Professional Documents
Culture Documents
Preface: (I) (Ii) (Iii) (Iv) (V) E.g.
Preface: (I) (Ii) (Iii) (Iv) (V) E.g.
S U P P O R T AND S P O N S O R I N G
(as of May 5, 2005)
The Organizers of the 3 rd FEZA Conference on Zeolites wish to thank various Institutions and
Companies for their support to this Conference. Their contributions allowed a reduced
registration fee for students and a bursary program.
INSTITUTIONS
and
COMPANIES
BASF
Bruker AXS
(~eskfi rafin6rsk~, a.s., Czech Republic
Elsevier
EniTecnologie
Eurosupport
ExxonMobil Chemical
Haldor Tops~e A/S
IFP
INEOS Silicas
Micromeritics
Quantachrome
Research Institute of Inorganic Chemistry, Czech Republic
Shell Chemicals
Sigma-Aldrich, Czech Republic
Total
Tricat Zeolites
UOP
Zeolyst
ix
ORGANIZING COMMITTEE
Chairman
J. (~ejka J. Heyrovsk2~ Institute of Physical Chemistry, Prague
Vice-chairman
P. Nachtigall Institute of Organic Chemistry and Biochemistry, Prague
Secretary
N. Zilkovfi J. Heyrovsk2~ Institute of Physical Chemistry, Prague
Treasurer
J. Ondrfi~kovfi J. Heyrovsk~ Institute of Physical Chemistry, Prague
Members
G. Ko~ovfi J. Heyrovsk~ Institute of Physical Chemistry, Prague
I. Nekoksovfi J. Heyrovsk~ Institute of Physical Chemistry, Prague
J. Klisfikovfi J. Heyrovsk3i Institute of Physical Chemistry, Prague
Z. Pavla6kovfi J. Heyrovsk2~ Institute of Physical Chemistry, Prague
J. Pawlesa J. Heyrovsk~ Institute of Physical Chemistry, Prague
P. Topka J. Heyrovsk~ Institute of Physical Chemistry, Prague
D. Nachtigallovfi Institute of Organic Chemistry and Biochemistry, Prague
J. Ku~era Institute of Organic Chemistry and Biochemistry, Prague
J. Nedv6dickfi Orgit, Ltd.
Z. Ml~,nskfi Orgit, Ltd.
R. Halfar Orgit, Ltd.
L. Bla2 Orgit, Ltd.
In this paper we will contrast the exciting initial work of finding new molecular sieves and the
follow-up items needed to address moving forward with application development of a
material. Principally, to arrive in the marketplace, the economics must be attractive for
introducing new technology. A key hurdle in the production of high silica zeolites,
metallophosphates, metalloorganic frameworks, and many novel silicogermantes, is the
management of the costs concerning the guest organo-cations and amines used as structure
directing agents (SDA's). In this discussion we will describe several novel routes attempting
to create a positive impact on this issue.
1. I N T R O D U C T I O N
There has been a continuous trend in breakthroughs in creating new molecular sieve
materials. It is an exciting frontier area for material science and nanotechnology as specific
applications are envisioned. A recent report on the field shows the increase in discovered
materials and the corresponding patent activity [ 1]. Even the increased diversity in the number
of materials commercialized supports the quest for more new structures and compositions.
The contrast of the number of three-dimensional frameworks recognized by the International
Zeolite Association (IZA) and the subsequent commercial use runs about a 15 or 20-1 ratio.
Only a small percent has had a chance to see commercial use.
Equally vital to the eventual commercial use will be the synthetic challenges of
controlling crystallite size, possibly morphology, surface features and elemental composition.
Another changing variable in determining the opportunity to commercialize a new molecular
sieve can be the value of the end use. Here we can imagine something of a cost continuum
along the lines of:
In this brief contribution we will highlight a few chemistries that have been particularly useful
of late in finding novel structures. These will primarily be excursions into the use of novel
guest molecules (SDA's) coupled with novel inorganic chemistry, focused on the use of
silicate systems with B, F and Ge as possible co-factors. The phosphate based molecular sieve
syntheses had previously shown a great diversity in structures and compositions but in those
systems, it was less clear that the SDA introduced as much of an influence on space-filling
[2]. Even more recently metal-organic molecular sieves, developed around di and tri
carboxylate ligands and metal centers (termed MOF) have produced an impressive array of
structure types [3-5].
Moving on from initial structure discoveries, we will show several strategies designed
to lower the cost of the synthesis. As we discuss these approaches we will occasionally
interweave the effect some of these changes would have on the cost of producing zeolite SSZ-
33 (CON in the structure code terminology used by the IZA). This zeolite is an intergrowth of
two pure polymorphs and possesses both 10 and 12-ring pores. It is made as a borosilicate
sieve [6]. The end member structures are CIT-1 [7] and ITQ-24 [8]. The structural details are
in Fig. 1. An aluminosilcate version is also known via direct synthesis and is zeolite SSZ-26
[9]. However, this synthesis procedure requires several steps to produce the requisite
dicationic propellane SDA that specifies this product. This material is of interest in our
discourse of the more general problem of efficient synthesis developments because it has
recently been shown to have desirable properties relative to commercially available beta
zeolite as a hydrocarbon trap in automotive emission control technology. The recent work
came from the labs of Davis at Caltech and Okubo at the University of Tokyo [ 10].
We have been encouraged by the rich diversity of novel zeolite structures obtained through
the use of boron as a substituting trivalent element inserted into the framework during
synthesis of silicate guest using SDA hosts. Table 1 shows a representation of mostly large
pore zeolites we have discovered using the borosilicate chemistry. More than half can also be
made by direct synthesis with aluminum instead of boron, but there remains some that can
only be made from borosilicate systems. In the list of materials, the 12/10 SSZ-33 appears as
well as novel 14-ring zeolites SSZ-53 and 59, and the new intersecting 10-ring zeolite SSZ-
58. Multidimensional 10-ring zeolites have proven to be very rare finds in this chemistry
despite the early discoveries of ZSM-5 and 11 more than 30 years ago. References for these
materials are given in Table 1. In our recent study of the synthesis of a selected system, SSZ-
42 (IFR) showed the possibility that boron may be added from solution to the growing zeolite
Table 1
Synthesis of Large Pore Zeolites
crystal and that the amorphous silicate gel material, when separated early in the reaction does
not contain boron. This is a great contrast to aluminosilicate systems and we showed this for
the synthesis kinetics of beta zeolite using the same SDA as for borosilicate SSZ-42 [ 17]. This
may provide a key as to why some novel SDA produce new structures in the presence of
borosilicate inorganic systems, but not when the chemistry is aluminosilicate.
For many catalytic and separations applications it is desirable to have Bronsted acid
sites which are strong, as in the case of A1, and contrasted with the known weak acidity of
boron sites. Fortunately, when the zeolite pores are large (> 6 A), it is possible to replace the
boron with aluminum, within the lattice, once the guest SDA has been removed. Surprisingly,
we have been able to carry out this substitution/re-insertion chemistry under acidic conditions
Si Si Si
I I I
O H(~) O ?r~
I H ,.~j
Si--O--B--O--Si ~ Si-OH HO-Si "= S i - - O - - A I - - O - - S i
H I
o o
I I I
Si Si Si
"Nest"
K1
(AI+a(H20)6 + H(~
K2
(K1/K2<<1)
Fig.2. An advantage in synthesizing new large pore zeolites as borosilicates
with the Al+ cations kept soluble. Fig. 2 shows a schematic of such a reaction. This approach
has been used to convert SSZ-33 into a material with strong acidic centers for use in such
applications as the hydrocarbon trap [18].
The laboratory at the Instituto de Tecnologia Quimica at the Universidad Polyt6cnica
in Valencia, Spain, has pioneered the breakthrough of two great chemistry innovations in
molecular sieve synthesis. The first, going back nearly a decade and building on the
pioneering work of the groups of Flanigen [19] and Guth and Kessler [20], and others,
Professor Corma's group (in Valencia) explored the use of fluoride anions in synthesis. But
the great breakthrough came in looking at reaction systems with much lower-than-usual
amounts of water. The reactions were very concentrated. This approach led to the discovery
of a number of all-silica materials with low framework density and high microporosity. This
had rarely been seen in silicate synthesis chemistry. The fluoride provides an important
structure-determining factor under these concentrated conditions and it is found in the product
[21, 22]. A number of these new phases and or compositions are listed in Table 2.
Table 2
Some novel structures from use of HF in molecular sieve synthesis
Structure Details Reference
ITQ-1 All SiO2 M W W 23
ITQ-2 Exfoliated M W W 24
ITQ-7 Multidimensional large pore 25
1TQ- 12 Multidimensional small pore 26
ITQ- 13 Multidimensional 9, 10 rings! 27
ITQ-29 All SiO2 LTA 28
We followed this lead of Professor Corma's group and using a large array of SDA we had
developed we explored this reaction. Table 3 shows some of the novel silicate compositions
we obtained in this chemistry. A greater discussion of these findings is contained in ref 29.
Mainly we were finding that we could make some very open framework materials as all-silica
for the first time. This has some impact on the use of silicates in membranes and in electronic
materials [30].
Table 3
New all-silica phases
Phase Structure Code Dimensionality
SSZ-33 CON Multi; 12x 10 Rings
SSZ-50 RTH Multi; 8x8 Rings
SSZ-55 ATS 1D, 12
EU-I EUO Multi; 10 Rings with Pockets
SSZ-37 NES Multi; 10 Rings with 12-~ag pockets
SSZ-61 + At Least one Large Pore (TEM)
The second breakthrough came from using Germanium as a silica substituent, in the presence
of Fluoride, and then later without it. With this approach, and trying a variety of SDA the
Corma group discovered a number of novel frameworks were produced. A key feature is the
influence of Germanium to form double 4-rings within the structures [31]. This feature is
almost never seen for silica alone. Some of the materials employed by this route are shown in
Table 4. Amongst the highlights is the discovery of a FAU material with 6 portals into the
cage instead of 4 (ITQ-22), and the first ever report of LTA as all-silica (ITQ-29).
Table 4
Some novel structures from use of Ge in molecular sieve synthesis
3. A T T E M P T S TO M A K E Z E O L I T E R E A C T I O N R E Q U I R I N G SDA M O R E
A T T R A C T I V E F O R SCALE-UP
, ~ N t CH3
+-~CH3 0.02 SSZ-35
"~
F,
,-N+-,.I 0.02 ZSM- 11
0.03 ZSM-12
N+
-CHa
CH3 0.05 SSZ-28
~CH3
144
,~1o8 t IIJ
a.
0.9.
.~ 72
SSZ-47 micropore volume = 0.06
.~-~,§ § >,,,,,,NH2
"~,- II M._J
Fig. 4. X-ray diffraction patterns tbr SSZ-47 (top) and 47B (bottom)
§ "~0 '
Q
Fig.5. Proof of concept
An ancillary benefit for this approach will be that if the SDA can be removed by a
solution (acidic) treatment, then a calcination step is avoided. This can have important
consequences for the state of Al+3 sites in the zeolite as high temperature treatments can pop
some of it out of the framework [42]. A solution treatment could avoid this.
In some more recent work Lee has made some interesting perturbations in the recyclable
system. The SDA is combined with small, pore-filling amines. While this lowers the cost of
the SDA, it also makes it easier to gain access to the pore system for the fragmentation step in
the loop in Fig. 5 [43]. Lee has gone on to prepare a number of ketal derivatives (Fig. 6) and
finds that multi-dimensional zeolite products have an easier time in fragmentation of the SDA
as contrasted with some one-dimensional zeolites like VPI-8 (VET) which are of considerable
interest [44 ].
4. S O M E K E Y ISSUES IN T H E T R A N S L A T I O N TO M A N U F A C T U R E
We use the term translation here because this is exactly what is needed. One takes the results
from discovery, synthesis, testing and characterization successes and then asks whether it can
be operable in a world of different parameters. Let's look at these translations by dividing our
problem into several linear segments.
/-5
If one develops an SDA not previously commercialized, there will be permit and testing
issues to be handled. These require both time and money. Even if we pass those checkpoints
there is the need to verify plant compatibility including issues of worker exposure and
corrosivity towards equipment.
5. S U M M A R Y
In this short review we have attempted to point out some of the new areas where silicate
based molecular structures are emerging. After doing so we concentrated on a range of
approaches to lower the cost of bringing a material to the marketplace by making cost-impact
changes concerning the most expensive component of the system, the SDA. Finally we
touched upon the changes in concerns as the scales become large for the synthesis. We
acknowledge the highly important contributions to this last step made by the chemists and
engineers who solve these problems behind closed doors in industry.
REFERENCES
[1] S. Auerbach, K. Carrado, P. Dutta, Handbook of Zeolite Technology (Marcel Dekker) 2003.
[2] E.M. Flanigen, B.M. Lok, R.L. Patton, S.T. Wilson, Proc. 7th Int. Zeol. Conf., Tokyo (1986)
103.
[3] O. Yaghi, Science, 295 (2002) 469.
[4] M. Sanselme, G.M. Greneche, M. Riou-Cavellec, G. Ferey, Solid State Sciences, 6 (2004) 853.
[5] A.K. Cheetham, G. Ferey, T. Loiseau, Angew. Chemie Intl. Ed., 38 (1999) 3268.
[6] R.F. Lobo, M. Pan, I.Y. Chan, H-X. Li, R.C. Medrud, S.I. Zones, P.A. Crozier, M.E. Davis,
Science (1993) 1543.
[7] R.F. Lobo, M. E. Davis, J. Am. Chem. Soc., 117 (1995) 3764.
[8] A. Corma, J. Am. Chem. Sot., 125 (2003) 7820.
[9] S.I. Zones, M.M. Olmstead, D.S. Santilli, J. Am. Chem. Soc. 114 (1992) 4195.
10
[10] S.P. Elangovan, M. Ogura, M.E. Davis, T. Okubo, J. Phys. Chem. B 108 (2004) 13059.
[11 ] http://www.iza-strucmre.org/databases/
[12] R.F. Lobo, M. Tsapatsis, C.C. Freyhardt, I. Chan, C.Y. Chen, S.I. Zones, M. Davis, J. Am.
Chem. Soc., 119 (1997) 3732.
[13] M.M. Helmkamp, M.E. Davis, Ann. Rev. Mater. Sci., 25 (1995) 161.
[14] S.]. Zones, D.S. Santilli, US Patent 5,656,149 (1997).
[15] G.S. Lee, Y. Nakagawa, S.I. Zones, U.S. Patent 20020027935 (2001).
[16] S. Elomari, US Patent 20020081262 (2002).
[17] S.I. Zones, S-J. Hwang, Microporous Mesoporous Mater., 58 (2003) 263.
[18] C.Y Chen, S.I. Zones, Proc. Intl. Zeolite Assoc., Montpellier, France (Elsevier, Amsterdam)
paper 11_P- 16 on CD-ROM (2001).
[19] E.M. Flanigen, R.L. Patton, US Patent 4, 073,865 (1978).
[20] J-L.Guth, H. Kessler, R. Wey, Stud. Surf. Sci. Catal., 28 (1996) 121.
[21 ] A. Corma. M.J. Diaz-Cabanas, J. Martinez-Triguero, F. Rey, J. Rius, Nature, 418 (2002) 514.
[22] G. Sastre, J. A.Vidal-Moya, T. Blasco, J. Rius, J.L. Lorda, M.T. Navarro, F. Rey, A. Corma,
Angew. Chem. Intl. Ed., 41 (2002)4722.
[23] M.A. Camblor, A. Corma, M.J. Diaz-Cabanas, J. Phys. Chem. B, 102 (1998) 44.
[24] A. Corma, V. Fomes, S.B. Pergher, T.L.M. Maesen, J.G. Buglass, Nature, 396 (1998) 353.
[25] http://www.iza-structure.org/databases/, + "Atlas of Zeolite Framework Types", + ~'ISV".
[26] P.A. Barrett, T. Boix, M. Puche, D.H. Olson, E. Jordan, H. Koller, M.A. Camblor, Chem.
Comm. (2003) 2114.
[27] A. Corma, M. Puche, F. Rey, G. Sankar, SJ. Teat, Angew. Chem. Int. Ed., 42 (2003) 2702.
[28] A. Corma, F. Rey, J. Ruis, M.J. Sabater, S. Valencia, Nature, 431 (2004) 287.
[29] S.I. Zones, S-J. Hwang, S. Elomari, I. Ogino, M.E. Davis, A.W. Burton, Compt. Rend.-Chem.
(2005) in press.
[30] A. Mitra, Z. Wang, T. Cao, H. Wang, L. Huang, Y. Yan, J. Electrochem. Soc., 149 (2002) B472.
[31 ] A. Corma, Proc. 14th International Zeolite Conf., Cape Town. South Africa, Plenary Lecture
(2004).
[32] A. Corrna, M.J. Diaz-Cabanas, F. Rey, S. Nicolopoulus, K. Boulahya, Chem. Comm. (2004)
1356.
[33] J.-L. Paillaud, B. Harbuzaru, J. Patarin, N. Bats, Science, 304, (2004) 990.
[34] A. Corma, M.T. Navarro, F. Rey, J. Rius, S. Valencia, Angew. Chem. Int. Ed., 40 (2001) 2277.
[35] A. Corma, M.J. Diaz-Cabanas, J. Marinez-Triguero, F. Rey, J. Rius, Nature, 418 (2002) 514.
[36] A. Corma, F. Rey, S. Valencia, J.L. Jorda, J. Rius, Nat. Mater., 2 (2003) 493.
[37] S.I. Zones, S-J. Hwang, M.E. Davis, Chem. Eur. J., 7 (2001) 1990.
[38] S.I. Zones, S-J. Hwang, Chem. of Mater., 14 (1) (2002) 313.
[39] S.I. Zones, Microporous Mater., 2 (1994) 281.
[40] R.M. Barrer, Hydrothermal Chemistry of Zeolites, Academic Press, (1982).
[41 ] H. Lee, M.E. Davis, S. I. Zones, Nature, 425 (2003) 385.
[42] H-K. Timken, O. Kuehl, Microporous Mesoporous Mater., 35-36 (2000) 521.
[43] H. Lee, M.E. Davis, S. I. Zones, J. Phys. Chem. B, 109 (2005) 2187.
[44i H. Lee, Ph.D Thesis, California Institute of Technology, 2005.
[45] J. Casci, Proceedings from the 2004 Summer School on Zeolites, "Zeolite Molecular Sieves:
Preparation and Scale-up", Microporous and Mesoporous Materials, (2005) in press.
[46] http://www.chevron.com/prodserv/refiningtechnology/zeolite_plus_8a.shtm
Studies in Surface Science and Catalysis, volume 158
J. Cejka, N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 11
The fast synthesis of zeolite L (LTL type structure) from a clear precursor solution (K20-
A1203-SiOz-H20) under microwave (MW) irradiation is described. The crystallization time of
LTL zeolite under MW was decreased substantially in comparison to conventional
hydrothermal heating, i.e., to 30 and 10 minutes at synthesis temperatures of 170 ~ and 190 ~
respectively. The degree of crystallinity is as good as from conventional heating method and
the size of particles is smaller. However the particle size distribution is increased with
escalating the synthesis temperature. The crystallization process has been followed with
various methods including XRD, DLS, TEM, nitrogen sorption, etc.
1. INTRODUCTION
A variety of crystalline framework type materials with pore diameters in the range of 0.3-1.2
nm have been prepared over the last few years by conventional hydrothermal treatment. In
addition, periodic mesoporous materials have been synthesized via self-assembly of long-
chain surfactant molecules and inorganic matrix components under quasi-ambient conditions.
However, there are only few ordered micro- and mesoporous materials prepared from
colloidal precursor solutions resulting in the formation of nanosized crystals with
hydrodynamic diameters of 10-100 nm [1-9].
The recent trend to develop rational synthetic strategies for preparation of nanosized
porous crystals with variable chemical compositions including pure silicates, aluminosilicates
and aluminophosphates is illustrated by the extensive work on the synthesis from
supersaturated precursor solutions containing always organic additives under specific
synthesis conditions, i.e., long crystallization time at room temperature (RT) or short
crystallization at elevated temperature under microwave radiation. The 'microwave-effect' is
recognized with the uniform and increased heating rate of the reaction mixture and enhanced
dissolution of the initial compounds leading to spontaneous nucleation and formation of
nanosized crystals with equal particle sizes. Via applying microwave heating the possibility to
synthesize zeolites for short crystallization times compared to classical synthesis route has
been demonstrated. Several theories describing the effect of microwave heating on the zeolite
crystallization process are discussed and being reviewed in the literature [ 10], however the so
called 'microwave-effect' is mostly attributed to the uniformed heating and increased heating
rate of the reaction mixture during the first several minutes. In addition, the superheating and
hot spots in the reaction mixtures may explain the faster crystallization of the zeolites.
Nevertheless, only some of the microporous materials have been synthesized under
microwave radiation.
12
It is well known that the synthesis route for preparation of zeolites and related
materials is govern by the formation of small amorphous entities in the precursor solutions or
gels. The type of the initial precursor solution and the conditions for hydrothermal treatment
are responsible for a high nucleation and crystallization rates, resulting in the formation of
stable nanosized particles with zeolite crystalline structures. By applying microwave heating
the microporous materials with structures such as EDI, LTA, MFI, FAU, BEA, AFI and AEI
have been prepared within several minutes [ 11-15].
The interest in the synthesis of LTL type zeolite is based on its wide applications in
conventional catalytic processes such as aromatization of n-hexane, chlorination of aromatic
organic molecules, etc. [16,17]. On the other hand, zeolite L is also used as hosts for the
incorporation of dyes and especially for light harvesting systems [18]. For most of these
applications the size and shape of the zeolite crystals as well as the high crystalline yield are
very important [ 19].
In this work we report the rapid synthesis of nanosized LTL zeolite using microwave
heating that leads to the formation of fully crystalline particles in the nanometer range.
Additional investigations concerning the crystallization process were performed, where the
temperature and time were varied leading to formation of nanoparticles with mono- versus
poly- modal particle size distribution.
2. EXPERIMENTAL
The initial precursor solutions used for further microwave treatment have the following
chemical composition: 5 K20 : 10 SiO2 : 0.5 A1203 : 200 H20. The solutions have been
prepared via dissolving 40.0 mmol aluminum hydroxide (3.9 g AI(OH)3, type A2100, 80 wt.
% A1203, 20 wt. % H20) and 0.4 mol potassium hydroxide (23.3 g KOH, Aldrich) in 86.0 g
double distilled (dd) water, and to this solution 0.4 tool colloidal silica (83.1 g SiO2 Ludox
SM30, Aldrich) was added. The resulting mixtures were stored for one day on an orbital
shaker and additionally were stirred for one more day to obtain water-clear precursor
solutions.
All syntheses are performed in a QWave 3000 Questron microwave oven (1200 w);
six Teflon lined autoclaves were filled with 25 ml initial precursor solutions. After the
microwave heating all samples were purified using high-speed centrifugation (60 min, 20 000
rpm, 3 times) and then the particles were redispersed in distilled water using an ultrasonic
bath. For simplicity, the samples abbreviated with L170-n' (n = 20, 30 and 90) are
corresponding to LTL samples exposed to MW irradiation at 170 ~ for 20, 30 and 90
minutes, respectively. Samples L190-n' (n-- 5, 10 and 40) are corresponding to LTL zeolite
synthesized at 190 ~ for 5, 10 and 40 minutes.
Crystallinity and purity of the samples were determined with a STOE STADI-P X-ray
diffractometer using Ge monochromated Cu K~ radiation. The change in the particle size of
the zeolites during crystallization process was determined with dynamic light scattering
(DLS) using a Malvern Zetasizer (Nano ZS) instrument in backscattering geometry; the
surface charge of the particles and their stability were determined from the zeta-potential
measurements. The transformation process of the amorphous into crystalline LTL particles
was followed by HRTEM using a JEOL T2100 microscope. The chemical compositions of
the final crystalline samples were determined based on inductively coupled plasma-atomic
emission spectrometry (ICP-AES). In addition, the porosity of the samples is proved by the
nitrogen sorption measurements performed with a Quantachrome Nova 4000e at liquid
nitrogen temperature.
13
The template- free alumininosilicate solutions were subjected to microwave heating at 170 ~
and 190 ~ in order to study the influence of both the type of heating and synthesis
temperature on the main characteristic of LTL zeolites such as particle size, polydispersity,
stability, and porosity. The crystallinity of the samples was confirmed by the XRD patterns
collected from purified powders after heating for different times in the MW oven. As can be
seen from Fig. l a, the amorphous aluminosilicate matter has been converted into LTL zeolite
within 30 min at 170 ~ no additional phase after longer time of microwave treatment was
formed. The intensity, position and width of the all reflections for samples L170-30' and
L 170-90' are nearly the same. Even faster crystallization of LTL zeolite was observed at 190
~ (Fig. l b). After five minutes of MW heating a broad Bragg reflection at 5.4 ~ 20 is
appeared similar to the L170-20' sample heated for 20 minutes. However, fully crystalline
zeolite particles were obtained for 10 rain at 190 ~ Further MW treatment for 40 min did
not lead to any change in the intensity of the Bragg reflections in sample L 190-40' compared
to sample L 190-10'.
400 400-
300 300-
- "~
ra~
= 200 200-
o
5 lO 15 20 25 30 35 40 45 ' 1'0' 1 ' 5 ' 2 ' 0 ' 2 ' 5 ' 3 ' 0 ' 3 ' 5 ' 4 ' 0 45
20/o 20/o
Fig. 1. XRD patterns of LTL samples synthesized at (a) 170 ~ and (b) 190 ~
The degree of crystallinity is an important factor for the zeolite nanoparticles in the
final crystalline suspension. Moreover the particle size distribution in the colloidal
suspensions during the entire crystallization process is a crucial point for the formation of
zeolites with a different degree of polydispersity. DLS study has been performed for all
samples subjected to MW treatment for various times and compared with the XRD data
revealing the ratios between the amorphous and crystalline matters. The size of the X-ray
amorphous particles in the initial solutions prior to crystallization varied from 40 nm to 180
nm. Approximately the same size of the particles is measured for the samples synthesized at
170 ~ (see Fig. 2). As can be seen the sample L 170-20', which is mainly X-ray amorphous
contains particles with a mean size of about 140 nm, and this value did not change
significantly for the fully crystalline sample L 170-30'. The final particle size of LTL zeolite is
about 90 nm in diameter (Fig. 2a). Similar observations where made for the LTL prepared at
190 ~ (Fig. 2b). Both sample L190-5' that is X-ray amorphous and the fully crystalline
sample L190-10' have nearly the same mean particle size o f - 2 1 0 nm. It is decreased
negligibly up to 170 nm under prolonged heating i.e. 40 min (see sample L190-40'). This
14
indicates that the slow dissolution of the particles takes places during the prolonged MW
treatment.
The DLS results reveal that LTL zeolite synthesized at 170 ~ have smaller crystals
with low degree of polydispersity in comparison to samples prepared at 190 ~ Similar
observation has been made for LTL synthesized in conventional oven where the
crystallization process took almost 4 days [9].
