vent without any decomposition. It can be pointed out that
(
NH4)4[PMo11VO40] is soluble in the water/dioxane sol-
vent and crystallization occurs only after complete neutral-
ization of the acid. Only one signal at −3.41 ppm is observed
CO32 is enough to partially decompose the polyanions
which are stable only in acidic aqueous solution. In order to
avoid partial hydrolysis of the heteropolyanion[27], bases of
lower strength can be used. Another way is to utilize mixed
−
31
V
on the P NMR spectrum of (NH4)4[PMo11V O40] after
dissolution in a water/dioxane solvent. Elemental analysis
of the crystals shows that neutral (NH4)4[PMo11VO40] is ef-
fectively obtained.

N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
175
Table 1
Structural model indicators (RB, R-Bragg factor; RF, R-structure factor)
of final Rietveld refinements with isotropic temperature factors (B ) of
iso
cations
a
Compound
Crystal model
indicators
Anionic
occupancy N or Cs
atoms
B
of
iso
RB
0.072 0.109
.099 0.118
0.101 0.077
.107 0.08
RF
V
(
(
(
(
NH ) [PMo V O ] (a)
0.75
4.2(8)
0.4(8)
4
4
11
40
b
0
1
V
NH ) H[PMo V O ] (b)
0.75
4.4(9)
0.2(9)
4
3
11
40
b
0
1
V
NH )
H
[PMo V O ] (c) 0.070 0.069
0.75
4.3(9)
−0.1(8)
4.8(7)
4
2.4 1.6
11
40
b
0.091 0.074
1
V
NH )
H
[PMo V O ] (d) 0.066 0.056
0.75
4
1.2 1.8
11
40
V
Fig. 2. Infrared spectrum of (NH ) [PMo V O ].
b
4
4
11
40
0.070 0.057
0.062 0.071
1
0(7)
V
Cs [PMo V O ] (e)
0.75
4.36(9)
1.99(8)
4
11
40
b
0.096 0.095
1
The salt is composed of fine crystals (2 µm average diam-
eter) forming aggregates and has a specific surface area of
a
(a) and (e): initial compounds. (b), (c), and (d):
V
◦
(NH ) [PMo V O ] treated at 220, 300, and 310 C, respectively,
4 4 11 40
in air flow and rehydrated in ambient atmosphere. Amount of ammonium
cations determined by TG.
b
2
−1
1
6 m g
.
V
The IR spectrum of (NH4)4[PMo11V O40] (Fig. 2)
−1
presents an intense band at 1062 cm and two shoulders
Rietveld refinements with anionic and cationic occupancies of 1 have
−1
at 1051 and at 1079 cm . As it is well known, such a
splitting of the antisymmetric P–O vibration corresponds to
the lowering of the symmetry of the central PO4 tetrahe-
been undertaken in order to localize additional cations necessary for charge
balance. These results are given for information.
dron to Cs symmetry due to the substitution of one Mo in
[
obtained for Cs4[PMo11VO40] considering the presence of
PMo12O40]3 by one vanadium atom. The other bands are
−
+
4−
3
4
6
Cs (all positions) and only 1.5 [PMo11VO40]
(
po-
classically observed in Keggin-type heteropolyanions [28]
and are not modified by this weak change.
sitions) in the unit cell with the experimental parameter
a = 11.7238(2) Å. These values can be compared to the cal-
V
−3
The X-ray diffraction pattern of (NH4)4[PMo11V O40]
culated value ρ = 4.65 g cm for 2 Cs4[PMo11VO40] in the
is indexed with a cubic unit cell (a = 11.6293(5) Å; space
unit cell. In the case of (NH4)4[PMo11VO40], the calculated
¯
−3
3
group Pn3m) similar to that of (NH4)3[PMo12O40] (a =
value is equal to 3.04 g cm if of the anionic positions are
4
1
1.6341(6) Å) [16]. The corresponding atomic coordinates
−3
occupied whereas the value is equal to 3.97 g cm if all the
anionic positions are occupied. The density of 3.22 g cm
of (NH4)3[PMo12O40] [29] have been used as the starting
model in the Rietveld refinement in which one molybde-
num atom is substituted by one vanadium. A priori, this
result raises a question since this compact structure can
only accommodate three cationic positions for each hete-
ropolyanion. For the location of the cation, two hypothe-
ses have been considered: (1) as proposed by Berndt et al.
a partial occupation of 3/4 for the anionic site is taken
into account; (2) no anionic defaults are considered and
some attempts to find an additional cation are undertaken.
