Isomerization of α-pinene over dealuminated ferrierite-type zeolites

Article abstract of DOI:10.1016/j.jcat.2007.10.001

Isomerization of α-pinene was performed on a series of dealuminated ferrierite (FER)-type zeolites in liquid phase at 363 K using a batch reactor. The course of zeolite dealumination was followed in detail using ²⁹Si, ²⁷Al, ¹H MAS NMR, XRD, FTIR, and sorption of nitrogen. The ammonium form of FER was dealuminated with aqueous solutions of HCl. While retaining the crystallinity of the zeolite particles, the treatments removed up to 53% of the tetrahedrally coordinated aluminum atoms from the FER framework. According to ²⁹Si MAS NMR studies, the framework aluminum atoms located at the 10-membered rings in the main channels of FER (T_B sites) were depleted preferentially from their positions. Even relatively mild dealumination of FER led to an active catalyst containing both Bronsted and Lewis centers, yielding up to 97% conversion of α-pinene at 363 K, in contrast to the 72% observed for the parent hydrogen form. Such catalytic behavior was discussed in terms of the conversion of a reactant inside micropores of the zeolite catalyst, on Bronsted acid centers with enhanced strength located probably in the vicinity of Lewis sites. The selectivity toward camphene and limonene changed smoothly with the dealumination level; thus, a higher selectivity toward limonene was observed at the expense of camphene formation with increasing the n_Si / n_Al ratio of the catalysts. The selectivity toward camphene and limonene was close to 85% for all of the materials studied. The initial rates of α-pinene transformations over FER-type materials exceeded those observed for other catalytic systems, heteropoly acid/SiO₂ and H₂SO₄/ZrO₂. This study demonstrates the successful application of a medium-pore zeolite for the catalytic transformation of α-pinene in liquid phase.

Full text of DOI:10.1016/j.jcat.2007.10.001

Journal of Catalysis 252 (2007) 161–170
www.elsevier.com/locate/jcat
Isomerization of α-pinene over dealuminated ferrierite-type zeolites
a
b
c
c
c
Rafal Rachwalik , Zbigniew Olejniczak , Jian Jiao , Jun Huang , Michael Hunger ,
a,∗
Bogdan Sulikowski
a
Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences, Kraków, Poland
b
Institute of Nuclear Physics, Polish Academy of Sciences, Kraków, Poland
c
Institute of Chemical Technology, University of Stuttgart, Stuttgart, Germany
Received 30 May 2007; revised 26 September 2007; accepted 2 October 2007
Abstract
Isomerization of α-pinene was performed on a series of dealuminated ferrierite (FER)-type zeolites in liquid phase at 363 K using a batch
29
27
1
reactor. The course of zeolite dealumination was followed in detail using Si, Al, H MAS NMR, XRD, FTIR, and sorption of nitrogen.
The ammonium form of FER was dealuminated with aqueous solutions of HCl. While retaining the crystallinity of the zeolite particles, the
29
treatments removed up to 53% of the tetrahedrally coordinated aluminum atoms from the FER framework. According to Si MAS NMR studies,
the framework aluminum atoms located at the 10-membered rings in the main channels of FER (T sites) were depleted preferentially from their
B
positions. Even relatively mild dealumination of FER led to an active catalyst containing both Brønsted and Lewis centers, yielding up to 97%
conversion of α-pinene at 363 K, in contrast to the 72% observed for the parent hydrogen form. Such catalytic behavior was discussed in terms
of the conversion of a reactant inside micropores of the zeolite catalyst, on Brønsted acid centers with enhanced strength located probably in the
vicinity of Lewis sites. The selectivity toward camphene and limonene changed smoothly with the dealumination level; thus, a higher selectivity
toward limonene was observed at the expense of camphene formation with increasing the n /n ratio of the catalysts. The selectivity toward
Si Al
camphene and limonene was close to 85% for all of the materials studied. The initial rates of α-pinene transformations over FER-type materials
exceeded those observed for other catalytic systems, heteropoly acid/SiO and H SO /ZrO . This study demonstrates the successful application
2
2
4
2
of a medium-pore zeolite for the catalytic transformation of α-pinene in liquid phase.
2007 Elsevier Inc. All rights reserved.
©
1
27
29
Keywords: Ferrierite; Dealumination; Isomerization of α-pinene; H, Al and Si MAS NMR; FTIR
1
. Introduction
Various acidic catalysts, such as oxides treated with acids
6,7], modified zeolites [4,8–10], activated clays [11,12], and
[
Renewable resources are becoming increasingly important
heteropoly acids [13–15], have been suggested in the literature
for the isomerization of α-pinene (Scheme 1). Basic catalysts
have been reported for isomerization of α- to β-pinene (alka-
line earth-metal oxides [16] and calcium amide [17]). Currently,
the main method for camphene production is based on the iso-
merization of α-pinene in the presence of the weakly acidic
for producing a variety of fine and bulk chemicals. Terpene
feedstocks are some of these raw materials. Of these, the alfa
and beta isomers of pinene are used to manufacture a number
of products for the pharmaceutical and chemical industries [1].
The isomerization of α-pinene, catalyzed by acid centers, is
of industrial significance. It gives bicyclic products, exempli-
fied by camphene, and monocyclic products such as limonene
and p-cymene (Scheme 1) [2–4]. Camphene is an intermedi-
ate compound for producing isoborneol, isobornyl acetate, and
camphor [5].
hydrated TiO catalyst [18].
2
In the present work, a medium-pore zeolite ferrierite (FER)
was chosen as the catalyst. FER, with slightly smaller pore
openings than ZSM-5 (MFI), is seldom found in nature. It is
produced on an industrial scale. FER is quite distinct from
most known zeolites as the four-membered rings are not
*
1 10
present in its structure. Its topological density, N –N , is
Corresponding author. Fax: +48 12 4251923.
