Journal of Catalysis 316 (2014) 47–56

Contents lists available at ScienceDirect

Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat

The variety of Brønsted acid sites in amorphous aluminosilicates

and zeolites
Maxime Caillot a,b, Alexandra Chaumonnot b, Mathieu Digne b, Jeroen A. van Bokhoven a,c,⇑
a
Institute for Chemical and Bioengineering, ETH Zurich, Wolfgang Paulistrasse 10, 8093 Zurich, Switzerland
b
Direction Catalyse et Séparation, IFP Energies nouvelles, Rond-point de l’échangeur de Solaize, BP 3, 69360 Solaize, France
c
Laboratory for Catalysis and Sustainable Chemistry, Paul Scherrer Institute, 5232 Villigen, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: Brønsted acid sites of similar strength to those in zeolites are found in ASAs prepared by cogelation, as
Received 18 December 2013 shown by thermogravimetric study of the dehydration of ethanol to ethylene. ASAs prepared by grafting
Revised 25 April 2014 and a commercial ASA (Siralox 30) show Brønsted acid sites that differ from those of zeolites. CO respec-
Accepted 3 May 2014
tively ethanol adsorption followed by infrared spectroscopy identified the OH stretching frequency of the
Available online 28 May 2014
Brønsted acid sites specific to Si/Al2O3 and Al/SiO2 materials obtained by grafting to be ca. 3600 cm1 and
3640 cm1, respectively, although the intensity of the corresponding band is weak. These sites, contrary
Keywords:
to those of a H-USY zeolite with very few extra-framework aluminum, are highly sensitive to the surface
ASA
Zeolite
hydration state: Their number decreases with increasing temperature. The surface structure of Siralox 30
Brønsted acidity resembles that of Si/Al2O3 prepared in anhydrous conditions, showing similar isosurface acidic properties
Siralox and distribution of silicon species.
TGA Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction Lewis acid sites [4]. The diversity of ASAs surfaces may originate
from the numerous techniques for synthesis [7]. We chose to ratio-
Aluminosilicates are widely used in the field of catalysis due to nalize this by directly comparing ASAs synthesized by grafting and
their Brønsted acidic properties that enable the conversion of cogelation. Molecular precursors of silicon and aluminum were
hydrocarbons. However, the reasons why crystalline zeolites are deposited on alumina and silica to obtain, respectively, Si/Al2O3
much more active than amorphous aluminosilicates (ASAs) remain and Al/SiO2 [8]. By tuning the synthesis conditions (activation tem-
unclear. It has been proposed that the lower catalytic activity of perature and water content during synthesis), we obtained materi-
ASAs is due to a smaller number of Brønsted acid sites (BAS) whose als showing a diversity of surfaces. Thanks to thermogravimetric
structure, i.e. a tetrahedral aluminum substituting a silicon atom in measurements and catalytic characterization; we found that the
a silica lattice, is the same in both ASAs and zeolites [1–4]. Alterna- grafting of silicon and aluminum species creates Brønsted acid
tively, specific structures of the site in ASAs are proposed: The sites and that the intrinsic activity of these is lower than that of
acidity of the proton of a silanol is enhanced due to the vicinity a H-USY zeolite.
of a coordinatively unsaturated aluminum atom, but not bridging IR spectroscopy enables the characterization of the surface
the aluminum atom like in zeolites [5]. CO and pyridine adsorption hydroxyl groups of ASAs and zeolites. The IR spectroscopy of
followed by infrared (IR) spectroscopy and their theoretical evalu- adsorbed CO yields the nature and relative amount of Brønsted
ation based on DFT also suggested a non-permanently bridging acid sites and strong and weak/medium Lewis acid sites [9–12].
nature of the Brønsted acid sites of ASAs [6]. A recent study made Sites of similar strength were found in ASAs and steamed zeolites
a distinction between strong Brønsted acid sites of zeolitic sites of [9], acknowledging the complexity of the Brønsted acidity in ASAs
ASAs and weaker Brønsted acid sites, which density is higher and and zeolites [12]. By evaluating ASAs (Siral from Sasol, ex Condea)
which might be formed by silanol groups interacting with strong from 0 to 100 wt.% SiO2, a mixed silica–alumina phase was found
for compositions below 30 wt.% SiO2, possessing bridged zeolitic
Brønsted acid sites, as well as the encapsulation of the alumina sur-
⇑ Corresponding author at: Institute for Chemical and Bioengineering, ETH Zurich,
face by silicon species above 30 wt.% SiO2 with decreasing acidity.
Wolfgang Paulistrasse 10, 8093 Zurich, Switzerland.
Thermogravimetric analysis after ethanol adsorption is a rele-
E-mail addresses: [email protected] (M. Caillot), alexandra.
[email protected] (A. Chaumonnot), [email protected] (M. Digne), vant tool for investigating the surface percentage of silicon and
[email protected] (J.A. van Bokhoven). aluminum species [13–18]. It enables the quantification of the

http://dx.doi.org/10.1016/j.jcat.2014.05.002
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.
48 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56

