Faujasite catalysts promoted with gallium oxide: A physicochemical study

Article abstract of DOI:10.1021/jp9604507

This paper examines the properties of a defected zeolite matrix with the faujasite structure, promoted with gallium oxide. The fate of β-Ga₂O₃ loaded onto ultrastable faujasite is followed by XRD, FT IR, and laser Raman spectroscopies, and also by solid-state multinuclear NMR. Gallium(III) oxide is reduced after the contact with the zeolite, which is demonstrated by XRD, FT IR, and Raman measurements. Reduced gallium oxide is transferred to the channel system of the faujasite matrix, and some of the gallium ions (up to 6/unit cell) are occupying the zeolite tetrahedral sites, as evidenced by ²⁹Si and ⁷¹Ga MAS NMR and IR studies. The Si/Al ratio of the matrix changes during this process. Gallium sites of the type Si-O(H)-Ga can be discerned in IR spectra as a 3600-3603 cm^-1 band of the OH group. Such hydroxyl groups, not present in the pure faujasitic matrix, interact with ammonia. The gallium-promoted catalysts were tested in the oxidative dehydrogenation of propane. It will be shown that the already few gallium ions introduced into the faujasitic matrix significantly increase the selectivity of propane conversion to propene.

Full text of DOI:10.1021/jp9604507

J. Phys. Chem. 1996, 100, 10323-10330
10323
Faujasite Catalysts Promoted with Gallium Oxide: A Physicochemical Study
B. Sulikowski,*,† Z. Olejniczak,‡ and V. Corte´s Corbera´n§
Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences, Niezapominajek 1,
30-239 Krako´w, Poland, Institute of Nuclear Physics, Radzikowskiego 152, 31-342 Krako´w, Poland, and
Instituto de Cata´lisis y Petroleoqu´ımica, C.S.I.C., Campus U.A.M.-Cantoblanco, 28049 Madrid, Spain
ReceiVed: February 13, 1996X
This paper examines the properties of a defected zeolite matrix with the faujasite structure, promoted with
gallium oxide. The fate of â-Ga2O3 loaded onto ultrastable faujasite is followed by XRD, FT IR, and laser
Raman spectroscopies, and also by solid-state multinuclear NMR. Gallium(III) oxide is reduced after the
contact with the zeolite, which is demonstrated by XRD, FT IR, and Raman measurements. Reduced gallium
oxide is transferred to the channel system of the faujasite matrix, and some of the gallium ions (up to 6/unit
cell) are occupying the zeolite tetrahedral sites, as evidenced by 29Si and 71Ga MAS NMR and IR studies.
The Si/Al ratio of the matrix changes during this process. Gallium sites of the type Si-O(H)-Ga can be
discerned in IR spectra as a 3600-3603 cm-1 band of the OH group. Such hydroxyl groups, not present in
the pure faujasitic matrix, interact with ammonia. The gallium-promoted catalysts were tested in the oxidative
dehydrogenation of propane. It will be shown that the already few gallium ions introduced into the faujasitic
matrix significantly increase the selectivity of propane conversion to propene.
Introduction
resistance to temperature. The conversion of propane in the
presence of oxygen has been chosen as a test reaction. Gallium-
Gallium-containing zeolites constitute a very interesting class
of solids from a catalytic standpoint. The gallium analogue of
ZSM-5 (MFI) is used for transformation of C3-C5 alkanes into
aromatic hydrocarbons.1 For example, conversion of propane
in the Cyclar process yields 63.1 wt % benzene, toluene, xy-
lene (BTX), hydrogen (5.9 wt %), and a fuel gas byproduct
containing mainly methane and ethane.2 The Ga3+ ions in
zeolites can occupy tetrahedral framework sites (T) and non-
framework (NF, or extraframework) cationic positions. Gallium
has been successfully introduced into numerous zeolite frame-
works by direct hydrothermal synthesis.3 Gallium can also
be inserted into solids Via postsynthesis treatment of silica
polymorphs or high-silica zeolites by using sodium gallate.
In this way galliation of Silicalite-II (MEL)4 and recently of
zeolite Beta (BEA)5 has been accomplished. The isomorphous
substitution of gallium into aluminosilicate zeolites results
in enhanced selectivity toward aromatic hydrocarbons. It is
interesting to note that not only deliberately synthesized
[Si,Ga]-ZSM-5 molecular sieves containing framework gallium
were the active catalysts. Dehydrogenation properties were also
induced by mechanical mixing of the standard ZSM-5 zeolite
and Ga2O3.6 Such mixtures after a proper pretreatment were
the active catalysts for dehydrogenation of lower alkanes.7 Most
of the research has been focused on the Ga2O3-MFI1,6-8 and
Ga2O3-MEL systems.4,9-10 The status of gallium in MFI is a
much discussed issue, because of considerations concerning the
mechanism of low-alkane aromatization. On the other hand,
there is a dearth of information on the interaction of Ga2O3 with
other zeolites. We wish to report on the physicochemical
properties of catalysts based on a zeolite having a pore system
wider than that of ZSM-5. For this purpose zeolite Y, a 12-
membered ring aluminosilicate, has been chosen. The acid-
extracted ultrastable form of zeolite Y was used due to its known
(III) oxide, a pure zeolitic matrix, and the matrix loaded with
different amounts of Ga2O3 were investigated in this process.