E-mail address: ncsuliko@cyf-kr.edu.pl (B. Sulikowski).
high (1021) compared with other zeolites [19]. Its structure
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.10.001

162
R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
Scheme 1. Products given by isomerization of α-pinene over dealuminated ferrierite-type zeolites.
is stable against thermal, hydrothermal, and chemical treat-
ments.
Silicon powder was used as the internal standard in a quantity
of 15 wt% for calibrating the diffraction angle and estimating
the crystallinity of the parent and dealuminated materials.
The objectives of the present work were to provide a detailed
29
account of FER dealumination by hydrochloric acid, using Si,
27
1
Al and H MAS NMR in tandem with FTIR, and to study
2.3. Sorption properties
isomerization of α-pinene catalyzed by the materials obtained.
The results allowed a comparison of the performance of FER
samples in relation to a heteropoly acid and sulfated zirconium
oxide.
Sorption of nitrogen was studied using the Quantachrome
Autosorb automated gas sorption system. Before the measure-
ments, the samples were outgassed at 473 K for 24 h. The
external surface area (SE), internal surface area (Sµ), total pore
volume (VT), and micropore volume (Vµ) were determined by
applying the t-method micropore analysis according to de Boer.
SiO2 was used as the reference material.
2
. Experimental
2
.1. Preparation of the samples
Adsorption of α-pinene was studied using potassium forms
of the starting sample and the most dealuminated one. Four-fold
ion exchange with 0.5 M KNO3 solution was applied at 353 K
for 2 h each to yield the samples labeled K-T0 and K-T3, re-
spectively. The samples were loaded into the two GC columns
and outgassed at 723 K for 1 h before the measurements. Ad-
sorption of α-pinene was studied using a gas chromatographic
method at 453 K, because the boiling point of α-pinene is
The ammonium form of labeled NH4-T was dealumi-
nated with aqueous solutions of hydrochloric acid applying
a multistep procedure. First, NH4-T was obtained by an ion-
exchange of a zeolite, K,Na-FER, purchased from Tosoh, Japan
(
nSi/nAl = 9.2). The ion-exchange process was repeated four
times using a 0.5 M aqueous solution of ammonium nitrate at
room temperature for 12 h. The NH4-exchanged sample was
washed and dried at 373 K overnight. The material thus ob-
tained was calcined at 773 K for 5 h to yield the hydrogen
form H-FER (T0), used as a reference catalyst. The parent ze-
olite NH4-T was dealuminated by a multistep procedure. First,
a 100-g portion of zeolite NH4-T was treated with 0.25 M HCl
4
29 K.
2
.4. MAS NMR measurements
29Si MAS NMR spectra were recorded on a homemade
NMR spectrometer at the magnetic field of 7.05 T using single-
−
solution at 368 K for 4 h, washed until no Cl ions could be
detected in the solution, and dried at 373 K overnight (step 1).
The resultant sample, T1, was dealuminated further applying
the same procedure, but with the HCl concentration in solution
progressively increased in consecutive steps, that is, 0.25 M
pulse excitation of 4 µs (π/2), a repetition time of 10 s, and a
1
27
sample spinning rate of 4 kHz. H and Al MAS NMR spec-
tra were measured using a Bruker MSL-400 spectrometer at a
magnetic field of 9.4 T, with single-pulse excitations of 2.2 µs
(
step 1) to 1.0 M (step 2) and, finally, to 2.0 M HCl (step 3).
(
π/2) and 0.6 µs (π/2), and repetition times of 30 and 0.5 s, re-
The treatments with acid in the second and third steps were
carried out at ambient temperature to yield samples labeled T2
and T3.
spectively; 160 and 1024 scans, respectively, were acquired at a
spinning rate of ca. 8.0 kHz. Decomposition and simulation of
the NMR spectra were performed using the Bruker WINNMR
and WINFIT software.
2
.2. XRD measurements
27Al and 29Si MAS NMR investigations were performed
on fully hydrated samples. For this purpose, the samples were
exposed to the vapor of a saturated Ca(NO3)2 solution at ambi-
ent temperature for 12 h. The 27Al MAS NMR spectra were
Powder X-ray diffraction (XRD) patterns of the hydrated
◦
samples for 2θ ranging from 5 to 50 were recorded on a
Siemens D5005 automatic diffractometer with CuKα radiation.

R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
163
normalized taking into account the acquisition numbers and
Table 1
Sorption properties of the reference sample T0 (H-FER) in comparison with the
dealuminated materials (T1–T3) as determined by nitrogen adsorption
the sample mass. For the quantitative evaluation, the signal in-
tensities of the tetrahedrally coordinated framework aluminum
atoms occurring at ca. 60 ppm were compared with that of the
well-defined parent material T0.
a
b
2
Sµc
d
Vµe
f
Sample
SSA
(
SE
VT
Vm
2
2
3
3
3
m /g) (m /g) (m /g) (cm /g) (cm /g) (cm /g)
1
Before the H MAS NMR experiments, the samples were
T0 (H-FER) 275
40
39
41
45
235
261
258
233
0.243
0.300
0.277
0.358
0.129
0.134
0.133
0.120
0.114
0.166
0.144
0.238
−
6
T1
T2
T3
300
299
278
dehydrated at 723 K for 12 h in a vacuum of p < 10 Pa,
and then filled into 4-mm MAS rotors in a glove box purged
1
with dry nitrogen. Quantitative H MAS NMR studies were
a
b
c
SSA—specific surface area.
SE—external surface area.
Sµ—internal surface area.
performed using normalized spectra and comparing the sig-
nal intensities with that of an external intensity standard. The
standard sample was a dehydrated zeolite H,Na-Y (exchange
degree, 35%) with 1.776 mmol OH groups/g and a weight of
d
e
VT—total pores volume.