The surface charge of the individual amorphous particles formed in the solution is
important for their stability during the entire synthesis process. DLS is used to resolve the
electrophoretic mobility of the particles in the solutions. By applying the Huckel equation the
zeta potential has been determined, and the magnitude of the zeta potential value corresponds
to the stability of the colloidal particles in the suspensions. If the zeta potential is high (> + 30
mV), the particles do not tend to agglomerate since the repulsive forces in-between the
individual crystals are dominant. The zeta potential measurements were performed for all
solutions with pH=7 after washing and redispersion in dd H20, while for the initial precursor
solution the zeta potential was measured before (pH=14) and after washing (Fig. 3).
(a) ib)
o,,.~
.l...a
L 190-40'
L 170-90'
L170-20' L190-5'
. . . . . . . . | . . . . . . . . | .... , . . . . . . . ..... ' . . . . . . . i . . . . . . . . | 9
Fig. 2. DLS data of LTL samples synthesized at (a) 170 ~ and (b) 190 ~
The zeta potential value of the as prepared solution is varied f r o m - 190 mV to + 170
mV, which is due to the presence of various amorphous aluminosilicate particles and excess
of potassium hydroxide (Fig. 3a). The maximum of the zeta potential curve is at about zero,
while in the purified solution the value is - 50 mV (Fig. 3b), which is reflecting the negative
surface charge of the particles present in the initial precursor solution at pH=7. The negative
surface charge of the amorphous particles stabilized in H20 is conterminous with repulsive
forces between the individual particles that explain their stability. An agglomeration of the
particles was not observed for a period for at least three months. The zeta potential values of
the fully crystalline samples L170-30' and L190-10' obtained after complete consumption of
the amorphous particles were measured; the values for the crystalline LTL nanoparticles are
always in the negative range, i.e.,- 50 + - 30 inV. However the LTL nanocrystals prepared at
170 ~ have more negative zeta potential value and this probably explains the small particle
size of the crystals synthesized at this temperature and lower degree of agglomeration. The
zeta potential values of the completely crystalline LTL particles are similar to the one
measured for particles in purified initial precursor solution demonstrating high stability and
15
high electrostatic forces of repulsion between the individual nanoparticles. However the
particles synthesized at 170 ~ have most negative zeta potential value in comparison with
those synthesized at 190 ~ This might me due to the chemical composition of the different
samples. Although the Si/A1 ratio in the precursor solution was 10, then for the extracted
amorphous particles from the solutions prior to MW heating was measured to be 3, while the
K/A1 ratio was shifted from 10 to 1.56 from non-purified to purified amorphous particles.
The low potassium amount in the initially formed amorphous particles demonstrates that the
excess of potassium is removed after purification, and the high amount of KOH is needed
only to adjust the pH of the solution and fully dissolve the initial reagents. However, it is
interesting to note that the chemical composition of purified amorphous particles is
comparable with that for the fully crystalline LTL zeolite. This result suggests that the
amorphous agglomerates have not only the same particle size but also the same chemical
composition as the crystalline LTL zeolite nanoparticles. The ratios Si/AI and K/A1 were
measured for the particles extracted from amorphous precursor solutions, for all intermediates
having different degree of crystallinity and for the completely crystalline samples. It has been
found that the Si/A1 ratio is decreased from 2.86 to 2.55, and the K/A1 from 1.38 to 1.14 for
samples L170-20' to L170-40', respectively. It has been also observed that the synthesis
temperature does not affect substantially on the Si/A1 ratio in the crystalline LTL zeolite
particles.
(c) (d)
9 v,,,,~
Fig. 3. Zeta potential measurements in initial precursor solution before (a) and (b) after
purification, and in purified samples (c) L-170-30' and (d) L-190-10'.
The nitrogen sorption data for the amorphous, intermediates and all crystalline samples were
collected and depicted in Fig. 4. With increase the degree of crystallinity, the microporosity
and the specific surface area of the samples are enlarged, i.e. the BET surface area calculated
for the amorphous particles, L190-5' and L-190-40' were 237 m2/g, 275 m2/g and 345 m2/g,
respectively. Additionally, t-plot analysis of the nitrogen sorption data is performed; the shape
of the t-plot curves where the adsorption isotherm is plotted in terms of the volume of gas
adsorbed v s the statistical thickness of the adsorbate can be used to distinguish between
microporous, non-porous and bi-modal micro/mesoporous solids (Fig. 4b). The extrapolation
of these curves to y-axis gives volume occupied by the micropores. The samples L190-5' and
L-190-40' show positive intercepts while the amorphous sample shows no microporous
16
"7
t-q
~o 300 ~0 60-
200. ~ 40-
20-
100
0-
, ,
Fig. 4. (a) Nitrogen sorption isotherms and (b) T-plot data from purified amorphous particles, and
samples L 190-5' and L- 190-40'.
The crystallization process of LTL zeolite was followed with TEM, where was
possible to distinguish between amorphous and crystalline areas in the individual particles
formed spontaneously after mixing of all compounds. The dried amorphous nanoparticles in
the precursor solution prior to MW treatment do not show any crystalline structures. Under
MW heating, the amorphous nanoparticles do not change their morphology, however, some
crystalline fringes that correspond to the LTL type structure occur (Fig. 5a). As can be seen
the starting points for the nucleation are located inside of the amorphous agglomerates as well
as at the periphery. The distance between the crystalline fringes is comparable with the
channel diameter of LTL zeolite (-7.1 A). Similar observations have been reported for
zeolites synthesized from clear precursor solutions, where the nucleation has been observed in
the middle and at the solid-liquid interface for LTA and FAU nanozeolite, respectively
[20,21]. In addition, bigger agglomerates with separate parts having amorphous and
crystalline appearances are recognizable in some of the intermediate samples (see Fig. 5b).
The high degree of condensation/densification is observed in the aluminosilicate particles
obtained after prolonged heating, where a total transformation of the amorphous particles into
single and polycrystalline LTL zeolite is visible. The amorphous layers at the grain boundary
between the individual LTL nanoclusters or amorphous shells have been seen only in the
samples heated for very short time. In addition, the fully crystalline LTL samples synthesized
at 190 ~ consist of many aggregates composed of single nano-particles. The individual
crystals have very rigid morphological appearance and cannot be described with one defined
shaped. Under prolonged crystallization, the small nanoparticles are further growing and also
formed bigger entities in which some of the crystalline fringes are well aligned or very
randomly distributed.
17
Fig. 5. HRTEM images of (a) L 190-5', (b) L 190-10' and (c) L 170-30' and (d) L 190-40'.
The silica amorphous particles together with the aluminum and potassium are forming
the amorphous agglomerates with a size of about 50-200 nm. After a short MW treatment, the
first crystalline LTL zeolite nanoclusters are appearing in the amorphous aggregates having a
size in the range of 5-10 nm. Further MW heating leads to the formation of bigger crystalline
particles composed of partially aligned crystalline fringes and amorphous boundaries
between. After prolonged MW heating almost a total transformation from amorphous into
crystalline matter is observed. At a temperature of 170 ~ mainly single crystals are obtained,
whereas at 190 ~ the particles composed of nanosized multi-clusters are formed. The Si/A1
and K/A1 ratios in the amorphous and crystalline particles are very similar, however the
excess of potassium is needed in order to keep the pH of the solution high and all the
compounds dissolved in the clear solutions during the entire crystallization process, and to
prevent the fast agglomeration and sedimentation of big particles.
18
4. C O N C L U S I O N S
Highly crystalline LTL zeolite was successfully synthesized via microwave heat treatment
using potassium - aluminosilicate clear precursor solutions at temperatures 170 ~ and 190 ~
for 30 rain and 10 min, correspondingly. Very fine and uniform single crystals are produced
that appear to aggregate into polycrystals under prolonged crystallization time. The LTL
zeolite synthesized at 170 ~ has a narrow particle size distribution, and the increase of the
microwave heating temperature yields crystals with different polycrystalline morphologies.
However, the mechanism of crystallization for LTL zeolite prepared in conventional and
microwave ovens is the same, and only the crystallization kinetics is different, i.e., the
microwave heating decreases the time of synthesis in respect to the traditional heating under
the same conditions. The reduction of the synthesis time was due to the uniformity of the
heating stage by microwaves.
ACKNOWLEDGEMENT
his work was supported by the bilateral DFG/CNRS, BFHZ and Procope (DAAD).
REFERENCES
[1] B.J. Schoeman, J. Sterte, J.E. Otterstedt, Zeolites, 14 (1994) 208.
[2] J. Kecht, B. Mihailova, K. Karaghiosoff, S. Mintova, T. Bein, Langmuir, 20 (2004) 5271.
[3] A.E. Persson, B.J. Schoeman, J. Sterte, J.E. Otterstedt, Zeolites, 14 (1994) 557.
[4] A.E. Persson, B.J. Schoeman, J. Sterte, J.E. Otterstedt, Zeolites, 15 (1995) 611.
[5] Z. Guangyu, J. Sterte, B.J. Schoeman, Chem. Mater., 9 (1997) 210.
[6] S. Mintova, N. Petkov, K. Karaghiosoff, T. Bein, Mater. Sci. Eng., C19 (2002) 111.
[7] B.J. Schoeman, J. Sterte, J.E. Otterstedt, Zeolites, 14 (1994) 110.
[8] S. Mintova, M. Reinelt, T.H. Metzger, J. Senker, T. Bein, Chem. Commun., (2003) 326.
[9] M. Tsapatsis, M. Lovallo, T. Okubo, M.E. Davis, M. Sadakata, Chem. Mater., 7 (1995) 1734.
[10] C.S. Cundy, Collect. Czech. Chem. Commun., 63 (1998) 1699.
[11 ] X. Querol, A. Alastuey, A. Lopezosler, F. Plana, M.J. Andres, R. Juan, P. Ferrer, C.R. Ruiz,
Environ. Sci. Technol., 31 (1997) 2527.
[12] a) L. Banaccorsi, E. Proverbio, Mater. Res. Innovat., 8 (2004) 53; b) Y. Han, H. Ma, S. Qiu, F.-
S. Xiao, Microporous Mesoporous Mater., 30 (1999) 321.
[13] a) A. Arafat, J.C. Jansen, A. R. Ebaid, H. van Bekkum, Zeolites, 13 (1993) 162; b) O.G.
Somani, A.L. Choudhari, B.S. Rao, S.P. Mirajkar, Mater. Chem. Phys., 82 (2003) 538; c) L.
Banaccorsi, E. Proverbio, Mater. Res. Innovat., 8 (2004) 53.
[14] H. Katsuki, S. Furuta, S. Komarneni, J. Porous Mater., 8 (2001) 5.
[15] D.S. Kim, J.-S. Chang, J.-S. Hwang, S.-E. Park, J.M. Kim, Microporous Mesoporous Mater., 68
(2004) 77.
[16] R.M. Barrer, H.Z. Villinger, Z. Krsitallogr., 128 (1969) 352.
[17] a) G. Jacobs, W.E. Alvarez, D.E. Resasco, Appl. Catal. A: General, 206 (2001) 267; b) J.-L.
Dong, J.-H. Zhu, Q.-H. Xu, Appl. Catal. A: General, 112 (1994) 105; c) M.J. Den Exter, H. van
Bekkum, C.J.M. Rijn, F. Kapteijn, J.A. Mailijn, H. Schellevis, C.I.N. Beenakker, Zeolites, 19
(1997) 13.
[18] M.C. Hausladen, R.C. Cyganovich, H.Y. Huang, C.R.F. Lund, Appl. Catal. A: General, 219
(2001) 1.
[19] a) G. Calzaferri, S. Huber, H. Maas, C. Minkowski, Angew. Chem. Int. Ed., 42 (2003) 3732; b)
H. Maas, A. Khatyr, G. Calzaferri, Microporous Mesoporous Mater., 65 (2003) 233; c) M.
Yatskou, M. Meyer, S. Huber, M. Pfenniger, G. Calzaferri, Chem. Phys. Chem., 4 (2003) 567.
[20] S. Mintova, N.H. Olson, V. Valtchev, T. Bein, Science, 283 (1999) 958.
[21] S. Mintova, N.H. Olson, T. Bein, Angew. Chem. Int. Ed., 38 (1999) 3201.
Studies in Surface Science and Catalysis, volume 158
J. t2ejka, N. 2;ilkov~iand P. Nachtigall (Editors)
9 2005 Publishedby Elsevier B.V. 19
Zeolite MCM-22 can form via a layered intermediate, which combines the potential of two
classes of materials: zeolites and layered solids. Various discrete products with different
arrangement of the MCM-22 monolayers have been recognized. A comprehensive overview
of the resultant MCM-22 family is presented with the emphasis on synthetic routes and
characterization, especially by X-ray diffraction, which allows identification of the various
family members. One of the discussed materials, MCM-56, is identified as comprising a
disordered collection of MCM-22 monolayers. It is the only delaminated zeolite available via
a convenient one-step hydrothermal synthesis. MCM-56 may be a substrate for novel zeolitic
materials with designed and tailored structures.
1. INTRODUCTION
2. EXPERIMENTAL
Hydrothermal syntheses of the MCM-22 materials with sodium hydroxide as base and
hexamethyleneimine (HMI) as template were carried out in stirred stainless steel autoclaves
under autogeneous pressure. The details of synthesis and characterization procedures have
been described elsewhere [3-8]. Typical MCM-22-P synthesis mixture has the following
molar composition: Si/A1 =15, OH/Si=0.18, H20/Si = 45 and HMI/Si=0.35.
20
10-ring sinusoidal
12-ring surfac / intra layer po res
pocket ~ ~
Potential ~ ~ ~ ~ ~ ~
T OH s~te
Fig. 1. Structure of the MCM-22 monolayer (top; side view) and its simplified representation
showing the pore system and potential terminal TOH and interlayer bridging sites.
9" ~
:~ ABd
roa
v i
I As-synthesized
oco. o
Calcined
5 10 15 20 25 30
2-theta (o)
Fig. 2. XRD patterns and schematic structures of MCM-22-P, MCM-49 and MCM-56. The (hkl)
Miller indices of selected reflections are provided.
Hydrothermal synthesis of zeolite MCM-22 with sodium and hexamethyleneimine can result
in two related but different end products, or their mixtures, depending on Si/A1 ratio and
21
basicity [3-5]. Higher Si/A1 favors a layered material, MCM-22 precursor (MCM-22-P), with
a c lattice parameter---27 N. It consists of MCM-22 monolayers stacked and aligned in
registry with hydrogen bonding between surface silanol groups. The precursor condenses into
the 3-dimensional MCM-22 framework, designated MWW, upon calcination, which is
accompanied by contraction of c to around 25 A. The second possible end product, MCM-49,
is the fully connected 3D zeolite framework (MWW) formed directly at lower Si/A1 ratio of
the synthesis mixture. Its lattice constants and consequently the XRD pattern are similar to
those of calcined MCM-22. As shown in Fig. 2, both as-synthesized products (MCM-22-P
and MCM-49) can be readily distinguished based on X-ray diffraction patterns.
The direct assembly of the MWW framework as MCM-49 proceeds via a distinct
crystalline intermediate designated MCM-56 [7]. It is also layered and based on detailed
characterization (see below) MCM-56 is identified as agglomerate of MCM-22 monolayers
without in-registry alignment in the c direction and no interlayer links.
MCM-56 can be recognized as a discrete MCM-22 family member based on its unique
XRD pattern (see Fig. 2). In the low angle region it resembles MCM-49 by having one
distinct peak at 6-7.5 ~ 20 due to coincidental overlap of (100) and (002) reflections. In
contrast, for MCM-22-P the expanded unit cell results in (002) reflection located at 6.5 ~ 20
and resolved from the (100) peak at 7.1 ~ 20. In the region around 26-29 ~ 20 MCM-56 looks
like MCM-22-P with only one distinct peak with a smooth descending tail, as shown in Fig. 4
in Ref. [5]. MCM-49 has 3 peaks with explicit (hkl) indices. The most significant feature of
the MCM-56 XRD pattern is the peculiar broad peak stretching from 8 to 10~ 20 (will be
referred to as a 'band'). It replaces the resolved (101) and (102) interlayer reflections seen in
MCM-22-P and MCM-49 and is attributed to lack of order in the corresponding c axis
direction. This 'band' is crucial for identification of MCM-56 as a delaminated zeolite, which
is shown in the section 1.4 below.
MCM-56 will be discussed further after presenting two topics concerning the MCM-
22 family, i.e. preparation of the pillared derivative MCM-36 by exfoliation of the layered
precursor and catalytic activity from the perspective of the MCM-22 monolayer structure.
rl?
MCM-56
Time
,,.-~
~o
tD
5 10 15 20 25 30
2-Theta(~
Fig. 4. XRD patterns during MCM-56 synthesis and the onset of conversion to MCM-49.
3.5. MCM-56 - p o t e n t i a l as a c a t a l y s t
MCM-56 can be processed and used as a catalyst in a manner similar to other zeolites.
Often such standard approach does not result in MCM-56 showing advantage in comparison
to MCM-22 and MCM-49. However, when strong active sites on the surface and an open
structure may play an important role in a catalytic process MCM-56 is found desirable. An
25
example to this effect is observed with the ethylbenzene synthesis in liquid phase, where
MCM-56 showed higher activity than both MCM-22 and MCM-49 [22].
Compared to other zeolites MCM-56 offers new opportunities but one must be aware
of accompanying challenges and pitfalls. Typical zeolites have majority of active sites inside
crystals, which provides a degree of protection and reduced sensitivity to the outside
environment. MCM-56's unique activity potential is associated with the external surface,
which is quite vulnerable to blockage (by detritus or layer overlap) or degradation, such as
increased dealumination. In order to take advantage of the unique attributes of MCM-56 may
require new post-synthesis processing methodology.
The biggest opportunity seems to lie with the post-synthesis structural and
compositional modification of MCM-56. As indicated, conventional zeolites are not
particularly amenable to structural improvements once crystallization is complete. For MCM-
56 the manipulation of orientation and spacing of monolayers offers large pore and/or
tailored structures with strong acid sites. The pillared MCM-36 [8,23] and the delaminated
derivatives [13,14] illustrate this but other more refined and possibly better designed
materials can be envisioned. Other zeolites, such as ferrierite, have been found to form
layered precursors too. They can be delaminated, swelled and pillared, resulting in a zeolite
family similar to MCM-22 [24]. However, MCM-56 remains unique as the only delaminated
zeolite obtained by a direct one-step synthesis.
4. CONCLUSIONS
REFERENCES
[1] (a) T.F. Degnan, Jr., C.M. Smith and C.R. Venkat, Applied Catalysis, A: General 221 (2001)
283. (b) J.S. Beck, A.B. Dandekar and T.F. Degnan, Jr., in M. Guisnet and J.-P. Gilson (eds.),
Zeolites for Cleaner Technologies, Imperial College Press, London, 2002, 223.
[2] J. (~ejka and B. Wichtlerovfi, Catal. Rev., 44 (2002) 375.
[3] M.K. Rubin and P. Chu, Composition of Synthetic Porous Crystalline Material, Its Synthesis
and Use, US Patent No. 4 954 325 (1990).
[4] M.E. Leonowicz, J.A. Lawton, S.L. Lawton and M.K. Rubin, Science (Washington, DC) 264
(1994) 1910.
[5] S.L. Lawton, A.S. Fung, G.J. Kennedy, L.B. Alemany, C.D. Chang, G.H. Hatzikos, D.N. Lissy,
M.K. Rubin and H.-K.C. Timken, J. Phys. Chem., 100 (1996) 3788.
[6] S.L. Lawton, M.E. Leonowicz, R.D. Partridge, P. Chu and M.K. Rubin, Microporous
Mesoporous Mater., 23 (1998) 109.
[7] A.S. Fung, S.L. Lawton, W.J. Roth, Synthetic Layered MCM-56, Its Synthesis and Use, US
Patent No. 5 362 697 (1994).
[8] W.J. Roth, C.T. Kresge, J.C. Vartuli, M.E. Leonowicz, A.S. Fung and S.B. McCullen, in H.K.
Beyer, H.G. Karge, I. Kiricsi and J.B. Nagy (eds.), Catalysis by Microporous Materials, Stud.
Surf. Sci. Catai. Vol. 94, Elsevier, New York, 1995, 301.
[9] S.M. Auerbach, K.A. Carrado and P.K. Datta, Hanbook of Layered Materials, Marcel Dekker,
New York, 2004.
26
[10] C.T. Kresge, J.C. Vartuli, W.J. Roth and M.E. Leonowicz, in O. Terasaki (ed.), Mesoporous
Crystals and Related Nano-structured Materials, Stud. Surf. Sci. Catal. Vol. 148, Elsevier, New
York, 2004, 53.
[11] W. J. Roth and J. C. Vartuli, in A. Sayari, M. Jaroniec and T.J. Pinnavaia (eds.), Nanoporous
Materials II, Stud. Surf. Sci. Catal. Vol. 129, Elsevier, New York, 2000, 501.
[12] W. J. Roth and J. C. Vartuli, in A. Sayari, and M. Jaroniec (eds.), Nanoporous Materials III,
Stud. Surf. Sci. Catal. Vol. 141, Elsevier, New York, 2002, 273.
[13] A. Corma, V. Fornes, S.B. Pergher, T.L.M. Massen and J.G. Buglass, Nature 396 (1998) 353.
[14] A. Corma, U. Diaz, V. Fornes, J.M. Guil, J. Martinez-Triguero, E.J. Creyghton, J. Catal., 191
(2000) 218.
[15] G.R. Meima, M.J.M. Van der Aalst, M.S.U. Samson, J.M. Garces and J.G.Lee JG, Jahrgang,
112(1996) 315.
[ 16] J.A. Horsley, Chemtech, October (1997) 45.
[17] H. Du and D.H. Olson, J. Phys. Chem. B, 106 (2002) 395.
[18] C.T.W. Chu, A. Husain, K.M. Keville and D.N. Lissy, Catalyst and Process for Isoparaffin-
Olefin Alkylation, US Patent No. 5 354 718 (1994)..
[19] W.J. Roth and D. N. Lissy, 13th Int. Cong. Catal., Book of Abstracts, p. 118 (P1-050).
[20] N. Xu, T. Wu, D. Wang, P. Wu, Q. Kan, J. Sun, Gaodeng Xuexiao Huaxue Xuebao, 22 (2001)
1898.
[21 ] G.G. Juttu and R.F. Lobo, Microporous Mesoporous Mater., 40 (2000) 9.
[22] J.C. Cheng, A.S. Fung, D.J. Klocke, S.LK. Lawton, D.N. Lissy, W.J. Roth, C.M. Smith and
D.E. Walsh, Process for Preparing Short-chain Alkylaromatic Compounds (with a Zeolite
Catalyst). US Patent No. 5 453 554 (1995).
[23] J.-O. Barth, A. Jentys, J. Komatowski and J.A. Lercher, Chem. Mater., 16 (2004) 724.
[24] A. Corma, U. Diaz, M.E. Domine and V. Fornes, J. Am. Chem. Soc., 122 (2000) 2804.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 27
A novel method for the direct synthesis of mesoporous Fe-MFI zeolite is reported. Mesopore
formation was observed over a whole range of as-synthesised Fe-MFI zeolites with varying
iron (0.075 - 0.6 wt%) and aluminium (0 - 1.1 wt%) concentrations. The main difference of
this method to conventional synthesis of Fe-MFI zeolites is that instead of adding iron salts
(eg Fe(NO3)3) to the solution gel, we prepared a synthesis gel starting from metallic iron foil
that was dissolved in nitric acid. Transmission electron micrographs show a broad mesopore-
size distribution in all as-synthesised Fe-MFI zeolite samples. The mesopores are retained
after calcination as well as after subsequent mild steaming.
1. INTRODUCTION
Fe-MFI has attracted increasing attention due to its ability to catalyse various reactions such
as the selective catalytic reduction (SCR) of NO• [1,2], the direct decomposition of NzO [3],
and recently the one-step hydroxylation of benzene to phenol using N20 as oxidant [4,5].
However, due to the microporous character of the zeolite, the presence of intracrystalline
diffusion limitations sometimes limits its catalytic performance. Due to the limited zeolite
micropore size (about 0.55 nm), even the reaction of small organic molecules is often
hindered by diffusion limitations. To improve the efficiency of zeolite-based catalysts,
diffusion-related problems can be minimised by introducing mesopores in the zeolite crystal.
These mesopores improve access to the zeolite pore system by increasing diffusion velocities.
A recently reported example of the use of such secondary mesopores is the improved
performance of mesoporous MFI zeolite in the catalytic cracking and isomerisation of n-
hexadecane [6].
Several methods have been investigated to alter the properties of zeolites to create
additional porosity without destroying their intrinsic properties. A conventional method is by
extraction of the tetrahedrally coordinated aluminium (dealumination) [7], from the zeolite
framework by steam treatment followed by acid leaching. However, the dislodgement of
framework aluminium unavoidably alters the acidic properties of the zeolite. A much better
28
2. EXPERIMENTAL
2.2. Characterisation
2.2.1. Elemental analysis
The elemental analysis of the as-synthesised materials was carried out by inductively coupled
plasma optical emission spectroscopy (ICP-OES), using a Perkin-Elmer Plasma 40 (Si) and
Optima 3000DV (axial).
Elemental analysis data for the as-synthesised Fe-MFI zeolites are given in table 1.
Comparison of the molar ratios for Si/Fe and Fe/A1 of the as-synthesised Fe-MFI samples to
their corresponding synthesis gels shows that the ratio as weighed in into the synthesis gel is
reproduced in the final Fe-MFI crystals. This indicates that our new synthesis method is
efficient as well as reproducible.
Table 1
Elemental analysis of the as-synthesized Fe-MFI zeolites
Si AI Fe Si/A! Si/Fe Si/AI Si/Fe
sample (w%) a (w%) a (w%) a (zeolite) b (zeolite) b (gel) c (gel) c
X-ray diffraction measurements [15] of these samples show that all as-prepared
zeolites have the MFI-type structure with a high degree of crystallinity. No other crystalline
phases are detected in the X-ray diffractograms.
SEM images of the as-synthesised materials show morphologies as expected for MFI-
type zeolites: almost spherical clusters of small rods and large twinned or multiple-twinned
crystals. A typical SEM image for each sample is shown in Fig. 1.
30
Fig. 1. SEM images of Fe-MFI. Top left: (1:4); top right: (1:8); bottom left: (1:16); bottom right: (1:0).
TEM shows that all as-prepared Fe-MFI samples contain a substantial amount of
mesopores with a broad pore size distribution in addition to the ordinary MFI pore system.
Next to zeolitic material, no amorphous debris was observed either in TEM or SEM
investigation. Fig. 2 shows a representative TEM image of the 1:4 sample. It contains well-
formed mesopores throughout the zeolite crystals. Fig. 3 shows a representative TEM image
of the 1:8 sample. It contains less well-formed mesopores. Fig. 4 shows a representative TEM
image of the 1:16 sample. This sample contains well-formed mesopores again. Fig. 5 shows a
representative TEM image of the 1:0 sample. Even the Fe-only (ie, Al-free) material contains
many well-formed mesopores. The mesopore size distribution is very broad for all samples,
with pores ranging up to 100 nm.