number of Brønsted acid sites, which enables to calculate the turn- The samples (20–30 mg) were put in crucibles and inserted into
over frequency of Brønsted acid-catalyzed reactions, such as the a Mettler Toledo apparatus (TGA/SDTA851e) and activated in a
isomerization of m-xylene. This gives the intrinsic acidity of the flow of helium (60 mL min1) at 500 °C (heating rate 10 °C min1)
Brønsted acid sites. On grafted Si/Al2O3, the turnover frequency is for 2 h, to eliminate physisorbed water. After cooling to room
1.4  103 s1 site1, on grafted Al/SiO2 4.3  104 s1 site1 and temperature, ethanol was adsorbed by passing helium through a
on an H-USY zeolite with very few extra-framework aluminum saturator of ethanol until saturation of the weight gain (approx.
3.1  102 s1 site1. The IR study of adsorbed ethanol is useful 30 min). The weakly adsorbed ethanol molecules were desorbed
for evaluating the intermediate that leads to ethylene upon dehy- by purging for 1 h at room temperature, and the strongly adsorbed
dration [19]. molecules were desorbed during the thermo-programmed
In the present study, we demonstrate the variety of Brønsted desorption step (TPD) by heating the sample from room tempera-
acidities in zeolites and ASAs. The model Si/Al2O3 and Al/SiO2 ASAs ture to 400 °C. This resulted, when calculating the first derivative of
that were previously prepared by grafting and characterized are the weight loss during TPD with respect to temperature, in two
compared to ASAs prepared by cogelation. A cogel ASA, prepared main desorption features corresponding to ethanol and ethylene
by mixing molecular precursors of silicon and aluminum species, release. The amount of dehydrated ethanol was calculated as
and whose structure is expected to differ from that of the grafted follows:
samples, shows a heterogeneous structure and the presence of zeo-
litic sites that are not found in the ASAs made by grafting. A Siralox Ethanol dehydrated ðnm2 Þ ¼ Ethanol retained after purge ðnm2 Þ
reference from Sasol (Siralox 30, 30 wt.% SiO2) is compared to Si/ Apeak ethylene
Al2O3 samples that we made by grafting. We uniquely give evidence  ð1Þ
Apeak ethylene þ Apeak ethanol
of the presence of specific structures for the BAS of ASAs made by
grafting thanks to CO and ethanol adsorption followed by IR. where Apeak i is the area of the peak corresponding to the desorption
of species i when calculating the derivative of the weight loss of
2. Experimental ethanol during TPD.
For the m-xylene isomerization test, the samples were first acti-
The characteristics and synthesis conditions of the materials vated in a flow of air at 350 °C, and gaseous m-xylene (0.6 cm3 h1)
evaluated in this study are given in Table 1. Grafted samples, which was reacted with the bed of catalyst at 350 °C in a flow of nitrogen.
synthesis is described in [8], were prepared by contacting, in con- The products were analyzed by online chromatography by means
trolled conditions, a silicon precursor (tetraethyl orthosilicate, TEOS) of a Hewlett Packard 6850 apparatus equipped with a flame ioniza-
to alumina, leading to Si/Al2O3 materials (entries 3–7), and an alumi- tion detector (FID) and an FFAP column. Because the catalysts
num precursor (aluminum isopropoxide) to silica, leading to Al/SiO2 deactivated fast when submitting to m-xylene flow, the rate of con-
materials (entry 8). Both Si/Al2O3 and Al/SiO2 materials underwent version of m-xylene to its products was calculated after 10 min on
final calcination at 550 °C. The commercial ASA reference was a stream [4]. The value was given by:

Pspecies Ai 1
1 2 i number of carbonsi flow rate ðcm3 h Þ  densitym-xylene ðg cm3 Þ NA
Rate ðmolecule h nm Þ ¼ Pspecies A
 1
 ð2Þ
Ai
þ m-xylene M m-xylene ðg mol Þ  mcatalyst ðgÞ SBET ðnm2 g1 Þ
i number of carbonsi 8

Siralox 30 from Sasol (ASA(comm.), entry 10) obtained by impregna- where Ai is the area of the GC peak of species i (i = toluene,
tion of alumina with a silica gel. The cogel sample (entry 9) was p-xylene, o-xylene, trimethylbenzenes), NA the Avogadro number
prepared by mixing molecular precursors of both aluminum (alumi- (6.02  1023 mol1), and SBET the surface area of the catalyst.
num sulfate) and silicon (sodium metasilicate), followed by washing,
drying, and calcination at 550 °C. The zeolite reference (entry 11) 3. Results
was an ultra-stabilized Y zeolite, H form, from Zeolyst.
CO-infrared studies were performed on a Nexus Fourier trans- 3.1. Ethanol adsorption and dehydration followed by
form apparatus (resolution 4 cm1). The samples, pressed into pel- thermogravimetric analysis
lets, were pretreated in situ under vacuum (105 mbar) for 10 h at
450 °C (heating rate of 5 °C/min) including a plateau at 150 °C for Fig. 1 gives the derivative of the weight loss during TPD on
1 h. In order to evaluate the influence of the activation temperature various ASAs and H-USY. The first peak at around 80 °C corre-
prior to CO adsorption, two of the grafted ASAs were activated at sponds to the desorption of ethanol; its position is the same for
250 °C instead of 450 °C. Spectra were taken before and after all samples. The second peak above 150 °C corresponds to the
adsorption of increasing doses of CO, at liquid nitrogen tempera- release of ethylene from dehydration of ethanol. No release of
ture, up to a total amount of 600 lmol. Ethanol-infrared studies diethylether was detected. On Si/Al2O3 materials prepared under
were performed on a Bio-RAD Fourier transform spectrometer anhydrous conditions (Fig. 1a), the ethanol dehydration peak first
(resolution 4 cm1). The samples were pressed into pellets and shifts to higher temperatures and then to lower temperatures,
pretreated in situ under high vacuum (108 mbar) for 10 min at which is indicative of the consumption of LAS and to the formation
200 °C or 400 °C (heating rate of 2 °C/min). Spectra were taken of BAS, respectively [17]. In terms of intensity and position of the
after adsorbing increasing doses of ethanol at room temperature peaks, ASA(comm.) (entry 10) shows ethanol and ethylene desorp-
into the chamber where the sample was hold. Ethanol was then tion features very similar to those of SA(12/anh) (entry 4). Fig. 1b
desorbed by gradually heating the sample up to 400 °C. gives the results for ASAs and H-USY. The increasing ethanol dehy-
The thermogravimetric study of ethanol adsorption and dration temperature from H-USY to SA(17/anh) and AS(25/anh), for
desorption followed the protocol given by Kwak et al. [20] that which ethanol dehydration takes place exclusively on BAS, relates
we previously used for Si/Al2O3 [17] and Al/SiO2 [18] materials. to a decreasing strength of the BAS [18]. ASA(cogel) shows a low
 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56 49

Table 1
Synthesis conditions and main characteristics of the materials [7].