In order to increase the atomic contrast to eventually obtain
clearer difference Fourier maps, the stoichiometric cesium
−3
is rather in agreement with the anion-deficient model.
3
.2. Thermal behavior of the ammonium salt
Three weight losses are seen on the thermogravimetric
curve of (NH4)4[PMo11VO40] (Fig. 3, curve a) recorded
V
salt Cs4[PMo11V O40] has also been prepared. Its X-ray
diffraction pattern has been indexed with a slightly higher
¯
unit cell (a = 11.7238(2) Å; space group Pn3m). The re-
sults of Rietveld refinements (Table 1) are in favor of the
first hypothesis and no additional cations are evident by dif-
ference Fourier calculation either in ammonium salt or in
cesium compound.
Density measurements were performed for the ammo-
nium (NH4)4[PMo11VO40] and cesium Cs4[PMo11VO40]
salts. The experimental values are respectively ρ = 3.55 and
Fig. 3. Thermogravimetric curves recorded in air flow of (NH ) [PMo
-
4
4
11
−3
V
V
◦
3
.22 g cm . A very good agreement between experimen-
V O ] (a), (NH ) [PMo V O ] thermal treated in air flow at 220 C
4
0
4 4
11
40
−3
−3
V
(
b), and (NH ) [PMo V O ] impregnated by 3 CsNO (c).
tal ρ = 3.55 g cm and calculated ρ = 3.57 g cm was
4 4 11 40 3

176
N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
◦
in air flow. The first one, between 25 and 100 C, is at-
tributed to three adsorbed water molecules and leads to
the formation of anhydrous salt [29]. The second one, be-
(NH4)4[SiMo12O40] was prepared by neutralization of the
acid by NH4HCO3 in water. This compound has also the
same structure as the cubic (NH4)3[PMo12O40] salt, with
◦
¯
tween 200 and 300 C, and the third one, between 300
and 500 C, are not well separated; the total loss corre-
sponds to 118 g mol . Considering the presence of 0.5%
of carbon shown by elemental analysis and under presumed
dioxane form (17 g mol ) the corrected total loss corre-
a = 11.6128(3) Å; space group Pn3m, like (NH4)4[PMo11-
◦
V
V O40] salt. In this salt, the anionic charge results from the
−1
V
IV
substitution of the central P by Si instead of the substitu-
VI
V
tion of one Mo by one V . The thermogravimetric curve
recorded in air flow shows three weight losses as in the case
−1
−1
V
sponds to 101 g mol , in good agreement with the value
04 g mol calculated for decomposition into oxides of
of (NH4)4[PMo11V O40]. The first one, between 25 and
100 C, is attributed to eight water molecules and leads to
−1
◦
1
(
NH4)4[PMo11VO40].
the formation of the anhydrous salt. The second one, be-
tween 200 and 300 C, and the third one, between 300 and
◦
A thermal treatment of the initial salt was performed in
◦
◦
air flow at 220 C. Elemental analysis of the treated solid
indicates 3 NH4 per mole and only one weight loss of
450 C, are not well separated; the total loss corresponds to
+
−1
.
103 g mol
A thermal treatment of the initial salt was performed in air
flow at 220 C and then up to 600 C. The first loss at 220 C
corresponds to 1 NH3 per mole and the second weight loss
of 89 g mol at temperatures higher than 270 C is assigned
to 3NH3 + 2H2O and corresponds to the decomposition of
the acidic salt (NH4)3H[SiMo12O40] into oxides.
−1
◦
8
9 g mol is seen at temperatures higher than 280 C on
the TG curve (Fig. 3, curve b). It is assigned to 3NH3 +
H2O corresponding to the decomposition of the acidic
salt (NH4)3H[PMo11VO40] into oxides (theoretical value
◦
◦
◦
2
−1
◦
−1
8
7 g mol ). The first weight loss observed with the initial
◦
salt after 200 C is thus attributed to the formation of the
acidic salt (NH4)3H[PMo11VO40].