Vµ—micropores volume.
5
8.5 mg.
f
Vm—mesopores volume (comprises mesopores volume + intercrystalline
space between ferrierite crystals).
2
.5. FTIR spectroscopy
acid centers are governed strictly by the number of framework
aluminum atoms [21]. Numerous dealumination methods have
been developed to vary the composition of the zeolite frame-
work at will [22]. One of these methods is based on the use of
mineral acids. When considering such a treatment of any alu-
minosilicate, the major factor is the framework’s resistance to
acid. Some zeolites are not stable even against mild acids; thus,
our first concern was the stability of the zeolite framework un-
der treatment with hydrochloric acid of different concentration.
Detailed inspection of the XRD diffraction patterns of the FER-
type samples after dealumination revealed neither collapse of
the zeolite framework nor any significant change in the integrity
of the zeolite crystals. Only small and anisotropic unit cell con-
tractions did occur during this treatment, producing pronounced
shifts of the reflections characteristic of atoms arranged along
the a-axis of the unit cell. This is in accordance with our earlier
study of this type of zeolite, albeit of a different origin (Ze-
olyst) [23].
FTIR measurements (for adsorption of both ammonia and
CO) were performed on self-supported discs of zeolites (about
2
1
0 mg/cm ) on a Bruker 48 PC spectrometer equipped with an
−1
MCT detector and with a spectral resolution of 2 cm . In the
most cases, 100–200 scans were recorded. FER samples were
activated in situ in the IR cell at 803 K for 2 h under a vacuum
−
6
of p < 10 Pa. Concentration of Brønsted acid sites was deter-
−
1
mined by adsorption of ammonia, using a 1450-cm band of
ammonium ions and an extinction coefficient of 14.8 cm/µmol.
The strength of acid sites was estimated by NH3 thermod-
esorption and CO adsorption. Thus the ratios of the integral
intensity of the IR band of chemisorbed ammonium ions at
−1
1
(
450 cm for the samples calcined at 650 K (A650) and 453 K
A0) were evaluated. The adsorption of CO was performed and
studied at the temperature of liquid nitrogen.
FTIR spectra were normalized, taking into account specific
density of a self-supported wafer.
Sorption studies of nitrogen on the parent and dealuminated
samples are summarized in Table 1. The specific surface areas
were essentially the same for all of the samples under study.
The micropore volume was largely retained even after the third
2
.6. Catalytic studies
The conversion of α-pinene was carried out at atmospheric
3
pressure in a batch-type glass reactor equipped with a reflux
condenser, a stirrer, and a temperature controller. In a typical
run, 20 mL of α-pinene (Fluka) was heated to 363 K. The cat-
alyst (1.0 g) was calcined at 673 K for 5 h in a helium flow,
cooled in He to ambient temperature, and then added to the
reaction mixture. The reaction was carried out under isother-
mal conditions at 363 K. After selected times, 10-µL aliquots of
the reaction mixture were obtained for GC analysis. The reac-
tion products were analysed by an HP 6890 gas chromatograph
equipped with a thermal conductivity detector, a homemade
packed column (3 m, 2.0 mm i.d., with 8% bentone-34, 6%
didecyl phthalate, and 1% silicon oil A on Chromosorb W) [20].
dealumination step, that is, 0.129 and 0.120 cm /g for sam-
ples T0 and T3, respectively. In addition, pores in the mesopore
range (Table 1, last column) started to develop slightly after
mild treatment with HCl; however, this increase was not sig-
nificant for the T1 and T2 samples. Only after using 2 M HCl
did the sample contain twice the amount of mesopores as the
parent sample T0. The acid solution removes some aluminum
atoms from framework position and washes them out of the
pore system, leading to the formation of secondary pores in the
mesopore range (vide infra). However, the average crystals size
is not affected by dealumination with HCl (SEM, not shown),
thus supporting earlier findings [24].
To acquire information on α-pinene sorption on FER, ad-
sorption of the reactant was carried out by gas chromatography.
Because the isomerization of sorbate proceeds at 363 K, the
hydrogen forms of the samples could not be used. Instead, the
potassium-exchanged forms of the starting material and of the
most dealuminated material were prepared, thereby avoiding
catalytic transformations of α-pinene during the experiment.
3
. Results and discussion
3
.1. Dealumination of FER
The nSi/nAl ratio in zeolites is of paramount importance
for catalysis, because the number and strength of the Brønsted

164
R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
Sorption was studied at 453 K. (The boiling point of α-pinene
is 429 K.) The amount of α-pinene adsorbed on the potassium
form K-T0 was 88 mg/g; this increased to 109 mg/g on the
most dealuminated sample, K-T3. Note the large difference in
+
+
diameter between H and K ions; sorption on the hydrogen
forms in fact would be higher.
29Si and 27Al MAS NMR spectroscopy was used to study the
location and chemical surroundings of silicon and aluminum
atoms in the parent and dealuminated zeolites. The exact distri-
bution of silicon and aluminum in the framework gives rise to
physical and chemical properties of the solid, with sorptive and
catalytic properties likely to be the most important for applica-
tions. For example, the siting of Brønsted acid centers, required
for the overwhelming majority of reactions catalyzed by zeo-
lites, is inherently connected with the aluminum distribution at
29
the T sites. Whereas Si MAS NMR spectroscopy gives the
distribution of tetrahedrally coordinated silicon atoms with dif-
ferent numbers of aluminum atoms or hydroxyl groups in their
vicinity, quantitative estimation of the distribution of aluminum
between framework and non-framework positions is a major
concern in the 27Al NMR spectroscopy in zeolites.