In Fe-containing zeolites, one of the main concerns is the prevention of formation of
crystalline clusters of iron oxide. These clusters are often formed upon severe steaming, as is
required to obtain mesopores. However, mild steaming is already sufficient to activate the
desired isolated iron species that give rise to catalytic activity. In our materials, the direct
generation of mesopores in the zeolite crystals upon zeolite synthesis means that severe
steaming to create mesopores is no longer necessary. A mild steaming treatment to activate
the isolated iron species suffices and clustering of iron into inactive crystalline iron oxide
nanoparticles is prevented.
31
.....,~..... ~..,
.---~-~" ....."~;.:~'~"::)i:.~;
:~..:;..,
~: ~:~.~.:.
.......
:!.iiI.~;::,
,,~::i:...
"i"~
.:,5..;i ~
i ~!~5 ;7
~i,~i~ ,,
i
~ .,
.. :
The presence of only isolated iron species in these samples without any discrete
crystalline nanoparticles has been proven using M6ssbauer spectroscopy [15]. The absence of
crystalline nanoparticles in these materials is now also shown using TEM at very high
magnification. Fig. 6 shows a representative TEM image of these materials. Only amorphised
zeolite is visible for our samples. This is due to the fact that high magnification TEM imaging
requires high electron beam intensity, which generally amorphises zeolite structures within
seconds. For conventionally prepared Fe-MFI zeolites, however, crystalline nanoparticles are
observed in high resolution TEM images [16]. This is due to the fact that occluded species
such as metal particles or, in this case, possibly present iron oxide crystallites are embedded
in the amorphous debris of the zeolite after amorphisation and are locked in at their original
location in the zeolite pore system.
All samples described in this paper have been subjected to calcination followed by
mild steaming. Both these treatments leave the pore system of the materials intact, opening up
a world of possibilities for improved catalytic performance. Detailed characterisation of the
calcined, H + and steamed forms of these materials is published elsewhere [ 17].
:.... !:~i,i::!:
' .:~.~,~.......
: .........:,.,~:.:,~:,.,~. ......~.i~
~..:,.. ~ . . . . . . .
........ ~: '.~::I~),,
~ ~,. ~%:..
i:
Fig. 6. Typical high magnification TEM image of Fe-MFI showing the absence of small crystallites of
iron oxide.
4. CONCLUSIONS
A new synthesis method for iron-containing MFI zeolite yields a hierarchical pore system of
large mesopores in addition to zeolite micropores by direct synthesis. As mesopores are
already present in the as-synthesised materials, standard calcination to remove the template
34
followed by mild steaming to activate the isolated iron species suffices to obtain catalytically
active materials. The previously required severe steaming in order to create mesopores to
overcome diffusion limitations during catalytic processes is no longer necessary and thus
sintering of iron species into iron oxide crystalline nanoparticles can be prevented.
REFERENCES
[1] A.A. Battiston, J.H. Bitter, D.C. Koningsberger, Catal. Lett. 66 (2000) 75.
[2] L.J. Lobree, I. Hwang, J.A. Reimer, A.T. Bell, Catal. Lett. 63 (1999) 233.
[3] F. Kapteijn, G. Marbfin, J. Rodriguez-Mirasol, J.A. Moulijn, J. Catal. 167 (1997) 256.
[4] G.I. Panov, Cattech 4 (2000) 18.
[5] A. Ribera, I.W.C.E. Arends, S. de Vries, J. P6rez-Ramirez, R.A. Sheldon, J. Catal. 195 (2000)
287.
[6] C.H. Christensen, I. Schmidt, C.H. Christensen, Catal. Comm. 5 (2004) 543.
[7] M. MfiIler, G. Harvey, R. Prins, Microporous Mesoporous Mater. 34 (2000) 135.
[8] J.C. Groen, J.C. Jansen, J.A. Moulijn, J. P6rez-Ramirez, J. Phys. Chem. B 108 (2004) 13062.
[9] J.C. Groen, L.A.A. Peffer, J.A. Moulijn, J. P6rez-Ramirez, Microporous Mesoporous Mater. 69
(2004) 29.
[10] T. Suzuki, T. Okuhara, Microporous Mesoporous Mater. 43 (2001) 83.
[11 ] M. Ogura, S.H. Shinomiya, J. Tateno, Y. Nara, E. Kikuchi, M. Matsukata, Chem. Lett. (2000)
882.
[12] I. Schmidt, A. Boisen, E. Gustavsson, K. St~hl, S. Pehrson, S. Dahl, A. Carlsson, C.J.H.
Jacobsen, Chem. Mater. 13 (2001) 4416.
[13] Y. Tao, H. Kanoh, K. Kaneko, J. Am. Chem. Soc. 125 (2003) 6044.
[14] A.H. Janssen, I. Schmidt, C.J.H. Jacobsen, A.J. Koster, K.P. de Jong, Microporous Mesoporous
Mater. 65 (2003) 59.
[15] J.B. Taboada, A.R. Overweg, M.W.J. Craj6, I.W.C.E. Arends, G. Mul, A.M. van der Kraan,
Microporous Mesoporous Mater. 75 (2004) 237.
[ 16] J. Perez-Ramirez, G. Mul, F. Kapteijn, J.A. Moulijn, A.R. Overweg, A. Domenech, A. Ribera,
I.W.C.E. Arends, J. Catal. 207 (2002) 113.
[ 17] J.B. Taboada, A.R. Overweg, P.J. Kooyman, I.W.C.E. Arends, G. Mul, J. Catal. (2005) (in
press).
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 35
bSchool of Chemistry, University of Manchester, P.O. Box 88, Manchester M60 1QD, United
Kingdom.
A semi-continuous flow reactor was used to mimic gel supersaturation conditions at both
intermediate and terminal stages of zeolite batch synthesis. Flow of nutrient into the reactor
was repeatedly switched on and off. Analysis of the crystals by atomic force microscopy
(AFM) reveals that samples removed under simulated conditions of gel depletion are
topographically similar to samples prepared by normal batch synthesis, with very low levels
of surface nucleation. Samples removed under conditions of high gel supersaturation appear
to be growing by the same mechanism however rates of surface nucleation are increased by
up to two orders of magnitude. Crystal growth resumes unperturbed upon reconnection of the
nutrient feed. Ability to exert control over surface nucleation rates has important ramifications
for the potential control of defect inclusion.
1. INTRODUCTION
In recent years great progress has been made towards improving our understanding of zeolite
crystal growth mechanisms using the techniques of AFM [1-11], SEM [12-14], TEM [15,16],
and simulation [16,17]. Crystal growth most probably occurs via the condensation of
monomeric and small oligomeric species until a surface energy minimum is reached, resulting
in terrace formation. Surface energy minima and thus terrace heights are generally
commensurate with some aspect of the unit cell. Analysis of experimental data reveals lateral
terrace spreading rates are far greater than successive terrace nucleation rates - this concurs
with our general perception of zeolite crystal facets as being effectively flat. However, it is
important to note that the wealth of data published to date has been acquired from batch
synthesised crystals.
SEM studies of sizes of zeolite crystals prepared by batch synthesis reveal an initial
induction period followed by a sustained period of linear growth [ 18] which eventually tails
off, as shown in Fig. 1. Linear growth occurs under conditions of high gel supersaturation
when the rate of nutrient dissolution is greater than the rate of nutrient uptake onto the
crystals. Over time, decrease of the former coupled to increase of the latter, as the crystals
grow in surface area, results in decreasing supersaturation levels, gel depletion and the slow
down and ultimate cessation of crystal growth.
In this paper we present a comparative AFM study of crystals recovered from gels
under carefully controlled conditions of high and low supersaturation. The principle aims are
to provide insight into earlier stages of zeolite crystal growth and to assess the possibility of
using gel supersaturation as a means of influencing crystal growth.
36
crystal size
Fig. 1. The evolution of crystal size and gel supersaturation during batch synthesis of a zeolite.
2. EXPERIMENTAL
2.1 Synthesis
1.6 pm long, un-twinned, lozenge-shaped silicalite seed crystals were synthesised as
detailed elsewhere [19]. Seed crystals at a concentration of 9.5 g L 1 in their spent mother
liquor were placed in a specially modified 1 L autoclaved reactor stirred at 300 rpm. The
slurry was heated to 130 ~ over a period of 30 mins and then left to equilibrate for 3 h, after
which nutrient feed was switched on at an initial rate of 90 mL h -1. The nutrient feed
comprised silica sol, tetrapropylammonium bromide and sodium hydroxide with the
following composition [Si] -- 0.48M, Si:OH = 8.72 and Si:TPA = 10.00. Samples were
removed at regular intervals in order to maintain an effectively constant working volume in
the reactor. In the absence of secondary nucleation, the assumption of effective stirring to
create homogeneous crystal distribution enabled calculation and monitoring of the required
nutrient feed to produce constant linear crystal growth rates of 0.4 ~tm h -~. Three times during
the course of the reaction the nutrient feed was disconnected for periods of c a . 16 h allowing
gel depletion akin to the latter stages of a normal batch synthesis. Quoted reaction times refer
to the total period for which the nutrient feed was switched on. Periods of disconnection are
not quantified, however, with reference to samples withdrawn prior to and immediately after
such periods of disconnection an asterisk is used to distinguish the latter.
Solution and gel silica concentrations were measured by chemical means for each sample
removed from the reactor. Solution silica comprises reactive monomeric and small oligomeric
species. Gel silica comprises larger oligomeric and polymeric silica that may be present as
37
@.07 111.70
~.--/~ t
11oi10
... ,. ~. / ' i
0.. \ .. , ........,. V'~_+<~.~ ,
1=
oO'~ v t / "........ 11.40
~ ~
11.111
11.10
o.! i i " i i ' l'o\~--~li ~~ ~ ,ii "
elapsed react#on ~me / h
Fig. 2. Reaction monitoring. The solid circle trace records pH. The upper hollow circle trace records
the concentration of solution silica whilst the lower one records the concentration of gel silica. The
grey sections (not shown to scale) represent times at which the feed was disconnected i.e. at 4.3, 9.6
and 16.4 h. No readings were recorded immediately prior to reconnection of the feed after the first
disconnection, consequently only estimated values are shown.
visible gel in a separable solid phase or as colloid dispersed in the form of a sol. Extensive
details of these measurements are quoted elsewhere[20]. Sample pH was also recorded and all
the data is plotted in Fig. 2. The gradual drop in sample pH as the reaction proceeds may be
attributed to mineralisation of the silica out-stripping condensation of silica onto the crystals.
This trend is reversed during each period of gel depletion as much of the silica condenses onto
the crystals.
E 8.0
::k
7.0
i m
N
m 6.0
~,~ 5.0
e~ 4.0
=E a.o i i
,,
i i i i i i i i
7 8 9 10 11 12 13 14 15 16 17
Time / h
Fig. 3. Mean sample crystal sizes, estimated from scanning electron micrographs, plotted against
growth time.
38
7.0urn 7.0/Jm
6.0 . . . . . ~! ..... 6.0
,~.0
4.0 1!~:'~ !:,~
:iiiii
;i ..o
3.0
2.o .....
1.0 i
Fig. 4. Mean sample crystal sizes estimated from scanning electron micrographs. Asterisks denote
samples removed after periods of disconnected nutrient feed.
Nutrient feed rates were calculated from Eq. (1) with further modification to correct
for the finite concentrations of silica in the exit stream. Symbols are defined as follows: x is
the [001 ] crystal length and A is the total surface area of the crystal population, both at time t,
u is the mass feed rate (e.g. mass SiO2 / unit time) and k is a constant dependent on crystal
geometry and density. A full derivation of the equation is given elsewhere [ 19].
dX=kU (1)
dt A
A simple computer program was used to calculate approximate values for A as the
reaction progressed, from known seed crystal size and experimental mass balance data. The
crystal growth rate was thus kept fairly constant at 0.4 pm h ~, as shown in. Fig. 3. Fig. 4
further reveals pauses in growth each time the nutrient feed is switched off, as evidenced by
height similarity between the asterisked crystals and their non-asterisked counterparts. This
result is of great significance as it reveals that very little volume is added to the growing
crystals in the latter stage of a typical zeolite batch synthesis, when supersaturation levels in
the gel are very low.
Fig. 5 shows AFM deflection images of typical (010) crystal surfaces. The two images
of crystals removed from the reactor after periods of gel depletion, (c) and (f), show multiply-
nucleated, approximately circular terraces that get closer together towards the edge of the
crystal, with inherent parabolic height cross-sections. The terraces are clearly randomly
positioned however each terrace necessarily nucleates on one below with the effect of
drawing the most recent nucleation points towards the centre of the crystal. Terrace heights
are generally 1.0 + 0.1 or 2.0 + 0.1 nm corresponding to the height of one and two pentasil
chains respectively. Double height terraces are a product of the random nature of terrace
fronts and their ever diminishing separation towards the edge of the crystal. Detailed
examination of the images reveals many instances of pairs of consecutive terraces merging to
create double height steps. Essentially, these images correspond well with those of 15 tam
crystals also recovered from depleted gels, published elsewhere [10]. Furthermore, in terms of
the number of nucleations given the size of the crystal, these images are typical of all images
of synthetic zeolites published to date [1-11 ]. This owes to the generally accepted practice of
zeolite batch synthesis whereby crystals are recovered by filtration from depleted gels.
There is however, one small difference that provides a link to discussion of the other
four images shown in Fig. 5. The difference, which may be seen in both images though it is
marginally clearer in image (c), is speckling of the surface. Cross-sectional analysis of the
speckled areas reveals the height of these small features to be identical to the height of the
39
Fig. 5. AFM deflection images of (010) faces of the series of silicalite crystals synthesised in the
constant nutrient feed reactor. Crystals shown in (a), (b), (d) and (e) were recovered from the reactor
after 7.6, 9.6, 13.3 and 16.4 h of nutrient feed respectively. Crystals shown in (c) and (f) were
recovered from the reactor after periods of cessation of nutrient feed and are denoted 9.6 h* and
16.4 h*. The scale bars represent 2 gm.
terraces strongly suggesting these small features are individual terrace nucleations and not
amorphous surface debris.
Examination of images (a), (b), (d) and (e) of fig 5. reveals four similar surface
topographies. In each case, the surface of the crystal is covered in a plethora of tiny, randomly
shaped, though approximating to circular lumps, similar in nature to the speckles observed on
the two previously discussed images. Once again, cross-sectional analysis reveals the height
of these lumps to be 1 nm leading to the conclusion that each one is a separate terrace
nucleation. Initial analysis indicates nucleation rates are between one and two orders of
magnitude higher on these crystals, removed from nutrient fed gels, with elevated
supersaturation rates.
Fig. 6 shows AFM deflection images of typical (100) crystal surfaces. These images
mirror the phenomena observed on the (010) faces. Fig. 6 (c) and (f) show low levels of
nucleation with the terraces present on the surface getting closer together as they approach the
edge of the crystal. Figures 6 (a), (b), (d) and (e) show high levels of nucleation similar to
those observed on the (010) face.
During zeolite synthesis, high levels of gel supersaturation cause high surface
nucleation rates. These nuclei then spread rapidly and coalesce into completed layers. Terrace
spreading rates are much faster than terrace nucleation rates and consequently it is surface
nucleation rates that govern how quickly a crystal grows in a given direction. Towards the end
of a synthesis, as the gel becomes depleted, crystal growth rates will decelerate. Being the
40
Fig. 6. AFM deflection images of (100) faces of the series of silicalite crystals synthesised in the
constant nutrient feed reactor. Letter assignments correspond identically with those of Fig. 4. The
scale bar in (d) represents 1 lain, all others represent 2 lam.
most energetically unfavourable, surface nucleation rates will tail off faster than terrace
spreading rates, leading to the wealth of previously published images showing relatively low
degrees of surface nucleation.
Of great interest is the ability of the crystal growth process to recover from periods of
quenching caused by gel depletion. In this present study the nutrient feed was stopped three
times. Crystal growth recovered unperturbed each time, with surface nucleation rates rapidly
jumping back to their previous high levels. Each period of gel depletion was accompanied by
a virtual halt in crystal growth, providing further evidence to link surface nucleation rates
with crystal growth rates. It is thus appears possible to control the degree of surface
nucleation, and hence crystal growth rate, by controlling gel supersaturation - even in the
limiting case where growth is completely stopped and restarted. Thus the growth may be
repeatedly switched on and off. One obvious goal of attempts to further understanding of
crystal growth phenomena in microporous materials is to exert control over a crystal growth
process. This has now been achieved.
The potential benefit of exerting control over such a fundamental aspect of zeolite
crystal growth may best be highlighted by example. The porous titanosilicate material ETS-10
contains the narrowest known quantum wires, in the form of titanate chains that run through
the material [21]. These titanate chains grow parallel to one another forming orthogonally
stacked layers. There are two possible nucleation sites for any new chain. The inherent
probability of chains in two separate, coalescing nuclei connecting correctly is only 50%.
Failure to connect correctly results in chain discontinuity and the generation of a structural
41
line defect. An ability to reduce surface nucleation rates for ETS-10 synthesis should result in
a decrease of such line defects and thus enhanced optoelectronic performance.
4. CONCLUSIONS
Design of a high pressure, semi-continuous flow reactor has enabled the controlled growth of
a range of silicalite crystals. Analysis of the surface of these crystals by atomic force
microscopy supports previously proposed mechanisms for zeolite crystal growth, based on
imaging of batch synthesised crystals. However, it reveals dramatic differences in growth
kinetics. Images of batch synthesised silicalite and other zeolites reveal a layer upon layer
growth mechanism whereby the rate of terrace nucleation on a crystal face dictates the rate of
growth of that face. Terrace spreading rates are significantly higher, with the area of any
given terrace being proportional to the time for which it has been growing. This leads to
commonly observed parabolic crystal faces. For batch synthesised crystals terrace nucleation
rates are generally quite l o w - apparent mononucleation is sometimes observed. Images of
silicalite crystals removed from highly supersaturated gels (designed to mimic those at
intermediate stages of batch synthesis) reveal far higher surface nucleation rates (between one
and two orders of magnitude higher). Cross-sectional analysis indicates that these surface
nuclei are essentially identical to, but far smaller versions of, the terraces observed during the
latter stages of growth.
When the nutrient supply to the reactor is cut, to mimic gel depletion during the final
stage of batch growth, the rate of surface nucleation, and hence crystal growth, is curtailed.
Being less energetically unfavourable, terrace spreading rates do not tail off as quickly,
resulting in spreading and coalescence of the nuclei present on the surface, such that surface
nucleation rates appear low. All previously published AFM images of zeolite crystals fall into
this category. After reconnection of the nutrient supply evelated supersaturation levels cause a
resumption of crystal growth and the zeolite surface is once again covered in nuclei. This
process of switching growth on and off may be repeated. It is therefore possible to control
surface nucleation rates by controlling gel supersaturation levels.
REFERENCES
[1] M.W.Anderson, J.R. Agger, J.T. Thornton and N. Forsyth, Angew. Chem., Int. Ed., 35 (1996)
1211.
[2] S. Sugiyama, S. Yamamoto, O. Matsuoka, T. Honda, H. Nozoye, S. Qiu, J. Yu and O. Terasaki,
Surf. Sci., 377 (1997) 140.
[3] J.R. Agger, N. Pervaiz, A.K. Cheetham and M.W. Anderson, J. Am. Chem. Soc., 120 (1998)
10754.
[4] M.W.Anderson, J.R. Agger, N. Pervaiz, S.J. Weigel and A.K. Cheetham (M.M.J. Treacy, B.K.
Marcus, M.E. Bisher and J.B. Higgins) Proc. 12th Int. Zeol. Conf., Baltimore, MRS, 1998,
pp1487.
[5] S. Sugiyama, S. Yamamoto, O. Matsuoka, H. Nozoye, J. Yu, G. Zhu, S. Qiu and O. Terasaki,
Microporous Mesoporous Mater., 28 (1999) 1.
[6] J. Warzywoda, M. Valcheva-Traykova, G.A. Rossetti Jr., N. Ba~, R. Joesten, S.L. Suib and A.
Sacco Jr., J. Crystal Growth, 220 (2000) 150.
[7] M.W.Anderson, N. Hanif, J.R. Agger, C.-Y.Chen and S.I. Zones, (A. Galerneau, F. Di Renzo,
F. Fajula and J. Vedrine) Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 2001,135, pp 141.
[8] J.R. Agger, N. Hanif and M.W. Anderson, Angew. Chem. Int. Ed., 40 (2001) 4065.
42
[9] S. Dumrul, S. Bazzana, J. Warzywoda, R.R. Biederman and A. Sacco Jr, Microporous
Mesoporous Mater., 54 (2002) 79.
[10] J.R. Agger, N. Hanif, C.S. Cundy, A.P. Wade, S. Dennison, P.A. Rawlinson, and M.W.
Anderson, J. Am. Chem. Soc., 125 (2003) 830.
[11 ] T. Wakihara, A. Sugiyama, T. Okubo, Microporous Mesoporous Mater. 70 (2004) 7.
[12] S. Bazzana, S. Dumrul, J. Warzywoda, L. Hsiao, L. Klass, M. Knapp, J.A. Rains, E.M. Stein,
M.J. Sullivan, C.M. West, J.Y. Woo and A. Sacco Jr. (R. Aiello, G. Giordano, F. Testa) Proc.
2nd Int. FEZA Conf., Taormina, Elsevier, Amsterdam, 2002, 142A, pp 117.
[13] J. Warzywoda, B. Yilmaz, P.Q. Miraglia and A. Sacco Jr., Microporous Mesoporous Mater., 71
(2004) 177.
[14] B. Yilmaz, P.Q.Miraglia, J. Warzywoda and A. Sacco Jr., Microporous Mesoporous Mater., 71
(2004) 167.
[15] O. Terasaki, J. Electron Microsc., 43 (1994) 337.
[ 16] B. Slater, C.R.A. Catlow, Z. Liu, T. Ohsuna, O.Terasaki, and M.A. Camblor, Angew. Chem.
Int. Ed., 41 (2002) 1235.
[ 17] T. Ohsuna, B. Slater, F.F. Gao, J.H. Yu, Y. Sakamoto, G.S. Zhu, O. Terasaki, D.E.W. Vaughan,
S.L. Qiu and C.R.A. Catlow, Chem. Eur. J., 10 (2004) 5031.
[18] S.P. Zhdanov, ACS Adv. Chem. Ser. 101 (1971) 20.
[19] C.S. Cundy, M.S. Henty and R.J. Plaisted, Zeolites, 15 (1995) 353.
[20] C.S. Cundy, M.S. Henty and R.J. Plaisted, Zeolites, 15 (1995) 343.
[21 ] A. Damin, F.X. Llabre's i Xamena, C. Lamberti, B. Civalleri,C.M. Zicovich-Wilson, and A.
Zecchina, J. Phys. Chem. B, 108 (2004) 1328.
Studies in Surface Science and Catalysis, volume 158
J. (2ejka,N. Zilkowiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 43
It is vital to develop a visible light sensitive photocatalyst. In this study, (Cr, Ti)-binary
oxide containing mesoporous silica (Cr/Ti-HMS) photocatalyst was prepared by an
application of a photo-assisted deposition (PAD) technique and characterized by XAFS and
photoluminescence studies. Using the PAD method, the tetrahedrally coordinated
Cr-oxide moieties can be anchored on Ti-HMS forming the direct interaction between
Cr-oxide and Ti-oxide moieties. The Cr/Ti-HMS exhibited the high efficiency and
selectivity in the photoepoxidation of propene with 02 under visible light irradiation. The
charge transfer excited state of the Cr-oxide moieties anchored on Ti-oxide moieties play a
significant role in photocatalytic reactions under visible light irradiation.
1. INTRODUCTION
The unique and fascinating properties of zeolites and mesoporous silica involving transition
metals within their cavities and framework have opened new possibilities for many
application areas not only in catalysis but also for various photochemical processes [ 1-5]. The
transition metal ions in these catalysts are considered to be highly dispersed at the atomic
level and also to be well-defined catalysts which exist in the specific structure of the zeolite
and mesoporous silica framework. Especially the titanium oxide species prepared within
the frameworks of mesoporous silica have revealed a unique local structure as well as a high
selectivity in various catalytic reactions. These Ti-containing mesoporous silica can also
exhibit efficient photocatalytic reactivity and high selectivity for NO decomposition [6], CO2
reduction with H20 [7-9], and partial oxidation of alkenes and alkanes [10]. However, these
Ti-contaning catalysts can operate as the photocatalysts only under UV-light irradiation.
Recently, we have found that Cr-containing mesoporous silica can demonstrate the
44
photocatalytic reactivity even under visible light irradiation [11,12], while the selectivities
were not sufficient. To establish the clean photocatalysis system it is vital to develop the
photocatalysts which can operate efficiently under visible light irradiation with high
selectivity [5,13,14].
In the present study, the (Cr, Ti)-binary oxide containing mesoporous silica
photocatalyst was prepared by the application of a photo-assisted deposition (PAD) technique
which can realize the direct interaction between Cr-oxide and Ti-oxide moieties. The
photocatalytic reactivity of (Cr, Ti)-containing mesoporous silica under visible light
irradiation for the epoxidation of propene with 02 has been investigated.
2. EXPERIMENTAL
2.3. Characterization
The diffuse reflectance absorption spectra were recorded with a Shimadzu
UV-2200A spectrometer at 295 K. The photoluminescence spectra were measured at 77 K
using a Shimadzu RF-5000 spectrophotofluorometer. The X-ray absorption fine structure
(XAFS) spectra, X-ray absorption near edge structure (XANES) and extended X-ray
absorption fine structure (EXAFS), were measured at the BL-9A facility [ 16] of the Photon
Factory at the National Laboratory for High-Energy Physics, Tsukuba. A Si(111) double
45
crystal was used to monochromatize the X-rays from the 2.5 GeV electron storage ring. The
Ti K-edge and Cr K-edge absorption spectra were recorded in the fluorescence mode at 295
K. The normalized spectra were obtained by a procedure described in previous literature
[17] and Fourier transformation was performed on k3-weighted EXAFS oscillations in the
range of 3-10 A -1.
3.1. Characterization
Fig. 1 shows the diffuse reflectance UV-VIS absorption spectra of original Ti-HMS
and Cr/Ti-HMS prepared by the PAD method. Ti-HMS exhibit absorption band in the
wavelength regions of 220-260 nm, shifting into the shorter wavelength regions as compared
to the bulk TiO2 catalyst. This absorption band is attributed to the LMCT (ligand-to-metal
charge transfer) band of the highly dispersed Ti-oxide moieties having a tetrahedral
coordination [18,19]. As shown in Fig. 1, Cr/Ti-HMS can absorb visible light in the range
of 400-600 nm, while Ti-HMS absorbs only UV light shorter than 300 nm. The absorption
bands observed with Cr/Ti-HMS at around 270, 350, 450 nm can be assigned to the charge
2- 6+
transfer from O to Cr of the tetrahedrally coordinated Cr-oxide moieties [5,11,12]. The
absorption bands assigned to the absorption of the dichromate of Cr=O3 cluster cannot be
observed above 550 nm [12], indicating that tetrahedrally coordinated Cr-oxide moieties
exist in the isolated state on Cr/Ti-HMS. The intensity of absorption band at around
220-260 nm assigned to the LMCT band of the tetrahedrally coordinated Ti-oxide moieties
become weaker with Cr/Ti-HMS than that of Ti-HMS indicating that Cr-oxide was deposited
on the Ti-oxide moieties by the PAD method.