The DSC curve of (NH4)4[SiMo12O40] shows two en-
◦
The DSC curve (Fig. 4, curve a) can be compared with the
dothermic peaks at 90 and 215 C and an exothermic peak
◦
◦
weight losses seen on the TG curve. The first peak at 60 C is
endothermic and corresponds to the loss of hydration water
molecules. The second peak at 265 C is strongly exothermic
and can likely be associated to the weight loss recorded be-
tween 200 and 300 C. The exothermic peaks above 440 C
correspond to the loss and, probably, oxidation of the re-
maining ammonia molecules and to oxide crystallization.
Otherwise, the exothermic peak at 265 C is not observed
on the DSC curve of the heated salt at 220 C which has al-
at about 400 C, but not the strong exothermic peak at
265 C (Fig. 4, curve c) observed on the DSC curve of
(NH4)4[PMo11V O40]. In conclusion, the exothermic peak
at 265 C must be correlated to the presence of a vanadium
◦
◦
V
◦
◦
◦
atom in the Keggin structure and to the departure of the first
ammonia molecule.
V
Evolution of the X-ray pattern of (NH4)4[PMo11V O40]
◦
with temperature is shown in Fig. 5. The cubic phase is
◦
◦
observed up to 410 C and then the peaks of molybdenum
oxide appear. But a slight shift of all diffraction lines toward
ready lost one ammonia molecule (Fig. 4, curve b).
◦
smaller θ is observed above 250 C; it corresponds to an in-
In order to understand the origin of the exothermic trans-
◦
crease of the lattice parameter with temperature.
formation at 265 C, the ammonium 12-molybdosilicate
3
.3. Characterization of the treated ammonium salt
X-ray diffractometry and TG measurement reveal that
ammonia molecules are eliminated whereas the cubic struc-
ture is maintained. Could all ammonium cations be released
to form the acid with the same cubic structure previously
depicted by Berndt et al.? Thermal treatments of the initial
V
salt (NH4)4[PMo11V O40] have been performed (for sev-
eral days) at different temperatures (220, 280, 290, 300, and
◦
3
10 C) in air flow and the samples have been characterized
by X-ray diffraction, IR spectroscopy, and thermal analysis
after rehydration in ambient atmosphere.
The amount of ammonium cations decreases as tempera-
ture increases, showing that ammonia is progressively elim-
inated. The infrared spectra show that the relative intensity
of the ammonium vibration decreases as ammonia is elimi-
nated and that the shoulders of the P–O vibration at 1051 and
−1
at 1079 cm are present but less resolved than for the ini-
tial salt. On the contrary, the other characteristic bands of the
heteropolyanion are not modified up to 300 C, showing that
Fig. 4. Differential Scanning Calorimetry curves recorded in air flow of
V
V
◦
(
NH ) [PMo V O ] (a), (NH ) [PMo V O ] heated in air flow at
4
4
11
40
4 4
11
40
◦
2
20 C (b), and (NH ) [SiMo
O
12 40
] (c).
4
4
the Keggin structure is kept. The X-ray diffraction patterns

N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
177
◦
◦
◦
V
Fig. 5. Evolution of the XRD patterns with temperature from 30 C up to 610 C recorded every 20 C for (NH ) [PMo V O ] (the asterisk indicates
4
4
11
40
spirious diffraction line of the sample holder).
ence of vanadium species out of the Keggin structure [10].
New peaks observed by X-ray diffraction could not be at-
tributed to a known phase. All this suggests a transformation
of the Keggin anion [PMo11VO40]4 at this temperature.
−
The specific surface area of the initial ammonium salt
2
−1
2
−1
is low (16 m g ) but a large increase to 143 m g oc-
curs when the first ammonia molecule is eliminated in air at
◦
2
20 C; it increases only slightly at higher temperatures (173
2
−1
+
m g for 2.4 NH4 ).