Estimation of Al location at the framework/non-framework
zeolite positions is by no means straightforward, because Al
is a quadrupolar nuclei with spin of 5/2. In faujasite series,
for example, the values for the non-framework aluminum were
consistently lower than those obtained from the silicon spectra,
and the presence of so called “invisible aluminum” in zeolites
has been reported [25,26]. The amounts of framework and non-
framework aluminum can be estimated correctly provided that
certain experimental conditions are met [27]. First, the sam-
ples must be fully hydrated. Second, the radiofrequency pulses
must be short; otherwise, a considerable loss of spectral inten-
sity leads to serious underestimation of the aluminum in the
samples. Al in a highly distorted coordination might become
NMR-silent. Thus, quantitative measurements were performed
with fully hydrated samples, and only the signals of the well-
defined tetrahedrally coordinated Al species were evaluated.
Fig. 1 shows the Al MAS NMR spectra of all the sam-
ples, in which signals at ca. 54 ppm are attributed to tetrahe-
drally coordinated framework aluminum atoms. In the spectra
of the dealuminated zeolites, the signal at 0 ppm indicates the
presence of a small number of octahedrally coordinated extra-
framework aluminum species, possibly acting as Lewis acid
sites. For the reference sample T0 (H-FER), a strong signal at
Fig. 1. Monocyclic, bicyclic and tricyclic products given by isomerization of
α-pinene.
Table 2
A comparison of the n /n ratios and numbers of aluminum atoms, N
determined by elemental analysis and Si and Al MAS NMR spectroscopy,
for the parent T0 (H-FER) and dealuminated samples T1–T3
,
Al
Si Al
29
27
Sample
n /n
Si Al
N (Al/uc)
Dealumination (%)
a
b
a
b
29
b
27
c
Bulk
NMR Bulk
NMR
Si NMR
Al NMR
T0 (H-FER) 8.9
9.1
12.6
18.2
20.5
3.6
2.7
2.0
1.8
3.6
2.6
1.9
1.7
0
0
T1
T2
T3
12.5
17.1
19.3
38
47
53
30
43
53
a
b
c
Determined by elemental analysis.
29
Determined by Si MAS NMR spectroscopy.
27
Calculated from Al MAS NMR signal loss.
27
tallographically nonequivalent positions TA and TB. Takaishi
et al. [28] proposed a model of the distribution of aluminum
atoms (6 Al per unit cell) in the FER framework depicted in
Scheme 2. According to this model, the framework aluminum
atoms are located at the crystallographic positions labeled T1,1
,
T1,10, T2,1, T2,2, T2,4, and T2,9. There are 3.6 Al/uc in the refer-
ence sample T0 (Table 2, columns 4 and 5). This situation was
represented in Scheme 2 by considering 4 Al/uc. The silicon
atoms at the aforementioned positions TA and TB are located
in the neighborhood of framework aluminum atoms located at
T1,1, T1,10, T2,2, and T2,11 or T1,1, T1,10, T2,4, and T2,9 posi-
tions, respectively (Schemes 2a and 2b).
5
4 ppm and a very weak signal at 0 ppm indicate that nearly
all the aluminum atoms are tetrahedrally coordinated in the
framework. The intensity of this signal decreases stepwise on
treatment with hydrochloric acid solutions; the dealumination
levels calculated from 27Al NMR spectra are given in Table 2,
last column. For more strongly dealuminated materials, such as
samples T2 and T3, the signal of tetrahedrally coordinated alu-
minum atoms is broadened (high-field shoulder). This indicates
that some of the framework aluminum atoms in the dealumi-
nated materials exhibit lower symmetry, thus giving rise to a
distorted local structure.
The 29Si MAS NMR spectrum of the ammonium form
NH4-T consists of four signals at −105.0, −109.0, −115.5,
and −112.0 ppm due to Si(1Al) TA, Si(1Al) TB, Si(0Al) TB,
and Si(0Al) TA species, respectively (Fig. 2a). An additional
2
9
3
Si MAS NMR signal at −103 ppm is attributed to Q silicon
In the case of FER-type zeolites, the 29Si MAS NMR spectra
are a superposition of signals due to silicon atoms at the crys-
atoms (SiOH) at framework defects or the outer surface of the
zeolite particles [29].

R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
165
Scheme 2. Model of the distribution of Al atoms (4 Al per unit cell) in the FER framework, following the Takaishi et al. paper [28].
The 29Si MAS NMR spectra of the reference sample T0
and the dealuminated FER materials were deconvoluted using
the aforementioned line positions. Analysis of the deconvoluted
spectra indicates that framework aluminum atoms located in the
vicinity of silicon atoms at TB positions (Figs. 2b–2e) are pref-
erentially depleted during the dealumination process. These TB
positions contribute to the 10-membered oxygen rings of the
zeolite main channels. Thus it can be concluded that the alu-
minum atoms located in the vicinity of Si(1Al) TA sites are
much more resistant to dealumination. Such a behavior is prob-
ably because of their hidden location exclusively in the five-
and six-membered oxygen rings of the zeolite structure and thus
decreased accessibility for attack by hydrochloric acid, and/or
higher stability of such sites in the zeolite framework.
Table 3
Concentration and strength of Brønsted acid sites in parent T0 (H-FER) and
dealuminated ferrierite samples (T1 and T3)
Sample
ν
Concentration of Brønsted
acid sites (mmol/g)
Strength of
acid sites
OH
−
1
b
(cm
)
1
b
c
H MAS FT IR
Average
A /A ꢀν
650 0 OH
a
−1
)
NMR
(cm
T0 (H-FER) 3603
T1
T3
1.389
0.986
0.380
1.182
0.845
0.260
1.29
0.92
0.32
0.29
0.45
0.51
286
291
299
3603
3604
a
b
c
1
Determined by H MAS NMR spectroscopy.
Using ammonia and extinction coefficient 14.8 cm/µmol [34].
Shift observed after CO adsorption.
to hydroxyl groups bound to the extra-framework aluminum
species (AlOH) [32]. Such species give rise to Lewis acidity.