C
.9
o A) Ti-HMS ~ -
E 1.i4+
~m"
.s
B) Cr/Ti-HMS
Fig. 1. UV-Vis spectra of (A) Ti-HMS and (B) Cr/Ti-HMS prepared by the PAD method.
46
t~
C:
Fig. 2. Photoluminescence spectra of (a) Ti-HMS (b) Cr-HMS and (c-e) Cr/Ti-HMS prepared
by the PAD method. Excited at (a) 260 nm, (b) 500 nm, (c) 260 nm, (d) 290 nm, (e) 500 nm.
o!~ hv "! hv
,cr . ................................ , c r
i hv" t
Cr/Ti-HMS exhibits photoluminescence band at around 640 nm under not only UV light
irradiation (260, 290, 3"70 nm) but also visible light irradiation (415,500 nm), while Ti-HMS
exhibits photoluminescence band at 4"70 nm only under UV irradiation (260 nm). Even
under visible light irradiation the photoluminescence bands on Cr/Ti-HMS were easily
quenched by the addition of the reactant gas (propylene and oxygen). These results
indicated that Cr/Ti-HMS can be excited not only under UV light irradiation but also under
visible light irradiation, while Ti-HMS can be excited only under UV light irradiation.
Fig. 3 shows the XANES spectra of the original Cr-HMS and Ti-HMS catalysts and
the Cr/Ti-HMS and Ti/Cr-HMS prepared by the PAD method. The XANES spectra of the
Ti and Cr containing compounds at the Ti K-edge and Cr- K-edge show several well-defined
preedge peaks that are related to the local structures surrounding the Ti atom and Cr atom,
respectively. These relative intensities of the preedge peaks provide useful information on
the coordination number surrounding the Ti atom and Cr atom [5,20,21 ]. As shown in Fig. 3,
these catalysts exhibit a single preedge peak, while its intensity depends on the catalysts.
47
Because tetrahedrally coordinated Ti and Cr was found to exhibit an intense single preedge
peak due to the lack of an inversion center in the regular tetrahedron structure [5,20,21 ], the
observation of an intense single preedge peak indicates that the tetrahedrally coordinated
Ti-oxide and Cr-oxide were incorporated within the framework of mesoporous silica in
Ti-HMS and Cr-HMS, respectively. In the case of Cr-oxide moieties, two terminal Cr=O
bonding is included in the tetrahedral coordination [5,11,12]. Although the Ti-HMS
exhibits an intense preedge peak at Ti K-edge spectra indicating the presence of the
tetrahedrally coordinated Ti-oxide moieties, the intensity observed with the Cr/Ti-HMS
becomes much smaller. In the case of Cr K-edge, the intensity of preedge peak with
Cr/Ti-HMS is as strong as that of Cr-HMS. These results indicate that the terahedrally
coordinated Cr-oxide moieties having two terminal Cr=O bonds was deposited on Ti-oxide
moieties making distorted Ti-oxide moieties by the direct interaction between Cr-oxide and
Ti-oxide moieties. On the other hand, the intensity of preedge peaks with Ti/Cr-HMS
prepared by the PAD method are very small indicating that tetrahedrally coordinated
Ti-oxide and Cr-oxide are heavily distorted.
Cr K-edge Ti K-edge
8
AZI
O
c)
Ti/C__yr-HMS K/ Ti/Cr-HMS
Fig. 3. XANES spectra at Cr K-edge (left) and Ti K-edge (right) of (A) Cr-HMS, (a) Ti-HMS,
(B,b) Cr/Ti-HMS, (C,c) Ti/Cr-HMS.
These spectroscopic studies indicate that in the Cr/Ti-HMS prepared using the PAD
method the tetra-Cr-oxide moieties was anchored on the tetra-Ti-oxide included in the
frameworks of mesoporous silica having the direct interaction between the Cr-oxide and
Ti-oxide moieties as shown in Scheme 1. In this scheme, the tetrahedrally coordinated
Cr-oxide moieties are deposited on the Ti'oxide tetrahedral moieties excited under the UV
light irradiation.
48
I Cr ions
.,Tp.:,,
. I~- O-..
o~Tr~- C T.
" 9 2- "
,i,-
Scheme 1. Design of Cr, T binary oxide by the PAD method.
Table 1
The products in the photocatalytic oxidation of propene with 02 on Cr-oxide and Ti-oxide
containing catalysts under light irradiation (~,>270 nm, )~>450 nm).
Catalyst )~ Conv. PO-yield Selectivity / %
/nm /% /% PO AL PA AC AA CH COx
CrS-1 450 3 0.1 4 45 2 14 8 2 25
Cr-HMS 450 10 1 12 18 11 10 24 4 21
Ti'-HMS 250 7 1.5 22 14 12 26 5 11 10
Ti-HMS 450 0 . . . . . . . .
PO: propylene oxide, AL: acrolein, PA: propanal, AC: acetone, AA: acetaldehyde, CH: hydrocarbon,
COx: CO + C02
The charge transfer excited state of the tetrahedrally coordinated Cr-oxide moieties anchored
on Ti-oxide moieties and the large pores of the mesoporous silica play a significant role in
these highly selective and effective photocatalytic reactions. On the other hand, Ti/Cr-HMS
hardly exhibited reactivity. The formation of distorted coordination of Cr-oxide and
Ti-oxide moieties observed on Ti/Cr-HMS by the spectroscopic techniques is not suitable for
the photocatalytic reaction.
Table 2
The products in the photocatalytic oxidation of propene with 02 on Cr-oxide and
Ti-oxide containing catalysts under visible light irradiation ()~>450 nm).
Catalyst Conv. PO-yield Selectivity / %
/% /% PO AL PA AC AA CH COx
Cr-HMS 10 1 12 18 11 10 24 4 20
Cr/HMS 4 1 19 17 19 12 23 4 6
Cr/Ti-HMS 40 5 44 15 11 8 13 1 8
Ti/Cr-HMS 1 0 0 0 34 31 23 11 1
PO: propylene oxide, AL: acrolein, PA: propanal, AC: acetone, AA: acetaldehyde, CH:
hydrocarbon, COx: CO + CO2
4. CONCLUSION
Using the photo-assisted deposition method, the tetrahedral Cr-oxide moieties can be
anchored on Ti-HMS forming the direct interaction between Cr-oxide and Ti-oxide moieties.
It has been found that the Cr/Ti-HMS exhibited the higher efficiency and the higher
selectivity in the photoepoxidation of propene with O2 under visible light irradiation.
ACKNOWLEDGMENT
This work was supported by the Grant-in-Aid Scientific Research from the Ministry of
Education, Science, Culture, and Sports of Japan (Grants 16656251). This work is also partly
performed under the project of collaborative research at the Joining and Welding Research
Institute (JWRI) of Osaka University with helpful advice from Prof. H. Fujii. The X-ray
adsorption experiments were performed at the Photon Factory of KEK (2004G295) with
helpful advice from Prof. M. Nomura.
REFERENCES
[5] H.Yamashita and M. Anpo, Current Opinion in Solid State & Maters. Sci., 7 (2004) 471.
[6] H. Yamashita, Y. Ichihashi, M. Anpo, C. Louis and M. Che, J. Phys. Chem., 100 (1996) 16041.
[7] M. Anpo, H. Yamashita, Y. Ichihashi, Y. Fujii and M. Honda, J. Phys. Chem. B, 101 (1997)
2632.
[8] K. Ikeue, H. Yamashita, M.Anpo and T. Takewaki, J. Phys. Chem. B, 105 (2001) 8350.
[9] H. Yamashita, K. Ikeue, T. Takewaki and M. Anpo, Top. Catal., 18 (2002) 95.
[10] H. Yamashita, K. Kida, K. lkeue, Y. Kanazawa, K. Yoshizawa and M. Anpo, Stud. Surf. Sci.
Catal., 146 (2003) 597.
[11] H. Yamashita, K. Yoshizawa, M. Ariyuki, S. Higashimoto, M. Che and M. Anpo, Chem.
Commun., 435 (2001).
[12] H. Yamashita, M. Ariyuki, K. Yoshizawa, K. Kida, S. Ohshiro and M. Anpo, Res. Chem.
lntermed., 30 (2004) 235.
[13] E.P. Reddy, L. Davydov and R G. Smirniotis, J. Phys. Chem. B., 106 (2002) 3394.
[14] H. Yamashita and M. Anpo, Catal. Suv. Asia, 8 (2004) 35.
[15] W. Zhang, R T. Tanev and T. J. Pinnavaia, J. Chem. Soc., Chem. Commun., (1996) 979.
[ 16] M. Nomura and A. Koyama, J. Synchrotron Rad., 6 (1999) 182.
[17] H. Yamashita, M. Matsuoka, K. Tsuji, Y. ShioyaandM. Anpo.,J. Phys. Chem., 100(1996) 397.
[18] H. Yamashita, S. Kawasaki, Y. Ichihashi, M. Harada, M. Anpo, G. Stewart, M. A. Fox, C. Louis
and M. Che, J. Phys. Chem. B, 102 (1998) 5870.
[19] H. Yamashita, M. Honda, M. Harada, Y. Ichihashi, M. Anpo and Y. Hatano, J. Phys. Chem. B,
102 (1998) 10707.
[20] J.M. Thomas and G. Sankar, Acc. Chem. Res., 34 (2001) 571.
[21 ] G. Sankar and J.M. Thomas, Top. Catal., 8 (1999) 81.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 51
blnstituto de Tecnologia Quimica, U P V - CSIC. Avda. de los Naranjos s/n. 46022 Valencia.
Spain.
1. INTRODUCTION
Tailored zeolite synthesis which would make feasible the design of specific microporous
topologies for specific applications is a much wanted objective in zeolite science. This is due
to the fact that the macroscopic properties of zeolites materials are narrowly tied to the
microscopic topology of ordered pores. Nevertheless, nowadays the crystallisation
mechanism is insufficiently understood for this purpose. Commonly, searching for new
structures and different chemical compositions is carried out by using a wide variety of
synthesis conditions and different structure-directing agents (SDA).
The SDA establish no-bonding interactions with the zeolite and with inorganic species
present in the gel [ 1, 2], the nature and strength of which depends on the chemical properties
of the SDA. Hereby it would be possible to change these interactions by modifying the
chemical properties of the organic species acting as SDA. A way to explore the effect of
modifying the organic-inorganic interactions is provided by the use of fluorinated compounds,
which allow to change the physicochemical properties of the SDA without changing its
geometry and size [3, 4]. We have reported a first study of this type by using
dibenzyldimethylammonium cation and its fluorinated derivatives [5].
In this work a new type of SDA, the 1-benzyl-l-methyl-hexamethylenammonium
(BMHM) cation and its fluorinated derivatives have been used for the synthesis of all-silica
zeolite materials in fluoride media. This study shows that fluorine-containing organic
52
compounds can be considered as a new family of SDA which can help in the search of new
zeolite topologies or improve the synthesis of known phases.
2. E X P E R I M E N T A L
Table 1
Synthesis conditions and phases obtained; FX: unidentified phase.
SDA Exp. T SDAOH HF SiO2 HzO Result*
Beta
BMHM H-1-A 150 0.53(+0.26**) 0.54 1 4.62
NON
Beta
M-FBMHM M-1-A 150 NON
FX
0.54(+0.26) 0.54 1 4.91
Beta
M-FBMHM M-1-B 135 NON
FX
Beta
M-FBMHM M-I'-A 150 0.80 0.80 1 4.90 NON
FX
BMHM H-2-A 150 ~" 0.54 0.54 1 4.60 Beta
BMHM H-2-B 140 Beta
O-FBMHM O-2-A 150 0.54 0.54 1 4.61 Beta
Beta
M-FBMHM M-2-A 150 0.54 0.54 1 4.68 NON
FX
P-FBMHM P-2-A 150 0.54 0.54 1 4.55 NON
BMHM H-3-A 150 Beta
0.44 0.44 1 4.72
BMHM H-3-B 135 Beta
BMHM H-4-A 150 Betas
0.44 0.44 1 7.93
BM H M H-4-B 135 Beta
O-FBMHM O-4-B 135 0.44 0.44 1 7.92 Beta,[.
M-FBMHM*** M-5-A 150 0.54 0.54 1 7.93 NON
* All samples obtained at 30d, except, H-I-A, P-2-A, M-5-A at 36d, and H-4-A, H-4-B, O-4-B at
63d.
** Iodide is added to increase the SDA content without increasing fluoride content.
***Contains hexamethylenimine as impurity, because ternary amine not was distilled
The synthesis of quaternary ammonium cations was carried out by methylation of the
tertiary amines obtained previously with a solution of CH3I (50% exc., Fluka) in ethanol. This
solution was stirred for 5d at room temperature, the ethanol was removed under vacuum at
60~ and the solid products obtained, iodide of 1-benzyl-l-methyl-hexamethylenammonium
(BMHMI) (,--70% yield) and iodide of 1-fluorobenzyl-l-methyl-hexamethylenammonium (o-
FBMHMI, m-FBMHMI, p-FBMHMI) (--- 90% yield) were washed with ethyl ether and
characterized by ~3C NMR and chemical analysis. These iodide salts were converted into the
corresponding hydroxide, BMHMOH or FBMHMOH, by ion-exchange with an Amberlyst
IRN78 resin (exc. cap: 4 me/g, Supelco). The BMHMOH and FBMHMOH solutions were
titrated with HC1 1N (Panreac) and phenolphthalein (Aldrich).
Finally, the synthesis gel was prepared by adding tetraethylorthosilicate (TEOS, Merck) on a
BMHMOH or FBMHMOH solution under continuous stirring. Once the ethanol is
evaporated, HF (48 wt%, Panreac) is added drop by drop. The resulting thick gel is then
introduced into 20ml Teflon-lined stainless steel autoclaves which were heated statically at
135-150~ for selected periods of time. The obtained solid product was filtered, washed and
dried overnight at 70~ Chemical composition and temperature of different synthesis gels are
given in table 1. Where in the nomenclature employed the first letter correspond to the cation,
H of hydrogenated or fluorinated in O, ortho-, M, meta- or P, para- position, number
correspond the gel composition, and last letter differences the temperature of synthesis.
The presence of fluorine in the BMHM molecule strongly affects the ability of this cation for
the nucleation of zeolite beta, and a new not yet identified phase appears if the meta-F
derivative is used.
' 1(I ' 15 ' 2[) ' 25 ' 3() ' 35 ' 40 ' 10 ' 1'5 ' 2() ' 25
2G 2O
Fig. 1. XRD patterns of samples a) H-2-A, b) O-2-A, c) M-2-A, d) P-2-A, Enlargement of the
diffractogram region 28=5~176 e) of M-2-A, f) of M-l-B, g) of M-1 '-A. The arrows show the peaks
corresponding to unidentified phase.
d) e) f)
Fig. 2. SEM of samples synthesised with BMHM a) beta and nonasil H-I-A, b) beta H-2-A, c) beta H-
4-B, d) beta crystals O-2-A, e) M-l-A, unidentified phase, f) M-5-A, nonasil phases, spherical
aggregates and crystal with hexagonal faces.
The crystal morphology of beta is influenced by the SDA content in the synthesis gel.
The most concentrated gel gives isolated crystal of nearly prismatic shape, H-1-A (average
length 18~tm, aspect ratio 1.4), Fig 2. Diluted gel leads to longer isolated crystal of well-
defined sharp-pointed truncated bipyramid typical of beta, H-4-B (average length 32~tm,
aspect ratio 2.6). Therefore the growth rate of beta along a, b and c axes varies differently
according to the crystallisation conditions. In addition, there is a noticeable change in the
55
morphology of beta crystals due to the presence of fluorine, since H-2-A shows unusual
spherical aggregates whereas O-2-A shows isolated crystals.
Nonasil appears both as spherical aggregates and as isolated crystals with hexagonal
morphology, and lamellar aggregates belonging to the unidentified phase are observed.
a)~ lOO
2 90 .2 90
80 --&
. .,,.~
8o
70 70
60
200 400 600 800 20o400 o0o soo
T~ Toc
Fig. 3. Thermogravimetry analyses of a) beta H-3-A b) nonasil P-2-A, c) beta, nonasil and
unidentified phase M-2-A, and d) beta, nonasil and unidentified phase M-l-A, samples.
Nonasil phase, Fig. 3b, shows a Thermogravimetry pattern with a weight loss between
400-674~ and at 900~ not all the organic has been removed, since chemical analyses shows
an organic content (-10%) higher than the weight loss up to 900~ In addition, the C/N ratio
(7.3 for P-2-A, 7.7 for M-5-A) indicates that cation fragments such as methyl-
hexamethylenimine, might occupy the [5 s 612] nonasil cages. Indeed, the carbon content (4
molec./u.c.) suggests a full occupancy of these cavities by the amine.
Thermogravimetry of the samples that contain beta, nonasil and the unidentified samples,
Fig. 3c and 3d, are very interesting because the weight loss occur in the same T range of beta
and nonasil, but the total weight loss is higher. This suggests a high content of organic
material occluded into the unidentified phase. In addition, the C/N ratio (12.3 for M-2-A, 12.5
for M-l-A, 13.4 for M-l-B, 12.1 for M-I'-A) indicates that most of the occluded organic
material would correspond to the unaltered cation. The similarity of the TG patterns of beta
and the sample containing the unidentified phase suggests that these cations would occupy
large channels or voids also in the latter.
56
h ~
3' 2-3 d ~ 1,, ,-,
sd ~ '4 c
' : sd .~d
iiiil~t'rl~ 2-4 1 v}
i
.~h J.i 1-3,~4 c e 0,,~
"" " - : I[1
2~0 " 1~i0 ' 1 89 ' 8b ' 4b ' 4 2do ' i~o ' I~o ' 8b ' 4b ' d
,B, l:,pm ~ppm
Fig. 4. ~H/~3CCP MAS NMR spectra of, i) o-FBMHMI, ii) O-2-A, iii) m-FBMNMI, iv) M-1-A,v) M-
2-A, and vi) M-5-A.
90 ppm of fluoride anions in different cages, and an intense signal at -114ppm of fluorine in
aromatic ring Fig.5iv and 5v. But the most noticeable feature is the presence of a very intense
resonance signal at -119 ppm, most probably due to F anions compensating the organic
cations [12]. However, this signal might arise as well from F atoms of the ring interacting
with the framework, since the incorporation of the cation inside the zeolite cavities modifies
the chemical shift of the aromatic carbon atoms.
_ _
-30 ' -60 ' -90 ' -120' -_~0 -;0 -90 -i~.0 -30 -;0 -90 -120
6 ppln 6 ppm 6 ppm
Fig. 5. 19F MAS NMR spectra of samples: i) Nonasil M-5-A, ii) beta H-2-A, iii) beta O-2-A, iv) beta,
nonasil, unidentified phase M-1 'A at 30d, enlargement on top, v) beta, nonasil, unidentified phase M-
l-A, enlargement on top.
iv) ...~-.ak_
ifi)
9
ii) ........
i)
-100 ' - 1 i 0 -12,0 ' - 1 3 0 ' - 1 4 0 -150
5 p pin.
Fig. 6.29Si MAS NMR as prepared zeolites, i) beta sample H-3-B, ii) beta, nonasil, unidentified phase
M-2-A, iii) beta, nonasil, unidentified phase M-1-A iv) nonasil M-5-A.
58
4. C O N C L U S I O N S
The BMHM cation is an efficient structure directing agent for the synthesis of pure silica beta
zeolite, but a narrow set of conditions are required to avoid degradation of the cation and the
crystallisation of nonasil. The behaviour of BMHM as structure directing agent is affected by
the substitution of only one hydrogen atom by fluorine in the aromatic ring, and this effect
also depend upon the fluorine position in the ring. Changes of crystal size and morphology
and the crystallisation of a unidentified open phase occur by using fluorinated BHMH cations.
All-hydrogenated as well as fluorine-containing derivatives are found unbroken within
the zeolite cavities, and the C-F bond stability under the synthesis conditions used in this
work has been evidenced by ~3C and 19F MAS NMR
Therefore, the change in the chemical character of the SDA brought about by the
replacement of H by F atoms, which implies a change in the interactions between organic
species and the zeolite framework, is a new factor that it is necessary bear in mind in zeolite
synthesis.
ACKNOWLEDGEMENTS
REFERENCES
Synthesis and catalytic properties of zeolites MCM-58 and MCM-68 were investigated. The
MCM-58 was synthesized using N-benzylquinuclidinium bromide (BQ) as a structure
directing agent (SDA) in a wide range of initial ratios nsi/nAI 20 - 40 and nsi/nl:e = 20 - 40.
=
1. INTRODUCTION
The alkylation reactions are widely used in chemical industry for the production of important
alkyl aromatic intermediates (ethylbenzene, p-ethyltoluene, p-xylene, cumene, cymene) [ 1-3].
Many industrial alkylation processes are still performed with classical catalysts (HF, solid
phosphoric acid) showing drawbacks (handling, safety, corrosion and waste disposal). In
order to avoid these problems many efforts have been devoted to search catalysts more
selective, safe and environmentally friendly, regenerable, and reusable. These problems of
classical catalysts can be avoided when zeolites are used as catalysts [4]. Zeolites constitute a
very important group of inorganic microporous materials subservient to applications like
adsorption, ion exchange and especially heterogeneous catalysis [5,6]. The catalytic activity
and selectivity of zeolites is attributed to the large internal surface with active sites that are
accessible through uniformly sized pores. Acid strength is one of the most important factors
for controlling reactivity and selectivity of zeolite catalysts. Properties like geometry,
chemical composition and aluminum distribution can be correlated with acid strength
of zeolites [7]. Thus, reactions are classified from the point of view of acid strength needed
(alkylation requires sites of medium-strong acidity). To optimize the acid strength and
geometry of zeolites strong effort is devoted to the synthesis of new structural types of
zeolites with tailored pores systems. Accurate preparation of synthesis using novel types of
SDA resulted recently in a successful synthesis of new types of zeolites (ITQ-15 [8] with
14- and 12-membered rings, SSZ-58 [9] zeolite with previously unseen double 5-ring
subunits, and micro/mesoporous composite molecular sieves [10].
In this contribution, synthesis and characterization of zeolites MCM-58 [ 11 ] as an A1- and
Fe-silicates (IFR) and aluminosilicate MCM-68 [12] was carried out. The MCM-58 topology
is characterized by an undulating, one-dimensional channel system with 12- membered rings.
MCM-68 has a unique crystal structure, which contains one channel system, in which each
60
channel is defined by a 12-membered ring and two further channels, in each of which the
channel is defined by a 10-membered ring. MCM-58 and MCM-68 were characterized using
XRD, XRF, FTIR and SEM. Catalytic properties of these zeolites were tested in the
alkylation of toluene with isopropylalcohol. The desired product p-cymene is the starting
material for production of some fungicides, pesticides, and flavors.
2. EXPERIMENTAL
Thereafter, zeolites were heated to 540 ~ with the same heating rate and kept for 2 h at this
temperature. Finally, the material was calcined for 24 h in a stream of air at 540 ~
The ammonium forms of both zeolites were prepared by four times repeated ion exchange
of the zeolite in 0.5 M aqueous solution of ammonium nitrate at ambient temperature.
The A1- and Fe-MCM-58 zeolites were synthesized from the reaction gel with-following
molar compositions: (0.016-0.025) A1203: SiO2:38 H20:0.30 K20:0.10 BQ for A1-MCM-
58 and (0.0125-0.025) Fe203: SiO2:38 H20:0.30 K20:0.20 BQ for Fe-MCM-58. The BQ,
which is used as a structure directing agent in the MCM-58 synthesis, is relatively expensive.
Thus the amount of BQ in the reaction mixture for A1-MCM-58 was lowered from the
original molar ratio BQ : SiO2 = 0.2 to 0.1. By optimizing the synthesis conditions and time at
170 ~ it was possible to reproducibly synthesize highly crystalline zeolite A1-MCM-58 (Fig.
1A) with initial ns~/nAl ratios from 20 to 40. In the case ofFe-MCM-58 it was not possible to
62
>~
t~
r
I ~ I , I ~ I ~ ~ I , I ~ I , ~ I , I ~ I
0 15 30 45 0 15 30 45 0 15 30 45
2 Theta, o
Fig. 1. X-ray powder diffraction patterns of as-made (a) and calcined (b) zeolites A1-MCM-58,
nsi/nAl 26 (A), Fe-MCM-58, nsi/nve = 26 (B), and A1-MCM-68, nsi/nAl = 9 (C).
=
synthesize this zeolite as a pure phase under the same conditions used for A1-MCM-58. The
sample was always contaminated with a-quartz. To prevent the formation of ot -quartz, the
reaction temperature for the synthesis of F e - M C M - 5 8 was lowered to 155 ~ and the
synthesis time was optimized.
The optimum synthesis time was found to be between 4.5 and 5.5 days independent of the
ns~/nFe ratio in the gel (Fig. 1B). A1-MCM-68 was readily synthesized after 13 days at 160 ~
The powder X-ray diffraction pattern of as-synthesized and calcined A1-MCM-68 are
depicted on the Fig. 1C.
The scanning electron micrographs (Fig. 2 ) of A1-MCM-58 (nsi/nAl = 26), F e - M C M - 5 8
(nsi/nFe = 26), and A1-MCM-68 (nsi/nAi-- 11) show that no amorphous phase is present.
Crystals of A1-MCM-58 (Fig. 2A) are represented by elongated particles with length from 1
to 5 ~tm depending on the nsi/nAi ratio in the synthesis gel. The size of crystals increased with
decreasing amount of aluminum in the reaction mixture.
The crystals of F e - M C M - 5 8 (Fig. 2B) are smaller in comparison with A1-MCM-58 with
sizes between 0.5 and 3.0 ~tm. Some agglomeration of F e - M C M - 5 8 crystals to form stars
was found, which is also typical for Fe-ZSM-12 [15].
A B C
Fig. 2. Scanning electron micrographs of zeolites A1-MCM-58 with nsi/nAl ratio = 26 (A), Fe-
MCM-58 with nsi/n~e ratio = 26 (B) and AI-MCM-68 with nsi/nAi ratio = 11 (C).
63
A1-MCM-68 crystals appear like agglomerates of particles smaller then 0.5 lam (Fig. 2C).
The type and concentration of the acid sites were determined after adsorption of probe
molecules such as d3-acetonitrile and pyridine on the activated aluminum and iron zeolites,
respectively. Spectra of A1-MCM-58 and A1-MCM-68 in the region of OH groups before and
after adsorption of d3-acetonitrile are depicted in Fig. 3A,a and Fig. 5A, a, respectively. The
absorption bands at 3746 and 3745 cm -~ belong to terminal Si-OH groups of A1 MCM-58 and
A1-MCM-68, respectively, and two bands at 3630 and 3564 cm -~ for A1-MCM-58 and 3610
and 3575 cm -~ for A1-MCM-68 are characteristic for bridging OH groups. The band at 3676
cm -~ in the spectra of A1-MCM-68 (Fig. 5A,a) was ascribed to OH groups on nonframework
aluminum species. After adsorption of d3-acetonitrile this band vanished. After the interaction
with d3-acetonitrile, new bands of Lewis sites at 2327 cm ~, and Broensted sites 2297 cm -~
(A1-MCM-58) and 2298 cm j (A1-MCM-68) are formed (Fig. 3B,b and Fig. 5B,b).