3
.4. Preparation and characterization of mixed
Fig. 6. Evolution of the lattice parameter in function of the amount of re-
maining ammonium cations per heteropolyanion after thermal treatments
ammonium–cesium salts
V
of the initial salt (NH ) [PMo V O ] at different temperatures (220 to
4
4
11
40
The mixed ammonium–cesium salts have been obtained
◦
3
10 C) in air flow.
+
+
by solid-state cationic Cs /NH4 substitution which oc-
curs during a thermal treatment of the ammonium salt im-
pregnated by cesium nitrate at incipient wetness [16]. The
exchange is promoted by the decomposition of ammonium
of the treated samples have all been indexed with a cubic
¯
unit cell (space group Pn3m). As the amount of ammonium
◦
decreases, the lattice parameter increases (Fig. 6). Rietveld
refinements have been performed on treated samples at dif-
ferent temperatures (Table 1). Since X-ray scattering factors
of oxygen and nitrogen are close, we did not consider a
nitrate at temperatures higher than 210 C. The thermo-
V
gravimetric curves of (NH ) [PMo V O ] (a) and of the
4
4
11
40
V
(NH ) [PMo V O ] + 3CsNO mixture (c) are shown in
4
4
11
40
3
Fig. 3. The weight loss observed at temperature higher than
180 C on curve c is attributed to the decomposition of am-
+
◦
statistic occupation of the cationic position by NH4 and
+
H3O . In the hypothesis of full occupation of cationic and
monium nitrate. It can be seen that a temperature as high as
◦
anionic sites, difference Fourier maps did not allow the loca-
300 C is needed to obtain a quantitative decomposition of
+
tion of additional H3O and no site with coherent hydrogen
NH NO . The treatment was then performed in two steps,
4
3
◦
bonds distances was found for hydronium in this compact
structure.
first at 250 C in order to eliminate more than 80% of the
◦
ammonium nitrate and then at 300 C in order to complete
It is not possible to eliminate the four ammonia molecules
at 300 C whatever the lengthening of thermal treatment,
and a limiting composition (NH4)0.8H3.2[PMo11VO40] has
been obtained without any modification of the Keggin struc-
ture and of the cubic lattice. Indeed, additional peaks are
the cationic exchange. Under these conditions, no decom-
position of the heteropolyanion was observed as shown by
infrared spectroscopy and only one cubic phase was seen by
X-ray diffraction.
◦
The infrared spectra of the solids xCs are given in Fig. 7.
observed on the X-ray pattern. After a long treatment at
For all samples, the shoulders of the ν P–O vibration at
as
◦
−1
3
10 C, additional peaks are observed on the X-ray pattern
1051 and at 1079 cm are present but less resolved than for
−1
V
and a new band at 1035 cm is present on the IR spectrum.
This new band has been previously correlated with the pres-
the initial (NH4)4[PMo11V O40] ammonium salt. The rela-
−1
tive intensity of the δ N–H vibration at 1410 cm decreases

178
N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
Fig. 7. Infrared spectra of the xCs solids obtained after thermal treatment in air flow (the asterisk indicates a KBr impurity).
Table 2
Amount of ammonium cations per heteropolyanion in the xCs samples obtained after thermal treatment in air flow before and after use in reaction determined
◦
by elemental analysis (a) and estimated by the losses between 250 and 600 C measured by thermogravimetric analysis (b)
Compounds
0 Cs
1 Cs
2 Cs
2.7 Cs
3 Cs
3.2 Cs
3.5 Cs
Before reaction
(
(
a)
b)
3.1
3
1.07
1.65
0.3
0.58
Trace
0.19
0
0.06
0
a
0
a
After reaction
(
(
a)
b)
0.75
0.5
0.38
a
0
a
0
a
0
a
0
a
0
a
a
Under the limit of analysis.
when x increases from 1 to 3 and, for x = 3, this band dis-
−1
appears. For x = 3.5, a fine band at 1385 cm appears
and corresponds to an excess of cesium nitrate. Therefore
only three on the four ammonium cations can be replaced
by cesium cations. Thermogravimetric analysis and elemen-
tal analysis show that the amount of ammonium cations is
always smaller than the expected amount (Table 2). This
shows that decomposition of ammonium cations occurs dur-
+
+
ing the cationic Cs /NH4 exchange. The X-ray diffraction
patterns of all samples have been indexed in the cubic sym-
metry. The lattice parameter increases with x up to about 2
and remains constant for higher x values (Fig. 8). Neverthe-
less exchange is confirmed for 2 ꢀ x ꢀ 3 by some important
variations of several line intensities on diffraction patterns,
especially for (200) and (220) reflections which decrease and
vanish for x = 3 and for (211) and (332) which increase up
to x = 3.
Fig. 8. Variation of the unit cell parameter of cubic phase of the xCs samples
obtained after thermal treatment in air flow versus the amount of impreg-
nated CsNO .
3
occurs between 3 and 3.2 Cs: the conversion of IBA and the
selectivity to MAA fall down sharply while acetone becomes
the major product. Propene is formed in a very small quan-
tity for x = 3 and the proportion of CO and CO2 increases
hereafter.