An additional signal at ca. 6 ppm indicated the occurrence of
SiOH groups, which are hydrogen-bonded to neighboring oxy-
In Table 2, columns 2 and 3, the framework nSi/nAl ra-
tios of the materials under study as determined by 29Si MAS
NMR spectroscopy are compared with the results of the ele-
mental analysis. As evidenced by the quantitative evaluation of
1
gen atoms [33]. The H MAS NMR signal at 2.8 ppm indicated
29
the Si MAS NMR spectra (Fig. 2 and Table 2, last column),
up to 53% of the initial aluminum atoms were removed from
the FER framework by the treatment with hydrochloric acid
solutions. Dealumination of the FER crystals corresponds to a
change in the nSi/nAl framework ratio from the initial value of
that some of the aluminum atoms, depleted from the zeolite
framework positions, remained in the channels; that is, they
were not extracted from zeolite particles by the aqueous so-
lution of hydrochloric acid. This observation was confirmed
27
by Al MAS NMR, where the spectra of the dealuminated
9
.1 to the final value of 20.5 after the final dealumination step.
samples T1–T3 exhibited a signal at ca. 0 ppm attributed to
extra-framework aluminum species (Fig. 1).
The intensities of the H MAS NMR signals at 4.2 ppm
were used to calculate the concentration of SiOHAl groups
This is a substantially higher increase in the framework nSi/nAl
ratio compared with those achieved in earlier attempts based on
the zeolite treatment with, for example, ammonium hexafluo-
rosilicate [30].
1
(
exhibiting Brønsted acidity) in the FER samples. The results
1H MAS NMR investigations were performed to gain quan-
titative insight into the distribution and amount of the OH
groupings. The spectrum of the reference sample T0 consists
of two lines with chemical shifts of 1.8 and 4.2 ppm, which
were assigned to the isolated silanol groups and the bridging
OH groups (SiOHAl) acting as Brønsted acid sites in the FER
framework, respectively (Fig. 3) [31]. After dealumination of
T0, a new signal appeared at 2.8 ppm, the intensity of which
increased further on treatment with HCl. This signal is due
are listed in Table 3, column 3. The results can be compared
with expected amount of acid sites for the T0 sample, equal
to 1.67 mmol H /g. Interestingly, the concentration of Brøn-
+
sted acid centers in the dealuminated FER samples is signifi-
cantly lower than that of the tetrahedrally coordinated frame-
work aluminum atoms (compare Table 3, column 3, and Ta-
ble 2, columns 4 and 5). This points to a compensation of
some of the negatively charged framework aluminum tetrahe-
−
dra [AlO4] by cationic extra-framework aluminum species.

166
R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
29
Fig. 2. Si MAS NMR spectra of the parent zeolite NH -T (a), the reference catalyst T0 (b), and the dealuminated samples T1–T3 (c–e, respectively).
4
1
Fig. 3. H MAS NMR spectra of dehydrated zeolites T0, T1, and T3.

R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
167
−
neously, a significant decrease of the main peak at 3604 cm
occurs. Peixoto et al. [35] showed that the band at 3650 cm
1
As expected, the average concentration of Brønsted acid sites
decreases on dealumination, from 1.29 to 0.32 mmol H /g.
+
−1
FTIR spectroscopy was applied as a complementary tool for
characterizing the hydroxyl coverage of the FER zeolites under
study. The spectrum of the dehydrated zeolite T0 (H-FER) con-
in the FTIR spectra of FER zeolites could be related to hy-
droxyl groups bound to extra-framework aluminum species. In
27
present work, this explanation is supported by the Al MAS
NMR signals of the octahedrally coordinated extra-framework
aluminum species in the samples T1–T3 (Fig. 1). In addition,
−
1
sists of two distinct bands at 3748 and 3603 cm , a very small
−
1
−1
band at 3650 cm , and a shoulder at 3550 cm (Fig. 4). The
1
−
1
the H MAS NMR spectra of dealuminated samples (Fig. 3) re-
band at 3748 cm corresponds to the terminal SiOH. Accord-
−
1
vealed a signal at 2.8 ppm, assigned to the AlOH groups formed
at extra-framework aluminum species.
ing to Datka et al. [34], the bands near 3600 and 3550 cm are
due to bridging SiOHAl groups located inside 10-membered
ring channels and inside FER cages, respectively. The position
of the main OH band essentially does not change on dealumi-
nation (Table 3).
Finally, the strength of the SiOHAl acid sites was estimated
by thermodesorption of ammonia and shifts of the OH stretch-
ing vibrations on adsorption of CO. Both methods indicated an
increasing strength of the Brønsted acid sites on dealumination
with HCl (Table 3), similar to that seen for other zeolites [22].
The additional band at 3650 cm−1, barely seen in T0, in-
creased after treatment with HCl (Fig. 4, sample T3). Simulta-
3
.2. Isomerization of α-pinene
Isomerization of α-pinene can proceed by two parallel
routes, one giving rise to the formation of monocyclic prod-
ucts exemplified by limonene, terpinenes, terpinolene, and
p-cymene, and the other yielding bicyclic and tricyclic prod-
ucts, of which camphene is the most important (Scheme 3).
The first step in the reaction involves protonation of the dou-
ble bond giving the pinanyl cation. The second involves a
Wagner–Meerwein-type rearrangement leading to isobornyl
and p-menthenyl cations. Further rearrangements of isobornyl
cations yield camphene and other bicyclic and tricyclic prod-
ucts. In contrast, p-menthenyl cations are transformed into
limonene, p-cymene, and other monocyclic compounds.
Fig. 5 shows the conversion of α-pinene as a function of time
at 363 K. The main products obtained are camphene, limonene,
Fig. 4. FTIR spectra of dehydrated zeolites T0 and T3.
