Application of ammonia intermittent temperature-programmed desorption to evaluate surface acidity of tungsten oxide supported on activated carbon.

Abstract

Intermittent temperature-programmed desorption of ammonia was used to study the strength and population of surface acid sites of tungsten oxide supported on activated carbon pretreated at 350 and 700 degrees C. Catalysts pretreated at 350 degrees C showed two types of surface acid sites and desorption occurred with free readsorption until a temperature of around 300 degrees C was reached. Pretreatment at 700 degrees C produced three different states of ammonia adsorbed on the catalysts and desorption occurred with free readsorption.

Full text

Page 1

Journal of Colloid and Interface Science 260 (2003) 449–453
www.elsevier.com/locate/jcis
Note
Application of ammonia intermittent temperature-programmed desorption
to evaluate surface acidity of tungsten oxide supported on activated carbon
M.A. Alvarez-Merino,aJ.P. Joly,bF. Carrasco-Marín,cand C. Moreno-Castillac,∗
aDepartamento de Química Inorgánica y Orgánica, Universidad de Jaén, 23071 Jaén, Spain
bLaboratoire d’Application de la Chimie a l’Environnement, UMR 5634 CNRS-Université Claude Bernard Lyon 1, 69622 Villeurbanne cedex, France
cDepartamento de Química Inorgánica, Facultad de Ciencias, Universidad de Granada, 18071 Granada, Spain
Received 4 October 2002; accepted 21 December 2002
Abstract
Intermittent temperature-programmed desorption of ammonia was used to study the strength and population of surface acid sites of
tungsten oxide supported on activated carbon pretreated at 350 and 700◦C. Catalysts pretreated at 350◦C showed two types of surface
acid sites and desorption occurred with free readsorption until a temperature of around 300◦C was reached. Pretreatment at 700◦C produced
three different states of ammonia adsorbed on the catalysts and desorption occurred with free readsorption.
 2003 Elsevier Science (USA). All rights reserved.
Keywords: Tungsten oxide catalysts; Carbon support; Surface acid strength
1. Introduction
Temperature-programmeddesorption (TPD) of ammonia
is a useful method to determine the surface acidity of
solids [1]. In principle, the strength of acid sites can be
evaluated by the activation energy of desorption (Ed) or the
heat of adsorption (?H). However, it is difficult to deduce
Ed (or ?H) distribution from TPD spectra, because the
surfaces under study are generally strongly heterogeneous
and the resolution of the spectra is very poor.
Several authors attempted to assess surface acidity by
using the deconvolution of the TPD spectrum [2–4] but
had to make numerous prior assumptions on the mechanism
of desorption and on the distribution of the sites. Thus,
these authors assumed that ammonia is adsorbed in a small
number of states with a Gaussian distribution of Ed. The
so-called preexponential factor was assumed to be equal
for all states and was determined in a separate experiment.
They used the classic desorption kinetic model with no
readsorption during the TPD runs.
The approach chosen in the present work is very dif-
ferent. The difficulties of interpreting TPD spectra were
*Corresponding author.
E-mail address: cmoreno@ugr.es (C. Moreno-Castilla).
solved experimentally using the method of intermittent
temperature-programmed desorption (ITPD), with no need
for prior assumptions [5]. This method was recently applied
to CO and CO2desorption from activated carbons [6].
Characterization of the total surface acidity of tungsten
oxide supportedon activatedcarbonby TPD of ammonia[7]
has shown that the shapes of the desorption profiles are
asymmetric, indicating the presence of surface acid sites
of different strengths. The present work aimed to utilize
the ammonia ITPD technique to evaluate the strength and
population of the surface acid sites of two representative
tungsten oxide catalysts supported on activated carbon.
2. Materials and methods
The TPD-MS apparatus used was described previous-
ly [8]. The equivalent conductivity of the tubes linking
the sample cell to the vacuum system was about 2 ×
10−5m3s−1. TPD runs were performed under dynamic
vacuum (4 × 10−5Pa). The mass spectrometer was set at
m/e = 15 amu in order to avoid the interference of water
fragmentation masses.
Two tungsten oxide catalysts supported on activated car-
bon were studied: one prepared from ammonium tungstate
(W14.9) and the other from tungsten hexacarbonyl
0021-9797/03/$ – see front matter  2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0021-9797(02)00253-9