3
.5. Catalytic behavior
Infrared spectra of the catalysts after use in reaction are
given in Fig. 10. For x < 2, the shoulder of the P–O vibra-
In the first 5 h of reaction the conversion and the selec-
−1
tivity to MAA improve and then a slow decrease of the con-
version but a little improvement of the selectivity to MAA
are observed for x ꢀ 3. A quasi steady state is reached and
the catalytic activity of the xCs samples is reported in Fig. 9.
The yield of MAA changes only slightly for 0 ꢀ x < 3 (48.2
to 45.6%) but a drastic evolution of the catalytic behavior
tion at 1062 cm disappears, this band becomes symmet-
ric, and a small band presumably due to cationic vanadyl
−1
species [10] is observed at 1035 cm . When x > 2, the
shoulders of the P–O vibration are still present after reaction
and are better defined for increasing x values. The relative
N–H vibration intensity decreases to disappear for x = 2.

N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
179
+
+
Rb , Cs ), less hydrated, are cubic. Recently, it has been
shown by Berndt et al. [19] that, in addition to the triclinic
phase, a small fraction of thin crystals of the acid has the cu-
¯
bic Pn3m structure as well and that all the members of the
series of salts CsxH4−x[PMo11VO40]·nH2O (x = 2–4) have
this cubic structure.
The ammonium cation has been chosen in this study
+
because (i) it is similar to the hydrated proton (H3O or
+
H2O5 ), (ii) its decomposition by heating treatment leads,
after rehydration of the solid in ambient air, to hydrated pro-
tons; (iii) it can be substituted by cesium by a solid-state
thermal treatment in presence of cesium nitrate.
In water/dioxane v/v solution, the neutralization of
the acid by (NH4)HCO3 is complete without partial de-
composition of the heteropolyanion and before the crys-
tallization occurs. In consequence, the true neutral salt
Fig. 9. Catalytic performance after 18 h on reaction of the treated ammo-
nium salt (0 point) and of the xCs samples (F, IBA conversion; selectivity
to MAA, 2; acetone, Q; propene, "; acetic acid, !; COx ꢀ).
(
NH4)4[PMo11VO40] is obtained. Whereas the anionic
charge of the heteropolyanion increases by substitution of
one molybdenum atom by one vanadium, a cubic unit cell
similar to that of (NH4)3[PMo12O40] is found. The Rietveld
refinement converged to better crystal model indicators for
an anionic occupation of 3/4 and thermal factors of nitrogen
atoms are also more realistic (Table 1). In the case of full
occupancy of the cationic and anionic sites, no additional
cations could be located by difference Fourier maps even for
the Cs [PMo VO ] sample. Therefore, these results con-
The remaining amount of ammonium cation has been cal-
culated for each sample by thermogravimetry and elemental
analysis (Table 2). The release of ammonia in the conditions
of reaction is very important; xCs samples with x ꢁ 2 are
free of ammonium cations after reaction. A new large band
−1
at 628 cm , characteristic of reduced molybdenum oxide
18], appears when x < 2 and its relative intensity decreases
[
4
11
40
when x increases. The samples are deep blue when x < 2.7.
For x = 2.7 and 3, the samples display a pale green color.
firm the anionic default proposed by Berndt et al. for cesium
salt.
Could this conclusion be applied to other four ammonium
or cesium salts? The 12-molybdosilicate anion has the same
negative charge as the 1-vanado-11-molybdophosphate and
the ammonium salt (NH4)4[SiMo12O40] has been obtained
by the same method. As expected, these two neutral am-
monium salts are isostructural with about the same unit cell
parameter. Therefore, it seems to be possible to extend the
amazing conclusion that the cubic structure could accommo-
date four monovalent cations by an anionic default [19]. But
for five ammonium cations, the symmetry is different [17].
4
. Discussion
4
.1. Structure of the catalysts and thermal behavior
Structures of heteropolyacids and salts are strongly de-
pendent on the nature of the counterions and on the amount
of hydration water molecules. Highly hydrated acids are tri-
clinic but salts obtained with large alkaline cations (K ,
+
Fig. 10. Infrared spectra of xCs samples after use in reaction (the asterisk indicates a KBr impurity).

180
N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
◦
2
−1
The thermal treatment of (NH4)4[PMo11VO40] at 220 C
VO40], which develops a specific surface area of 150 m g
(after evacuation at 200 or 350 C for 2 h).