Experimental Section
Catalyst Preparation. The ultrastable sample was prepared
by a procedure combining hydrothermal treatment of zeolite
Na-Y,11,12 followed by acid leaching.13-15 Zeolite Na-Y,
highly crystalline to X-rays, with Si/Al ) 2.47 and Na/Al )
1.09 determined by atomic absorption (AA), was used as the
parent material. The zeolite was twice ammonium exchanged
by stirring with 10 wt % NH4Cl at 368 K for 1 h. The
ammonium form (0.76 NH4, Na-Y, 25.81 wt % loss of ignition)
was heated in a horizontal quartz reactor at 813 K for 4 h under
self-steaming conditions. The ion exchange was then repeated,
and the sample was calcined at 1093 K to yield the ultrastable
form (US-Y) with Si/Al ) 2.55 and Na/Al ) 0.003 (bulk,
determined by AA). This form was treated twice with 1 dm3
of 0.1 and 0.2 M HCl per 100 g of the sample at 464 K for 1
h, washed with water, and dried. The acid-extracted ultrastable
zeolite (US-Y-ex) had Si/Al ) 8.91 (bulk) and Na/Al ) 0.003
and was used as a matrix (M). The matrix was ground in an
agate mortar with spectroscopically pure â-Ga2O3 (Bi, Cr, Zn,
Sn, Al, Co, and Fe: the content of each metal < 10-4 wt %;
In, Mg, Mn, Cu and Pb < 5 × 10-5 wt %; BETAr ) 2.90 m2/
g). R-Ga2O3 from Aldrich (containing ca. 5 wt % â-phase,
BETAr ) 27.5 m2/g) was used in some experiments. The hybrid
catalysts containing 4.8-44.4 wt % â-Ga2O3 were calcined in
air or reduced in hydrogen at 473-873 K prior to characteriza-
tion. The samples for reduction were heated at a rate of 3 K/min
in a nitrogen flow (30 cm3/min). The reduction with hydrogen
(20 cm3/min) was performed for 2 h followed by cooling the
samples to ambient temperature in a dry nitrogen flow. Mixed
samples will be labeled hereafter for brevity as x Ga/M, where
x denotes the initial amount of â-Ga2O3 in the sample (wt %)
and M represents the zeolitic matrix US-Y-ex.
* Corresponding author. Telephone: (+48 12) 252841. FAX: (+48 12)
251923. E-mail: ncsuliko@cyf-kr.edu.pl.
† Polish Academy of Sciences.
Characterization. Powder XRD patterns of the fully
hydrated samples were acquired on Seifert XRD 3000P and
Dron-2 diffractometers using Ni-filtered Cu KR radiation.
‡ Institute of Nuclear Physics.
§ Instituto de Cata´lisis y Petroleoqu´ımica C.S.I.C.
X Abstract published in AdVance ACS Abstracts, May 15, 1996.
S0022-3654(96)00450-9 CCC: $12.00 © 1996 American Chemical Society

10324 J. Phys. Chem., Vol. 100, No. 24, 1996
Sulikowski et al.
High-resolution solid-state magic-angle-spinning (MAS) NMR
spectra were recorded at 6.3 T (53.713 and 70.456 MHz for
29Si and 27Al, respectively). 29Si MAS NMR spectra were
acquired with π/2 pulses and a repetition time of 10 s. 27Al
MAS NMR spectra were obtained using very short (1 µs) pulses
with 25 kHz B1 amplitude to ensure that they are quantitatively
reliable.16,17 The recycle delays were 0.5 s. Typically, 1000
and 4000 transients were accumulated for silicon and aluminum
spectra. The rotors were spun in air at 4 kHz. 71Ga MAS NMR
spectra were acquired on a Bruker ASX 500 spectrometer
operating at 152.514 MHz. Samples were spun in air at 14
kHz, and 1 µs radio frequency pulses were applied with 1 s
recycle delay. Chemical shifts are quoted in ppm from external
tetramethylsilane (TMS), aluminum nitrate, and (NH4)Ga(SO4)2
for silicon, aluminum, and gallium, respectively.
The IR spectra in the zeolite framework vibration region
were obtained with a Nicolet 800 FT spectrometer (resolution
1 cm-1) using the KBr technique. The same spectrometer
was used for recording the spectra of self-supported disks
and to monitor the interaction of ammonia with the hydroxyl
groups. All the samples were evacuated at 600 °C and ca. 10-7
Torr prior to measurement. Laser Raman FT spectra of the
hydrated samples were collected with the Raman module of
the Nicolet 800 spectrometer. A Nd:YAG laser was used as
the excitation source (λ ) 1046 nm), and the beam power on
the sample was adjusted as necessary (50-600 mW). The
sample was placed in a Raman cell, and the scattered light was
detected with a liquid-nitrogen-cooled germanium detector. The
correction of background due to the Rayleigh scattering and
the correction for white light was performed by use of Nicolet
software.
Argon sorption at the temperature of liquid nitrogen was
measured in a volumetric sorption unit of standard design. The
samples were outgassed at 623 K before sorption studies.
Catalytic Tests. Catalyst particles (ca. 0.1 g, 0.25-0.42 mm)
were loaded into a tubular down-flow quartz reactor. SiC
chips were placed above the zeolite particles up to a total bed
volume of 1 cm3. The catalyst was preheated in a flow of dry
helium (100 cm3/min) at 773 K for 1 h. The propane (99.9%)-
oxygen (99.99%) mixture (2:1 molar ratio) diluted with helium
was fed into the reactor to give a total flow of 100 cm3/min
and W/F ) 4 g cat h/(g mol C3). The conversion of propane
was studied at temperatures in the range 773-873 K under
atmospheric pressure. Special precautions have been undertaken
to suppress the homogeneous reaction of propane. Reactants
and products were analyzed on-line with a Varian 3400 gas
chromatograph equipped with a TCD detector. Two columns
loaded with Porapak QS and molecular sieve were used for
analysis.
Figure 1. 29Si (a, d, f) and 27Al (b, c, e) solid-state MAS NMR spectra
of Na-Y (a, b), US-Y-ex (M) (c, d), and 4.8 Ga/M (e, f). The 4.8
Ga/M sample was reduced with hydrogen prior to measurement (as in
Figure 2).
4
I
∑
n
n)0
(Si/Al)NMR
)
I4 + 0.75I3 + 0.5I2 + 0.25I1
where In denotes the intensity of the NMR signal corresponding
to the Si(nAl) grouping. The formula provides a reliable
quantitative method for the determination of framework com-
positions of crystalline aluminosilicates (cf., for example, ref
20). The (Si/Al)NMR ratios of the framework, calculated from
the above formula, are listed in Table 3.
The parent sample Na-Y had 55.6 Al in the framework, as
evidenced by NMR spectra (Figure 1a,b). After hydrothermal
treatment, 22.4 Al atoms have been expelled into extraframe-
work positions to give sample US-Y with 33.2 aluminum ions
in the framework (Table 1). The US-Y sample was additionally
treated with HCl to remove nonframework aluminum and to
extract more tetrahedral Al. Mineral acids can be used for
extraction of aluminum from high-silica aluminosilicates (e.g.,
clinoptilolite),21 and faujasite is also amenable to acid treat-
ment.22 Acid treatment of the US-Y sample, yielding the US-
Y-ex matrix, did not affect its XRD crystallinity, as revealed
by examination of the corresponding diffractograms. Also, the
BET surface area of US-Y-ex (599 m2) remained the same in
comparison with the parent Na-Y material (Table 1). Thus,
the procedure applied led neither to the amorphization of the
solid nor to a decrease in its sorption capacity.