Entry Sample nature Sample wt.% SBET Synthesis technique Pretreatment conditions Synthesis conditions
name SiO2 (m2 g1) (temperature, mode) (temperature, media)
(1) Alumina c-Alumina 0.0 235 – – –
(2) Silica Silica 100.0 550 – – –
(3) ASA (Si/Al2O3) SA(7/anh) 7.3 200 CLD anhydrous 40 °C, vacuum 110 °C, toluene
(4) SA(12/anh) 12.2 180 CLD anhydrous (x2) 40 °C, vacuum 110 °C, toluene
(5) SA(17/anh) 17.1 160 CLD anhydrous (x3) 40 °C, vacuum 110 °C, toluene
(6) SA(17/ 16.5 175 CLD aqueous (4 eq. water) 40 °C, vacuum 110 °C, toluene
4eqW)
(7) SA(17/ 17.2 165 CVD 250 °C, air flow 250 °C, air
CVD250)
(8) ASA (Al/SiO2) AS(25/anh) 24.8 330 CLD anhydrous (x2) 30 °C, vacuum 110 °C, toluene
(9) ASA (cogel) ASA(cogel) 20.0 160 Cogelation – –
(10) ASA (comm. ASA(comm.) 30.0 315 Impregnation of alumina – –
reference) with silica gel
(11) Zeolite H-USY 94.6 780 – – –

Table 2
Surface density of ethanol retained after purge and undergoing dehydration (nm2)
following a pretreatment at 500 °C.

Entry Sample Ethanol retained Percentage of free Ethanol undergoing

after purge alumina surface dehydration
(nm2) (col. 3) (%) (col. 4) (nm2)a (col. 5)
(1) c-Alumina 2.8 100 0.83
(2) Silica 1.3 0 0.00
(3) SA(7/anh) 2.4 70 0.27
(4) SA(12/anh) 2.2 57 0.23
(5) SA(17/anh) 1.7 29 0.17
(8) AS(25/anh) 2.2 56 0.61
(9) ASA(cogel) 1.1 0 0.12
(10) ASA(comm.) 2.1 55 0.27
(11) H-USY 1.4 – 0.18
a
Based on the total surface area.

3.2. Infrared spectroscopy

3.2.1. Infrared spectra of the OH stretching region upon ethanol

adsorption
Fig. 2 gives the difference spectra of the IR stretching region,
recorded after adsorption of ethanol on alumina and Si/Al2O3 sam-
ples, following a pretreatment at 200 °C. Infrared spectra of the OH
stretching region of the materials are shown in a previous paper
Fig. 1. Derivative of TPD weight loss of ethanol on ASAs and H-USY after [17]. Ethanol adsorption on the OH groups of the support causes
pretreatment at 500 °C: (a) Si/Al2O3 prepared by anhydrous CLD [16] (entries 3– a decrease in intensity in the OH stretching region (3600–
5) and commercial reference (entry 10); (b) Si/Al2O3 (entry 5), Al/SiO2 (entry 8), and 3800 cm1) and the appearance of peaks in the CH2–CH3 stretching
cogel (entry 9) ASAs and H-USY (entry 11). Numbers in brackets correspond to region (2800–3000 cm1) [13]. H-bonding between ethanol mole-
entries in Table 1.
cules and hydroxyl groups of the support is characterized by a
broad positive band in the 3000–3600 cm1 [21]. On alumina, eth-
anol molecules are mainly adsorbed on Al(VI)–OH groups of the
amount of ethanol desorbing, and its ethanol dehydration peak has (1 0 0) surface, as indicated by the negative peak at 3770 cm1.
the same position as that of H-USY. The weight loss above 250 °C is Upon heating and dehydration of ethanol, these OH groups are
ascribed to dehydroxylation. restored (not shown here). These observations confirm the role of
Table 2 gives the amount of ethanol retained after purge and the the (1 0 0) surface of c-alumina in the dehydration of ethanol
amount of ethanol undergoing dehydration. ASA(cogel) and c-alu- [17,20]. To a lesser extent, Al–OH groups of the (1 1 0) surface (at
mina show, respectively, the minimum and maximum values of 3790 cm1 and 3730 cm1) also contribute to the adsorption of
ethanol retained after purge (col. 3). The calculation of the percent- ethanol. On SA(7/anh), ethanol adsorption takes place on silanol
age of free alumina surface, in column 4, is based on the preferen- groups (3745 cm1) and OH groups of the (1 1 0) surface of alumina
tial adsorption of ethanol on alumina surfaces compared to the (3730 cm1). The OH groups of the (1 0 0) surface are grafted with
silica surface [17,18]. Given its specific crystalline structure, this silicon species and can no longer adsorb ethanol [17], hence the
calculation has no meaning for H-USY. The low amount of ethanol absence of a negative peak at 3770 cm1. On SA(12/anh), due to
retained on ASA(cogel) is evidence of a highly dehydroxylated sur- grafting of the (1 1 0) surface with silicon species [17], the intensity
face, without sites typical of alumina, such as LAS. Except for silica, of the negative bands corresponding to the OH groups of the (1 1 0)
all samples possess ethanol dehydration sites (col. 5). We demon- surface (3790 cm1 and 3730 cm1) decreases. The ethanol dehy-
strated that these sites are LAS on alumina, BAS on AS(25/anh), dration sites of SA(17/CVD250) are exclusively BAS [17]. On
SA(17/anh) and H-USY, and a sum of LAS and BAS on SA(7/anh) SA(17/CVD250), silanols are the main adsorption sites, but the neg-
and SA(12/anh) [17,18]. ative band extends until 3650 cm1.
50 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56

Fig. 2. Infrared spectra (OH and CH2–CH3 stretching region) of Si/Al2O3 after adsorption of ethanol, following pretreatment at 200 °C. (x): entries in Table 1; (1 0 0) or (1 1 0):
surfaces of c-alumina.