Page 2

450
M.A. Alvarez-Merino et al. / Journal of Colloid and Interface Science 260 (2003) 449–453
(HW15.1). Tungsten loading was 15% in both catalysts.
Their preparation and surface characteristics are reported in
detail elsewhere [7].
The catalysts (0.3–1.0 g) were pretreated in the TPD
apparatus under a dynamic vacuum at 10◦Cmin−1up to
either350or700◦Cfor4h.Thesampleswerethencooledto
100◦Candexposedtoammoniaata pressureof1.6×103Pa
for 30 min. After this time, the catalysts were outgassed
for 30 min at the same temperature to remove physisorbed
ammonia. Finally, the samples were cooled to 25◦C and
either TPD or ITPD of chemisorbed ammonia was carried
out under dynamic vacuum.
The TPDrunswere carriedoutat 10◦Cmin−1. TheITPD
consisted of a series of interrupted TPDs at 4◦Cmin−1
obtained by applying a sawtooth heating program [6,8]. The
successive temperatures at which the linear heating program
was stopped were separated in increments of 25◦C.
3. Results and discussion
Table 1 shows the total amount of ammonia desorbed
from the supported catalysts, obtained from the area under
the TPD profiles. These results indicate that total surface
acidity decreased with higher pretreatment temperature.
In addition, the HW15.1 catalyst had the highest surface
acidity, derived from its much greater dispersion compared
with the W14.9 catalyst [7].
Information on the strength and population of the surface
acid sites was obtained by ITPD. Figure 1 depicts, as an
example, the ITPD of NH3 from HW15.1 pretreated at
350◦C. Each partial TPD was started at a temperature low
enough to make the desorption rate lower than the detection
threshold at the higher sensitivity of the mass spectrometer.
Thus, the onset of these desorptions was known with a
higher sensitivity than that shown in Fig. 1A.
Figure 1B shows the Arrhenius transforms of the curves
presented in Fig. 1A. In the lower rate region, where
desorption occurs at quasi-constant coverage, the Arrhenius
law was obeyed. The slope of the straight lines provides the
apparentactivation energy of desorption(E). The upper part
of these curves was not at quasi-constantcoverage;therefore
the downward concavity was due to a depletion of NH3
species from the surface.
Thevariationof E with theamountofNH3desorbed,cal-
culatedfromcurvessuchasthatinFig.1B,is showninFig.2
for HW15.1 pretreated at 350 and 700◦C. Similar curves
were obtained for W14.9. These curves exhibited plateaus
separated by rapid changes in E, which correspond to the
successive steps of ammonia desorption. Table 1 shows the
correspondingvalues of E, the apparent preexponential fac-
tors (A), and the widths of the plateaus. These widths are
related to the population of ammonia molecules involved in
the desorption steps.
The sum of the amounts of ammonia desorbed by ITPD
matched the amount determined from the TPD spectrum,
with an error lower than 5%. Furthermore, the performance
of TPD after ITPD showed no significant differencewith the
initial TPD carriedout beforeITPD.This indicatesthat there
was no irreversible transformation of the catalysts during
ITPD.
It is worthnotingthatthevaluesoftheapparentactivation
energy of desorption were obtained without recourse to a
theoretical model and without needing to make assumptions
about the desorption kinetics, except for the Arrhenius
law, whose application was experimentally verified at lower
desorption rates.
The possibility of free readsorption of ammonia under
our experimental conditions was investigated by an over-
shoot method [8]. In the course of TPD or ITPD runs, the
sample holder valve is closed for a short time and then
rapidly opened in order to reestablish the usual TPD condi-
tions. When the valve is closed, a sharp decrease of the mass
spectrometer signal occurs. When the valve is opened, two
types of mass spectrometer signal can be observed: the first
is a strong overshoot, typical of an irreversible desorption,
and the second corresponds to a mass spectrometer signal
that returns to the expected value with no overshoot. This
is the case for free readsorption. Application of this method
showed that readsorption occurs freely for all the states re-
Table 1
Apparent activation energies of desorption, preexponential factors, and relative population of the adsorbed ammonia states at 100◦C on tungsten oxide/carbon
catalysts obtained by ITPD
Catalyst and heat treatment
T range
(◦C)
100–200
200–300
100–250
250–325
100–200
200–325
325–400
100–175
175–275
275–400
NH3desorbed
(%)
EA
Total amount desorbed
(µmolg−1) (kJmol−1)
104.2 ±0.2
120.8 ±1.1
104.0 ±0.8
115.1 ±0.9
119.8 ±0.3
130.8 ±0.2
139.6 ±0.2
101.9 ±0.2
115.7 ±0.2
142.6 ±0.6
(s−1)
5×109
3×109
4×109
2×109
4×109
2×109
9×109
5×109
2×109
1×109
W14.9–350 0–62
62–96
0–80
80–96
0–44
44–95
95–100
0–26
26–80
80–100
100.8 (99.6)a
HW15.1–350
189.6 (181.7)
W14.9–700
82.7 (74.8)
HW15.1–700
100.1 (98.8)
aValues in parentheses were obtained by TPD.