Results and Discussion
The hydrogen form of zeolite Y is unstable thermally;18
therefore we prepared for our experiments the matrix with an
increased silicon to aluminum framework ratio. This was done
by use of a procedure described in the Experimental Section.
29Si MAS NMR spectra of parent zeolite Na-Y, the ultrastable
matrix, and the matrix loaded with different amounts of â-Ga2O3
and reduced by hydrogen are shown in Figures 1 and 2. The
29Si spectra in these figures consist of up to five signals,
depending on composition. The signals correspond to five
different possible environments of a silicon atom Si(nAl), where
n (n e 4) denotes the number of aluminum atoms connected,
Via oxygens, to a central silicon atom. The silicon to aluminum
ratio of the zeolitic framework can be calculated directly from
the 29Si MAS NMR spectrum using the formula19
As seen from Table 1, most of the extraframework aluminum
generated during hydrothermal treatment was indeed removed.

Faujasite Catalysts Promoted with Gallium Oxide
J. Phys. Chem., Vol. 100, No. 24, 1996 10325
TABLE 1: Characteristics of the Parent Material and
Gallium-Loaded Zeolitic Catalysts
lattice
BET (argon)
(m2/g)
AlF/uc
(average)
sample
Si/Ala constant (Å)b
Na-Y
2.47
2.55
24.690(2)
24.51(1)
24.373(5)
24.382(5)
24.406(4)
24.431(3)
24.50(2)c
599
55.6
33.2
14.0
8.3
9.4
11.1
8.0
US-Y
US-Y-ex (M)
4.8 Ga/M
9.1 Ga/M
16.7 Ga/M
28.6 Ga/M
599
432
580
472
403
8.91
}
a Bulk, by wet chemical analysis (AA). b Lattice parameters were
calculated for the fully hydrated samples. c After reduction with
hydrogen at 873 K for 2 h.
TABLE 2: Number of Aluminum Atoms in the
Reduced Catalysts, Occupying Framework (F) and
Nonframework (NF) Sites, As Derived from 27Al MAS
NMR Studies
calculated from 27Al NMR
Al/uc not
sample
AlF/uc AlNF/uc (AlF + AlNF)/uc seen by NMR
Figure 2. 29Si MAS NMR spectra of (a) 9.1 Ga/M, (b) 16.7 Ga/M,
and (c) 28.6 Ga/M. All the samples were reduced with H2 at 600 °C
for 2 h prior to measurement.
4.8 Ga/M
9.1 Ga/M
16.7 Ga/M
28.6 Ga/M
10.0
10.5
10.7
8.4
3.5
3.4
3.0
3.4
13.5
13.9
13.7
11.8
5.9
5.5
5.7
7.6
Moreover, some aluminum atoms were extracted from tetra-
hedral sites, thus creating vacancies in the aluminosilicate
framework. Approximately 19 AlF atoms per unit cell were
extracted from US-Y by HCl, and hence the same number of
defects was introduced into the US-Y-ex matrix (note that M
was not healed under hydrothermal conditions prior to loading
with gallium oxide). It was supposed that such a defected matrix
would easily react with or accommodate gallium oxide. The
total silicon to aluminum ratio equals to 8.91 for US-Y-ex (M);
this corresponds to 19.4 Al per unit cell. Aluminum in the
sample is distributed between tetrahedral and octahedral posi-
tions. The number of aluminum atoms in the framework of
zeolites X and Y affects their unit cell parameters as well as
most of the mid-infrared frequencies, largely because Si-O
bonds are shorter than Al-O bonds (1.60 Å and 1.75 Å).
Empirical formulas were therefore derived that allow estimation
of the number of Al atoms occupying framework sites in
faujasite.23-25 All the formulas, despite the method used for
aluminum depletion,26 give reasonable results.
The number of framework aluminum atoms in the matrix was
calculated from the unit cell parameter contraction, the shift of
the Si-O main asymmetric stretching band in IR, and decon-
volution of 29Si MAS NMR (cf. Figure 1d). The sample
contains 11.3 and 11.5 AlF according to IR,23,24 15.7 and 14.5
from XRD measurements,23,25 12.7 and 16.7 according to 27Al
and 29Si MAS NMR spectra, respectively. The average number
of aluminum occupying framework positions of the matrix,
estimated from four independent methods, is therefore 14.0/uc
(uc ) unit cell).
TABLE 3: Number of Aluminum and Gallium Ions at
Framework Sites in the Catalysts, As Calculated from NMR
Studies
AlF/uc
GaF/uc
sample
Na-Y
US-Y
US-Y-ex (M)
4.8 Ga/M
9.1 Ga/M
16.7 Ga/M
28.6 Ga/M
27Al NMR
29Si NMR
average
29Si NMR
55.6
nd
12.7
10.0
10.5
10.7
8.4
55.6
33.2
16.7
6.6
8.3
11.5
7.5
55.6
33.2
14.0a
8.3
9.4
11.1
8.0
3.6
2.2
5.9
3.6
a Mean value, calculated from 27Al and 29Si MAS NMR, IR, and
XRD.
In the 29Si spectrum of the matrix an additional signal at
-114.7 ppm is clearly seen (Figure 1d), with an intensity of
17.5%. This line is assigned to nonordered Si(0Al) groupings
of amorphous silica, which are formed during the matrix
preparation. The silicon signals in this range have also been
observed in zeolites dealuminated by SiCl4.28,29 After loading
with gallium oxide and reduction with hydrogen this signal
vanishes (Figures 1f and 2a,c). Silicon and gallium species are
therefore competing for healing the matrix defects; and as will
be discussed below, it is mostly silicon which is inserted into
tetrahedral zeolitic sites. In the gallium-containing samples
dealumination is proceeding further. Approximately 5-7 AlF
atoms are expelled into nonframework positions, and vacancies
formed are healed with silicon coming from amorphous-like
silicon species giving rise to a signal at -114.7 ppm. Close
inspection of the x Ga/M 29Si spectra (Figures 1 and 2) revealed
that, besides lines of Si(1Al) and Si(2Al) groups at -102 and
-96 ppm, weak signals at -100 and -94 ppm were present.