On alumina, the CH2–CH3 stretching region is characterized by ethanol first adsorbs on OH groups of the supercages (SiOHAl HF,
four main peaks. The peaks at 2967 cm1 and 2926 cm1 are 3630 cm1) and sodalite cages (SiOHAl LF, 3564 cm1) [22], then
ascribed to the asymmetric stretching vibration of CH3 (ma(CH3)) finally on silanols, which confirms that ethanol is a good probe of
and CH2 (ma(CH2)), respectively; the peaks at 2895 cm1 and the BAS. As for ASAs, H-bonding of ethanol is visible and the
2868 cm1 are ascribed to the symmetric stretching vibration of CH2–CH3 stretching region is dominated by the peak of ma(CH3).
CH2 (ms(CH2)) and CH3 (ms(CH3)), respectively [13]; H-bonding Thus, this stretching region has a characteristic shape when etha-
confirms the adsorption of ethanol on OH groups. From alumina nol is adsorbed on BAS of ASAs and zeolites. However, contrary
to SA(7/anh), the four peaks are blueshifted, and the intensity of to H-USY, there is no wavenumber specific to the OH stretching
ms(CH3) decreases. When the silica loading increases, ma(CH3) is of BAS of ASAs. These might be hidden because of the appearance
redshifted, the intensity of ms(CH3) decreases, and H-bonding is of the broad H-bonding band.
more pronounced.
The temperature of activation of alumina impacts the type of 3.2.2. Infrared spectra of the OH stretching region and CO stretching
adsorption of ethanol (Fig. 3). Upon ethanol adsorption after a pre- region upon CO adsorption
treatment at 400 °C, no-H-bonding is visible. The negative signal in Fig. 5 gives the evolution of the OH and CO stretching regions of
the OH region only appears for high ethanol coverage, and it is not H-USY upon CO adsorption following a pretreatment at 450 °C. Due
centered at 3770 cm1, which indicates that most of the ethanol to the formation of the OH–CO bond, the stretching wavenumber
molecules are not adsorbed on OH groups. The absence of distur- of the adsorbing OH groups is shifted toward lower values. On a
bances, in the OH stretching region, between 3750 cm1 and difference spectrum, the wavenumber shift due to adsorption on
3800 cm1, confirms that the (1 0 0) surface has no hydroxyl group. OH groups results in a negative peak indicating the original wave-
In fact, upon dehydroxylation of the (1 0 0) surface by pretreatment number of the OH group and a positive band at lower wavenumber
at high temperature, LAS form and become the adsorption sites of relating to the new OH–CO complex [12]. The intensity of the shift
ethanol instead of OH groups. However, whatever the activation (Dm) is proportional to the acidic strength of the OH group [9]. CO
temperature, the four peaks in the CH2–CH3 stretching region have first adsorbs on strong Lewis acid sites (2230 cm1, r) (Fig. 5a).
the same ratios of intensity and the same positions. This means This causes a shift of the OH stretching peak of isolated silanols
that the shape of the CH2–CH3 stretching region is not sensitive (3745 cm1) and SiOHAl HF (3632 cm1) toward higher wavenum-
to the nature of the adsorption site (BAS or LAS), but rather to bers. This phenomenon was observed by Crépeau et al. and it was
the local environment of the site (aluminum or silicon atoms). ascribed to thermal effects caused by the temperature gap between
Fig. 4 compares the results of ethanol adsorption on ASAs CO and the temperature of the cell [11]. One can also postulate that
(SA(17/CVD250) and AS(25/anh)) to that on a zeolite (H-USY), fol- silanol groups are disturbed due to the introduction of the first CO
lowing a pretreatment at 200 °C. On AS(25/anh), the OH and CH2– molecules. Further adsorption of CO molecules is characterized by
CH3 stretching regions are very similar to that of SA(17/CVD250), a peak at 2180 cm1 (CO–OH(a)) in the CO region (s). The simul-
and silanols are the main adsorption sites of ethanol. On H-USY, taneous appearance of a negative band at 3632 cm1 in the OH
 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56 51

Fig. 3. Infrared spectra (OH and CH2–CH3 stretching region) of c-alumina after adsorption of ethanol, following pretreatment at 200 °C and 400 °C.

Fig. 4. Infrared spectra (OH and CH2–CH3 stretching region) of Si/Al2O3, Al/SiO2, and H-USY after adsorption of ethanol, following pretreatment at 200 °C. Numbers in brackets
correspond to entries in Table 1.

stretching region confirms the adsorption on strong BAS (SiOHAl have failed to locate the original wavenumber of BAS, which has
HF, OH(a)). Upon CO adsorption, this OH band is shifted to lead to various hypotheses aiming at explaining its absence: struc-
3280 cm1 (CO–OH(a)). The CO adsorption on silanols (t) causes ture of BAS differing from that of zeolites [5,11] or low number of
the appearance of several peaks in the CO stretching region (CO– bridged sites [10], for example. Here, we observe both the shifted
OH(b) and (c)). The silanol peak (OH(b–c)) is shifted to wavenumber of the CO–OH complex starting at around
3660 cm1 (CO–OH(b–c)) upon CO adsorption. 3400 cm1 (CO–OH(A)), the original wavenumber of the BAS at
Fig. 6 gives the parallel evolution of the OH and CO stretching 3640 cm1, (OH(A)) and the stretching frequency of CO on BAS at
region of AS(25/anh) upon the adsorption of CO following a pre- 2177 cm1 (CO–OH(A)). In agreement with the observations of
treatment at 450 °C. The first CO molecules are adsorbed on strong Cairon et al. and Crépeau et al. [9,11], CO is then chemisorbed on
(2230 cm1, r) then weak/medium LAS (2200 cm1, s) (Fig. 6a) weaker OH groups (u), hence higher wavenumber for CO–OH(B)
[9–11,23]. Further CO adsorption causes the appearance of a peak (3550 cm1) and lower wavenumber for CO–OH(B) (2157 cm1).
at 2177 cm1, which is associated with adsorption on BAS (t) [11]. The right part of the silanol peak (3745 cm1) becomes negative,
As for zeolites (Fig. 5), negative and positive bands are visible in the which suggests that weak-acidic silanols in interaction with each
OH stretching region. Studies about ASAs have shown the presence other are the adsorbing sites (OH(B)). Physisorption of CO causes
of positive bands corresponding to perturbed OH bands, but they the appearance of a third component in the OH stretching region
52 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56