In the case of Fe-MCM-58, absorption bands with maxima at 3746 cm -~ for terminal
Si-OH groups and 3646 cm -I and 3520 cm -1 for bridging Si-OH-Fe groups (Fig. 4A,a) were
detected. Pyridine adsorption leads to the consumption of all bridging Si-OH-Fe groups
(Fig. 4A,b) and to the formation of a band at 1545 cm -I characteristic for the presence of the
pyridinium ion, which reflects the interaction of pyridine with Broensted sites. The band at
1448 cm -] originates from coordinatively bonded pyridine on Lewis sites (Figure 4B,b). The
determined concentrations of Broensted and Lewis acid sites for A1- and Fe-MCM-58, and
A1-MCM-68 are summarized in Table 1.
The concentration of the Broensted acid sites in A1-MCM-58 and A1-MCM-68 is
practically constant despite the nsi/nAI ratio, while it slightly decreases with increasing nsi/nFe
ratio. In other words, the higher the concentration of trivalent atoms in the zeolite, the higher
the concentration of Lewis sites. The concentration of Broensted sites strongly depends on the
type of the trivalent cation.
A J3746 A B 1448
3746
3630
e-
~c] 3646 "~.
,I0
O
a
J ~,
3564
i la /~ ~ 3500 \
J2
I ~ I ~ ~ I J I ,
Table 1
nsi/nAi and nsi/nFe ratios of zeolites A1-MCM-58, Fe-MCM-58, and A1-MCM-68 in the
initial gel and in the synthesized zeolite determined by XRF and FTIR and relative amount
of Lewis and Broensted acid sites.
While about 50 % of the acid sites are of the Broensted type in the Al-containing MCM-58
samples, only about 8-16 % of the total acid sites are Broensted sites in Fe-MCM-58. On
one side, the concentration of trivalent cations in the synthesized sample is larger for Fe than
for A1, on the other side, the concentration of Broensted sites is significantly lower for Fe-
MCM-58 than for A1-MCM-58. Due to the relatively large ionic radius of Fe 3+ (0.063 nm)
compared to Si 4+ and A13+ (0.040 nm and 0.053 nm, respectively), the incorporation of iron
into the framework is expected to be much more complicated.
The toluene alkylation with isopropylalcohol was investigated over zeolites A1-MCM-58,
Fe-MCM-58 and A1-MCM-68 possessing different nsi/nAl(Fe) ratios. The main product of the
isopropylation of the toluene is a mixture of cymenes. However, the reaction over aluminum
zeolites is accompanied by competitive reactions yielding benzene, xylenes, ethyltoluenes,
and n-propyltoluenes. The conversion of toluene was quite stable with T-O-S for all tested
zeolites (Fig. 6). The toluene conversion was high for aluminum zeolites MCM-58 and
MCM-68 (between 14 and 19 %) which possess relatively high amount of Bronsted sites.
n-Propyltoluenes and xylenes (except cymenes) were formed in significant concentrations
over aluminum containing zeolites. After 15 rain the selectivity to xylene was 1 4 - 23 % and
selectivity to n-propyltoluene between 6 and 9 % for alumosilicates. It was shown that the
inner structure of zeolites controls the formation of n-propyl benzenes or n-propyltoluenes in
alkylation reactions [6]. It was proposed that n-propyltoluene formation is taking place via a
bimolecular reaction between isopropyltoluene and excess of toluene in the reaction mixture.
The reaction between isopropyltoluene and toluene can take place on the channel intersection
of zeolites with three dimensional channel systems as in ZSM-5 or MCM-68. On the contrary,
this reaction is sterically hindered in zeolites with one dimensional channel system as
mordenite [ 16,17].
However, a significant amount of n-propyltoluene is also formed in MCM-58 that has one
dimensional channel system. It was explained by the size of the channel system (12 MR) and
by the location and geometry of SDA used for the synthesis. The SDA structure is similar to
65
A 232~
m k-
lOO
o
o~20 ,<
3
r
80 (1)
=
r
2.
3610 l__l/I 60 o
9
<_.
s
40
g
5 i 20
I J I i I i I , i , I , I , I J 0
Fig. 5. IR spectra of hydroxyl vibration region of Fig. 6. T-O-S dependence of toluene conversion
A1-MCM-68 with nsi/nAi = 9 before (a) and after and cymenes selectivity over A1-MCM-58,
(b) d3-acetonitrile adsorption (A) and spectra of nsi/nAi ratio = 26 (m), Fe-MCM-58, nsi/nvc ratio =
acetonitrile region before (a) and after (b) its 26 (A), and AI-MCM-68, nsi/nAi ratio = 19 (.).
adsorption (B).
4. C O N C L U S I O N S
Acknowledgement
Financial support of the Grant Agency of the Academy of Sciences of the Czech Republic
(B4040402). I would like to thank Dr. Libor Brabec for recording scanning electron
micrographs and Dr. Jifi (~ejka for motivation and helpful discussion.
REFERENCES
[1] M.C. Camblor, A. Corma and L.A. Villaescusa, Chem. Commun., 8 (1997) 749.
[2] W.W. Kaeding, L.B. Young and C.C. Chu, J. Catal., 89 (1984) 267.
[3] C. Perego and P. Ingallina, Catal. Today, 76 (2002) 87.
[41 J. 0ejka, G.A. Kapustin and B. Wichterlovfi, Appl. Catal. A, 108 (1994) 187.
[s] C.S. Cundy and P.A. Cox, Chem. Rev., 103 (2003) 663.
[61 J. 12ejka and B. Wichterlovfi, Catal. Rev., 44 (2002) 375.
[7] A. Corma, G. Sastre and P.M. Viruela, J. Mol. Catal. A, 100 (1995) 75.
[8] A. Corma, M.J. Diaz-Cabanas, F. Rey, S. Nicolooulas and K. Boulahya, Chem. Commun.
(2004) 1356.
[9] A. Burton, S. Elomari, R.C. Medrud, I.Y. Chan, C.Y. Chen, L.M. Bull and E.S. Vittoratos, J.
Am. Chem. Soc., 125 (2003) 1633.
[10] P. Prokegovfi, S. Mintov~i, J. Cejka and T. Bein, Microporous Mesoporous Mater., 64 (2003)
165.
[ 11 ] S. Ernst, M. Hunger and J. Weitkamp, Chem. Ing. Tech., 69 (1997) 77.
[12] D.C. Calabro, J.C. Cheng, R.A. Crane, C.T. Kresge, S.S. Dhingra, M.A. Steckel, D.L. Stern and
S.C. Weston, US Patent No. 6 049 018 (2000).
[13] B. Wichterlovfi, Z. Tvarfi~kovfi, Z. Sobalik and P. Sarv, Microporous Mesoporous Mater., 24
(1998) 223.
[14] C.A. Emeis, J. Catal., 141 (1993) 347.
[15] G. Kogov/t and J. 0ejka, Collect. Czech. Chem. Commun., 67 (2002) 1760.
[16] J. {2ejka, N. Zilkovfi, J. E. Sponer and B. Wichterlovfi, Collect. Czech. Chem. Commun., 63
(1998) 1769.
[17] J. Sponer, J. Sponer, J. (~ejka and B. Wichterlovfi, J. Phys. Chem. B, 102 (1998) 7169.
[18] J. {2ejka, A. Krej6i, N. 2;ilkovfi, J. Kotrla, S. Ernst and A. Weber, Microporous Mesoporous
Mater., 53 (2002) 121.
Studies in Surface Science and Catalysis,volume 158
J. 0ejka, N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 67
Nanostacked zeolite having MFI-structure contained titanium metal ions on its framework,
has been prepared via microwave irradiation. Nanofabrication of titanium containing MFI
zeolitic materials under microwave irradiation was successfully implemented by the adhesion
of crystallites through the condensation of hydroxyl groups on surface via selective adsorption
of microwave energy onto high dielectric metal species including Ti species as a nanoglue.
This microwave technique gave births of both the uniform size of hockey puck shaped
crystals and the fibrous stacked morphology of MFI crystals. This material gave superior
adsorption selectivity of linear hydrocarbons such as n- hexane, n-octane and both increased
of activity and selectivity in the epoxidation of styrene into styrene epoxide.
1. INTRODUCTION
Microwave techniques have attracted growing attention for the rapid synthesis of nanoporous
materials requiring several days to prepare under hydrothermal conditions [1-10]. Potential
advantages include a rapid and more simultaneous nucleation and growth, homogeneous
heating throughout the reaction vessel, superheating, and suppression of undesired phases
compared with conventional hydrothermal techniques. To date, several types of mesoporous
materials as well as microporous zeolites have been synthesized by microwave irradiation.
Microwave techniques in the synthesis of inorganic materials are generally known to be faster
and simpler than conventional methods [11]. Energy transfer from microwaves to the
materials is believed to occur either through resonance or relaxation, which results in rapid
heating. Furthermore, the employment of the microwave technique in the synthesis of
nanoporous materials have been shown to provide versatile effects, for instance, short heating
times, inductive heating through the conducting properties of the synthesis mixture, specific
energy dissipation via microwave energization of the hydroxylated surface or associated
wafer molecules in the boundary layer, and formation of the high potential of the active water
molecules [2, 6, 8].
Zeolite molecular sieves are good candidates for designed microscopically and
nanoscopically engineered materials due to their well-defined pore structures [12-13]. Among
them, silicalites, zeolites having the MFI framework topology, have been studied extensively
with regards to the selective separation of hydrocarbons [14] and the development of
electrode materials [15], membranes [16], and catalysts [17]. Because many emerging
applications of porous materials require precise control of crystal size, shape, and orientation
[ 18, 19], the development of new synthesis strategies is increasingly important. For example,
the properties of spherical mesoporous silica particles with a narrow particle size distribution
as a material for column chromatography are superior to those of curved hexagonal rods with
68
the same structure [19]. Recently, it has been reported that the microwave synthesis method
could provide an efficient way to control particle size distribution [20], phase selectivity [21 ],
and macroscopic morphology [22] in the synthesis of porous materials as well as result in fast
crystallization [23]. However, to our knowledge, there are few reports of microwave
techniques being utilized to control the morphology of porous materials [22]. Here, we report
on new applications of the microwave method for the fabrication of MFI-type zeolite crystals
incorporating Ti and other metal species. By using the method, we prepared fibrous MF!
zeolites that exhibited useful physicochemical properties due to their self-stacking. Pure and
metal-incorporated MFI crystals were synthesized by microwave-heating precursor solutions
at 438 K. Our recent results reveal that the TS-1 can be used as a convenient catalyst for the
shape selective production ofp-xylene [24].
In this report, nanostacked Ti-MFI zeolite was successfully prepared by microwave
irradiation. It was demonstrated adsorption of linear aliphatic hydrocarbons such as n-hexane
and n-octane in liquid phase as a superior catalyst in the selective epoxidation reaction of
styrene into styrene epoxide.
2. E X P E R I M E N T A L
Nanostacked Ti-MFI zeolite was synthesized using tetrapropyl orthotitanate (TPOT, Aldrich,
98 %), TEOS (Tetraethyl orthosilicate, TCI, 98 %), isopropyl alcohol (IPA, TCI) and
deionized water. Tetrapropylammonium hydroxide (TPAOH) was used as a structure-
directing agent. Typically two kinds of different solution A and B were prepared respectively.
The total molar ratio of gel composition is 0.1TEOS: 0.0014 TPOT: 0.02 TPAOH: 0.1IPA:
2.22H20. In case of A solution, firstly 20.35mL (0.02mol) of TPAOH was added in 22.2mL
of deionized water and then 22.38mL (0.1tool) TEOS were added in this mixture with
continuously stirring to give the homogeneously mixed sol-gel solution. In a separate vessel B,
TPOT was dissolved in 6 mL of isopropyl alcohol by continuously stirring. Finally solution B
was poured into the A with vigorous stirring and heated up to 80 ~ to remove isopropyl
alcohol for 1 hour. The precursor gel of 30-60 g was loaded in a 100 ml Teflon
autoclave, which was sealed and placed in a microwave oven (CEM, Mars-5, maximum
power of 1200 W). The precursor gel was heated to the reaction temperature of 180~ in less
than 2 min (microwave power was 600 W) and kept for 1 or 2 h (microwave power was 300
W). After the reaction, the autoclave reactor was cooled to room temperature, and the solid
product was separated with filtration and washed with deionized water for several times. It
was dried overnight at 105 ~ There was no extra separation procedure to remove impurities
or amorphous gel. The molecular sieves were calcined at 550~ in an electric furnace.
The crystallinity of stacked Ti-MFI sample was determined with an X-ray diffractometer
(Tigaku, D/MAX IIIB) using Cu Ka radiation, and titanium contents were determined with an
inductively coupled plasma (ICP, Jovin Yvon Ultima-C). FT-IR spectra were obtained with a
Nicolet Impact 410 spectrometer in the range of 500-4000cm ~. The measurement of
chemisorption property upon these linear hydrocarbons was carried out with a Chemisorb
2705 Micromeritics.
Styrene epoxidation was carried out using a glass flask equipped with a reflux condenser,
a magnetic stirrer, a thermometer and a septum to withdraw samples periodically for analysis.
Catalyst of 0.35g, 18 mL acetone as solvent, 5 mL styrene (99%, Aldrich) and 1.3 mL
aqueous hydrogen peroxide solution (30 wt%) were loaded into the flask. The molar ratio of
styrene/H202 was 3.4. The reaction was performed at 70 ~ under ambient atmosphere. The
69
liquid mixture was separated from catalyst, and analyzed using a gas chromatography
(YoungLin GC, Korea) equipment with a flame ionization detector and a capillary column
(DP-sil-5 CB). The column temperature was 100 ~ and the carrier gas was helium..
MFI-type zeolite crystal incorporated with Ti species was prepared by microwave irradiation
and applied to both the oxidative styrene epoxidation reaction and the length shape-selective
sorption reaction with two kinds of linear hydrocarbons such as n-hexane and n-octane. The
powder X-ray diffraction (XRD) patterns of Ti-MFI sample which is prepared by microwave
and hydrothermal treatment show the characteristic peaks for the MFI structure without any
impurities. The IR spectra of the calcined Ti-MFI samples exhibit an absorption band at 960
cm -i for the stretching mode of a [SiO4] unit bonded to a Ti4+ ion (O3SiOTi). UV/Vis diffuse
reflectance spectra, collected in vacua after evacuation ( _ 10.5 torr) at 627 K, revealed a
strong peak around 210-240 nm, assignable to the ligand-to-metal charge-transfer absorption
band from the oxygen to tetrahedral titanium (IV) atom in the zeolite framework.
Fig. 1. SEM images of (a) fibrous Ti-MFI zeolite with microwave irradiation (stacked Ti-MFI-MW)
and (b) Ti-MFI with microwave irradiation (non-stacked Ti-MFI-MW, nonstaking) (c) hockey-puck
shape Ti-MFI zeolite with hydrothermal heating (Ti-MFI- HT).
Fig. la shows both the uniform size of hockey puck shaped crystals and the stacking of
these crystals into fiber or rod types in the synthesis of zeolitic materials such as MFI type
zeolites. The formation of these nanostacked MFI-type zeolite crystals having nanorod shape
was ascribed to the presence of metal species, which selectively absorbed microwave
irradiation. Self-assembled zeolite nanostructures can be also obtained by modification of the
zeolite surface by putting metal species as nanoglue.
The nanofabricated materials like the stacked Ti-MFI zeolite showed significantly
enhanced shape selectivity in adsorption. Fig. 2 shows the temperature-programmed
desorption (TPD) profiles of linear hydrocarbons after vapor-phase adsorption of an
equimolar mixture of n-hexane and n-octane over stacked Ti-MFI-MW and Ti-MFI-HT at 100
~ The linear hydrocarbons TPD patterns are similar over the two adsorbents, that is, two
peaks for n-hexane and n-octane, while the fibrous Ti-MFI-MW reveals a remarkably
enhanced adsorption selectivity on n-hexane. Moreover, it is notable that stacked Ti-MFI-MW
selectively desorbs n-hexane in the temperature range 100-200 ~ (Fig. 2a), implying that this
material can be applied in producing a selected n-hexane from a linear hydrocarbon mixture
when combined with temperature swing desorption.
70
119
(a) 183
0
'E
\ ~ 266
Temperature (~
106
239
C
O
Temperature (oC)
Fig. 2. TPD profiles of linear hydrocarbons by using (a) stacked Ti-MFI-MW and (b) non-stacked Ti-
MFI-HT after vapor-phase competitive adsorption of an equimolar mixture of n-hexane (red full line)
and n-octane (black full line). Adsorbent weight: 0.25g; carrier gas: nitrogen (flow rate = 30 mLmin -l)
The MFI structure was known to offer three different sites for adsorption: straight and
sinusoidal channels along the [010] and [ 100] directions, respectively, and their intersections.
The diffusivity in the straight channel is reportedly much higher than in the sinusoidal channel.
Because of the crystal stacking along the [010] direction, the straight channels of Ti-MFI-MW
are longer and their openings are much fewer than those of the non- stacked Ti-MFI-HT [25].
Fig. 3 shows the rate of styrene conversion in oxidative epoxidation of styrene over Ti-MFI
zeolite at 373K, which were prepared by microwave irradiation and hydrothermal treatment.
All samples show initial induction periods in the first 30 min of the reactions, in which the Ti-
MFI-HT catalyst has a relatively lower conversion rate than those of stacked or non-stacked
Ti-MFI-MW. After this initial period, the oxidative styrene conversion rate over all samples
gradually increases. The highest rate at about 3 hour: the conversion rate of styrene over Ti-
MFI-HT and stacked Ti-MFI-MW is nearby 37% and 40% respectively, while stacked Ti-
MFT-MW catalyst reaches around 150%. In addition, with the increase of reaction time, the
71
tendency of styrene conversion over these two samples, Ti-MFI-HT and non-stacked Ti-MFI-
MW, is almost the same. In case of stacked Ti-MFI-MW, the conversion of styrene, however,
gives higher value from 2 hours.
70t ~ / / u ~ n ~ u
60-
50,
u) /
56 /
i 40.
54 "/
o
30, 52 ~ ,,/'~/
1- 50J --n-- Ti-MFI-MW(stacking)
42 ' I ' I ' I ' I ' I ' I ' I ' I ' I ' I ' I '
0 ' i ' i , I ' I ' i '
0 1 2 3 4 5 6 7 8 9 10 11 12
0 2 4 6 8 10
At the same time the total selectivity on aromatic styrene epoxide through the stacked Ti-
MFI-MW (65 %) is higher than non-stacked Ti-MFI-MW (62%) and Ti-MFI-HT (55%).
Catalytic activities of the stacked Ti-MFI-MW are increased after catalytic reaction for 6
hours, while non-stacked Ti-MFI-MW and Ti-MFI-HT are gradually decreased at the same
time. It is clear that the reason of high conversion of stacked Ti-MFI-MW in comparison with
the other two samples is likely to induce from the nanofabrication such as nanostaking which
was formed through hydration reaction between hydroxyl group existing on the surface of
microporous material and adsorption of microwave energy. In addition, it seems that high
surface area through crystal growth of c- direction and the metal incorporation at a suitable
concentration level by selective adsorption of microwave energy will be able to enhance
catalytic property.
In summary, metal elements, which are located on the surface of zeolite crystals, behave
as a nanoglue between the crystals through selective absorption of microwave energy.
Nanostacked Ti-MFI zeolite by microwave is of great interest for various applications, for
example, in the oxidative epoxidation reaction between styrene and hydrogen peroxide as
well as selective separation of linear hydrocarbon from a mixture. Nanofabrication of zeolite
crystals show several interesting properties such as lower packing density, higher
hydrophobicity, enhanced catalytic properties, and higher shape-selectivity in adsorption as
compared to non-fabricated MFI zeolite.
72
REFERENCES
[1] A. Arafat, J.C. Jansen, A.R. Ebaid, H. van Bekkum, Zeolites 13 (1993) 162.
[2] C.S. Cundy, R.J. Plaisted, J.P. Zhao, Chem. Commun., (1998) 1465.
[3] Y.K. Hwang, J.-S. Chang, Y.-U. Kwon, S.-E. Park, Stud. Surf. Sci. Catal., 146 (2003) 101.
[41 S.H. Jhung, J.-S. Chang, Y.K. Hwang, S.-E. Park, J. Mater. Chem., 14 (2004) 280.
[5] S.H. Jhung, J.-S. Chang, J.S. Hwang, S.-E. Park, Microporous Mesoporous Mater., 64 (2003)
33.
[6] S.-E. Park, Y.K. Hwang, D.S. Kim, S.H. Jhung, J.S. Hwang, J.-S. Chang, Catal. Surv. Asia, 8
(2004) 91.
[7] C.G. Wu, T. Bein, Chem. Commun., (1996) 925.
[8] S.-E. Park, D.S. Kim, J.-S. Chang, W.Y. Kim, Catal. Today, 44 (1998) 301.
[9] H.M. Sung-Suh, D.S. Kim, Y.K. Park, S.-E. Park, Res. Chem. Intermed., 26 (2000) 283.
[10] D.S. Kim, J.M. Kim, J.-S. Chang, S.-E. Park, Stud. Surf. Sci. Catal., 135 (2001) 333.
[11] (a) J. Zhu, O. Polchik, S. Chen, A. Gedanken, J. Phys. Chem., B104 (2000) 7344; (b) K.J. Rao,
B. Vaidhyanathan, M. Ganguli, P.A. Ramakrishnan, Chem. Mater., 11 (1999) 882.
[12] (a) G. A. Ozin, Adv. Mater. 1992, 4, 612; b) T. Bein, Chem. Mater., 8 (1996) 1636.
[13] M.E. Davis, Nature, 417 (2002) 813.
[14] (a)W. Yuan, Y. S. Lin, W. Yang, J. Am. Chem. Soc., 126 (2004) 4776; (b) .S. Li, C.
Demmelmaier, M. Itkis, Z. Liu, R. C. Haddon, Y. Yan, Chem. Mater., 15 (2003) 2687.
[15] S. Li, X. Wang, D. Beving, Z. Chen, Y. Yan, J. Am. Chem. Soc., 126(2004)4122.
[16] G. Xomeritakis, Z. Lai, M. Tsapatsis, Ind. Eng. Chem. Res., 40 (2001) 544.
[17] (a) B. Notari, Adv. Catal. 41 (1996) 253; (b) V. Ramaswamy, B. Tripathi, D. Srinivas, A. V.
Ramaswamy, R. Cattaneo, R. Prins, J. Catal., 200 (2001) 250.
[18] C. Thoelen, J. Paul, I. F. J. Vankelecom, P. A. Jacobs, Tetrahedron: Asymmetry, 11(2000) 4819.
[ 19] Z. Lai, G. Bonilla, S. Diaz, J.G. Nery, K. Sujaoti, M.A. Amat, E. Kokkoli, O. Terasaki, R. W.
Thompson, M. Tsapatsis, D.G. Vlachos, Science, 300 (2003) 456.
[20] (a) C. S. Cundy, Collect. Czech. Chem. Commun., 63 (1998) 1699; (b) C. Gabriel, S. Gabriel, E.
H. Grant, B. S. J. Halstead, D. M. P. Mingos, Chem. Soc. Rev., 27 (1998) 213.
[21 ] S.H. Jhung, J.-S. Chang, J.-S. Hwang, S.-E. Park, Microporous Mesoporous Mater., 64 (2003)
33.
[22] (a) S. H. Jhung, J.-S. Chang, Y. K. Hwang, S.-E. Park, J. Mater. Chem., 14 (2004) 280; (b) Y. K.
Hwang, J.-S. Chang, Y.-U. Kwon, S.-E. Park, Microporous Mesoporous Mater., 68 (2004) 280.
[23] (a) B. L. Newalkar, S. Komameni, H. Katsuki, Chem. Commun., 2389 (2000); (b) S. Mintova,
S. Mo, T. Bein, Chem. Mater., 10 (1998) 4030.
[24] Y.K. Hwang, J.-S Chang, S.-E. Park, D. S. Kim, Y.-U. Kwon, S. H. Jhung, J.-S. Hwang, and
M. S. Park, Angew. Chem. Int. Ed., 44 (2005) 557
[25] S. Nair, M. Tsapatis, J. Phys. Chem. B, 104 (2000) 8982.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 73
c IEM UMR-563 5 CNRS, 1919 route de Mende, 34293 Montpellier, Cedex 5, France
The synthesis and the detail investigation of the formation of LTA- and FAU-type zeolites at
room temperature (RT) are reported. The complete transformation of the amorphous
precursors into zeolite A and X took place for 3 and 21 days, respectively. The events taking
place at the nanometer scale were studied by high resolution transmission electron
microscopy (HRTEM). The study was completed by small angle X-ray scattering (SAXS) and
X-ray diffraction (XRD).The sluggish crystal growth kinetics at ambient conditions permitted
to track down the entire sequence of crystallization events from the formation of the initial gel
to the complete transformation into a zeolite-type material. Liquid inclusions denoted as
"negative crystals" were found within the volume of the gel particles for both systems.
Further, the crystal growth was followed by HRTEM, which revealed that the crystallization
in the two systems is governed by different crystallization mechanism.
1. INTRODUCTION
There are continuous efforts in understanding the mechanism of zeolite crystallization [1-5].
The interest is driven from both fundamental and practical point of view [6]. The
development of new powerful characterization techniques allows the constant acquisition of
new data about the crystallization mechanism of different zeolite systems. However, the
zeolite crystallization is complex and unique for a particular zeolite type and no universal
mechanism could be proposed. Even for well-known zeolites like NaA and NaX, which are
amongst the most studied zeolites, there are still uncertainties, regarding the mechanism of
their formation [7-10].
The events in the early stages are of paramount importance in determining the course of
the subsequent crystallization and it is clear that a detailed understanding of these phenomena
should be obtained if the goal of a rational control of the synthesis of these materials is to be
achieved. Different sets of complementary techniques have been used to gain an insight into
the complex processes occurring during the nucleation [11, 12]. However, a serious obstacle
in studying the zeolite formation are the invasive post-synthesis sampling procedures that may
provoke redistribution of the (alumino)silicate species, particle growth, and possibly
74
nucleation. Part of these difficulties were circumvented by in situ investigations, for instance
NMR [13,14] IR/Raman [15] EXAFS [16] optical and electron microscopy [17], light-
scattering techniques [18,19] X-ray diffraction [20-22] and combinations of SAXS/SANS
[23,24], SAXS/WAXS [25,26] and synchrotron X-ray radiation [27], which provided valuable
information about the zeolite crystallization mechanism. Nevertheless, important questions
like for instance where the first unit cells of zeolite structure are formed, their spatial
arrangement in the gel system, and what are the species depositing on the growing crystals
remain to be elucidated.
We have found a simple and effective way to study zeolite the crystallization by
synthesizing zeolites at room temperature. The slow crystal growth kinetics at room
temperature allows to discriminate the different stages of the zeolite formation and to study
the products of crystallization with minimum invasive post-synthesis treatments. The present
study reports on the synthesis and investigation of the nucleation and crystal growth
mechanisms of LTA- and FAU-type zeolites under ambient conditions.