The signals were assigned to Si(1Ga) and Si(2Ga) groupings,
respectively, taking into account published data on the as-
prepared gallosilicates with the faujasite structure.30 Gaussian
deconvolution of the silicon spectra gave Si/Al and Si/Ga ratios
in the samples. Distribution of aluminum between framework
and nonframework positions was also calculated independently
from 27Al spectral data. The data thus obtained are summarized
in Tables 1-3. As seen, there is a fairly good agreement
between the quantitative estimation of silicon, aluminum, and
As revealed by 27Al NMR, the matrix still contains a few
atoms (5.4 AlNF) of nonframework aluminum in the unit cell,
which have not been extracted by HCl under the conditions
specified. Distribution of the AlNF species in zeolite Y depends
on the temperature of hydrothermal treatment.27 Treatment at
573 K gave samples with octahedrally coordinated aluminum
entities located in the large cavities, while increasing the
temperature to 773 K yielded the sample with AlNF located in
both the large cavities and sodalite cages. Our sample was
calcined at, higher temperature, so it is reasonable to assume
that some or all of the nonframework aluminum is occupying
hindered positions (sodalite cages and/or hexagonal prisms) and
thus not amenable to extraction by mineral acid.

10326 J. Phys. Chem., Vol. 100, No. 24, 1996
Sulikowski et al.
gallium ions in the framework of catalysts, carried out by
different approaches.
The characterization of the starting materials and the gallium-
containing samples is given in Table 1. The crystallinity of all
of the gallium-loaded samples was largely retained upon thermal
treatment in air and hydrogen. Thermal treatment of the sample
up to 673 K did not affect the crystallinity of the matrix. For
the 28.6 Ga/M sample reduced at 873 K the decreased XRD
crystallinity of the matrix would correspond to the decreased
BET surface area from 599 to 428 m2/g. The experimentally
found value was 403 m2/g (Table 1). Similarly, for 16.7 Ga/M
we would expect a BET area of 498 m2/g, because of the
decreased content of the zeolitic phase in 1 g of sample, and
we found 472 m2/g. Therefore, we conclude that gallium oxide
species do not restrict the access of probe molecules to the
internal zeolite pore system, as was observed for ferrierite.31
The latter zeolite has, however, a much smaller pore system
(3.5-5.4 Å).
The amount of â-Ga2O3 in the matrix after various treatments
was estimated from XRD diffraction patterns. The sum of peak
heights at 30.28°, 33.34°, and 37.26° 2θ corresponding to
â-Ga2O3 was divided by the sum of peak heights of the zeolite
with hkl indices of 331, 440, 533, and 642; this was proportional
to the ratio â-Ga2O3/zeolite. In this way ratios of 0.28 and 0.05
were found for the 28.6 Ga/M and 4.8 Ga/M samples, in very
good agreement with the amount of â-Ga2O3 initially loaded
into the catalysts. For the 28.6 Ga/M sample after calcination
in air and reduction with H2 at 673 K, the ratio decreased to
0.15 and 0.20, respectively (the calculated values were corrected
for the loss of crystallinity of the zeolite). According to XRD,
considerable amounts of the bulk â-Ga2O3 lost their long-range
ordering. Part of the gallium ions was used to fill the vacancies
of the matrix. After thermal treatment of 28.6 Ga/M (both in
air and hydrogen) the development of minute amounts of an
unidentified phase was observed.
Figure 3. 71Ga MAS NMR spectra of (a) 4.8 Ga/M, (b) â-Ga2O3, and
(c) R-Ga2O3.
The unit cell of â-Ga2O3 consists of eight Ga3+ ions.34 The
gallium ions in this structure adopt tetrahedral and octahedral
coordination. Average interionic Ga-O distances are 1.83 Å
for tetrahedral and 2.00 Å for octahedral coordination, respec-
tively.35 Each GaI3+ is surrounded by a distorted tetrahedron
of oxygen ions, and each GaII3+ is placed in a highly distorted
octahedron of oxygen ions.35 Those two crystallographically
nonequivalent galliums in the unit cell of â-Ga2O3 should give
rise to two distinct NMR signals. In Figure 3b the 71Ga 14
kHz magic-angle-spinning NMR spectrum of â-Ga2O3 taken
on a 500 MHz spectrometer is shown. As seen, the spectrum
is dominated by spinning sidebands, and contrary to expectation
no two distinct gallium signals, corresponding to two different
environments of Ga3+, can be readily discerned. We note that
the central band is not separated from the sidebands. The
appearance of the NMR spectrum for R-Ga2O3, contaminated
with ca. 5% of the â form, is similar, but slight differences are
seen upon close inspection (Figure 3c). R-Ga2O3 is isostructural
with R-corundum, and therefore, all the gallium ions adopt
octahedral coordination. As found for â-Ga2O3, the octahedral
coordination gives rise to the signal around 0 ppm, onto which
the small, sharp signal of the â-phase around 15 ppm is
superimposed. The spinning sidebands for R-Ga2O3 are also
much less intense. There is also a shoulder seen at 105 ppm
and assigned to â-phase contamination. This lends support to
the assumption that the 105 ppm line in Figure 3b is also a
superimposition of a spinning sideband and a real signal coming
from tetrahedrally coordinated gallium in â-Ga2O3. The
spectrum of 4.8 Ga/M catalyst, shown in Figure 3a, is different.