◦
in air flow leads, as expected from the thermogravimetric
curve, to the acidic salt (NH4)3H[PMo11VO40]. So, decom-
position of one of the four NH4 cations occurs at relatively
+
+
+
4.2. Exchange Cs /NH4
low temperature and would lead to a more stable salt in
which, following the model of Berndt, both cationic and
anionic sites are totally occupied. Rietveld refinements of
the diffraction patterns of (NH4)4[PMo11VO40] and treated
samples are comparable (Table 1). The results are in bet-
ter agreement with the model of anionic vacancies with all
occupied cationic positions. Taking into account that no dif-
It has been proven that cesium counterions change phys-
ical and chemical properties of the heteropolyanion such
as stability, porosity, acid strength, and oxidizing ability
[1] which are relevant for an application in heterogeneous
catalysis. Preparation of mixed cesium–ammonium salts
(NH4)4−xCsx[PMo11VO40] has been tempted by a solid-
+
+
+
+
ference could be observed between NH4 and H3O by
powder X-ray diffraction, this result is in agreement with a
state Cs /NH4 exchange starting from the ammonium salt.
Figs. 3 and 7 show that the driving force of the solid-state
exchange between ammonium and cesium cations is the de-
composition of ammonium nitrate and that a temperature of
+
substitution of a NH4 by a proton during thermal treatment;
+
this proton is then rehydrated and is in the form of H3O .
◦
It could be concluded that the anionic default is conserved
whereas it is not in agreement with the model proposed by
Berndt et al. for acidic cesium salts. Furthermore, this model
is valid for all treated samples with variable amount of re-
maining ammonium.
300 C is sufficient in order to obtain a quantitative elimina-
−1
tion of NH4NO3. Apparition of a small band at 1385 cm
only for x > 3 shows that nitrate is really eliminated for
lower x values and thus that the cationic Cs /NH4 ex-
+
+
change is quantitative.
+
+
The important exothermic phenomenon observed on the
Up to about x = 2 the Cs /NH4 substitution gives rise
to an increase of the unit cell parameter which remains con-
stant for higher x values (Fig. 8). Actually, at the temper-
ature needed for the cationic exchange, decomposition of
ammonium cations associated to a loss of ammonia occurs
simultaneously with the Cs /NH4 exchange. The exact
composition of the solid depends on the relative rates of
ammonium decomposition and cationic exchange. As pre-
viously seen, the first ammonium cation is decomposed very
◦
DSC curve at about 265 C is very likely due to the oxidation
of the released ammonia molecule to nitrogen and/or nitro-
gen oxides. Interestingly, no exothermic phenomenon is ob-
served under the same conditions with (NH4)4[SiMo12O40]
even if the acidic salt (NH4)3H[SiMo12O40] is obtained un-
der the same experimental conditions. This means that, in
this case, the released ammonia molecule is not oxidized.
This result clearly shows that the vanadium atom in the Keg-
gin structure enhances strongly the oxidizing ability of the
polyoxometalate which is thus able to catalyze the oxidation
+
+
◦
quickly at temperatures of 250 and 300 C required for the
exchange and it results that no more than three ammonium
cations can be expected to be substituted by cesium cations.
But substitution of ammonium by either hydrated protons or
cesium cations gives rise to an increase of the lattice para-
meter. So, the increase of the lattice parameter up to x = 2 is
very likely due to substitution of ammonium cations by both
cesium and hydrated protons and, for x > 2, since no ammo-
nium cations remain after the thermal treatment, the lattice
parameter is not strongly dependent on the value of x in sam-
ples. This is in agreement with the values of 11.804, 11.807,
and 11.785 Å obtained for x = 2, 3 and 4, respectively, with
acidic salts obtained by the precipitation method [19].
◦
of NH3 by oxygen at 220 C.
A series of salts (NH4)4−yHy[PMo11VO40] has been ob-
tained with 0 < y < 3.2 by long thermal treatments (several
days) of the ammonium salt performed in air flow at temper-
◦
atures ranging from 220 to 300 C. It has not been possible
to decompose quantitatively the ammonium cations in or-
der to obtain the cubic acid H4[PMo11VO40] because a par-
tial transformation of the heteropolyanion occurs at higher
temperature. The important point is that the cubic phase
is maintained whatever the composition. The unit cell pa-
rameter increases quite linearly as the ammonium amount
+
decreases, below 3 NH4 /HPA, suggesting that ammonium
cations and hydrated protons are statistically distributed in
the cationic sites (Fig. 6). Extrapolation of the unit cell pa-
rameter for x = 0 gives a value of 11.71 ± 0.01 Å which
can be compared with the value obtained by Berndt et al.