Fig. 5. Progressive adsorption of CO on H-USY (entry 11) followed by infrared spectroscopy after pretreatment at 450 °C.

(CO–OH(C), 3660 cm1) and in the CO stretching region (CO– AS(25/anh) (3640 cm1), and which are hardly visible at high acti-
OH(C), 2135 cm1). The global silanol peak becomes negative, vation temperature. The CO dose that is adsorbed on the BAS of
which shows that both isolated and interacting silanols undergo SA(17/4eqW) leads to a more intense CO–OH(A/B) band in the
the physisorption of CO. CO region, at a higher wavenumber than for the following CO doses
The OH and CO stretching regions of SA(17/anh), SA(17/4eqW) (2173 cm1 versus 2156 cm1), as previously observed [12].
and SA(17/CVD250) (entries 5, 6, and 7, respectively) upon CO
adsorption after pretreatment at 450 °C are similar to the OH and
3.3. Isomerization of m-xylene
CO stretching region of AS(25/anh). However, no negative band
for BAS in the OH region nor specific peak for BAS in the CO region
Table 4 gives the rates of conversion of m-xylene. These and the
is detected.
turnover frequency (TOF) of the BAS of Si/Al2O3 (entries 3, 4, and
The relative number of LAS and BAS of each sample is compared
5), Al/SiO2 (entry 8), and H-USY (entry 11) were determined based
by integrating the corresponding band, in the CO stretching region,
on the amount of ethanol dehydration exclusively occurring on
at saturation of its intensity. The values, reported in Table 3, are
BAS, as previously calculated [17,18]. The density of BAS is
normalized by the weight of the wafer and the surface area. Entries
assumed to correspond to the density of adsorbed ethanol that is
4–6 and 10 show no peak that can be exclusively ascribed to BAS;
converted to ethylene upon temperature-programmed desorption.
hence, no values are reported. ASA(comm.) has the same number
ASA(cogel) shows the highest isoweight and isosurface rates of
of strong LAS as SA(12/anh) and the same amount of weak/medium
conversion, and the highest TOF of the ASAs. However, compared
LAS as SA(17/anh). Consistent with its low amount of extra-frame-
to SA(17/anh) and AS(25/anh), showing the best catalytic perfor-
work aluminum species, H-USY shows almost no LAS.
mance of their respective category, the number of BAS per surface
On ASAs, there is an impact of the activation temperature of the
unit is low. Furthermore, the level of conversion and the TOF are
sample on the amount of CO adsorbed on BAS and LAS. Fig. 7 gives
much lower than for H-USY. The isosurface conversion rate of
the results for AS(25/anh) and SA(17/4eqW), which possess the
ASA(comm.) is intermediary between that of SA(12/anh) and that
highest surface density of BAS of Si/Al2O3 and Al/SiO2, respectively
of SA(17/anh); but due to a higher surface area, its isoweight con-
[17,18], on the OH and CO stretching regions. At high activation
version rate is much higher. Considering that the surface structure
temperature, the number of LAS (2180–2240 cm1) and BAS
of ASA(comm.) can be rationalized as that of a Si/Al2O3 material
(OH(A)) of AS(25/anh) (Fig. 7a) is lower; physisorption (OH(C)) is
(vide infra), hence by using the same TOF, the number of BAS per
constant. Integration of the peaks (Table 3) confirms a lower num-
surface unit can be estimated. The value is, again, between those
ber of strong LAS (69%), weak/medium LAS (31%), and BAS
of SA(12/anh) and SA(17/anh).
(30%) at 450 °C versus 250 °C. Although their quantification is
more difficult in the absence of a proper peak to integrate in the
CO region, the number of silanol groups of SA(17/4eqW) 4. Discussion
(Fig. 7b), measured during physisorption of CO, is constant what-
ever the activation temperature. On SA(17/4eqW), the number of 4.1. Zeolitic Brønsted acid sites in ASAs
strong and weak/medium LAS increases (+115% and +32%, respec-
tively) when the activation temperature increases; the number of Table 5 summarizes the main conclusions about Brønsted acid-
silanol groups of SA(17/4eqW) also increases (+30%). The nega- ity of ASAs and zeolite. The ethanol dehydration temperature of
tive peak at around 3600 cm1 in the OH stretching region indi- ASA(cogel) is lower than that of other ASAs and equal to that of
cates the presence of BAS at a lower wavenumber than for H-USY. It means that, contrary to the other ASAs, ASA(cogel) has
 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56 53

Fig. 6. Progressive adsorption of CO on Al/SiO2 (AS/25anh, entry 8) followed by infrared spectroscopy after pretreatment at 450 °C.