2. E X P E R I M E N T A L
2.1. Synthesis
In order to obtain LTA- and FAU-type zeolite in a reasonable period of time under
ambient conditions, a very reactive systems were employed. A silica source containing low-
mass silica species was used and further de-polymerized by adding sodium hydroxide so as to
obtain a completely transparent initial solution. The latter was mixed with a clear aluminate
solution where AI(OH)-4 is the dominant anionic species [1]. The zeolite precursor mixtures
were produced by vigorous mixing of the alkaline silicate and aluminate solutions, prepared
as described above, where all components were expected to be relatively homogeneously
distributed. The compositions of the gels employed for the syntheses were as follows:
Zeolite A: 6Na20 : 0.2A1203 : 1.0SiO2 : 150H20
Zolite X: 4Na20 : 0.2A1203 : 1.0SiO2 : 200H20
The reactants used for the preparation of the gel were: sodium hydroxide pellets (97%,
(Aldrich), sodium aluminate (54.3% A1203, 44.5% Na20, Riedel-de HaEn), sodium silicate
solution (14 % NaOH, 27% SiO2, Fluka) and distilled water. The mixture kept its transparent
character for several seconds and after that turned into a milky-white low viscosity gel. The
initial mixtures were stirred for 90 min and transferred into polypropylene bottles. The sealed
bottles were stored in an oil bath (T-25 +I~ The milky-white color and the observed
sedimentation revealed that the gel comprised relatively large particles. Small aliquots of the
mixtures were taken at different periods of time. Prior to taking aliquots, the samples were
shaken in order to disperse the settled particles. During the first day, three samples were taken
from the system yielding FAU-type material, immediately after homogenization of the gel
(1.5 h), after 6 h and after 24 h. Thereafter, further samples were taken every 2-3 days.
Samples from the zeolite A yielding system were taken after 0, 6, 21, 29, 44, 52, 67, 73 and
96 h. The solids were recovered by high speed centrifugation (g=50 000) and re-dispersion in
distilled water under ultrasonic radiation. Part of each solid was dried at room temperature
and the other part was kept as a suspension.
2.2. Characterization
The structural evolution of the gel species was tracked down with a high-resolution
transmission electron microscopy (HRTEM), in a FEI-Philips CM300 microscope in low-
dose mode operating at 300 kV accelerating voltage, equipped with a LaB6 electron gun, and
75
EDAX energy dispersive X-ray spectrometer (EDS). Samples were prepared by diluting the
suspension of the reaction products in distilled water, ultrasonicating them, and depositing a
drop onto Cu grids coated with a thin (5 nm thickness) holey carbon support film. In order to
verify that the zeolite crystals do not collapse under the high vacuum in the TEM, some of the
samples were studied at liquid nitrogen temperature using Gatan Inc. cryo-system, model 626.
No apparent increase in crystallinity or size distribution was observed, which allowed us to
conclude that the utilization of low-dose technique at room temperature is an adequate
method for characterization.
To complement this study, X-ray diffraction (XRD) and small angle X-ray scattering
(SAXS) were performed. The XRD diagrams of the samples were recorded on a STOE
STADI-P diffractometer in Debye-Scherrer geometry equipped with a linear position-
sensitive detector (6 ~ in 20) and employing Ge monochromated Cu K~, radiation. SAXS
measurements were done with a high resolution spectrometer (Bonse Hart Camera).
The XRD studies of the product solids showed that zeolite A and X were successfully
synthesized under RT conditions. The crystallization of zeolite X took much longer time than
that of zeolite A. No changes in the X-ray crystallinity of the FAU-type material were
observed after 3 weeks of RT synthesis, whereas zeolite A reached maximum crystallinity
only after 3 days of treatment. On the other hand, the induction period for zeolite A was much
longer in respect to zeolite X if we consider the ratio between the induction and crystallization
time. Thus, the induction period for zeolite A was about 2/3 (44 h), while for zeolite X was
1/2 (ca. 10 days) of the total crystallization time. The difference in the relative induction times
is surprising having in mind the higher alkalinity and lower water content of the zeolite A
yielding system. These differences in the composition should provide a higher nucleation rate
in the case of zeolite A. Besides the ratio between the induction and the total crystallization
times, the two zeolites seem to differ in their particle size. As can be seen, the XRD peaks of
the FAU-type material are relatively broad indicating that the product consists of very small
crystallites (Fig. 1A). In contrast, the peaks of the LTA-type zeolite do not suggest very small
particle size (Fig. 1B). Both systems were subjected to a detailed study in order to reveal the
crystallization mechanism, which controls the formation of the two zeolites. The nanoscale
events leading to the transformation of the amorphous precursors into zeolite X and A were
followed by means of HRTEM.
Our previous investigation [28] showed that the gel yielding zeolite X had a grain-like
structure (Fig. 2A). The grain-like appearance of the gel phase was due to the presence of
A B
=.
v v
>,
c- c-
(D
c-
15 25 35 4 14 24 34
Fig. 1. XRD patterns of zeolite X (A) and A (B) synthesized at RT for 21 and 3 days, respectively.
76
brighter, relatively isometric areas, which were almost homogeneously distributed within the
volume of the gel particles (Fig. 2B). Based on EDS analysis and contrast differences, these
areas in the gel phase were attributed to silica-rich liquid inclusions trapped within the
amorphous mass (Fig. 2B). The faceted shape of these structures observed in the TEM images
at high magnification (not shown) suggests the presence of ordered structure at the interface
dense gel/liquid inclusion. The bright areas in the gel phase were denoted as "negative
crystals" [28], which is a term employed to describe faceted cavities inside solid materials or
crystal filled with mother medium, which could be in the form of solution or gas [29].
The structure of the zeolite A precursor was somewhat different, namely the gel grain-
like appearance characteristic for zeolite X was not observed (Fig. 2C). Careful inspection of
the gel confirmed that the initial gel phase was relatively uniform and did not contain areas
with different contrast (Fig. 2D). Such areas appeared after 6 h of RT synthesis. Their number
gradually increased during the induction period and after 29 h all gel particles contained such
spots. Besides the number, the structures of these areas changed with the
Fig. 2. TEM micrographs showing the structure of the gel material after 90 min mixing of the
reactants: (A, B) zeolite X and (C, D) zeolite A precursors.
77
crystallization time. The round shape was typical for the bright spots at the beginning of the
induction period, while at the end of the process these spots contained straight edges
suggesting some level of organisation in these areas. This evolution is obviously a part of the
gel transformation, which precedes the onset of the crystallization. Hence, the "negative
crystals" observed in the solid amorphous precursors are related with the process of
nucleation. Something more, the fact that their structure changes along the induction period
suggests that important processes take place in these areas. These findings make us think that
the "negative crystals", in particular at the interface solid/liquid, are most likely the places for
zeolite nucleation.
It is somewhat strange, however, that the negative crystals were observed in the gel
yielding zeolite X just after 90 min of mixing of the reactants, whereas these structures
appeared during the long induction period in the case of zeolite A. These differences in the
evolution of the systems suggest different mechanisms of formation or at least different
duration of the crystallization events. Indeed, the combined TEM/synchrotron XRD analysis
at the initial stage of crystallization revealed that zeolite nuclei were formed during the
homogenization (90 min) of the initial reactants [28]. The zeolite A yielding system, was
carefully inspected by means of HRTEM supported by XRD and SAXS analyses and no
evidences for any nuclei were found. Obviously, the nucleation takes place during the long
induction period and the first crystalline particles were detected both by XRD and HRTEM
after about 40 h of RT synthesis. Therefore, the time interval when the "negative crystals" and
the zeolite nuclei appeared match well, which is another proof that these events are related.
The evolution of the zeolite A yielding system includes the following main stages. After
the homogenization of the reactants, a gel composed of aggregates with globular form was
obtained (Fig. 2C and D). These aggregates maintained their form all over the induction
period. After 44 h of RT synthesis the structure of the gel changed, i.e., became more open
and angular particles could be observed. Together with the amorphous zeolite A precursor
particles, crystallites with a size of 10-15 nm were detected in the lattice images with a d-
spacing of 12 A. The presence of X-ray crystalline LTA-type material at this stage was
confirmed by XRD. Samples taken after 52, 67 and 73 h of RT synthesis showed continuing
reorganization of the solid and decreasing of the amount of the material with amorphous
appearance. The material synthesized for 73 h contained mainly single crystals with sizes
between 100 and 400 nm. Limited number of aggregates built of smaller zeolite crystal was
also observed.
Further, the morphology and interface of the zeolite A precursor gel and the samples
obtained after 1, 2, 3 and 4 days of treatment were characterized by SAXS. For all samples,
the scattering curves present a similar feature in two parts, where the slow decrease of the
scattering intensity is followed by a marked decrease (Fig. 3). The linear slopes obtained for
the. first decrease were similar for all the samples (around-2). The value of the second slope
varies from-3.3 for the initial gel t o - 4 for the sample synthesized for 4 days. The values of
the slope at high q values (Porod domain) are indicative of the surface roughness of the
particles. The particles obtained after one and two days had rough surfaces, while those
synthesized for 3 and 4 days were smooth. This behaviour is confirmed by plotting the I(q)*q4
data as a function of q. Fig. 4 shows the scattering curves in the Porod domain plotted under
the form I(q)*q4 vs. q. The curves corresponding to 3 and 4 days treated samples exhibited
oscillations in the intensity around the Porod law, whereas the curve corresponding to the
sample treated for 2 days did not show such oscillations. The oscillations clearly indicate that
thes particles are well-defined in size and shape. In contrast, for the samples treated for less
78
e
1011
d v_ v_vv~
10 7
10 s ' ' ' ' ' 1 ' ' ' ' ' ' ' ' 1 '
1 0 .3 10 .2
Q (A-')
Fig. 3. Scattering curves of the zeolite A samples treated at various times, a=initial gel, b-1 day, c=2
days, d=3 days and e= 4 days.
1,5x10 ~ -
1,0xl0 -~ o%
o ~
[ 2 ] -
o
o C
13"
D
~o~ ~ ~ o
oO ~ ~
~oO
~ 1 7 6qOo
v
D • c~oo~O%c~ od Oo o
5,0x10 -2
[]~o-~ %~:r~~~ ~ ~ b
+o~ +4+++++++++~_~+~4+++§ a
0,0 ' I ' I ' I
q (A-')
Fig'. 4. Plot of I(q)*q 4 as a function of q for zeolite A samples treated at various times, a=3 days, b=4
days and c=2 days.
than 3 days, the absence of oscillations indicates that the size of the zeolite A precursor
particles is not well-defined and their surface is not smooth.
The SAXS data for zeolite A are in agreement with the main crystallization stages
depicted by HRTEM, i.e., (i) the initial gel possess the characteristic features of amorphous
material and any traces of crystalline particles were not found; (ii) the induction period
comprises the first two days; (iii) the crystallization is accomplished after 3 days of RT
synthesis.
79
Fig. 5. Zeolite A (A) and X (B) synthesized at RT for 3 and 21 days, respectively.
4. CONCLUSIONS
We have performed a room temperature synthesis of zeolites A and X for relatively short
periods of time. The slow crystallization kinetics allowed a detailed study of all stages of the
zeolite formation, from mixing of the reactants to the complete transformation of the gel into
a crystalline zeolite-type material. The growth mechanisms governing the crystallization of
the two zeolites were found different providing products with fairly different morphological
features. Zeolite X crystallized as 100-300 nm spherical aggregates built of 20-30 nm
nanocrystals, whereas 100-400 nm individual crystals dominated in the zeolite A product.
Although the consequence of the events in the two systems was not similar, the
formation of negative crystals was observed in both gels. Their formation precedes the
nucleation of the system. The interface between the highly hydrated inclusions and the solid
gel in these areas is considered as the place, where the zeolite nucleation takes place.
80
This study demonstrates the utility of room temperature syntheses for studying the
zeolite crystallization. A further advantage of the room temperature synthesis is the formation
of zeolite nanocrystals with a high yield from organic-free gel systems.
ACKNOWLEDGEMENTS
The financial support from the CNRS-DFG and Procope bilateral programs is gratefully
acknowledged.
REFERENCES
Institute for Chemical and Bioengineering, Swiss Federal Institute of Technology (ETH), CH-
8093 Zurich, Switzerland
1. INTRODUCTION
Since their discovery in the early 90's [1,2] ordered mesoporous molecular sieves have
attracted much attention because of their potential use as catalysts, catalyst supports and
adsorbents. It has turned out that their acidity and hydrothermal stability is relatively low
compared with conventional zeolites [3]. This prevents their industrial application in catalytic
reactions [4]. The relatively low acidity and hydrothermal stability has been attributed to the
amorphous nature of the pore walls. Several strategies to enhance the acidity and
hydrothermal stability of mesoporous aluminosilicates by introducing some crystallinity into
the pore walls have been developed. Recently the assembly of preformed nanosized zeolite
precursors with surfactants has been reported by Liu et al. and Han et al. [5-9]. Kremer et al.
reported the synthesis of a material called ,,Zeogrid" [10-12]. Zeogrid was obtained by
precipitation of a silicalite-1 precursor solution, which had been aged at room temperature,
with a surfactant and subsequent calcination.
We report the synthesis of materials obtained by combining the above-mentioned
strategies. A clear ZSM-5 synthesis solution is aged at room temperature. After addition of
the surfactant (cetyltrimethylammoniumchloride) the resulting suspension is hydrothermally
treated at 100 ~ We investigated the effect of the aluminium content, the time of aging and
the duration of the hydrothermal treatment on the structure and on the catalytic activity of the
final materials.
82
2. E X P E R I M E N T A L
2.1. Synthesis
The chemicals used for the syntheses were tetraethoxy silane (TEOS) (>98%, Fluka),
tetrapropylammonium hydroxide (TPAOH) (1.0 M in water, Aldrich), tetramethyl-
ammonium hydroxide (TMAOH) (2.8 M in water, Fluka), cetyltrimethylammonium chloride
(CTAC1) (25 wt% in water, Fluka), aluminium isopropoxide (>98%, Aldrich), sodium
hydroxide and distilled water. First a clear solution with the composition 9 TPAOH : 25
TEOS : x A1203 : 0.03 Na20 : 420 H 2 0 w a s prepared (x = 0, 0.125, 0.25). This solution was
aged at room temperature. Addition of the surfactant (CTAC1) resulted in a suspension with
the composition 9 TPAOH : 25 TEOS : x A1203 : 0.03 Na20 : 3.1 CTAC1 : 840 H20. This
suspension was hydrothermally treated at 100 ~ under static conditions. In a typical
synthesis 1.5 g H20, 62 mg NaOH, 306 mg AI(;OPr)3 and 27.54 g TPAOH were mixed
together and stirred at room temperature for 2 h. Simultaneously a mixture of 31.38 g TEOS
and 27.54 g TPAOH was prepared and stirred at room temperature for 30 min. Both solutions
were mixed together and aged at room temperature for 48 h. A surfactant solution containing
23.94 g CTAC1 (25 wt% in water) and 27.42 g H20 was prepared. To 14.67 g of the aged
zeolite clear solution 8.56 g of this surfactant solution was added. The resulting white
suspension was hydrothermally treated at 100 ~ for 2, 8, 24, 48 or 98 h. After filtration and
washing with water calcination in air at 540 ~ for 5 h gave the final materials in the sodium
form. The ammonium form of the materials was obtained by treating the samples three times
with 0.1 M NH4NO3 in water at room temperature for 10 h.
The samples are labelled the following way: TPAx_y_z where x stands for the Si/AI
ratio, y for the time of aging and z for the duration of the hydrothermal treatment. The label
TMAx_y_z is used when TMAOH was used as the structure-directing agent instead of
TPAOH.
2.2. Characterisation
Nitrogen physisorption measurements were performed at liquid nitrogen temperature
with a Micromeritics TriStar 3000 apparatus. Prior to the measurements the samples were
degassed at 400 ~ at 10 Pa for at least 4 h. Powder XRD patterns were measured with a
Siemens D5000 powder diffractometer (CuKotl) with Bragg-Brentano geometry. IR
measurements were performed on a Bio-Rad Excalibur FTS 3000 IR spectrometer. For the
pyridine adsorption the samples were activated at 400 ~ and then treated for 10 min at 100~
with a flow of 30 ml/min He which was saturated at 0 ~ with pyridine (vapour pressure 550
Pa). After the adsorption the samples were kept under a constant flow of He for 1 h at 100 ~
for 1 h at 200 ~ and for 1 h at 300 ~ The IR spectra were recorded at the end of each
heating period. Si/A1 ratios have been determined by atomic absorption spectroscopy on a
Varian SpectrAA 220 FS spectrometer equipped with an air/C2H2 or a N20/C2H2 burner.
Raman Spectra were recorded on a Bruker Equinox 55 FRA 106/S Spectrometer.
catalytic test they were activated in situ at 400 ~ for 1 h in a stream of He. For the catalytic
test a stream of 24.0 ml/min He was saturated at 15 ~ with cumene (320 Pa). This stream
was diluted with a second stream of He leading to a total flow of 51 ml/min and a partial
pressure of 150 Pa of cumene. In the case of TPB as reactant a total flow of 30 ml/min of He
saturated at 69 ~ with TPB (150 Pa) was used. For comparative purpose a commercial ZSM-
5 (Zeocat PZ-2 (100), Chemie Uetikon) with a particle size of---2 ~m was also tested as
catalyst. First order rate constants k were calculated by the formula k = -ln(1-x) "Fp -1 "ml (x is
the conversion, F the total flow, p the total pressure and m the amount of catalyst).
3. R E S U L T S
13_
500:
!--
O3
I= 100 and 50
~E 400
o
"O
.Q
300
"13 ..(" Si/AI = oo
E 200
m
O
>
,;f
d
|
0.0 0'.2 0'.4 0'.6 018 lO
relative pressure (P/Po)
Fig. 1. N2 isotherms of samples with Si/A1 = 100, 50 and oo.
84
k
13_
~- 500-
03
ar -agingtime=24h
E
o 400-
"(3
..{3
"-o 3oo- aging time = 48 h
,r
E 200-
0
>
98h
1.2 ~48h
' "24h
o 2h
c0.8
t~
0
if)
..Q
~0.4 "
0
900 800 700 600 500
w a v e n u m b e r (cm -~)
Table 1
Rate constants and amount of Bronsted acid sites of the samples used for the cracking
experiments.
Sample Rate constant a)/mol -1 h-i bar-~ Bronsted acid sites c) Si/A1
Cumene TPB b) /pmol g-~
ZSM-5 5.5 0.34 281 r 23
TPA80 48 2 0.042 2.22 8 70
TPA80 48 8 0.042 2.17 8 70
TPA80 48 24
I __
0.044 2.12 8 68
TPA80 48 48m
0.051 2.40 8 68
TPA80 48 98
I __
0.134 2.75 13 66
TMA80 48 48 0.022 1.19 6 67
TMA80 48 96 0.018 0.93 5 75
a) The given value for the rate constant is the average of the measured values over a period of 200 min.
The measured rate constants were constant during this time. b)TPB = 1,3,5-triisopropylbenzene.c)
Determined by IR after pyridine adsorption at 100 ~ and desorption for lh at 300 ~ The ZSM-5 sample
was used as a standard, d) Determined by lH-NMR after pyridine adsorption.
86
9 times higher rate constant for the TPB cracking than the ZSM-5 sample meaning that there
is less size restriction in case of the TPA80_48_98 sample.
The amount of strong Bronsted acid sites is also given in Table 1. The ZSM-5 sample has
the highest concentration. The samples prepared with TPAOH have a higher concentration of
strong acid sites than the samples prepared with TMAOH. The highest acidity is found in the
sample TPA80_48_98 that shows some zeolitic characteristics in XRD.
4. DISCUSSION
The presence or absence of aluminium has a huge influence on the behaviour of the synthesis
mixture. Aluminium containing mixtures sustain their mesoporosity for a longer time during
the hydrothermal synthesis before zeolitisation occurs. On the other hand, purely silicious
samples rapidly lose a big part of their initial mesopore volume. Fig. 4 shows the correlation
of the loss in pore volume (calculated from N2 isotherms) with the area of the band at 550 cm-
i in the IR and the area of the two MFI reflexes at 8.8 and 7.9 ~ 2| in the XRD. For the series
of samples with Si/A1 = 80 (Fig. 4A) the decrease in pore volume correlates nicely with the
increase in the amount of zeolitic phase which can be seen in the increase of the IR and XRD
peak areas. The first significant loss in pore volume can be seen after 48 h. At the same time
we observe the first increase of the areas of the IR and XRD peaks. For the purely silicious
sample (Fig. 4B) the pore volume already drops before XRD and IR peaks of a zeolite phase
appear. The composition of our synthesis mixture should lead to the formation of a hexagonal
mesophase, if conventional silica sources were used. The preorganisation of silica by the TPA
template during the aging period leads to the formation of plate-like shaped precursors of
silicalite (nanoblocks). It was reported that these nanoblocks form a lamellar phase when
added to a solution of CTAC1 [11,12]. During calcination of this lamellar structure pores are
formed by the gaps between the particles of the individual layers. Although the surfactant
concentration in our synthesis is lower than in the preparation of zeogrid we speculate that a
lamellar phase is also formed in our system. A further condensation during the prolonged
hydrothermal treatment fills the gaps in the layers and thereby leads to a decrease in pore
volume. The formation of the zeolitic phase would then occur after the disappearance of the
supermicropores.
A) B)
0.7- "x'x -.-IR /
':,m 0.6
-.- pore volume ~ /
E
o...%0.6
(1.)
vgos. xX -*- XRD e-
Fig. 4. Correlation between changes in pore volume and areas of the peak at 550 cm1 in the IR and the
two XRD reflexes at 7.9 and 8.8 ~ 2| A) Series of samples with Si/A1 = 80. B) Series of samples with
Si/A1 = ~.
87
5. CONCLUSIONS
REFERENCES
[1] J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T.W. Chu,
D.H. Olson, E.W. Sheppard, S.B. McCullen, J.B. Higgins, J.L. Schlenker, J. Am. Chem. Soc.,
114 (1992) 10834.
[2] C.T.Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S. Beck, Nature, 359 (1992) 710.
[3] A. Corma, M.S. Grande, V. GonzalezAlfaro, A.V. Orchilles, J. Catal., 159 (1996) 375.
[4] A. Corma, Chem. Rev., 97 (1997) 2373.
[5] Y. Han, F.S. Xiao, S. Wu, Y.Y. Sun, X.J. Meng, D.S. Li, S. Lin, F. Deng, X.J. Ai, J. Phys.
Chem. B, 105 (2001) 7963.
[6] Y. Han, S. Wu, Y.Y. Sun, D.S. Li, F.S. Xiao, J. Liu, X.Z. Zhang, Chem. Mat., 14 (2002) 1144.
[7] Y. Liu, W.Z. Zhang, T.J. Pinnavaia, J. Am. Chem. Soc., 122 (2000) 8791.
[8] Y. Liu, W.Z. Zhang, T.J. Pinnavaia, Angew. Chem. Int. Edit., 40 (2001) 1255.
88
The structure of a novel large pore gallosilicate zeolite, TNU-7, has been refined from X-ray
powder diffraction data. The silicate framework is built up from two different types of layer
which are shown by X-ray and electron diffraction to be arranged in a fully ordered way. This
gives rise to 12-MR pores and asymmetric cation distributions. TNU-7 is prepared
hydrothermally over a very narrow compositional range and only in the presence of Ga. Other
zeolitic phases crystallise at lower and higher Ga contents confirming the unique structure-
directing ability of this element. The location of ion exchanged Cs + and Sr2+ cations in TNU-7
has been studied by high resolution X-ray powder diffraction.
1. INTRODUCTION
While many novel zeolite structures have been discovered through the use of structure-
directing agents (SDAs), recent advances have shown that introducing heteroatoms such as
Be, B, Zn and Ge into silica frameworks during crystallization is another way to obtain novel
topologies. This also appears to be the case for Ga, the element most closely similar to A1,
despite the general notion that the gallosilicate product is the same as for the aluminosilicate
crystallization[l]. Studies of the structure-directing effect of Ga in zeolite syntheses has
revealed that replacement of A13§ by Ga 3+ in low-silica gels can give microporous materials
with topologies different from those of aluminosilicate zeolites prepared under otherwise
identical conditions[2]. Examples include TNU-1 (type CGS)[3], ECR-34 (ETR), the first
gallosilicate having 18-ring pores[4] and the material described here, TNU-7[5].
TNU-7 is a novel gallosilicate prepared in the Na20-Ga20-SiO2 hydrothermal system from
gel compositions (Si/Ga) intermediate between those yielding Na-gallosilicate mordenite
(MOR) and Na-gallosilicate mazzite (MAZ). The crystallisation of TNU-7 depends critically
on the presence of Ga in the synthesis gel, as well as its concentration. Variation of the Ga
concentration results in the sequential formation of MOR, TNU-7 and MAZ implying a rather
unique structure-directing role of Ga in this system. TNU-7 is a regular intergrowth of the two
end-member phases (see Fig 1) as proposed for the aluminosilicate 'boundary phase' ECR-
1A[6,7]. Previously, no structural refinements were possible (and no 3-letter code assigned)
because the aluminosilicate version showed a high degree of disorder. However, TNU-7,
which is stable to high temperature and cation exchange, exhibits regular stacking, thus
90
permitting a study of the structure through high resolution powder diffraction using a
wavelength at which the effects of absorption are minimal. In this work we determine the
favoured Na + positions in dehydrated Na-TNU-7 and compare them to related sites in
gallosilicate mordenite and mazzite. Also, we report cation positions for TNU-7 exchanged
with a large univalent cation (Cs +) and also a divalent cation (Sr 2+) that establish the cation
distribution as a function of size and charge. We also determine whether there is a preference
for cations sitting in the MOR or MAZ structural layers.
I I
4i
...... , .......e.~.J..
..~..~...~ #,, )~III, ~.
9 ~ .O ~ I JI~" ~" q I
9 ~ ..~i~ .i. 9
I ~.
.o..o...~..-"i.."~,~
~.'~ ......~..
,~. ~ii~ ~L~... ". . . . . I ~ ~II e.~. ~I ~ ~-.~ ....
b
~. ~.~~i
i
I
I
"~ ..~: ~,..~.i, -
I
.....
cj
' MAZ MOR
Fig. 1. TNU-7 - a regular intergrowth of MAZ and MOR layers (black - Si/Ga, grey - oxygen)
2. E X P E R I M E N T A L
TNU-7, Na-Ga-mazzite and Na-Ga-mordenite and ion exchanged Cs- and Sr-TNU-7 were
dehydrated and sealed under a vacuum of 4x10 -4 Torr at 300~ in 0.7mm quartz glass
capillaries. The samples were analysed on station ID31 at the ESRF, Grenoble using
monochromated synchrotron X-rays of wavelength 0.63248A. Data were collected at room
temperature from 3 to 35 o 20, step size 0.01 o, over periods of 4-8 hours, and refined via the
Rietveld method using the GSAS package[9].
During the Rietveld refinements, background was fitted using the cosine Fourier series.