4.3. Reactivity
Catalytic behavior is another probe which can give in-
formation on the exact compositions of the solids. For low
x values, the catalysts have a deep blue color after reac-
tion, showing that heteropolyanion is partially reduced at the
steady state and thermogravimetric analysis shows that am-
monia is lost during the reaction (Table 2). Moreover, the
absence of the shoulder of the P–O vibration and a new band
(
12.0± 0.4 Å) from electron diffraction patterns of the small
amount of the acid which is cubic.
An important consequence of the progressive loss of
ammonia is the increase of the specific surface area from
2
−1
1
1
6 m g for (NH4)4[PMo11VO40] (crystalline sample) to
2
−1
−1
43 m g for (NH4)3H[PMo11VO40] (fine powder) show-
at 1035 cm suggests that vanadium atoms have moved to
ing that texture of the solid is strongly modified. Such
a high specific surface area has also been observed for
Cs3H[PMo11VO40] [19] and, in this study, for Cs4[PMo11-
a cationic position [10]. On the other hand, partial decom-
position of the heteropolyanions into partly reduced molyb-
denum oxide is proven by the observation of a large band at

N. Laronze et al. / Journal of Catalysis 220 (2003) 172–181
181
25 cm 1. Increasing amounts of cesium limits the reduc-
−
(NH4)5[PMo11V O40]. The initial state of vanadium seems
to have little influence. Indeed, conditions of reaction have
induced reduction and transformation of the catalysts lead-
ing to nearly the same compounds.
IV
6
tion and the decomposition since the relative intensity of the
−
1
6
25 cm band decreases.
The evolution of the catalytic performances in the first
hours of reaction could be due to the change of the catalyst
composition under the reactive steam. Selectivities and con-
version reach a quasi steady state; thus, they are reported
after 18 h of reaction.
References
The drastic variation of the catalytic behavior for x ≈ 3
has been previously observed [5,18]. For x > 3 the composi-
tion of the polyanion is not modified after use since infrared
spectra are unchanged (Figs. 7 and 10), but the catalysts are
inactive for MAA production. Lee et al. explain the selective
formation of acetone by the surface basicity caused by Cs
species (possibly oxides or carbonates). Another explanation
can be proposed without the requirement of new basic sites.
Actually, the yield of acetone is approximately the same be-
fore and after x = 3 and is likely due to the basicity of the
oxygen atoms of the heteropolyanion which could be in-
creased by the presence of cesium [30]. The high selectivity
in acetone can mainly be attributed to the fall of the rate of
formation of MAA and propene.
[1] M. Misono, Chem. Commun. (2001) 1141.
[2] G.P. Schindler, T. Ui, K. Nagai, Appl. Catal. A 206 (2001) 183.
[
[
3] F. Cavani, A. Tanguy, F. Trifirò, M. Koutrev, J. Catal. 174 (1998) 231.
4] M. Misono, N. Mizuno, T. Komaya, in: Proceedings, 8th International
Congress on Catalysis, Berlin, Vol. 5, Dechema, Frankfurt-am-Main,
1984, p. 487.
[5] K.Y. Lee, S. Oishi, H. Igarashi, M. Misono, Catal. Today 33 (1997)
183.
[6] N. Mizuno, M. Tateishi, M. Iwamoto, J. Catal. 163 (1996) 87.
[7] N. Mizuno, H. Yahiro, J. Phys. Chem. B 102 (1998) 437.
[8] J.B. Black, N.J. Claydon, P.L. Gai, E.M. Serwicka, J. Catal. 106 (1987)
1.
[9] M. Langpape, J.M.M. Millet, U.S. Ozkan, M. Boudeulle, J. Catal. 181
(1999) 80.
[
10] E. Cadot, C. Marchal, M. Fournier, A. Tézé, G. Hervé, in: M.T. Pope,
A. Müller (Eds.), Polyoxometalates, Kluwer Academic, Dordrecht,
1
994, p. 315.
11] Th. Ilkenhans, B. Herzog, Th. Braun, R. Schlögl, J. Catal. 153 (1995)
75.