Despite a lower signal-to-noise ratio, understandable in view
of the much lower concentration of gallium species in the
catalyst, two features are seen. First, the signal resembling lines
of the pure â-Ga2O3 around 0 ppm is seen. Also, the high-
intensity spinning sidebands are lost. Second, and most
interesting, a line at about 180 ppm, which coincides neither
with the spinning sideband at 200 or 105 ppm in pure â-Ga2O3,
with an intensity comparable to that of the line around 0 ppm,
is evident. We assign this signal to tetrahedrally coordinated
gallium in the zeolitic matrix, in accord with conclusions reached
We observed the expansion of the lattice parameter of the
matrix after the loading of â-Ga2O3. It is known that incorpora-
tion of gallium into the faujasite structure led to the enlarge-
ment of the unit cell parameters, due to the longer Ga-O bonds
(as compared to Al-O bonds). Such an expansion was
observed by Ku¨hl32 and Szostak.33 The plot of the lattice
parameter a0 of hydrated synthetic Ga-FAU as a function of
the number of Ga atoms per unit cell shows two discontinuities,
even more pronounced that found for Al-FAU, but the
relationship within each region is linear.32 In the region <64
Ga/uc a linear regression analysis of the data by Ku¨hl gives the
expansion of the unit cell as 0.01508 Å/1 Ga. The experimen-
tally observed expansion of a0 for 28.6 Ga/M was in the range
0.157-0.187 Å and corresponded therefore to the insertion of
10.4-12 Ga per unit cell. On the other hand, the expansion of
the unit cell of the 4.8 Ga/M sample corresponded to the
insertion of only 0.6 Ga/uc. It should be noted, however, that
for highly siliceous gallosilicates, similar to aluminosilicates,
the plot becomes nonlinear and the estimation of the number
of Ga in this region by XRD is inaccurate (the error may be
equal to even a few Ga/uc). The lattice of 4.8 Ga/M expanded
slightly upon reduction at 873 K, and only traces of the bulk
â-Ga2O3 were found in this sample by XRD. This confirmed
that most of the gallium oxide in the precursor disappeared upon
reduction. The amount of gallium oxide in the precursor
corresponded to 6.8 Ga/uc, assuming that all Ga2O3 would be
accommodated by the matrix. As XRD has shown only minute
amounts of â-Ga2O3 after the reduction, we conclude that about
5 Ga/uc were found in the 4.8 Ga/M catalyst at the tetrahedral
sites.

Faujasite Catalysts Promoted with Gallium Oxide
J. Phys. Chem., Vol. 100, No. 24, 1996 10327
from IR, Raman, and XRD studies of this sample. NMR studies
of gallosilicates, using the 69Ga and/or 71Ga resonances, are
scarce. Solid-state NMR of gallium nuclei is much more
difficult than of 27Al, because the former gives significantly
broader line widths. At the same magnetic field strength, the
relative ratio of the line widths for 27Al, 69Ga, and 71Ga is 1:6.46:
2.01.36 Therefore, of the two gallium nuclei, 71Ga is the more
attractive one. The observed chemical shift of tetrahedral
gallium is from 150 to 184, depending on the zeolite structure.
For as-prepared gallosilicates with the faujasite structure,
containing much more gallium than our sample, the shift was
172 to 174 ppm.36 The spectrum depicted in Figure 3a
constitutes therefore direct evidence that some gallium ions
indeed occupy tetrahedral framework sites of the zeolite matrix.
Finally, we conclude that strong quadrupolar interactions of the
71 gallium nuclei, observed in our experiments, cannot be
effectively removed even using the highest currently available
field (500 MHz for protons) and a very fast spinning of the
rotor (14 kHz). The quantitative analysis of such spectra is
possible only by numerical simulation of the line shape. This
work is now in progress.
Raman spectroscopy has been scarcely used for characteriza-
tion of aluminosilicates, partly due to the high fluorescence
encountered when using conventional helium or argon lasers.
Fluorescence which masks the weak Raman signals can
sometimes be suppressed, e.g., by repeating calcination of a
sample before measurement. The introduction of the Nd:YAG
laser, with an exciting line at λ ) 1046 nm, significantly reduced
the fluorescence that so hampers conventional excitation sources.
The Raman spectrum of the zeolitic matrix is shown in Figure
4c. Contrary to Na-X, Na-Y, and other zeolites, the two
strongest bands were observed at 506.5 and 488.8 cm-1 in the
acid-extracted US-Y-ex sample. In sodium faujasite the single
band at 515 for Na-X and 505 cm-1 for Na-Y has been
assigned to the motion of the oxygen atom in the plane
perpendicular to the T-O-T bond.37 There is a shift of 10
cm-1 upon increasing the Si/Al ratio in a faujasite substitutional
series (zeolites X and Y). In the case of ultrastable zeolite Y
the strongest Raman signal was split into two bands.
Figure 4. FT laser Raman spectra of (a) spectroscopically pure
â-Ga2O3, (b) a stoichiometric mixture of â-Ga2O3 + Ga after reduction
at 600 °C in a vacuum-sealed ampule, (c) US-Y-ex (M), and 28.6 Ga/M
as-prepared (d), and after reduction with hydrogen at 400 °C (e) and
600 °C (f).
lines characteristic of bulk â-Ga2O3 (416, 653, and 766 cm-1
)
The position of a band around 500 cm-1 is a function of the
average T-O-T angles for different zeolites.38 Accordingly,
the band at 506.5 cm-1 corresponds to the 142.0° and the band
at 488.8 cm-1 to the 146.9° T-O-T angle. In faujasite there
are four nonequivalent oxygen atoms, and hence each Si atom
has four different Si-O-T angles (T ) Si or Al). In natural
faujasite39 these are 138.6°, 139.7°, 145.3°, and 147.4°, while
in modified faujasite40 the four angles are 136.1°, 142.4°, 146.0°,
and 149.8°. There are thus two pairs of similar angles, the
smaller and the larger angles. The average angle of each pair
is 139.2° and 147.1°. The splitting of the band around 500 cm-1
is consistent with the Si-O-T angles of faujasite. Obviously,
due to the width of the bands, separate signals for each of the
four angles in faujasite were not observed. It follows that
Raman spectroscopy is a very sensitive tool and can provide
accurate structural information on the subtle changes in the
structure of modified faujasite, in addition to X-ray39 or neutron
diffraction studies.40
also disappeared completely. The signal at ca. 350 cm-1 was
broadened. The new broad and weak signals appear in the
spectrum upon reduction at 568, 827, and 957 cm-1. A separate
TPR experiment has shown that the bulk â-Ga2O3 was nonre-
ducible by hydrogen at 873 K. However, the same features as
found for the reduced 28.6 Ga/M sample were observed for
gallium oxide reduced deliberately by pure gallium under
vacuum at 873 K. â-Ga2O3 gives lines at 346, 416, 476, 630,
653, and 766 cm-1 (Figure 4a). The bands in the range 300-
600 cm-1 correspond to bending vibrations, while the bands
above 600 cm-1 are due to the Ga-O4 tetrahedral stretching
modes.41 The signals of the reduced oxide are seen at 354,
565, 831, and 948 cm-1. The appearance of the bands is the
same as in 28.6 Ga/M reduced with H2 (cf. Figure 4e,f).