Page 3

M.A. Alvarez-Merino et al. / Journal of Colloid and Interface Science 260 (2003) 449–453
451
Fig. 1. (A) Intermittent TPD of ammonia from catalyst HW15.1 heat treated at 350◦C. (B) Arrhenius transforms of the curves presented in (A).
ported in Table 1. Nevertheless, absence of readsorptionwas
only observed during the final ITPD of the samples pre-
treated at 350◦C. At temperatures above 330◦C, this TPD
corresponded to the highest point of the curve depicted in
Fig. 2. This observation is likely associated to a dissocia-
tive adsorption of ammonia,as reported by Auroux and Ger-
vasini [9]. Thus, in catalysts pretreated at 350◦C, free read-
sorption occurred during ammonia desorption up to a value
of 96%, with only the last 4% occurring without readsorp-
tion (Table 1).
A further insight into the desorption mechanism can be
gained by comparing our results with those of classical
TPD models. Thus, the desorption rate under vacuum
with possible readsorption of the adsorbate is classically
written [8]
(1)
−qmdθ
dt=
qmνθnexp?−Ed
1+qmka
RT
?
C(1−θ)nexp?−Ea
RT
?,
where qm, θ, t, ν, n, ka, C, Ed, and Ea respectively
denote, for a given adsorption state, the total amount of

Page 4

452
M.A. Alvarez-Merino et al. / Journal of Colloid and Interface Science 260 (2003) 449–453
Fig. 2. Variation of the apparent activation energy of desorption with
the amount of NH3desorbed in sample HW15.1 pretreated at different
temperatures: ", 350◦C and !, 700◦C.
sites, coverage, time, frequency factor, kinetic order of
desorption,adsorptionrateconstant,conductivityof thetube
connecting the catalyst sample cell to the vacuum system,
activation energy of desorption, and activation energy of
adsorption. Equation (1) has two limit forms (Eqs. (2)
and (3)) depending on the value of the denominator [8]:
(2)
−dθ
dt= νθnexp
Equation (2) is used for the desorption rate without
readsorption. In this case, E (obtained at quasi-constant
coverage)is equal to the activation energy of desorption Ed.
The second form is
?n
This expression corresponds to the case of free readsorp-
tion, where ?H = Ea− Ed. In this case, E is equal to
−?H.
These considerationsallowedthe preexponentialfactor A
to be calculated from Eq. (4):
?
−Ed
RT
?
.
(3)
−qmdθ
dt=Cν
ka
?
θ
1−θ
exp
??H
RT
?
.
(4)
−dθ
According to Eq. (2), in the case without readsorption
dt= Aexp
?
−E
RT
?
f(θ).
(5)
A = ν
while from Eq. (3), in the case of free readsorption:
and
f(θ) = θn,
(6)
A =
Cν
kaqm
The assessment of the order of magnitude of A from our
experimental results was performed using Eq. (4). Because
only the order of magnitude of A is of interest, f(θ) may
be considered close to 1 in any case, provided that θ is
not close to 0 or 1. For thermally desorbing states without
readsorption, A (or ν according to Eq. (4)) was found to
be equal to 2 × 1012s−1and 4 × 1013s−1, respectively,
and
f(θ) =
?
θ
1−θ
?n
.
for samples W14.9 and HW15.1 pretreated at 350◦C. These
figures, close to 1013s−1, are typical of desorption without
readsorption.
In the case of free readsorption,
(7)
kaqm
C
which means that
? 1,
(8)
A =
Cν
kaqm
? 1013s−1,
as was experimentally found. Thus, our ITPD results are
consistent with the classical physicochemical approach of
adsorption and, consequently, the values of E in Table 1 are
heats of adsorption.
Results compiled in Table 1 indicate that the catalysts
pretreated at 350◦C had two states of adsorbed ammonia
with the same E value, with the population of each state
being the only difference. Thus, the low desorption energy
state was more populated in the HW15.1 catalyst.
When the catalysts were heat treated at 700◦C, they
clearly exhibited up to three desorption states for the NH3
adsorbed; new states appeared at a desorption energy of
between 130 and 140 kJmol−1.
The differential heat of adsorption of ammonia on bulk
WO3as a functionof the surface coveragehas been reported
to display discontinuous adsorption heterogeneity [9,10],
showing four steps between 30 and 170 kJmol−1. The
values obtained in the present work are within this range.
The ammonia ITPD technique does not provide any
information on the nature of the surface acid sites of the
tungsten oxide catalysts supported on activated carbon used
in this work. However, this has been studied for other
tungsten oxide supported catalyst. For example, in the case
of WO3/ZrO2 catalysts, which have both Brønsted and
Lewis acid sites, an increase in the calcination temperature
creates double metal oxygen bonds [11]. The increase in
theirnumbermakesstrongerthe oxoacid.So,the appearance
of strong acid centers, with ammonia desorption energy
higher than 130 kJmol−1after the heat treatment of the
catalystsat700◦C,mightberelatedtotheformationofthese
acid sites. In addition, the increase in treatment temperature
of the WO3/C catalysts reduced the oxidation state of
tungsten [7], so it is expected that the Lewis acid centers
will vary with the heat treatment.
Catalysts studied in this work were used in the skeletal
isomerization of 1-butene [12]. Results showed that when
they were pretreated at 350◦C, their activity to produce
isobutene did not decrease with reaction time and no
dimerization byproducts (C3-C5and C2-C6) were obtained.
Onthecontrary,whenthecatalystswerepretreatedat700◦C
there was a large decrease of activity with reaction time,
with the appearance of byproducts from the dimerization
of butenes. These findings might be explained by the
results found in this work. Thus, the stronger acid sites
with activation energy higher than 130 kJmol−1, created