BAS of zeolitic strength. The presence of such strong BAS was gel) (ca. 20 nm) compared to the micropores of the zeolite and
already demonstrated in ASAs prepared by cogelation and homo- amorphous structure, resulting in a lower confinement and fitting
geneous deposition of aluminum on silica using selective poisoning of the reactant molecule to the pore and the active site, lowers the
of these sites by lutidine during m-xylene isomerization [4]. How- heat of adsorption. The low amount of ethanol retained on the sur-
ever, the TOF for m-xylene isomerization of ASA(cogel) is five times face of ASA(cogel) after purge (Table 2) and the large amount of
lower than that of H-USY. ASA(cogel) and H-USY have the same water lost during pretreatment (20% of the total sample weight)
intrinsic activation energy for catalytic reactions but different rates indicate that the surface of the sample is sensitive to heating and
because of different surface structures affecting the adsorption and highly dehydroxylated upon activation at 500 °C. The study of
the number of active sites [3]. The higher pore diameter of ASA(co- the OH stretching region in IR spectroscopy confirms the low

Table 3
Number of sites probed by CO (a.u. isosurface data: intensity of the band in the CO region after integrating and dividing by the weight of the wafer and the surface area of the
sample).

Entry Sample Activation temperature (°C) Strong LAS Weak/medium LAS BASa Silanols
(4) SA(12/anh) 450 39 374 –
(5) SA(17/anh) 450 76 272 –
(6) SA(17/4eqW) 250 34 333 – 605
(6) SA(17/4eqW) 450 73 441 – 802
(8) AS(25/anh) 250 88 219 781 468
(8) AS(25/anh) 450 27 152 547 456
(10) ASA(comm.) 450 38 278 –
(11) H-USY 450 8 14 1250
a
Entries 4–6 and 10: no peak specific to BAS. Entry 11: integration of all the bands corresponding to CO adsorption on OH groups.
54 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56

Fig. 7. Impact of the activation temperature on CO adsorption: (a) AS(25/anh); (b) SA(17/4eqW). Dotted lines correspond to CO adsorption on BAS.

Table 4
m-Xylene conversion over ASAs and zeolites.

Entry Sample Rate of m-xylene converted @ 10 min TOF (103 s1 site1) BAS (nm2)b
1
Isoweight (mmol h g1
cat) Isosurface (molecule h 1
nm 2
)
(3) SA(7/anh) 7 21.8 1.4 0.04
(4) SA(12/anh) 16 54.0 1.4 0.11
(5) SA(17/anh) 25 94.1 1.4 0.19
(8) AS(25/anh) 43 78.9 0.4 0.52
(9) ASA(cogel) 69 259.6 6.0 0.12
(10) ASA(comm.) 42 80.0 1.4a 0.16
(11) H-USY 2640 2034.8 31.4 0.18
a
Based on the hypothesis that ASA(comm.) is similar to Si/Al2O3 material, for which the TOF is 0.14 s1 site1.
b
Based on the TOF.
 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56 55