Gallium oxide loaded onto ultrastable faujasite was therefore
amenable to the reduction by hydrogen, in a way resembling
the behavior of the Ga2O3/ZSM-5 system.42,43
The zeolite matrix containing 28.6 wt % â-Ga2O3 gave a
typical IR spectrum with the frequencies corresponding both
to the matrix and to the oxide. Upon reduction the signal of
the bulk â-Ga2O3 at 668 cm-1 decreased, while the main
asymmetric stretching vibration of the zeolite at 1072 cm-1 was
shifted to 1070 and 1053 cm-1 after the reduction at 400 and
600 °C, respectively. Such a shift to lower wavenumbers is
expected upon the formation of the less siliceous framework.24
The shift of the main stretching frequency might be interpreted
Other much weaker bands were seen at 299, 356, 665, and
815 cm-1. Upon mixing with gallium(III) oxide both bands of
the oxide and those of the matrix were found. The bands related
to the bulk gallium oxide were found (Figure 4d) at 347, 416,
653, and 766 cm-1
. After the reduction in hydrogen the
appearance of the spectrum changed significantly. The features
characteristic of the matrix disappeared. However, the zeolite
retained its crystallinity as seen from IR and XRD data. The

10328 J. Phys. Chem., Vol. 100, No. 24, 1996
Sulikowski et al.
Figure 6. FT IR difference spectra in the hydroxyl vibration region
after ammonia adsorption at 0.001 T on (a) US-Y-ex (M), (b) 4.8 Ga/
M, and (c) 9.1 Ga/M. Samples b and c were reduced ex situ with H2
prior to measurement.
Figure 5. FT IR spectra in the hydroxyl vibration region of (a) US-
Y-ex (M) and (b) 9.1 Ga/M reduced with hydrogen.
as isomorphous substitution of some gallium into the framework.
The ratio of the frequency for â-Ga2O3 at 668 cm-1 to the matrix
frequency at 832 cm-1 changed from 1.4 to 0.5, which
confirmed further the loss of some bulk gallium oxide in the
reduced sample and transport of gallium (probably as Ga2O42)
to faujasite cages.
TABLE 4: Oxidative Dehydrogenation of Propane on
â-Ga2O3 and Zeolite Catalysts. Conversion (mol %) and
Selectivity (%) to Hydrocarbons
selectivity (%) to
sample
temp (K) conversion (%) methane ethene propene
The IR spectrum of the faujasite matrix US-Y-ex (M) after
evacuation at 600 °C is shown in Figure 5a. Three well resolved
bands are clearly seen, superimposed onto a broader one. The
band at 3739 cm-1 is due to terminal Si-OH group vibrations,
while the bands at 3621 and 3564 cm-1 correspond to high-
(HF) and low-frequency (LF) bands of structural Si-O(H)-Al
vibrations. The HF band is shifted from its normal position in
zeolite Y to lower frequencies, as the Si/Al ratio of the zeolite
is increased due to dealumination (Table 1). The broad band
with the maximum at 3550 cm-1 arises from the interaction of
silanol groups via hydrogen bonding.44
â-Ga2O3
773
823
848
873
773
798
823
848
773
798
823
848
773
798
823
848
0.63
2.62
4.82
8.21
2.96
4.95
7.76
11.84
1.49
3.54
6.74
10.37
2.00
3.77
4.52
8.56
5.94
6.38
7.11
8.31
3.04
1.65
2.01
2.86
5.39
2.94
2.48
3.18
4.97
4.58
3.67
3.95
14.15
18.84
20.25
23.50
8.08
5.74
6.76
8.66
14.10
9.18
7.93
9.36
29.94
33.61
34.39
31.41
39.01
33.69
28.56
26.29
58.43
43.21
33.22
28.62
48.17
43.54
43.13
30.73
US-Y-ex
4.8 Ga/M
28.6 Ga/M
11.90
11.89
11.19
10.94
After loading with â-Ga2O3 followed by reduction with
hydrogen the IR spectra change significantly. This is exempli-
fied by 9.1 Ga/M (Figure 5b). The bands are now broader; the
band of the Si-OH groups is seen at 3742 cm-1, while the
bands of the structural Si-O(H)-Al groups at 3621 cm-1 and
hydrogen-bonded silanol groups are less intensive. The LF band
is seen as a shoulder superimposed on the band of the hydrogen-
bonded Si-OH groups. A novel band arises in the spectrum,
zeolitic matrix, and on the matrix loaded with various amounts
of gallium oxide. All the catalysts were reduced in hydrogen
before the tests. The homogeneous reaction of propane was
studied separately using a microreactor filled with SiC chips
while keeping all the other parameters constant. By redesigning
the catalytic setup and applying proper test conditions we were
able to suppress the homogeneous reaction of propane to 0.3-
3% at 773-873 K.
The main products formed from propane on Ga2O3 were
propene, ethene, methane, CO, and CO2. The total conversion
of C3H8 in the presence of oxygen was much lower than those
reported for anaerobic conditions.45
The results of catalytic tests are shown in Table 4. The
conversion level was kept below 10% to observe the primary
products of the reaction. The activation energy on all the
catalysts was in the range 24.2-34.8 kcal/mol, thus diffusion
effects did not affect the results. The conversion of propane
on pure gallium oxide is 2 times higher than that corresponding
to the homogeneous reaction of propane.
The activity of pure US-Y-ex (M) in the presence of oxygen
was relatively high under the conditions used, going from 3%
at 773 K to 12% at 848 K. The main products on M were the
same as on gallium oxide, but the selectivity to propene was
lower. Contrary to ZSM-5, not even traces of aromatic
with the maximum at about 3600-3603 cm-1
.
Adsorption of ammonia reveals more details in the OH
vibration region of the samples. The corresponding difference
spectra of M, 4.8 Ga/M, and 9.1 Ga/M, shown in Figure 6,
indicate that all the OH groups interact with NH3. The band at
3739 cm-1 in M (Figure 5a) splits into two at 3740 and 3746
cm-1, so at least two silanol groups are present in the gallium
oxide modified samples (Figure 6b,c). The other bands are
present at 3624, 3563, and 3600-3603 cm-1. The former two
correspond to the same groupings as in the pure matrix (Figure
5a). The band at 3600-3603 cm-1, not found in the matrix, is
assigned to structural groups of the type Si-O(H)-Ga, which
are formed by insertion of some gallium into tetrahedral
positions of faujasite. The intensity of this band does not change
with the content of gallium oxide in the precursor. This
confirms, in accord with NMR studies, that the amount of
tetrahedral gallium is essentially the same in 4.8 and 9.1 Ga/M
samples (cf. Table 3).