Scheme 3. Mechanism of the α-pinene isomerization.

168
R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
dealuminated sample T1. As is evident from Table 3, while a
large decrease in the acid centers occurred for T3, their strength
grew only slightly. Advanced dealumination, despite producing
a pronounced system of mesopores in the samples (Table 1),
facilitating transport of the reaction products, is not optimal for
the course of the reaction studied. Clearly, an optimum between
the amount of acid centers and their strength is needed to give
good catalytic performance. Such observations were described
earlier for ZSM-5 [36] and mordenite [37]. In the former case,
an optimum amount of Brønsted and Lewis acid sites, as deter-
1
27
mined by H and Al MAS NMR spectroscopy, is required to
achieve maximum catalytic activity in n-hexane cracking over
dealuminated ZSM-5 [36]. Similarly, ethylbenzene conversion
plotted against the Si/Al ratio in mordenite passes through a
maximum [37]. In conclusion, dealumination of FER is a favor-
able method for preparation of zeolitic catalysts, because both
the number and the strength of acid centers may be affected by
the HCl treatment [22].
Fig. 5. Conversion of α-pinene vs the reaction time for the isomerization reac-
tion on the reference zeolite T0 and the dealuminated catalysts T1–T3.
Table 4
Comparison of the initial rate of α-pinene transformation on ferrierite type
(
T0–T3), 12-tungstophosphoric acid (HPW)/SiO and H SO /ZrO systems
2
2
4
2
It also seems that extra-framework aluminum plays a non-
negligible role in catalysis. These species, directly evidenced
Sample
Acidity
(
Initial reaction rate, r
Reference
0
+
mmol H /g) (mmol α-pinene/(gcat min))
1
27
by H and Al MAS NMR, barely existing in the parent
hydrogen form T0, appear after the first dealumination step
a
T0
T1
T2
T3
1.29
0.92
–
4.28
5.01
2.82
2.31
This work
(
sample T1, Figs. 1 and 3), giving rise to the Lewis-type cen-
0.32
ters. These are known to possibly interact with the Brønsted
acid sites, enhancing their strength. Moreover, dealumination
changes the proportion between the two kinds of acid centers,
and the Lewis centers become significant when dealumina-
tion is advanced (Figs. 3b and 3c). For such samples (i.e., T2
and T3), lower conversion of α-pinene was found. The small
amount of relatively strong acid centers (25% in T3 compared
with T0) might be effectively poisoned by heavier reaction
products, leading to stable conversion after 40–60 min (Fig. 5).
Such site poisoning was considered by López et al. [10].
Moreover, it has been observed in other systems (e.g.,
H2SO4/ZrO2) that if only Lewis acid sites are present, then no
catalytic activity is evidenced, whereas pure Brønsted acidity
leads to low overall activity [14b]. The best catalytic perfor-
mance in the isomerization of α-pinene on H2SO4/ZrO2 was
demonstrated for catalysts having both Brønsted and Lewis acid
centers.
Generally, camphene and limonene are the main reaction
products of the conversion of α-pinene on FER catalysts. The
selectivity toward these compounds was close to 85% for all the
materials studied. The selectivity to camphene for zeolite T0
was 52% (Fig. 6a); simultaneously, the selectivity of limonene
formation was 35% (Fig. 6b). As can be seen from Fig. 6, the
selectivity toward camphene and limonene remained almost the
same with increasing reaction time, in contrast to the results of
Severino et al. [3]. This difference might be attributed to a very
dissimilar structure of the faujasite and FER-type zeolites, as
well as a significant difference in the number of OH groups in
both materials.
b
b
b
HPW/SiO -200 1.63
0.63
0.95
1.42
[13]
2
HPW/SiO -300 2.15
2
HPW/SiO -400 2.85
2
1
5% H SO /ZrO
[14b]
2
4
2
c
c
d
d
SZ
SZ250
a
0.24
0.20
1.47
1.53
Reaction temperature 363 K.
b
c
d
Reaction temperature 333 K, calculated from data in Ref. [13].
Expressed in mg of pyridine desorbed.
Reaction temperature 393 K, calculated from data in Ref. [14b].
α- and γ -terpinene, terpinolene, and p-cymene. As seen, the
conversion of α-pinene for all of the catalysts studied increased
at the beginning and continued to increase with time of the re-
action. The conversion of α-pinene on the hydrogen form of
FER approached 70% after 3 h. The catalytic performance of
the sample T1 (after the first dealumination step) was the best,
resulting in ca. 97% conversion of α-pinene. For the other sam-
ples, the conversion leveled up after 1 h (Fig. 5).
Dealumination of FER affected the initial rate of α-pinene
transformation (Table 4). The first dealumination step seemed
to be the most important for the course of the reaction. During
this step, ca. 1 Al was removed from the unit cell of FER, while
simultaneously the most significant increase in acid strength oc-
curred (i.e., parameter A650/A0 increased from 0.29 to 0.45;
Table 3). The initial reaction rate, r , was highest for the sam-
0
ple T1 after the first dealumination step, and then decreased
significantly on further depletion of aluminum from the FER
framework (Table 4).