Page 5

M.A. Alvarez-Merino et al. / Journal of Colloid and Interface Science 260 (2003) 449–453
453
during the pretreatment at 700◦C, would be responsible
for the production of the dimerization byproducts and the
deactivation of the catalysts.
4. Summary
Results given in this work show that the method of
ITPD of ammonia can give information on the strength
and population of the surface acid sites of tungsten oxide
catalysts supported on activated carbon. This technique has
the advantage that it does not need previous assumptions
of either the mechanism of desorption or the distribution
of surface acid sites. The knowledge of these acid sites
might explain the behavior of these catalysts in the skeletal
isomerization of 1-butene to isobutene.
References
[1] C.M. Hidalgo, H. Itoh, T. Hattori, M. Niwa, Y. Murakami, J. Catal. 85
(1984) 362.
[2] H.G. Karge, V. Dondur, J. Phys. Chem. 94 (1990) 765.
[3] H.G. Karge, V. Dondur, J. Weitkamp, J. Phys. Chem. 95 (1991) 283.
[4] F. Arena, R. Dario, A. Parmaliana, Appl. Catal. A 170 (1998) 127.
[5] J.P. Joly, C. R. Acad. Sci. Paris (Sér. II) 295 (1982) 717.
[6] S. Haydar, C. Moreno-Castilla, F. Carrasco-Marín, J. Rivera-Utrilla,
A. Perrard, J.P. Joly, Carbon 38 (2000) 1297.
[7] M.A. Alvarez-Merino, F. Carrasco-Marín, J.L.G. Fierro, C. Moreno-
Castilla, J. Catal. 192 (2000) 363.
[8] J.P. Joly, A. Perrard, Langmuir 17 (2001) 1538.
[9] A. Auroux, A. Gervasini, J. Phys. Chem. 94 (1990) 6371.
[10] A. Gervasini, A. Auroux, J. Therm. Anal. 37 (1991) 1737.
[11] P. Afanasiev, C. Geantet, M. Breysse, G. Coudurier, J.C. Vedrine,
J. Chem. Soc. Faraday Trans. 90 (1994) 193.
[12] M.A. Alvarez-Merino, F. Carrasco-Marín, C. Moreno-Castilla, J. Ca-
tal. 192 (2000) 374.