Oxidative dehydrogenation of propane in the presence of
oxygen was studied on the spectroscopically pure â-Ga2O3, the

Faujasite Catalysts Promoted with Gallium Oxide
J. Phys. Chem., Vol. 100, No. 24, 1996 10329
hydrocarbons were found in the products. The matrix contained
the Brønsted acid sites (3621 cm-1 OH groups); their strength
was, however, not high enough to catalyze aromatization of
ethene and propene. Consecutive reactions of olefins formed
on MFI zeolites lead to the formation of aromatic products. The
activation of propane on M is due to Brønsted acid sites, in
agreement with other studies.46,47 We have recalculated catalytic
results for M to compare with ZSM-5.48 The estimated
conversion of propane on M under the conditions used with
ZSM-5 was 27% as compared to 35% on the latter, thus the
activity of faujasite is slightly lower than that of ZSM-5.
The conversion of propane on two catalysts containing
gallium is compared in Table 4. It follows that introduction of
gallium affects the overall activity; there is, however, not much
difference between the two solids (their BET areas are similar,
cf. Table 1). The products of propane conversion are the same
as on gallium oxide and M. No aromatics were found in the
effluents. The most effective catalyst was 4.8 Ga/M, which
contained gallium dispersed in the framework of faujasite, and
on which traces of the bulk gallium oxide were found after the
reduction. Introduction of further amounts of gallium to give
the catalyst with a total of 10-12 Ga per unit cell did not lead
to increased activity and selectivity of propane conversion.
Introduction of gallium into the faujasitic matrix enhanced the
selectivity toward propene. The effect was already significant
for 4.8 Ga/M, and we conclude that the presence of small
amounts of dispersed gallium (ca. 6 Ga/uc) was responsible for
the enhanced selectivity to propene.
the matrix significantly increase the selectivity to propene. This
is due to the decreasing amount of strong acid centers Si-
O(H)-Al and forming simultaneously new hydroxyl groups Si-
O(H)-Ga. As further loading with active phase does not change
the reaction selectivity, we can conclude that the first gallium
ions introduced into the framework form the active centers for
transformation of propane. The gallium-containing catalysts
based on the ultrastable zeolite Y constitute therefore an
interesting system for a selective transformation of low molec-
ular weight alkanes to olefins.
Note Added in Proof. After submitting the manuscript, we
found a recent 69Ga and 71Ga NMR study of â-Ga2O3 by Massiot
et al. [Massiot, D.; Farnan, I.; Gautier, N.; Trumeau, D.;
Trokiner, A.; Coutures, J. P. Solid State Nucl. Magn. Reson.
1995, 4, 241]. The octahedral and tetrahedral sites of gallium
ions in this oxide have been fully characterized. That work
provides a further support for the interpretation of our results.
Acknowledgment. We are grateful to the State Committee
for Scientific Research, Warsaw, for support (Grant No. 2 P303
149 04). Thanks are also due to Dr. S. Steuernagel of Bruker
(Karlsruhe, Germany) for 71Ga MAS NMR spectra. The
excellent technical assistance of Ms. J. Krys´ciak is also
acknowledged.
References and Notes
(1) Mowry, J. R.; Anderson, R. F.; Johnson, J. A. Oil Gas J. 1985, 83,
128.
(2) Mowry, J. R.; Martindale, D. C.; Hall, H. P. Arab. J. Sci. Eng.
1985, 10, 367.
Conclusions
(3) Sulikowski, B.; Klinowski, J. J. Chem. Soc.,Chem. Commun. 1989,
1289 and references therein.
(4) Thomas, J. M.; Xinsheng, L. J. Phys. Chem. 1986, 90, 4843.
(5) Liu, X.; Lin, J.; Liu, X.; Thomas, J. M. Zeolites 1992, 12, 936.
(6) Kanazirev, V.; Price, G. L.; Dooley, K. M. J. Chem. Soc., Chem.
Commun. 1990, 712.
(7) Kanazirev, V.; Dimitrova, R.; Price, G. L.; Khodakov, A. Yu.;
Kustov, L. M.; Kazansky, V. B. J. Mol. Catal. 1991, 70, 111.
(8) Carli, R.; Bianchi, C. L.; Giannantonio, R.; Ragaini, V. J. Mol.
Catal. 1993, 83, 379.
(9) Zulfugarov, Z. G.; Suleimanov, A. S.; Samedov, Ch. R. In Structure
and ReactiVity of Modified Zeolites; Jacobs, P. A., Jaeger, N. I., J´ıru, P.,
Kazansky, V. B., Schulz-Ekloff, G., Eds.; Elsevier: Amsterdam, 1984; Stud.
Surf. Sci. Catal. 1984, 18, 167.
(10) Ione, K. G.; Vostrikova, L. A.; Petrova, A. V.; Mastikhin, V. M.
Ref 9, p 151.
(11) McDaniel, C. V.; Maher, P. K. In Molecular SieVes; Society of
Chemical Industry: London, 1968; p 186.
(12) Kerr, G. T.; Shipman, G. F. J. Phys. Chem. 1968, 72, 3071.
(13) Kerr, G. T.; Chester, A. W.; Olson, D. H. In Proceedings of the
Symposium on Zeolites, 11-14 September, 1978, Szeged, Hungary, Acta
Phys. Chem. Szeged, NoVa Ser. 1978, 24, 169.
(14) Scherzer, J. J. Catal. 1978, 54, 285.
(15) Sulikowski, B.; Rakoczy, J.; Hamdan, H.; Klinowski, J. J. Chem.
Interaction of pure gallium(III) oxide with the US-Y-ex matrix
is a complex phenomenon. Pure gallium oxide cannot be
reduced by hydrogen at temperatures up to 873 K. However,
gallium oxide is easily reduced after contact with the zeolite,
which is clearly seen from Raman, XRD, and IR measurements.