Further dealumination of FER is detrimental both for the
initial reaction rate and for the final conversion achieved, not
exceeding ∼35% after 3 h (Fig. 5). Thus, samples T2 and T3
were significantly less active than the reference sample T0 and
The selectivity toward camphene and limonene in the liquid-
phase isomerization of α-pinene was explored as a function of
the number of dealumination treatments. Fig. 7 illustrates selec-
tivity toward the main products obtained after a reaction time of
180 min, corresponding to maximum conversion of α-pinene on

R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
169
Of interest is the question of where the reaction proceeds,
in the internal micropore system or on the external surface
of crystallites/mesopores. We note that zeolites often can ad-
sorb molecules having theoretical dimensions somewhat larger
than the pore size of the molecular sieve, although the average
dimension should not exceed 3 Å [38]. Thus, liquid-phase ad-
sorption of alpha-pinene on ZSM-11 zeolite and its pure-silica
homologue Silicalite II was described, with ZSM-11 having
a pore dimension of 5.3 × 5.4 Å [19]. Similarly, 1,2,4- and
even large 1,3,5-trimethylbenzene can enter the pore system of
ZSM-5 [39]. Critical dimensions of these molecules are 7.4 and
8
.6 Å, whereas crystallographic dimensions of the larger chan-
nel in ZSM-5 are only 5.3 × 5.6 Å [19]. The molecular size of
α-pinene is comparable with that of benzene, with correspond-
ing data of 6.63 × 7.74 Å for benzene and 6.77 × 6.91 Å for
α-pinene [40]. Benzene, toluene, and xylenes are sorbed in FER
pore system [23,24]. Sorption of toluene on FER samples has
been reported to be 23–26 mg/g [24], whereas sorption of α-pi-
nene on samples T0 and T3 was much higher (88–109 mg/g).
Furthermore, there is no relationship between mesopore vol-
ume and the catalytic activity of the samples. The amount and
strength of Brønsted acid centers located inside the zeolite pore
systems, together with the presence of Lewis sites in their vicin-
ity, exert such a strong influence on the catalytic activity that it
is reasonable to assume that the reaction proceeds mainly inside
the micropore system of FER.
(a)
Finally, we compared the performance of FER-type mate-
rials with that of other systems containing acid centers: 12-
tungstophosphoric acid (HPW) supported on SiO2 and H2SO4
supported on ZrO2. We calculated the initial rate of α-pinene
transformations (r0) over FER catalysts, as well as for the other
two systems based on published data [13,14b], and found that
the reaction rate of α-pinene over zeolite catalysts was com-
parable with, and generally higher than those found for the
HPW/SiO2 and H2SO4/ZrO2 systems (Table 4).
(
b)
Fig. 6. Selectivity towards camphene (a) and limonene (b) as a function of re-
action time.
4
. Conclusion
We have developed a procedure for the dealumination of
FER-type zeolite that allows removal of up to 53% of the
framework aluminum atoms from the parent material. The
procedure involves treatment of the ammonium form of FER
(
NH4-T) with aqueous solutions of hydrochloric acid of dif-
ferent concentrations. The dealuminated materials retain their
crystallinity and good sorption properties, as demonstrated by
nitrogen and α-pinene adsorption studies. NMR and IR spec-
troscopy studies enabled us to closely follow the course of
dealumination of FER and, in particular, the distribution of alu-
minum in the samples.
Fig. 7. Selectivity towards camphene and limonene after a reaction time of
180 min vs the number of dealumination steps.
all samples. The selectivity to camphene decreased with each
dealumination step, from ca. 52% for the reference material
T0 to 41% for the FER zeolite obtained after the third dealu-
mination step. Simultaneously, the selectivity toward limonene
increased. Such behavior can be explained by an increased
strength of the Brønsted acid sites in the dealuminated FER
samples, which favors formation of limonene [3]. Tailoring the
reaction parameters (e.g., temperature, α-pinene-to-catalyst ra-
tio) can be expected to produce further improvements in selec-
tivity.
29Si MAS NMR spectroscopic studies indicated that the at-
tack of HCl caused selective removal of aluminum atoms lo-
cated at different crystallographic positions in the FER frame-
work. The framework aluminum atoms were preferentially re-
moved from their positions in the vicinity of silicon atoms
occupying T sites. These aluminum atoms contributed to the
B
10-membered oxygen rings of the main channels. In contrast,
aluminum atoms, bound via oxygen bridges to silicon atoms at

170
R. Rachwalik et al. / Journal of Catalysis 252 (2007) 161–170
the TA positions, were much more resistant to dealumination.
The latter are located exclusively in the five- and six-membered
oxygen rings of the zeolite structure, which are less prone to
attack by hydrochloric acid and possibly more stable.
Even relatively mild dealumination of FER with 0.25 M
HCl yielded the catalyst T1 exhibiting 97% conversion of α-
pinene at 363 K, compared with the parent hydrogen form of
FER, which exhibited only 72% conversion. This sample, con-
taining both Brønsted and Lewis acid centers (with the former
predominating), gave the best catalytic performance. When the
dealumination level was more advanced, catalytically less ac-
tive materials were obtained, due to decreased overall acidity
and the altered ratio between the Brønsted and Lewis acid cen-
ters.
[7] M.A. Ecormier, A.F. Lee, K. Wilson, Microporous Mesoporous Mater. 80
2005) 301.
8] A.I. Allahverdiev, G. Gündüz, D.Y. Murziny, Ind. Eng. Chem. Res. 37
1998) 2373.
9] A. Allahverdiev, S. Irandoust, D.Y. Murzin, J. Catal. 185 (1999) 352.
10] C.M. López, F.J. Machado, K. Rodríguez, D. Arias, B. Méndez, M. Hase-
gawa, Catal. Lett. 62 (1999) 221.
11] D.R. Brown, C.N. Rhodes, Catal. Lett. 45 (1997) 35.
12] C. Breen, R. Watson, J. Madejova, P. Komadel, Z. Klapyta, Langmuir 13
(
[
[
(
[
[
[
(1997) 6473.
[
13] O. Masini, L. Grzona, N. Comelli, E. Ponzi, M. Ponzi, J. Chil. Chem.
Soc. 48 (2003) 101.
[14] (a) N.A. Comelli, L.M. Grzona, O. Masini, E.N. Ponzi, M.I. Ponzi, J. Chil.
Chem. Soc. 49 (2004) 245;
(b) N.A. Comelli, E.N. Ponzi, M.I. Ponzi, Proc. 2nd Mercosur Congr.