Table 5 decreases when increasing the activation temperature, while the

Main conclusions on the BAS of grafted and commercial ASAs and zeolites. number of silanols remains constant (Fig. 7a). There is no clear
Entry Sample name Type of Evolution of the Characteristic explanation for this observation, but it confirms the role of the
Brønsted number of BAS wavenumber hydration state of the surface in the number of BAS. NMR on
acid site with increasing in IR (cm1) AS(25/anh) showed no change in the distribution of coordination
(BAS) temperature
of the aluminum atoms upon adsorption of ethanol, as it is
(5) SA(17/anh) expected for PBS.
(6) SA(17/4eqW)
(7) SA(17/
In the OH stretching region of the infrared spectra of AS(25/anh)
CVD250) upon CO adsorption, a Brønsted acid site with a stretching fre-
(10) ASA(comm.) ASA (Si/Al2O3) Decreases (upon 3600a quency of 3640 cm1 can be identified in Al/SiO2 (Fig. 6). Because
heating, of the lower surface area of AS(25/anh), the negative peak observed
formation
during CO adsorption is less intense than those of H-USY. As a con-
of a strong Lewis
acid site and a sequence, this peak is hidden by the H-bonding following the
silanol) adsorption of ethanol (Fig. 4). The broadness of the band at
(8) AS(25/anh) ASA (Al/SiO2) Decreases 3640 3640 cm1 in IR-CO indicates a certain variety of sites, which cor-
(11) H-USY relates with a broad ethanol dehydration peak (Fig. 1b), contrary to
(9) ASA(cogel) Zeolitic Stable 3630b
H-USY. The lower wavenumber of the BAS of Si/Al2O3 (3600 cm1)
a
As found in SA(17/4eqW), which contains the highest number of active sites, and the lower ethanol dehydration temperature (Fig. 1b) confirm a
following a pretreatment at 250 °C before CO adsorption. different nature and environment of the sites between the two
b
The stretching frequency of Brønsted acid sites of ASA(cogel) is supposed to be
types of ASAs.
the same as that of H-USY, but no evidence was found in IR-CO.
On H-USY, we show that both the dehydration of ethanol and
the adsorption of CO followed by IR probe the BAS of the supercag-
number of silanol groups (not shown here). Moreover, the surface es. These sites show constant ethanol dehydration temperature
contains almost no aluminum species that form LAS, which would and amount of ethanol dehydrated whatever the pretreatment
retain ethanol even after dehydroxylation. Thus, the cogelation temperature. The number of BAS probed by CO adsorption is sim-
synthesis does not lead to a homogeneous polymerization of spe- ilar after activation at 250 °C and 450 °C (not shown here), which
cies, and most of the minor component (silicon species) ends up confirms the high stability to temperature of the bridged BAS of
on the outermost surface of the material. The core contains the H-USY. The shift of the OH band (Dm) upon CO adsorption is higher
major component (aluminum species), which explains the high for the BAS of H-USY (Dm = 352 cm1) than for the BAS of AS(25/
pore diameter and low surface area. The few aluminum species anh) (Dm = 240 cm1) and for the silanol groups (Dm = 85 cm1
on the surface might be incorporated into the silica matrix and for both H-USY and AS(25/anh)), which is consistent with a higher
would lead to zeolitic BAS (Table 5), by bridging silanols. In con- acidity of the BAS of the zeolite [9]. A similar conclusion had been
trast, the surface of H-USY is more stable against thermal deactiva- obtained by calculating the turnover frequencies (TOF) of the BAS
tion (10% weight loss during activation) and shows a higher of these materials for m-xylene isomerization [18].
amount of ethanol retained after purge (Table 2); thus, its surface We found no impact of the type of zeolite (H-USY with and with
silanols condensate to a lesser extent than those of ASA(cogel) very few EFAL, H-ZSM-5) nor of the activation conditions (200 °C or
upon heating. The stretching frequency of the BAS of ASA(cogel) 500 °C) on the dehydration temperature of ethanol (ca. 210 °C, see
could not be determined upon CO adsorption followed by IR spec- Appendix A). However, the nature of the intermediate leading to
troscopy; this can be the consequence of the low surface area and ethylene, either a carbenium ion [24] or an alkoxy [15,20], is
number of BAS and the high amount of water despite debated. On alumina, the reactivity of ethanol depends on the acti-
pretreatment. vation temperature of the catalyst. After a pretreatment at 500 °C,
ethanol is adsorbed at room temperature on the LAS of the (1 0 0)
surface and is further dehydrated upon heating at ca. 250 °C
4.2. Diversity of Brønsted acid sites in zeolites and ASAs according to a mechanism of the E2-type, i.e. concerted [25]. Our
results show that after a pretreatment at 200 °C, ethanol is
In all ASAs other than ASA(cogel), the ethanol dehydration tem- adsorbed on the Al(VI)–OH groups of the (1 0 0) surface. Contrary
perature is higher than that of H-USY, which indicates higher acti- to the OH groups of silica [17], these sites are strong enough not
vation energy and lower strength of the site. Hence, they are of a to desorb ethanol upon heating at 200 °C. Above 200 °C, LAS are
different nature (Table 5). Such ASAs may, however, catalyze the formed, which enables the formation of ethoxy species. These spe-
isomerization of m-xylene to its products. This proves that sites cies are further transformed into ethylene upon heating at ca.
other than zeolitic BAS and weaker than zeolitic BAS may catalyze 270 °C. In a recent study, Phung et al. claim that the formation of
the isomerization of m-xylene, which contradicts former results ethoxy species on alumina occurs whatever the hydroxylation
[4]. state of the surface [19]. However, the more silicon atoms are
Based on computational data, sites that are specific to ASAs grafted on the surface of alumina, the smaller the effect of the pre-
have been proposed [6]. The temperature range, in which these treatment temperature on the ethanol dehydration temperature.
pseudo-bridging silanols (PBS) can be obtained is narrow, as a cer- This confirms that the active Lewis acid sites of the bare alumina
tain hydration state of the surface is required. In the present study, surface are responsible for the lower ethanol dehydration temper-
CO adsorption on Si/Al2O3 after activation at 250 °C or 450 °C ature after pretreatment at 500 °C. On Al/SiO2 materials, this effect
shows that the number of both strong LAS and silanols increases of the pretreatment temperature is limited (between 1 °C and
and BAS are less visible when the activation temperature increases 7 °C).
(Fig. 7b). Based on our model of BAS for Si/Al2O3 [17], it means that
the interaction between the aluminum species and the silicon spe- 4.3. Grafted ASAs as a model for the variety of commercial ASAs
cies, which can be assimilated to a PBS structure, is weaker at high
temperatures, leaving the two species isolated and causing the Our ASA(comm.) is prepared by impregnation of alumina,
increase in the strong Lewis acidity and the decrease in the obtained by hydrolysis of an aluminum alkoxide, with orthosilicic
Brønsted acidity. On Al/SiO2, the number of both LAS and BAS acid [26]. The results obtained for this sample (percentage of free
56 M. Caillot et al. / Journal of Catalysis 316 (2014) 47–56

Table 6
Percentage of free (1 0 0) and (1 1 0) surfaces on Si/Al2O3 and ASA(comm.).

Entry Sample Part of free (1 0 0) Part of free (1 1 0)

surface (%) surface (%)
(3) SA(7/anh) 27 81
(4) SA(12/anh) 14 68
(5) SA(17/anh) 0 36
(10) ASA(comm.) 14 65

alumina surface, position of the ethanol dehydration peak, number

of strong and weak/medium LAS, and isosurface rate of conversion
of m-xylene) are systematically close to those of SA(12/anh), Fig. S1. Derivative of TPD weight loss of various zeolites after pretreatment at
despite dissimilar synthesis techniques. Thus, the deposition of sil- 200 °C and 500 °C. EFAL: extra-framework aluminium species.
icon species must proceed in a similar way that leads to the step-
wise grafting of the c-alumina facets [17]. Assuming that
ASA(comm.) has the same BAS as Si/Al2O3 materials (Table 5),
hence the same turnover frequency for m-xylene isomerization Table S1
Number of Brønsted acid sites (BAS).
(TOF), the distribution of species on the surfaces can be calculated
[17] (Table 6). The results confirm the partial coverage of the facets Zeolite BAS
of ASA(comm.) in a similar way as SA(12/anh). This conclusion con- Isosurface (nm2) Per Al atom (Al atom1)
firms and provides further clarification on the deposition process H-ZSM-5 0.75 0.51
of silicon species in Siral materials as described by Daniell et al. H-USY CBV 760 0.11 0.24
[10]. These materials do not possess bridged zeolitic BAS. H-USY CBV 720 0.18 0.21
NH4-USY CBV 712 0.45 0.20