Peak shape was modelled using the pseudo-Voigt profile function. Si and Ga atoms were
positioned on each T-site with fractional occupancies estimated from the bulk Si/Ga ratio,
assuming that the Ga atoms were disordered over all T-sites. For mazzite, Si/Ga = 3.03, the Si
occupancy was 0.752; for TNU-7 the values were 3.85 and 0.794; for mordenite, 5.89 and
0.85. T-O and O-O distances were constrained at values taken as weighted averages of Ga-O
and Si-O tetrahedral bond lengths to keep sensible framework geometries - these constraints
were maintained throughout. Extra-framework cations were located via Fourier difference
maps. Difference peaks located at reasonable distances from the framework were investigated
as sites by refinement of occupancy and position. Extra-framework cations were refined with
a single thermal parameter, as were T-sites and oxygen atoms. Na-Ga-mazzite was found to
contain Na-TNU-7 impurity so this data was analysed by a two phase refinement.
3. R E S U L T S AND D I S C U S S I O N
Refined lattice parameters and Rwpvalues for the refinements are shown in Table 1. Table 2
lists the Na-TNU-7 atomic parameters and Fig.2 shows the fit.
Table 1
Refined lattice parameters and associated Rwpvalues
Space
Material Group a (,~) b (/~) . . . . . c(A) Rwp
Na-TNU7 Pmmn 7.57214(9) 18.0739(2) 25.8438(5) 6.47
Sr-TNU7 7.57836(9) 18.0998(2) 25.9256(6) 5.79
Cs-TNU7 7.56920(9) 18.3157(3) 26.0812(7) 6.67
Na-Ga-mazzite P63/mmc 18.1314(2) 18.1314(2) 7.6854(1) 9.36
Na-Ga-mordenite Cmcm 18.0176(1) 20.2835(2) 7.49243(6) 7.05
Many zeolite structures are composed of a single type of complex silicate layer, joined to
adjacent layers by T-O-T bonds that are approximately perpendicular to the plane of the
layers. Mazzite and mordenite are two well-known examples of such frameworks (Fig. 5). In
addition, the MAZ and MOR structures are related so that not only do they share two
common axes (the axes defining the plane of the structural layer) but their layers are
connected in a similar fashion via chains of 5-MRs. This permits a highly regular 1:1 stacking
of alternating layers in the gallosilicate Na-TNU-7. The structure of TNU-7, refined here from
X-ray synchrotron data, is remarkable in that the large channels (6.6 x 7.4A) are asymmetric
in that different environments are present on the different sides near to the MAZ or MOR
layers. Furthermore, the wide channels are only connected to the intrachannel small pore
system via 8-membered ring (8MR) openings in the MOR layers - MAZ strips are separated
from the intrapore space by 5-MRs.
92
oL ....
"i
i
Fig. 2. Rietveld plot for ~='i--2 , ~ ~ ~ , ~ L .w. _~.,...,~
Na-TNU-7, shown for ~ , "" "~:-"
two theta 3- 15 ~ 11 - ~ ~ t ~ - ~ , ~
o ~ : i 1........................................ l ......................
4 6 8 10 12 14 o 20
Table 2
Refined atomic parameters of Na-TNU-7
c.j
Fig. 3. Refined extra framework cation sites in TNU-7
Sites A and B are situated in a puckered 8-MR within the MAZ and smaller inter-strip
channels. Site C is within the 8-MR window entering the MOR layer while D and E are in the
large channel associated with the MAZ strip. A further position F is in the 6-MR of the
distorted gmelinite cage of the MAZ layer. 11.7 of the 12.3 Na + required for charge balancing
are located. Studies of extra framework cation positions in Na-Ga-mazzite and Na-Ga-
mordenite reveal many similarities. In Na-Ga-mazzite, where 7.3 of the estimated 9 Na +
cations have been found, sites A and E are found (Fig. 4), while sites equivalent to B, C and D
are observed in Na-Ga-mordenite where 6.4 of the 7 Na+ cations have been located (Fig. 4).
Table 3
Refined Na + cation positions in Na-TNU-7
These locations agree with reports of Na + positions in these phases[ 10,11 ], although a third
position ('position 3') found in an organically-templated Ga- mazzite is not observed. Also,
site D of Na-Ga-mordenite is not found in Na-TNU-7. Fractional occupancies of sites within
the MOR region of Na-TNU-7 are comparable to those in Na-Ga-mordenite, while there are
94
more cations in the large channel of Na-Ga-mazzite compared with similar sites in the MAZ
region of TNU-7. Indeed, site A in the MAZ strip is particularly favoured in Na-TNU-7
compared to the other sites. Calculations of the extra framework cation charge per T-site in
the separate strips within Na-TNU-7 are similar to those for the individual mazzite and
mordenite phases (Table 4) and suggest a higher framework charge per framework unit
associated with the MAZ region of TNU-7 than with the MOR. This may be indicative of
'chemical zoning' within the TNU-7 structure implying a larger Ga content within the MAZ
region as expected from simple stacking of MAZ and MOR layers. Attempts to refine the Na-
TNU-7 structure with Ga-rich and Ga-poor strips did not improve the fit.
._b
,. ....
~. ~." '~-...~ Ilk
.,- %.. ~
,- ~ .......... -~_., <-, ~, .~...,. .....
,. o, ,, ,,. D :---* ,. ,,
I ~ "~* ~ ~ ~"~ ~"~I~"~I .... ~"~ ~ " Q ~ ~"~ ~I
Fig. 4. Projection of Na-Ga-mazzite (left) and Na-Ga-mordenite (right) along c showing refined Na+
positions. For Na in Ga-mazzite 9 A, (89 0, 0; occupancy 0.37(1))" E (0.2393(8), 0.1196(4), 0.25;
0.84(1)). For Na in Ga-mordenite; Bmor (0, 0, 0; 0.55(1)); C (0, 0.810(2), 0.112(3); 0.242(7)); D
(0.169(1 ), 0.969(1 ), %; 0.29(1))
Table 4
Extra-framework cation charge per T-site in materials studied in this work
planar 8-MR (Fig. 5). Similarly, positions C' and D within the large channels are further from
the framework oxygens due to the larger ionic radius of the Cs + cation (Fig. 5). While no
evidence was found to suggest that sites E and F observed for sodium are occupied, an
additional site G (Fig. 3) was found close to the central 4-rings of the three 4-ring unit of the
gmelinite cage. The increased cation size therefore has a strong influence on the cation
distribution. It is noticeable, however, that the framework charge per tetrahedral atom
associated with the MAZ layers remains higher than that for the MOR layers.
/ (
b b
Fig. 5. Local environment around sites A, A',C, C ' , D and D' in TNU-7.
Table 5
Refined Cs + sites in Cs-TNU-7
Site x y z Occ. Mult.
A' 0.25 0.0597(2) 0.0225(2) 0.867(6) 4
B' 0.25 0.25 0.2532(3) 0.974(7) 2
B" 0.75 0.25 0.349(4) 0.846(7) 2
C' 0.25 0.75 0.425(1) 0.284(7) 2
D 0.75 0.6171(3) 0.2606(2) 0.625(5) 4
G 0.25 0.235(6) 0.0422(2) 0.047(4) 4
Table 6
Refined Sr 2+ sites in Sr-TNU-7
Site x y z Occ. Mult.
A 0 0 0 0.563(6) 4
B 0.049(2) 0.25 0.2994(1) 0.485(5) 4
B" 0.25 0.75 0.664(3) 0.112(9) 2
B'" 0.371(8) 0.75 0.593(3) 0.071(5) 4
C' 0.19(1) 0.75 0.385(1) 0.065(4) 4
D 0.25 0.072(2) 0.758(2) 0.129(5) 4
96
Sr 2+ is intermediate in size between Na + and Cs +. 5.5 of the expected 6.2 Sr 2+ cations have
been located and found to occupy sites A, B, C and D and would also appear to occupy B "
and site B ' " . Partial occupancies of several close sites suggest that energies of Sr 2+ are very
similar in these different environments.
The charge per T-site in the MAZ and MOR layers of the ion-exchanged materials (Table
4) is also higher in the MAZ regions than in the MOR strips for this structure, although the
difference is reduced.
4. C O N C L U S I O N S
The structure of gallosilicate TNU-7 has been refined against synchrotron x-ray diffraction
data as a completely ordered intergrowth of gallosilicate mazzite and mordenite layers. Its
synthesis depends on the presence and concentration of Ga in the synthesis mixture.
Sequential formation of Na-Ga-mazzite, Na-TNU-7 and Na-Ga-mordenite with increasing
Si/Ga ratio indicates a unique structure-directing role of Ga. While the positions of extra
framework Na § cations in Na-TNU-7 show strong similarities with those observed in the
parent phases, the positions of ion-exchanged cations Cs + and Sr 2+ appear to be influenced
strongly by ionic radius. This is most apparent in the location of Cs + cations. Extra framework
cation charge per T-site implies a higher charge associated with the MAZ region indicating
possible chemical zoning within the structure.
ACKNOWLEDGEMENT
This work was partially supported by the Korea Science and Engineering Foundation under
the Grant No. R02-2003-10087-0 to S.B.H, and the EPSRC (S.J.W.). Synchrotron data were
collected at the ESRF, Grenoble, in collaboration with Prof A. Fitch.
REFERENCES
[1] R. Fricke, H. Kosslick, G. Lischke, M. Richter, Chem. Rev., 100. (2000) 2303.
[2] H.H. Cho, S. H. Kim, Y. G. Kim, Y. C. Kim, H. Koller, M. A. Camblor, S. B. Hong, Chem.
Mater. 2000, 12, 2292;
[3] S.B. Hong, Y.G. Kim, Y.C. Kim, P.A. Barrett and M.A. Camblor, J. Mater. Chem., 9 (1999)
2287.
[4] K.G. Strohmaier, D. E. W. Vaughan, J. Am. Chem. Soc., 125 (2000) 16035.
[5] S.J. Warrender, P. A. Wright, W Zhou, P. Lightfoot, M. A. Camblor, C. Shin, D. J. Kim and S.
B.Hong, Chem. Mater. (in the press)
[6] M.E. Leonowicz, D. E. W. Vaughan, Nature 1987, 329, 819.
[7] D.E.W. Vaughan, K. G. Strohmaier, Microporous Mesoporous Mater. 1999, 28, 233.
[8] W.C. Paik, M. A. Camblor, S. B. Hong, Stud. Surf. Sci. Catal. 2001, 135, [CD-ROM] P05-P-07
[9] A.C. Larson and R. B. von Dreele, Generalized Crystal Structure Analysis System, Los Alamos
National Laboratory, USA, 1988
[10] J.M. Newsam, R. H. Jarman, A. J. Jacobson, Mater. Res. Bull. 1985, 20, 125
[11 ] W.M. Meier, Z. Kristallogr. 1961, 115,439
[12] R.D. Shannon, Acta Crystallogr., A32, 1976, 751
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. 2;ilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 97
1. INTRODUCTION
Ordered mesoporous silica are formed with amphiphilic molecules acting as structure
directing agents. It is now a well-known fact that a large number of surfactants or amphiphilic
polymers have the ability, under the right chemical conditions, to, in cooperation with the
polymerizing silica, drive the formation towards a variety of mesoscopic structures. Here we
have used the triblock copolymers that go under the trade name Pluronics. These polymers,
consisting of two polyethylene (EO) blocks surrounding a central polypropylene (PO) block,
have proven to be versatile structure directors for mesoporous silica formation [1]. As the
block length can be varied a whole family of these polymers exist and are commercially
available. Prior to the discovery that these polymers can be used in mesoporous silica
synthesis they had already, for many years, been to a considerable interest for the surfactant
science community. Hence, a large knowledge base on the physicochemical properties of
Pluronics can be found in the literature [2-4]. For instance it is known that the phase
behaviour of the polymer can be influenced by several parameters, e.g. temperature, which
has the effect of increasing the hydrophobicity of the polymer with an increase in
temperature, length of the respective blocks, addition of additives such as alcohols as well as
addition of salt. Lately it has been shown that this phase dependence of pure polymer water
98
system is also reflected in the mesoporous silica formed with the polymer as structure
promoter [5-10].
In a typical synthesis of mesoporous SiO2 the concentration of the Pluronic polymer is in the
range of 2.5 weight% in an acidic aqueous solution. Under such conditions the polymer forms
micelles with a core consisting mainly of the PO-blocks and a palisade layer formed by the
more hydrophilic EO-blocks together with water molecules. When the silica source is added
the conditions for the polymer is severely altered. This leads to a phase change and a liquid-
liquid phase separation [ 11 ].
The driving force for the formation of the mesostructure is the positive interaction between
the (polymerizing) silica and the EO-groups. It has been claimed that this interaction to a
large extent is an electrostatic interaction based on the assumption that the EO-groups are
protonated at the acidic conditions prevailing in the synthesis solution [ 1]. The interaction has
been designated as (S~ ~~or +)), where S is the nonionic polymer, X- the counterion and
I+ the silica framework. It is thus inferred that the counterion, which in the absence of added
salt is the anion of the acid, affects the phase behaviour as it is electrostatically bound
between the positively charged EO-groups and the positively charged or neutral silica
framework. However this seams to be an unlikely scenario. Diethyl ether is a very weak base
and consequently the protonated form is a very strong acid with pKa-value of-3.5. By a
simple calculation based on a HC1 concentration of 1.6 M, which is a normal acid
concentration for these types of syntheses, the ratio of unprotonated to protonated EO groups
would be in the range of 2000. An electrostatic effect hence seems negligible. Further, the the
inorganic framework is assumed to be positively charged [1] but according to the same
reasoning as above it is unlikely that ortho silicic acid to any larger extent is protonated. The
effect observed for anions should therefore be sought elsewhere.
There have been several studies of the effects of simple inorganic salts on nonionic polymers.
One explanation of the effect is to regard it to be the result of an interfacial phenomenon [12].
Different salts, and it is the anion that plays the leading role, is more or less, depending on the
polarizability of the anion, adsorbed to, in our case, the EO-groups. The interaction is, as
explained above, not an electrostatic one. Anions will either be preferentially adsorbed to the
palisade layer of the micelles or desorbed. This is often referred to as the salting-out and
salting-in effect. If the anions are adsorbed they are said to be salting-in ions and if they are
desorbed they are called salting-out ions. Based on their salting-out strength, anions can be
arranged in the so-called Hofmeister series (with $042- having the strongest salting-out
effect):
SO42 ~>HPO42 > F- > C1- > Br- > NO3- > I- > C104- > SCN-
The effect of addition of salt to a micellar solution can be explained by two phenomena.
Firstly, addition of a salt increases the ionic strength of the aqueous solution, which becomes
more polar. The polymer will thus experience a more polar surrounding and behave, as
compared to the no salt case, in a more hydrophobic manner. The EO-chains will be less
associated with the solvent, that is they will be dehydrated, and will hence require less space.
This in turn leads to a larger aggregation number of the micelles. The effect is analogous to
what is observed for a temperature increase or a shortening of the EO chains. Secondly,
depending on the type of anion present a specific ion effect can be observed. This effect is in
99
00
00 .--.
00 v
:>,
00 e-
<1)
t'--
00-"
0.13
q(l/,~
Fig 1. In-situ SAXS/SAXD of the synthesis of SBA-15 with P 104 as structure director. No salt is
added to the synthesis mixture
opposition to the general salt effect, making the polymer more hydrophilic as anions are
adsorbed to the palisade layer.
In this work we describe the effects of the addition of salt on the formation process of SBA-
15 materials. We have followed the synthesis with and without salt by in-situ SAXS/SAXD
measurements. The effects observed in this study can be explained by the general ionic effect
(solvent increasing in polarity) described in the previous paragraph. In another investigation
we have studied the specific ion effect by adding a series of salts with different anions [13].
2. E X P E R I M E N T A L
30000
25000
zJ
2OOOO
v
15000
c
1o0oo c
5000
0
60
Fig 2. In-situ SAXS/SAXD of the synthesis of SBA-15 with P 104 as structure director. NaC1 (1.0
mol/l) is added to the synthesis mixture.
3. RESULTS
Fig. 1 and Fig. 2 show the time-resolved SAXS/SAXD pattern for the non-salt
synthesis (Fig. 1) and the synthesis with 1.0 M NaC1 (Fig. 2). Two differing zones can be
identified. At the beginning of the measurements there is diffuse scattering due to the
presence of micelles. As the reaction proceeds, a diffraction peak from the ordered
mesoscopic phase appears.
It is obvious that a dramatic change occurs quite early in the synthesis (13 min in Fig. 2 and
30 min in Fig. 1). It is at this point that the SAXD pattern for the 2 D hexagonal structure
(p6mm) starts to appear and concomitantly the diffuse scattering from the micelles becomes
less pronounced. The strongest peak (100) is most clearly seen. It starts out with rather low
intensity but quickly with time the intensity increases. Also with time the higher order
reflections (110), (200) and (210) become (more or less) apparent.
101
11.6 - ++
+
+
11.4 - + NaCl
~_ 0 n0 sal.__._~t
"~" 1 1 . 2 -
e,,, \
r 11.0-
~. 10.8-
In
10.6 -
10.4 -
r~
I I I I
50 1 O0 150 200
time (rain)
In the salt case (Fig. 2) the diffuse scattering disappears quickly whereas in the no salt case
this scattering lingers on much longer. For the system with NaC1, the ordering process occurs
very fast, seen from the fact that the Bragg reflection appears at shorter time, and also from
the higher intensity of this reflection, compared to the intensity reached by the system without
salt. It seems that for the aqueous solution the ordered hexagonal phase coexists with non-
ordered micelles much longer than for the brine.
The unit cell parameter of the initial hexagonal phase is also different for the two cases, 127
A for the no salt synthesis and 134 A for the NaCl-containing synthesis. Fig. 3 shows how the
d-spacing of the (100) reflection changes with time.
The diffuse scattering was used for fitting the form factors of micellar size and shape. It was
found that the form factor of prolate ellipsoids corresponded to the observed scattering. The
fitting was performed for several reaction times times for each synthesis mixture. Here, the
cell parameter obtained when the hexagonal phase was formed was used as a maximum limit
constraint for the micellar radius. Under these constraints, the data could not be fitted using
the form factor of a sphere with reasonable values for the micellar size. Without any fitting
constraints, the form factor of a sphere could be fitted to the data, but the micellar diameter
was then about 10% larger than the cell parameter of the hexagonal phase. The validity of the
introduced constraint is supported by the fact that the obtained values for Rsmall w e r e
coinciding with the values found by the calculation of the electron density distribution of the
same systems immediately after the formation of the hexagonal phase. Thus slightly
elongated micelles were successfully fitted. We note that one cannot distinguish between
slight elongation and polydispersity. However, in our case, polydisperse spheres would lead
to micelles with bigger radius than allowed by the hexagonal packing. Besides, the tendency
of growth was verified by the ratio between both dimensions of the micelle, and for this
reasons we do believe that we have prolate ellipsoids instead of polydisperse spheres in
solution. The fitting results obtained for two different reaction times: 10 minutes for both and
at the point where the first hexagonal structure appeared (13 min and 30 min respectively) are
presented in table 1.
102
Table 1
Parameters obtained by fitting of elongated micelles form factors at different times after addition
of TEOS. Rsmallis the smallest radius of the micellar hydrophobic core and Rbig the larger one.
The total micellar dimension, in both axes, is given by R plus the thickness of the polar region.
Thickness of
polar region / ~ 25 25 21 21
4. DISCUSSION
In the systems we are studying the ionic effect can be regarded as a general ionic
effect. In both syntheses we have 1.6 M of HC1, that is, there is a high concentration of the
salting-out CI anion present. In the salt containing system we increase the concentration of
C1- by adding NaC1 (1.0 M). The effects we observe are thus related to a general ionic effect.
(Cations are generally considered to have a much weaker effect on the salting-out behaviour.)
The concentration of the polymer is 2.5 weight% in an aqueous solution and this is above the
critical micelle concentration but far below the concentration necessary to form a liquid
crystalline phase. The polymers will thus be located in micelles dispersed in the solution.
Addition of NaC1 will make the solvent more polar and this will affect the hydrophobicity of
the polymer, in particular the EO-chains. This change in hydrophobicity as a consequence of
the salt addition is reflected in two observable effects in the formation of the silica material.
Firstly, the EO-chains will be dehydrated when NaC1 is added, hence they will have a more
hydrophobic behaviour and be more closely associated with the hydrophobic core (mainly
PO-groups) of the micelles. With no salt the EO chains will extend further out into the
solution and as a result require more space. This leads to a smaller aggregation number for the
micelles. The polarity of the solvent will thus have a direct effect on the micellar dimensions
(however the size of the micelles is dependent both on the aggregation number and the
extension of the EO-chains in the solvent). This effect has been both experimentally and
theoretically studied [ 12, 17]. In this study we can observe this size difference by analysing
the scattering data. In Table 1 the micellar dimensions in the two syntheses are listed. The
micelles are fitted with an elongated shape. The larger of the two radii (Rbig) reflects the
103
growth of the hexagonal phase and will be discussed below and the smaller radius (Rsmali) is a
more direct measure of how the salt has affected the micellar size. Rsmall is notably larger in
the NaC1 case in accordance with the reasoning above.
The second effect of the change of EO hydrophobicity is a change in the interaction between
micelles. There will be a stronger attraction between more hydrophobic micelles (NaC1 case)
than between the less hydrophobic ones (no salt case). This effect explains the faster kinetics
observed for the NaC1 containing synthesis where the hexagonal structure appeared 13 min
after the addition of TEOS as compared to 30 min for the no salt case. It thus evident that a
more polar surrounding will affect the kinetics of the aggregation. The silica polymerisation
on the other hand is most likely less influenced by the NaC1 addition. As far as we know there
is no catalytic effect of this salt on the silica reactions under acidic conditions. The silica
kinetics should therefore be more or less the same for the two cases. A support for this is the
observed decrease in the d-spacing. For the NaC1 case the decrease is much more pronounced
probably because the aggregation takes place when the silica is less polymerised. The silica
wall density will hence increase more in the NaC1 synthesis.
In previous in-situ studies we followed the formation of SBA-15 and found that the first silica
precipitate is built up of globular micelles in a silica matrix [11, 15]. With time the micelles
merged and long tubes were formed. The tubes packed in the 2-D hexagonal framework. Here
we can see this growth from the fitting of elongated micelles to the scattering data. The Rbig
parameter mentioned above is directly associated to the length of the micelles (total length is
given by Rbig+thickness of polar region). From Table 1 it is seen that 10 rain after the TEOS
addition Rbig and Rpar are different for both syntheses. The micelles are larger in the NaC1
synthesis. With time Rbig increases reflecting the fact that the micelles are growing, but they
still have finite size, as we see from the behaviour of the decay of the scattering at low q-
values. In the case of NaC1 the micelles are bigger, already after 13 min (65.6 A long), than
after 30 min in the no salt case (63.6 A).
It has been observed that the pores of SBA-15 are corrugated [18] whereas for the MCM-41
structure the pore walls are reported to be flat. It is possible that if the aggregation occurs with
silica having few cross-links, as in the NaC1 synthesis, the corrugation of the walls can be
diminished. When the aggregation occurs and the micelles merge into long tubes the silica is
still very flexible. Hence, there will be more opportunity for the tubes to form without
"waists", that is, the silica will have the possibility to rearrange and the corrugation effect
might be milder.
ACKNOWLEDGEMENTS
The authors thank Hgkan Wennerstr6m for valuable discussion. The Finnish Academy, the
Swedish SSF program of colloid and interface science and the EU large-scale facility
programme (Elettra) HPRI-CT- 1999-00033 are also acknowledged.
104
REFERENCES
[1] D. Zhao, Q. Huo, J. Feng, B.F. Chmelka and G.D. Stucky, J. Am. Chem. Soc. 120 (1998) 6024.
[2] P. Alexandridis and J. F. Holzwarth, Langmuir 13 (1997) 6074.
[3] G. Wanka, H. Hoffmann and W. Ulbricht, Macromolecules 27 (1994) 4145.
[4] P. Holmqvist, P. Alexandridis and B. Lindman, J. Phys. Chem. B 102 (1998) 1149.
[5] F. Kleitz, S. H. Choi and R. Ryoo, Chem. Commun. (2003) 2136.
[6] J. Tang, C. Yu, X. Zhou, X. Yah and D. Zhao, Chem. Commun. (2004) 2240.
[7] K. Flodstr6m and V. Alfredsson, Microporous Mesoporous Mater. 59 (2003) 167.
[8] K. Flodstr6m, V. Alfredsson and N. KNlrot, J. Am. Chem. Soc. 125 (2003) 4402.
[9] C.Yu, B. Tian, J. Fan, G.D. Stucky and D. Zhao, Chem. Commun. (2001) 2726.
[10] C.Yu, B. Tian, J. Fan, G.D. Stucky and D. Zhao, J. Am. Chem. Soc. 124 (2002) 4556.
[11] K. Flodstr6m, H. Wennerstr6m and V. Alfredsson, Langmuir 20 (2004) 680.
[12] A. Kabalnov, U. Olsson and H. Wennerstr6m, J. Phys. Chem. 99 (1995) 6220.
[13] C. Teixeira, P. Linton, H. Amenitsch, F. Kleitz, M. Linddn and V. Alfredsson, Manuscript in
preparation.
[14] M. Lind6n, S.A. Schunk and F. Schfith, Angew. Chem., Int. Ed. 7 (1998) 821.
[15] K. Flodstr6m, C.V. Teixeira, H. Amenitsch, V. Alfredsson and M. Lind6n, Langmuir 20 (2004)
4885.
[16] K. Flodstr6m, H. Wennerstr6m, C.V. Teixeira, H. Amenitsch, M. Lind6n and V. Alfredsson,
Langmuir 20 (2004) 10311.
[17] N.J. Jain, V. K. Aswal, P.S. Goya| and P. Bahadur, Coll. Surf. 173 (2000) 85.
[ 18] Z. Liu, O. Terasaki, T. Ohsuna, K. Hiraga, H. J. Shin and R. Ryoo, ChemPhysChem 2 (2001)
229.
Studies in Surface Science and Catalysis,volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 105
aDavy Faraday Research Laboratory, The Royal Institution of Great Britain, 21 Albemarle
Street, London W 1S 4BS, United Kingdom.
CESRF (European Synchrotron Radiation Facility), 6 Rue Jules Horowitz, BP 220, 38043
Grenoble CEDEX 9, France.
A combined computational and experimental study has been undertaken to resolve cation-
directing effects. Li-A(BW) and Na-J(BW) have different topologies but are synthesised from
almost identical starting ingredients, the alkali metal being the only variable. This exemplar
reaction therefore allows the effect of the cations upon crystallisation and nucleation
processes to be directly inferred. We report the minimum energy structures of Li-A(BW) and
Na-J(BW) obtained by application of an atomistic approach and the experimental synthesis of
Li-A(BW) and Na-J(BW). The synthesis was prompted by inconsistencies between the
calculated Na + cation positions of Na-J(BW) and those determined from previous
experimental work. With the support of molecular dynamics simulations, we present a re-
determination of cation positions as part of an ongoing investigation into the crystal growth
mechanism of these two materials.