Before x = 3, high catalytic activity is observed what-
ever the x value, whereas the composition of the initial
or used catalyst varied. Such a behavior is obtained in the
very similar ammonium–cesium salts obtained from the re-
[
2
[12] C. Marchal-Roch, R. Bayer, J.F. Moisan, A. Tézé, G. Hervé, Top.
Catal. 3 (1996) 407.
[13] J.K. Lee, J. Melsheimer, S. Berndt, G. Mestl, R. Schlögl, K. Köhler,
Appl. Catal. A 214 (2001) 125.
IV
duced (NH4)5[PMo11V O40] [18]. The drastic fall of the
catalytic behavior occurring between 3 and 3.2 Cs can be
explained by the nonpresence of small amounts of Lewis
acid sites on the surface. It must be noted that IR spectra of
the used catalysts are identical for each x value of both se-
ries. Indeed, conditions of reaction have induced reduction
and nearly the same compounds are obtained at the steady
state.
[
[
14] L. Marosi, G. Cox, A. Tenten, H. Hibst, J. Catal. 194 (2000) 140.
15] L. Marosi, C. Otero Areán, J. Catal. 213 (2003) 235.
[16] C. Marchal-Roch, N. Laronze, R. Villanneau, N. Guillou, A. Tézé, G.
Hervé, J. Catal. 190 (2000) 173.
[
[
[
17] C. Marchal-Roch, N. Laronze, N. Guillou, A. Tézé, G. Hervé, Appl.
Catal. 199 (2000) 33.
18] C. Marchal-Roch, N. Laronze, N. Guillou, A. Tézé, G. Hervé, Appl.
Catal. 203 (2000) 143.
19] S. Berndt, D. Herein, F. Zemlin, E. Beckmann, G. Weinberg, J. Schüt-
ze, G. Mestl, R. Schlögl, Ber. Bunseges. Phys. Chem. 102 (1998) 763.
20] P. Courtin, Rev. Chim. Min. 8 (1971) 75.
[
[
[
[
5
. Conclusion
21] R. Massart, Ann. Chim. 3 (1968) 507.
22] A. Boultif, D. Louër, J. Appl. Crystallogr. 24 (1991) 987.
23] A.D. Mighell, C.R. Hubbard, J.K. Stalik, NBS*AIDS80: A Fortran
Program for Crystallographic Data Evaluation., Nat. Bur. Stand. (US),
Technol. Note No 1141, 1981. NBS*AIDS83 is an expanded version
of NBS*AIDS80.
Stoichiometric ammonium or cesium salts of [PMo11-
VO40]4 or [SiMo12O40]
−
4−
have been prepared and they
¯
can display a cubic Pn3m structure. Rietveld refinements are
improved by attribution of a 3/4 occupancy factor for the
anionic position for (NH4)4[PMo11VO40] and Cs4[PMo11-
VO40] as it was already reported for the cesium salt.
[
[
[
24] J. Rodriguez-Carvajal, in: Collected Abstract of Powder Diffraction
Meeting, Toulouse, 1990, p. 127.
25] T. Roisnel, J. Rodriguez-Carvajal, Mater. Sci. Forum 378-381 (2001)
Cubic mixed acidic ammonium salts (NH4)4−yHy
118.
[
PMo11VO40], 0 < y < 3.2 have been obtained by air ther-
26] J.G. Highfield, J.B. Moffat, J. Catal. 98 (1986) 245.
mal treatment of the ammonium salt. Their unit cell parame-
ter increases as the amount of ammonium decreases.
[27] L. Pettersson, in: M.T. Pope, A. Müller (Eds.), Polyoxometalates,
Kluwer Academic, Dordrecht, 1994, p. 27.
[28] C. Rocchiggioli-Deltcheff, R. Thouvenot, R. Franck, Spectrochim.
Acta 32 A (1976) 587.
+
+
The solid-state exchange of Cs /NH4 from (NH4)4-
V
[
PMo11V O40] occurs simultaneously with an elimination
[29] J.C.A. Boeyens, J.G. McDougal, J. van Smit, J. Solid State Chem. 18
of ammonia. Thus no more than 3 Cs cations could be ex-
changed. The catalytic behavior of these salts is very sim-
ilar to those of the mixed Cs /NH4 salts issued from
(
1976) 191.
[30] P. Moriceau, A. Lebouteiller, E. Bordes, P. Courtine, Phys. Chem.
Chem. Phys. 1 (1999) 5735.
+
+