5. Conclusion

Zeolitic-like Brønsted acid sites are found on an ASA prepared References

by cogelation (20% SiO2), despite a heterogeneous core–shell struc-
ture exhibiting essentially silicon surface species. Grafted ASAs [1] E.J.M. Hensen, D.G. Poduval, P.C.C.M. Magusin, A.E. Coumans, J.A.R. van Veen, J.
Catal. 269 (2010) 201.
possess specific Brønsted acid sites that differ from those of zeo- [2] D.G. Poduval, J.A.R. van Veer, M.S. Rigutto, E.J.M. Hensen, Chem. Commun. 46
lites, with high sensitivity to temperature. The grafting approach (2010) 3466.
of Si/Al2O3 materials that we previously described, including step- [3] B. Xu, C. Sievers, J.A. Lercher, J.A.R. van Veen, P. Giltay, R. Prins, J.A. van
Bokhoven, J. Phys. Chem. C 111 (2007) 12075.
wise grafting of silicon species on the facets of c-alumina, is con- [4] E.J.M. Hensen, D.G. Poduval, V. Degirmenci, D.A.J. Michel Ligthart, W. Chen, F.
vincingly extended to a commercial ASA reference (Siralox). We Maugé, M.S. Rigutto, J.A.R. van Veen, J. Phys. Chem. C 116 (2012) 21416.
confirm the relevance of ethanol as a probe of the Brønsted acid [5] M. Trombetta, G. Busca, S. Rossini, V. Piccoli, U. Cornaro, A. Guercio, R. Catani,
R.J. Willey, J. Catal. 179 (1998) 581.
sites of ASAs and zeolites and OH and Lewis acid sites of c-alumina. [6] C. Chizallet, P. Raybaud, Angew. Chem. Int. Ed. 48 (2009) 2891.
[7] A. Chaumonnot, in: H. Toulhoat, P. Raybaud (Eds.), Catalysis by Transition
Acknowledgments Metal Sulfides: From Molecular Theory to Industrial Application, Editions
Technip, Paris, 2013.
[8] M. Caillot, A. Chaumonnot, M. Digne, C. Poleunis, D.P. Debecker, J.A. van
The authors thank Anne-Agathe Quoineaud, Emmanuel Soyer, Bokhoven, Microporous Mesoporous Mater. 185 (2014) 179.
and Laurent Lemaître from IFP Energies nouvelles for IR-CO mea- [9] O. Cairon, T. Chevreau, J.C. Lavalley, J. Chem. Soc. Faraday Trans. 94 (1998)
3039.
surements. Amaia Beloqui Redondo, Matthew A. Brown, and Daniel
[10] W. Daniell, U. Schubert, R. Glockler, A. Meyer, K. Noweck, H. Knozinger, Appl.
Fodor were very helpful in setting the equipment for IR observa- Catal. A – Gen. 196 (2000) 247.
tions of adsorbed ethanol. We acknowledge financial support from [11] G. Crepeau, V. Montouillout, A. Vimont, L. Mariey, T. Cseri, F. Mauge, J. Phys.
IFP Energies nouvelles. Chem. B 110 (2006) 15172.
[12] O. Cairon, Phys. Chem. Chem. Phys. 12 (2010) 6333.
[13] S. Golay, R. Doepper, A. Renken, Appl. Catal. A – Gen. 172 (1998) 97.
Appendix A. Ethanol adsorption and dehydration followed by [14] R. Schenkel, A. Jentys, S.F. Parker, J.A. Lercher, J. Phys. Chem. B 108 (2004)
15013.
thermogravimetry [15] J.N. Kondo, K. Ito, E. Yoda, F. Wakabayashi, K. Domen, J. Phys. Chem. B 109
(2005) 10969.
Fig. S1 gives the derivative of the weight loss during TPD on var- [16] R. Barthos, A. Szechenyi, F. Solymosi, J. Phys. Chem. B 110 (2006) 21816.
[17] M. Caillot, A. Chaumonnot, M. Digne, J.A. van Bokhoven, ChemCatChem 5
ious zeolites. The values of the derivatives are normalized by the
(2013) 3644.
surface area of each sample. The first peak at around 80°C corre- [18] M. Caillot, A. Chaumonnot, M. Digne, J.A. van Bokhoven, ChemCatChem 6
sponds to the desorption of ethanol. The second peak at around (2014) 832.
210°C corresponds to ethylene stemming from the dehydration [19] T.H. Phung, A. Lagazzo, M.Á.R. Crespo, V.S. Escribano, G. Busca, J. Catal. 311
(2014) 102.
of ethanol. The nature of the zeolite (H-ZSM-5 or USY) and the alu- [20] J.H. Kwak, D.H. Mei, C.H.F. Peden, R. Rousseau, J. Szanyi, Catal. Lett. 141 (2011)
minium content impact the amount of ethanol dehydrated, but the 649.
dehydration temperature is the same for all samples and activation [21] H. Knozinger, B. Stubner, J. Phys. Chem. 82 (1978) 1526.
[22] M. Niwa, K. Suzuki, K. Isamoto, N. Katada, J. Phys. Chem. B 110 (2006) 264.
temperatures. Table S1 gives the number of Brønsted acid sites [23] A. Zecchina, E.E. Platero, C.O. Arean, J. Catal. 107 (1987) 244.
(BAS), quantified by the amount of ethanol dehydrated. One every [24] C.B. Phillips, R. Datta, Ind. Eng. Chem. Res. 36 (1997) 4466.
two aluminium atoms of H-ZSM-5 forms a BAS and one every four [25] S. Roy, G. Mpourmpakis, D.Y. Hong, D.G. Vlachos, A. Bhan, R.J. Gorte, ACS Catal.
2 (2012) 1846.
or five aluminium atoms of USY forms a BAS. CBV 712, which pos- [26] Production of Fischer–Tropsch synthesis produced wax. [US20040186188],
sesses extra framework aluminium species (EFAL), shows the 2004.
lower ratio of BAS per aluminium atom.