Chem. Eng. & 4th Mercosur Congr. Process Sys. Engng., 14–18.08.2005,
Rio de Janeiro, Enpromer 2005.
The selectivity toward camphene and limonene was af-
fected by the dealumination level; a greater selectivity toward
limonene was observed at the expense of camphene formation
[
15] A.D. Newman, A.F. Lee, K. Wilson, N.A. Young, Catal. Lett. 102 (2005)
45.
[16] R. Ohnischi, K. Tanabe, Chem. Lett. 3 (1974) 207.
[17] A.F. Plate, M.W. Mil’vitskaya, Teknologiya 10 (1967) 1340.
[18] M. Gscheidmeier, H. Häberlein, H.H. Häberlein, J.T. Häberlein, M.C.
Häberlein, US Patent 5,826,202 (1998).
with an increasing n /n ratio of the catalysts. This finding
Si Al
can be explained by the increased strength of Brønsted acid
sites on dealumination. The selectivity toward camphene and
limonene remained essentially constant during reaction times
of up to 180 min. The overall selectivity toward camphene +
limonene was close to 85% for all of the FER samples studied.
Finally, we note that the initial rate of α-pinene transforma-
tions over FER-type zeolites in the liquid phase is comparable
with that of other systems, and generally exceeds that found for
[
19] Ch. Baerlocher, W.M. Meier, D.H. Olson, Atlas of Zeolite Frame-
work Types, Elsevier, Amsterdam, 2001, http://topaz.ethz.ch/IZA-SC/
atlas_pdf/.
[20] B. Sulikowski, Czasop. Techn. 195 (1976) 33.
[21] D.W. Breck, Zeolite Molecular Sieves, Wiley, New York, 1974.
[22] B. Sulikowski, Heterogen. Chem. Rev. 3 (1996) 203.
[23] R. Rachwalik, Z. Olejniczak, B. Sulikowski, Catal. Today 101 (2005) 147.
[24] Y.S. Shin, A. Auroux, J.C. Vedrine, Appl. Catal. 37 (1988) 1 (Part I), 21
(Part II).
the HPW/SiO and H SO /ZrO catalysts.
2
2
4
2
[25] J. Klinowski, S. Ramdas, J.M. Thomas, C.A. Fyfe, J.S. Hartman, J. Chem.
Soc. Faraday Trans. II 78 (1982) 1025.
Acknowledgments
[26] P.P. Man, J. Klinowski, A. Trokiner, H. Zanni, P. Papon, Chem. Phys.
Lett. 151 (1988) 143.
[
[
[
[
[
27] P.P. Man, J. Klinowski, J. Chem. Soc. Chem. Commun. (1988) 1291.
Financial support was provided by the European Union un-
der a Marie Curie Action grant TOK-CATA (no. MTKD-CT-
28] T. Takaishi, M. Kato, K. Itabashi, Zeolites 15 (1995) 21.
ˇ
29] P. Sarv, B. Wichterlová, J. Cejka, J. Phys. Chem. B 102 (1998) 1372.
2
004-509832), and by the grant of the Ministry of Science
30] P. Cañizares, A. Carrero, Appl. Catal. A Gen. 248 (2003) 227.
31] D. Freude, M. Hunger, H. Pfeifer, G. Scheler, J. Hoffmann, W. Schmitz,
Chem. Phys. Lett. 105 (1984) 427.
and Higher Education, Warsaw (no. PBZ-KBN-116/T09/2004).
The authors thank Professor J. Datka and Dr. K. Góra-Marek of
Jagiellonian University, Kraków, for the IR measurements.
[
[
32] M. Hunger, Catal. Rev.-Sci. Eng. 39 (1997) 345.
33] C. Doremieux-Morin, P. Batamack, J.M. Bregeult, J. Fraissard, Catal.
Lett. 9 (1991) 403.
References
[34] J. Datka, M. Kawałek, K. Góra-Marek, Appl. Catal. A Gen. 243 (2003)
293.
[1] J.L.F. Monteiro, C.O. Veloso, Top. Catal. 27 (2004) 169.
[35] D.P.B. Peixoto, S.M. Cabral de Menezes, M.I. Pais da Silva, Mater.
Lett. 4451 (2003) 1.
[36] E. Brunner, H. Ernst, D. Freude, M. Hunger, C.B. Krause, D. Prager,
W. Reschetilowski, W. Schwieger, K.-H. Bergk, Zeolites 9 (1989) 282.
[37] D. Seddon, Appl. Catal. 7 (1983) 327.
[38] A.E. Comyns, G.W. Morris, J.P. Sankey, US Patent 4,687,871 (1987).
[39] H.P. Röger, K.P. Möller, C.T. O’Connor, Microporous Mater. 8 (1997)
151.
[40] A. Berenguer-Murcia, A.J. Fletcher, J. Garcia-Martinez, D. Cazorla-
Amoros, A. Linares-Solano, K.M. Thomas, J. Phys. Chem. B 107 (2003)
1012.
[
2] V.P. Wystrach, L.H. Barnum, M. Garber, J. Am. Chem. Soc. 79 (1957)
5786.
[
[
[
[
3] A. Severino, A. Esculcas, J. Rocha, J. Vital, L.S. Lobo, Appl. Catal. A
Gen. 142 (1996) 255.
4] C.M. López, F.J. Machado, K. Rodríguez, B. Méndez, M. Hasegawa,
S. Pekerar, Appl. Catal. A Gen. 173 (1998) 75.
5] M. Gscheidmeier, R. Gutmann, J. Wiesmüller, A. Riedel, US Patent
5,559,127 (1997).
6] L. Grzona, N. Comelli, O. Masini, E. Ponzi, M. Ponzi, React. Kinet. Catal.
Lett. 69 (2) (2000) 271.