Reduced gallium oxide is transferred to the channel system of
faujasite. Some of the gallium ions are inserted into the zeolite
tetrahedral sites (thus healing defects created during the acid
treatment of the ultrastable zeolite), while the remaining ones
occupy extraframework positions. Further dealumination of the
matrix proceeds simultaneously during the high-temperature
reduction of the catalyst precursors. As a result about 5-7 Al/
uc are expelled into extraframework positions. The remaining
aluminum vacancies are filled with silicon coming from
amorphous silica. We evidenced it by observing the disap-
pearance of the NMR signal at ca. -115 ppm. When the
amount of gallium oxide mixed with the matrix is increased
further, the situation does not change significantly. More
aluminum is depleted in the framework, but the amount of
tetrahedral gallium remains essentially the same. The maximum
amounts of tetrahedral gallium which can still be accommodated
by the matrix is low and does not exceed about 6 Ga per unit
cell. Framework gallium gives rise to structural hydroxyl groups
Si-O(H)-Ga, seen in the IR spectra at 3600-3603 cm-1. Such
groups are not present in the pure zeolitic matrix.
Soc., Chem. Commun. 1987, 1542.
(16) Man, P. P.; Klinowski, J. J. Chem. Soc., Chem. Commun. 1988,
1291.
(17) Man, P. P.; Klinowski, J.; Trokiner, A.; Zanni, H.; Papon, P. Chem.
Phys. Lett. 1988, 151, 143.
(18) Maesen, T. L. M.; Sulikowski, B.; van Bekkum, H.; Kouwenhoven,
H. W.; Klinowski, J. Appl. Catal. 1989, 48, 373.
(19) Engelhardt, G.; Lohse, U.; Lippmaa, E.; Tarmak, M.; Ma¨gi, M. Z.
Anorg. Allg. Chem. 1981, 482, 49.
All the samples show activity in the oxidative dehydrogena-
tion of propane. The highest conversion level of propane was
observed for pure US-Y-ex zeolite. The matrix itself is very
active in the conversion of propane. Contrary to the H-ZSM-
5/Ga2O3 system no aromatics were found in the reaction
products under the conditions studied. Loading of the ultrastable
zeolite with increasing amounts of â-Ga2O3 gives rise to the
continuous decrease of the overall conversion level of propane,
while the enhanced selectivity to propene is still observed. It is
interesting that the already few gallium ions accommodated by
(20) Klinowski, J.; Ramdas, S.; Thomas, J. M.; Fyfe, C. A.; Hartman,
J. S. J. Chem. Soc., Faraday Trans. 2 1982, 78, 1025.
(21) Barrer, R. M.; Makki, M. B. Can. J. Chem. 1964, 42, 1481.
(22) Lee, F. T.; Rees, L. V. C. J. Chem. Soc., Faraday Trans. 1 1987,
83, 1531.
(23) Sohn, J. R.; DeCanio, S. J.; Lunsford, J. H.; O’Donnell, D. J.
Zeolites 1986, 6, 225.
(24) Sulikowski, B.; Klinowski, J. J. Chem. Soc., Faraday Trans. 1990,
86, 199.
(25) Fichtner-Schmittler, H.; Lohse, U.; Engelhardt, G.; Patzelova´, V.
Cryst. Res. Technol. 1984, 19, K1.
(26) Sulikowski, B. Heterogeneous Chem. ReV. 1996, 3 (3), 171.

10330 J. Phys. Chem., Vol. 100, No. 24, 1996
Sulikowski et al.
(27) Freude, D.; Fro¨hlich, T.; Hunger, M.; Pfeifer, H.; Scheler, G. Chem.
Phys. Lett. 1983, 98, 263.
(28) Klinowski, J.; Thomas, J. M.; Fyfe, C. A.; Gobbi, G. C.; Hartman,
J. S. Inorg. Chem. 1983, 22, 63.
(29) Ray, G. J.; Nerheim, A. G.; Donohue, J. A. Zeolites 1988, 8,
458.
(39) Olson, D. H.; Dempsey, E. J. Catal. 1969, 13, 221.
(40) Barrie, P. J.; Gladden, L. F.; Klinowski, J. J. Chem. Soc., Chem.
Commun. 1991, 592.
(41) Dohy, D.; Lucazeau, G.; Revcolevschi, A. J. Solid State Chem.
1982, 45, 180.
(42) Le Van Mao, R.; Carli, R.; Yao, J.; Ragaini, V. Catal. Lett. 1992,
(30) Engelhardt, G.; Michel, D. High-Resolution Solid-State NMR of
Silicates and Zeolites; Wiley: Chichester, New York, 1987; p 254.
(31) Corte´s Corbera´n, V.; Valenzuela, R. X.; Sulikowski, B.; Derewin´ski,
M.; Olejniczak, Z.; Krys´ciak, J. Fifth European Workshop Meeting on
Selective Oxidation by Heterogeneous Catalysis, Berlin, 1995.
(32) Ku¨hl, G. H. J. Inorg. Nucl. Chem. 1971, 33, 3261.
(33) Szostak, R. Molecular SieVes; Van Nostrand Reinhold: New York,
1989; p 214.
(34) Kohn, J. A.; Katz, G.; Broder, J. D. Am. Mineral. 1957, 42, 398.
(35) Geller, S. J. Chem. Phys. 1960, 33, 676.
(36) Timken, H. K. C.; Oldfield, E. J. Am. Chem. Soc. 1987, 109, 7669.
(37) Dutta, P. K.; Shieh, D. C. J. Phys. Chem. 1986, 90, 2331.
(38) Dutta, P. K.; Shieh, D. C.; Puri, M. Zeolites 1988, 8, 306.
16, 43.
(43) Price, G. L.; Kanazirev, V. J. Catal. 1990, 126, 267.
(44) de Ruiter, R.; Kentgens, A. P. M.; Grootendorst, J.; Jansen, J. C.;
van Bekkum, H. Zeolites 1993, 13, 128.
(45) Meriaudeau, P.; Naccache, C. J. Mol. Catal. 1990, 59, L31.
(46) Buckles, G.; Hutchings, G. J.; Williams, C. D. Catal. Lett. 1991,
8, 115.
(47) Derouane, E. G.; Abdul Hamid, S. B.; Ivanova, I. I.; Blom, N.;
Højlund-Nielsen, P.-E. J. Mol. Catal. 1994, 86, 371.
(48) Paredes Quesada, N. Ph.D. Thesis, Universidad Auto´noma, Madrid,
1993.
JP9604507