1. INTRODUCTION
Since the first commercialisation of zeolites, much of synthetic chemistry reported in the
literature has focused on empirical approaches to tailor-make zeolites, with less emphasis
upon why or how these phases are formed [ 1]. An array of factors influence which phase is
formed, arising from competing processes during zeolite nucleation and growth [1]. The
difficulty in deciphering exactly how microporous materials form is due in part to persistent
questions about the identity and role of solution species in directing growth. Several attempts
have been made to explore the solution species present during zeolite growth, both
experimentally, through spectroscopic methods [1,2], and computationally [3-6], exploring
minimum energy structures of possible precursor units. These works support theories of poly-
and mono-meric species crystallising from solution upon pre-formed solid surfaces in the
solution [ 1,7].
The broader aim of this work is to understand the role that the cation plays in influencing
the species present in solution and to predict how the cation influences the reaction energetics
and mechanisms of attaching species to the crystal surface. The surface structures and crystal
morphologies of zeolites can give great insight into how and why the growth of certain
106
structures occur [8]. To determine the crystal growth mechanisms, it is necessary to predict
the most stable surface structures, as these define the rate determining steps in the crystal
growth process. Here we present the first step in exploring the zeolite growth mechanisms by
modelling the minimum energy configurations of the bulk zeolite structures of Li-A(BW) and
Na-J(BW). Li-A(BW) and Na-J(BW) are synthesised from gels containing A1203, SiO2, H20
and alkali under identical hydrothermal conditions [9-12]. Media rich in lithium favours the
growth of Li-A(BW) [Li4(H20)41[AI4Si4016]whilst sodium rich solution favours the growth of
Na-J(BW) ]Na+3(H20)1.sI[A13Si3OI2] and critically, as is evident from the structural formulae,
the Si:A1 ratio is 1:1. Consequently, the ion present in solution must have a directing
influence on the zeolite growth process and hence the framework topology formed. A study of
the two zeolites therefore affords insight not only into the general mechanism of zeolite
growth but also the structure directing effects of the extra-framework cations Li + and Na +
independent of other structure directing influences. The cations are expected to affect the
distribution of oligomers and also the energy barriers of forming particular oligomers in
solution. Consequently, we expect to understand how these oligomers are incorporated into
the growing crystal surfaces, independent of the other synthesis parameters.
Although synthesised by Barrer and White in the early 1950s [13,14], the structure of Li-
A(BW) and Na-J(BW) was not reported until over 30 years later. Both Li-A(BW) and Na-
J(BW) were reported to comprise of 4, 6, and 8-membered rings, connected differently and
forming two distinct topologies, ABW and JBW, respectively. Hydrated cations lie within the
8-ring channels of both Li-A(BW) and Na-J(BW) and anhydrous cations are present only in
the 6-ring channels of Na-J(BW). The structural details of Li-A(BW) are well documented,
supporting the initial model proposed by Norby et al. [ 15-17]. The structure of J(BW) zeolites
containing germanium framework species and extra-framework potassium as well as sodium
interstitial cations have been reported [18,19], but the structure of pure Na-J(BW), containing
sodium cations within the channels has been reported with less confidence. The structure
proposed by Hansen et al. [20] identified two partially occupied hydrated sodium cation sites,
labelled as Na(3) and Na(31) respectively. The Na(31) sites were proposed to replace 20% of
the equivalent Na(3) sites. The structure of Na-J(BW) was also characterised by Ragimov et
al. [21], although the zeolite was identified simply as a sodium aluminosilicate and the
distribution of the two hydrated sodium cation sites within the 8-ring channels was noted to
be "distributed statistically".
In order to predict surface structure, an accurate representation of the zeolites bulk
structures is a pre-requisite. Here we present preliminary results of computational atomistic
modelling of the minimum energy configurations of the zeolite bulk structures.
Inconsistencies with published experimental data prompted the synthesis of Na-J(BW) and
modelling of the thermal behaviour of Na + cations from which a re-determination of the
cation positions was made.
The structural and thermodynamic properties of numerous zeolites have been extensively and
very successfully studied using static lattice methods [22,23]. In this instance, the minimum
energy configuration of the bulk zeolite structures of Li-A(BW) and Na-J(BW) were obtained
by application of the static energy minimisation program GULP [24]. Species within the 3D
simulation cells were described using the Born model of ionic solids, where each atom is
represented by a point charge and non-Coulombic interactions are described by van der Waals
attractive and repulsive forces under periodic boundary conditions. Zeolite materials possess
107
the highly polarisable oxygen ion within their framework and may contain water molecules
within their pores and channels. Thus, the polarisability of the oxygen ions can be
approximated by the shell model of Dick and Overhauser [25]. The intermolecular potentials
used to model the interactions between the component species of Li-A(BW) and Na-J(BW)
were those previously fitted to dense aluminosilicate zeolites and applied to describe a
number of dense zeolite frameworks by Lewis et al. [26-28] and are therefore expected to be
readily transferable to these systems. Water is described by potentials due to de Leeuw et al.
[29]. For each material, the proposed cell was relaxed at constant pressure such that the all
species and cell parameters are free to vary. Note that each structure is relaxed to mechanical
equilibrium at an effective temperature of OK.
(b)
Fig. 1. (a) Li-A(BW) experimental and (b) optimised model. Lithium ions are depicted as black
spheres, the framework as grey sticks and water molecules are displayed as cylinders (view along
[ool])
Table 1
Comparison of experimental [15, 17] and calculated Li-A(BW) Li + coordination sphere:
Li-O distances (A), and percentage difference between the calculated and experimental
model. O(1-4) are framework oxygen atoms whereas Ow denotes a water oxygen atom.
Li-O distances Experimental Model Optimised Model Difference (%)
comparison with the experimental model (Table 1). The distortion of the LiO4 tetrahedron
indicates possible failures of the fitted Buckingham potential which describes the Li-O
interaction. The pronounced decrease in the Li-Ow distance suggests that the potential
describing the Li-Ow interaction is too attractive. However, it is possible that this
overestimation is simply a thermal effect, a postulate that we will explore using molecular
dynamics. Nevertheless, the existing model is in an acceptable agreement with published data
and can therefore be used for (interatomic potential) surface structure predictions.
(b)
t
,!
Fig. 2. (a) Na-J(BW) experimental model with Na(3) sites fully occupied and (b) 1 in 5 Na(3) sites
substituted for an Na(31) site located as far apart as possible. Na+ cations in Na(3) sites are modelled as
black spheres and those to that in a Na(31) site as a grey sphere, with the framework and water
modelled as grey sticks.
109
3. E X P E R I M E N T A L P R O C E D U R E
A prerequisite for understanding cation directing effects is that both Li-A(BW) and Na-(JBW)
are synthesised under identical conditions, differing only in the alkali present in solution. The
approaches used to synthesise Na-J(BW) were therefore utilised to synthesise Li-A(BW) for
consistency.
A recently proposed synthesis method for both Na-J(BW) and Li-A(BW) [12] using meta-
kaolin, resulted in the successful synthesis of both zeolites. Here uncalcined kaolin (-~46%
SiO2, -~39% A1203) rather than the meta-kaolin precursor was used, preparing gels of
composition A1203"2SIO2"1.45Na20"31H20 and A1203"2SIO2-1.45Li20-31H20 treated under
hydrothermal conditions at 473K for 96h.
X-ray diffraction data obtained (at 298K) on the BM1B station (X = 0.7997 A) at the
European Synchrotron Radiation Facility, Grenoble, confirmed the framework type of our Li-
A(BW) and Na-J(BW) samples. Refinement of the structure of Na-J(BW) by application of
Rietveld refinement and Le Bail extraction for two impurity phases (which were assigned to
sodalite and vishnevite (Na6.sKl.zCa0.12(Si6A16024)(SO4)0.96(H20)2, isostructural to hydroxy-
cancrinite (Nas(Si6A16Oz4)(OH)l.4(CO3)'3(H20)6.35)) gave preliminary results which indicate
approximately 40% occupation of Na(31) sites, about double that reported by Hansen et al.
[20] (Table 2). These initial results give an indication of the positions and occupations of the
two hydrated Na + sites, as the presence of two other phases induces greater errors and
difficulty in refinement of a single phase within a sample.
Table 2
H~,drated sodium atomic positions and populations
Atom x y ,, z
Na(3) 0.210 0.510 0.325 0.6 + 0.024
Na(31) 0.212 -0.008 0.213 0.4 + 0.024
25000 -
20000
15000
E 10000
O
o
5ooo
0 I IIII IIIIIIIIIIIIIIHIIIIllllllIIIHIliiNlililllliniI
I I I I I Ill Ill Ill Ill II Illllllllllllllllllllllllllllllllllllllllllllll
I I I I II il IIII Ill IIIIIIIILIHIIliL IlllHIIlil Nlli IIIIIIIIIIIIIlillllllllllllllll IIIIlilllllllllllii
Since the experimental diffraction data was obtained at room temperature and that OK
atomistic energy minimisation ignores thermal factors present at ~273K, modelling the
dynamics of the system at higher temperatures was performed to understand thermal
sensitivity of cation site occupations.
4. M O L E C U L A R DYNAMICS SIMULATIONS
The dynamics of the framework and interstitial species were modelled by application of
molecular dynamics. Evaluation of Newton's laws of motion gives the position and velocities
of each atom at each time-step, allowing the evolution of the system over time to be observed.
Molecular Dynamics (MD) simulations were implemented using the DL_POLY code [30]. In
this study the isobaric-isothermal NPT ensemble was applied. A time-step of 0.5fs (0.0005ps)
was used for a period of up to 112.5ps. Simulations were performed within periodic boundary
conditions and the minimum image convention at temperatures of 100, 200, 300 and 400K.
The system was allowed to equilibrate at each temperature over a timescale of up to 37ps.
Once equilibrated to the target temperatures, the final high temperature configurations were
optimised by quenching to OK, to assess the stability of the cation site occupations.
The atomic trajectories sampled during the simulation were analysed to identify whether
migration of Na(3) to Na(31) sites occurred. Above 300K, cation motion was activated and
0.50-
4.0-
/"'i (b)
2< 0.45.}i. (a)
?< 3.5-
0.40-
0.35-1!!'I .-/"- . . . . . . . . . . . . . . .
/
::>.....
r
3.0 71/ ,,
,' lJ \\'i
lhii~ ,";.... ".... ...........................
.. "t,.,,\,
0
2.5
t~ \\
i5 2.0 \
9 0.20 ~,'i,",~:..:2;2:~:-~.:~~-~~~
"'."~t " ;~. , . i t '-~-- ' ,::+-~.~%-~-2:.:Ls _ :=~-:~ 1.5 \-..
""-.. .....
co
t-
1.0 v,~ ,A r, r..... _ . ....................................................
o. o
o os .....
0.5
0 . 0 0 ~ , , 0.0 ~ " , ,
0 87.5 0 Time / ps 82.5
Time / ps
12 t (c) ;- 35] (d)
F" ~ - ~ ~ : ~ ~ .....................
7< 30 / ' .... .,'-- ~ .......
8_i'
I ~ ---~
"~ .......
"~
9 254~/ ///"
0
13- ~- 6
._~ 6 A
1:3
t\
4 '\
0" u- 10 4 t
09
r-
\ m ~'~,;.... 56.5 112.5
2
, i 0 ' I
112.5 0 112.5
0 T i m e / ps T i m e / ps
Fig. 4. Mean square deviations (A2) (y axis) of hydrated sodium cations within the Na-J(BW) channels
over a time period (x axis) of approximately lOOps at (a) 100K, (b) 200K, (c) 300K and (d) 400K with
the MSD between 4-8A.2 and time 60-112.5ps highlighted inset. The x-axis in each depicts time in
time-steps of the evolution of the simulation.
lll
migration of hydrated sodium cations were observed along the a c plane, populating Na(31)
sites. To quantify which cations move from which sites, the sites were labelled and the mean
square displacement (MSD) of all 96 Na(3) sites within the simulation cell plotted as a
function of time (Fig 4.). Note that the average distance between the two hydrated Na(3) and
Na(31) is 2.89.A, and hence a total cumulative MSD of approximately 8.36~ indicates a
characteristic migration ofNa + from the Na(3) to the Na(3 l) site.
Although we do not observe an MSD of-8/~ until 300K, at 200K it can be seen that the
sodium ions are activated and vibrate around their mean Na(3) positions. At 300K clear
migration can be seen, approximately 5:91 (-1:18) Na + cations migrate to Na(31) sites. The
occupation of Na(31) sites are found to increase with temperature; at 400K the ratio of
Na(31)/Na(3) is observed to become 9:87(-1:9), approximately double that at 300K. In each
case, when quenched to 0K the Na(31) sites remain occupied with no migration back to the
original Na(3) site.
Analysis of the trajectories at 400K shows a linear translation within a channel (parallel to
c), in which a series of Na(3) cations migrate along the z axis, through the channel. The
MSDs were deconvoluted into cartesian components which indicated that the MSD in the z
axis was approximately 20 times larger than that of x and y. Over the duration of the
simulation, we only observed this migration once and hence longer timescales and a larger
simulation cell is required to accurately resolve how frequently this event occurs.
In Fig. 4(d), one cation is found to displace rapidly into an as yet uncharacterised location
-~5]~ from its original site. Almost concurrently, 4 cations become displaced into Na(31)
locations, which over the course of the simulation, percolate to neighbouring Na(3) sites
approximately 5.57A away from their original position, effecting a translational shift or
current. This observation suggests a potential co-operative diffusion mechanism that has been
observed in far more porous zeolites by, for example, Auerbach et al. [31 ].
5. DISCUSSION
Modelling of the Li-A(BW) phase gave structural geometries in agreement with previous
experimental studies. However, a modelling study of Na-J(BW) revealed discrepancies
between the predicted occupation at OK and reported experimental data. Structural refinement
of a 3 phase sample of Na-J(BW) yielded Na(3) and Na(31) positions and relative occupations
of 0.6 and 0.4 respectively, considerably higher than that previously reported. However, the
molecular dynamics simulations showed a pronounced increase in the occupation of Na(31)
sites with temperature. Moreover, once migration from Na(3) to Na(31) has occurred, the
reverse migration does not occur on a fast timescale and is not promoted by reducing
temperature. During synthesis therefore, activated Na(31) sites would be expected to remain
occupied. It is conceivable that fluctuations in temperature during synthesis due to heating
apparatus and local hot-spots in the synthesis mixture could lead to high occupations of
Na(31). Clearly further simulation and experimental work is needed to understand the thermal
dependence on occupation of the Na(31) site. However, these initial results indicate that
migration is activated between 200 and 300K giving a lower bound for the activation barrier.
Simulations carried out at a range of temperatures would allow us to extract the activation
barrier and hence deduce what temperature would give rise to the populations extracted from
the refinement, which will provide an additional test of the integrity of both the simulation
and the refinement.
In future work, predictions of the Li-A(BW) and Na-J(BW) surface structures will be
given. The templating effects of Na + and Li+ ions in aluminosilicate rich solution and their
112
effects upon the size and distribution of oligomers will be discussed, making use of the
methodology recently applied to study siliceous solution fragments [32]. Using constrained
optimisation procedures to estimate barrier heights of precursor condensation reactions, we
will present preliminary predictions of the solution species and how these units crystallise
from solution upon the crystal surface. The conclusions drawn from these computational
experiments will be used to provide preliminary insights into the growth processes and the
role of cation direction within solution and at the crystal surface during zeolite crystallisation.
REFERENCES
In situ infrared spectroscopy has been applied to follow the kinetics of the sol-gel
synthesis of mesoporous molecular sieves. A stirred reaction vessel has been adapted to a
horizontal attenuated total reflectance (ATR) accessory to carry out the analysis. Diamond has
been chosen as internal reflection element because of the hardness, thermal and chemical
resistance requirements that the synthesis conditions impose. The use of this tool to analyse
the formation of the surfactant-silicate mesophase and monitor the hydrolysis and
condensation reactions during the synthesis of MCM-41 materials is reported.
1. INTRODUCTION
Despite the intensive research on the synthesis of zeolites and ordered mesoporous materials,
little attention has been devoted to the use of in situ techniques to monitor the progress of the
synthesis of these molecular sieves. Nonetheless, the number of in situ studies of zeolite
crystallization, mainly using X-ray scattering techniques, has increased in recent years aiming
to provide physicochemical evidences on the synthesis mechanisms. A number of in situ
studies have also been reported concerning the synthesis of organized mesoporous materials.
In a recent review, Patarin et al. [ 1] concluded that, according to these in situ studies, the key
factor that controls the mesostructure is the interaction between the inorganic precursor and
the template. However, it was pointed out that each technique only probes one part of these
complex systems, and this leads to differences in the proposed mechanisms.
One technique that has shown a great potential in this field is attenuated total reflectance
Fourier-transform infrared (ATR-FTIR) spectroscopy. In the ATR cell, absorption of the
infrared beam takes place at the interface of the internal reflection element (IRE) and the
reaction mixture. In this way, a short and highly reproducible pathlength is achieved, which
allows obtaining quantitative data even in strongly infrared-absorbing aqueous solutions.
Previous studies on ATR-FTIR analysis of sol-gel processes [2-7] have been carried out using
zinc selenide as IRE. Although ZnSe is the most commonly used material due to its large
useful wavenumber range, it is chemically reactive, being attacked by acidic and alkaline
media, and does scratch. Germanium crystals are more chemically resistant and less prone to
scratching; however, this material becomes infrared opaque and can be permanently damaged
when heated above 50~ Therefore, taking into account the temperature and pH conditions
usually required for the synthesis of molecular sieves, we have considered diamond as the
most convenient IRE, among those commercially available, because of its hardness and
chemical resistance.
114
Using a specially designed ATR cell, we have carried out an in situ FTIR
characterization of MCM-41 synthesis mixtures. In this paper, the kinetics of MCM-41
formation and the interactions among the cetyltrimethylammonium surfactant molecules are
analysed.
2. E X P E R I M E N T A L
In situ FTIR spectra have been recorded using a specially designed glass cell attached on top
of a horizontal Attenuated Total Reflectance (ATR) accessory (SensIR Technologies
DurasamplIR, 9-reflection diamond sampling surface). The reaction vessel is designed to hold
from 50 to 100 ml of sol, equipped with a stirrer and can be heated at temperatures up to
100~ by using a circulating heating bath. Addition of reactants, solvents or catalysts, as well
of sampling of the reaction gel, can be carried out through an inlet/outlet port. Infrared
spectra in the range 650-4000 cm -~ were recorded at 4 cm -~ resolution, using a Thermo
Nicolet Nexus FTIR spectrometer provided with a liquid nitrogen-cooled MCT detector.
Happ-Genzel apodization function was applied.
Two synthesis of purely siliceous MCM-41 using different silica sources were carried out
in the ATR cell:
- Synthesis of MCM-41 using tetraethyl orthosilicate (TEOS).
Synthesis was performed at room temperature using a low surfactant concentration, by a
procedure adapted from that reported in ref. 7. The molar composition of gel mixture was: 1
SiO2 : 0.12 CTABr : 33 NH4OH : 257H20. The synthesis was as follows: 0.66 g of
hexadecyltrimethylammonium bromide (CTABr, Aldrich) were added under stirring to 76.01
g of a 25% ammonia aqueous solution (Aldrich), until the solution became clear; then, 3.13 g
of TEOS (98 wt%, Aldrich) were added to this solution and the obtained gel was stirred for 2
hours. TEOS addition was taken as zero time for the synthesis.
- Synthesis of MCM-41 using colloidal silica.
The synthesis was carried out at 75~ using the following gel composition: 1 SiO2 : 0.12
CTABr : 0.28 TMAOH : 26.2 H20. The synthesis was as follows: 7.28 g of CTABr was
dissolved at 35~ into 50.73 g of deionized water. When the solution was clear, 25.00 g of
colloidal silica (40 wt% suspension in water, Aldrich) were added under stirring until the
solution became homogeneous. Then, it was transferred into the ATR cell and heated to 75~
To this solution, 17.00 g of tetramethylammonium hydroxide (TMAOH) aqueous solution
(25wt%, Aldrich) were added and this time taken as zero time. The mixture was reacted for
24h.
X-ray powder diffraction (XRD) patterns were collected using Cu K~ radiation, using a
Seiffert XRD 3000P diffractometer with a curved graphite secondary monochromator, a
primary automatic divergence slit and a 0.2 mm detector slit. Patterns were recorded between
0 ~ and 10~ two-theta, with a step size of 0.020 ~ and a step time of 3 seconds per step.
3. RESULTS AND D I S C U S S I O N
3.1. S y n t h e s i s with T E O S
Synthesis of purely siliceous MCM-41 was carried out in the ATR cell at room
temperature using TEOS as the silica source. The pH of the synthesis mixture decreased from
an initial value of 12.2 to 11.7 after nearly 1.5 h of reaction. Continuous collection of the
series of FTIR spectra of the gel started as TEOS was added to the basic aqueous CTABr
solution. Fig. 1 shows selected ATR-FTIR spectra in the region corresponding to the
115
0.6 1040
"T
O E
E 0.5 A v
0
c 1035 c
O
.2 0.4
o O
c 96
mO
.s 63 ~ 0.3 , 1030
c
c, 25 ~ 0.2
9
1025-•
=~ o.1
, . , . , . , . , 0.0 i i , i , i , '1 ,
Fig. 1. Left: ATR-FTIR spectra of MCM-41 synthesis gel collected at room temperature. Labels
indicate reaction time, in min. Spectra have been corrected for absorption due to the aqueous ammonia
solution. Right: Vas(Si-O) band location (triangles) and ethanol concentration (line).
sol-gel reaction. The production of ethanol can be determined by integrating the isolated band
at 880 cm t. In order to carry out a quantitative analysis, the ATR intensity of this band was
calibrated using measured absorbances for known ethanol and water mixtures. The calculated
ethanol concentration (Fig. 1) shows a sharp increase as TEOS is added to the reaction cell
and reaches a constant level, slightly lower than that corresponding to complete hydrolysis of
TEOS, after one hour of reaction approximately.
Analysis of the infrared bands due to C-H stretching and deformation modes of the
surfactant can provide information on conformation and ordering of their molecules. Thus,
monitoring of the sol-gel reaction with the ATR-FTIR cell to follow changes in these bands
allows to evaluate the effect that growing of the CTA-silicate mesophase has on the packing
of surfactant molecules.
The C-H stretching and deformation regions of selected FTIR spectra have been plotted
in Fig. 3. It can be observed that, initially, the intensity of the CTA bands quickly increases
and reach a nearly constant level after several minutes of reaction. The rise of the CTA bands
intensity can be attributed to the growth of the surfactant-silicate mesophase. During the first
minutes of reaction, a fast growth of the CTA-inorganic oligomers mesophase takes place that
increases local CTA concentration in the gel region illuminated by the infrared evanescent
wave.
Four bands can be identified in the C-H stretching region. The bands at 2923 and 2853
cm -1 are assigned to the asymmetric and symmetric stretching of methylene groups of the
surfactant tail, and the weak bands at 2957 and 2872 cm -~ are assigned to the asymmetric and
symmetric stretching of terminal methyl groups. It is noticeable that there is a slight frequency
shift for these bands with reaction time. This shift can be more accurately determined for the
stronger bands, those belonging to methylene groups, by means of interferogram interpolation
and baseline correction. In this way, a small red-shift of ca. 0.5 cm -1 is calculated for the
methylene asymmetric stretching band after 5 min. of reaction. On the other hand, the
symmetric stretching band, which is less sensitive to changes in the tail conformation, shows
a negligeable shift. Therefore, although the frequencies of the methylene stretching bands are
consistent with a liquid-like environment of the CTA tails, the results suggest that subtle
changes of the molecular packing occur as soon as the mesophase forms. Indeed, a decrease
of the methylene stretching frequencies can be attributed to a decrease of the number of
gauche conformers in the methylene chains, that is, to a higher degree of ordering [8].
>.,
~
t""
t-
2 4 6 8 10
2e (o)
Fig 2. Powder X-ray diffraction pattern of as-made MCM-41 synthesized in the ATR cell. Synthesis
with TEOS at room temperature for 1.5 h.
117
o~ 0.1 10.02
o J6
, , ,
124 h
05 //~~--"~-~\\ //2 0 h
9 //////~X~-~\\\ ~///16 h
////III~\\~--~\X~ v Io h
/4.5 h
o~ /2.5h
//1.5h
~ / / ~ //// \ % ~ ~ ~ ~ 1.0 h
~---'////~ X X ~ / ~ 13 m in
/ // / \ \\ / 6 m in
/ // / \ \ ~ 4 m in
solution. Weak bands due to the surfactant can also be observed at ca. 965 and 908 cm -~.
When TMA is added to the cell, a band at ca. 950 cm -~ is also observed, that identifies the
tetramethylammonium species. As a result of addition of TMA, pH rises from 9.2 to 13.2. In
parallel with the change of pH, the maximum of the Si-O composite band shifts from 1106 to
1102 cm -~. This shift can be attributed to a significant change in the distribution of silicate
species (SixOy(OH)z"-), which average negative charge is expected to increase due to the large
increase of pH [9]. Formation of the mesostructured silicate phase takes place at a relatively
low rate following the addition of TMA, as indicated by the developing of bands at ca. 1030
and 1200 cm -~, characteristic of the Si-O asymmetric stretching fundamentals in solid
silicates. The intensity of these bands slowly increases with time. On the other hand, the
composite band of colloidal silica continuously decreases and narrows, indicating a decrease
of the concentration of the highly polymerised silicates species. The maximum of the
composite band also shifts toward higher wavenumber values with time. As the pH of the
mixture measured at the end of the reaction was 10.9, it is concluded that probably the
decrease of pH along the reaction largely contributes to the band shift. After several hours at
75~ the FTIR spectra change very slowly and still show the silicate composite band
superimposed to the mesostructured silicate bands. Even after 24 h, the FTIR spectrum
indicates that the mixture contains unreacted colloidal silica species. Furthermore, the shape
of the bands corresponding to the solid silicate suggests a poor ordering of the mesostructured
material. This is confirmed by the XRD pattern of the as-made sample obtained after 24 h of
reaction (Fig. 5). The diffractogram shows three peaks at low angle that identify the
hexagonal MCM-41 phase, although the low intensity of the peaks assigned to (110) and
(200) reflections evidences a poorly ordered structure.
Spectra in the region of C-H stretching and deformation fundamentals (Fig. 6) are
characteristic of the surfactant molecules in a micellar environment, as it has been discussed
above for the results obtained in the synthesis with TEOS at room temperature. However, in
this case, the strong 6as(CH3-N+) band due to TMA at 1488 cm -1 can also be observed,
overlapping with the ~Sas(CH3-N+) doublet of CTA molecule. It is interesting to note that the
intensity of the TMA band at 1488 cm -I slowly increases with time during the first hour,
which indicates a progressive incorporation of TMA + cations into the CTA-silicate
119
e--
2 4 6 8 10
20 (o)
Fig 5. Powder X-ray diffraction pattern of as-made MCM-41 synthesized in the ATR cell. Synthesis
with colloidal silica at 75~ for 24 h.
mesophase. After one hour, the relative intensities of TMA and CTA bands remains constant.
This result suggests that the relative concentration of TMA + and CTA + cations in the silica
mesophase reaches a steady level during the first hour, but remains constant at longer times
despite the growth of the mesophase.
The analysis of the evolution of infrared bands corresponding to the surfactant reveals
more marked changes in the case of the synthesis with colloidal silica at 75~ The spectrum
of the solution that was recorded before TMA addition shows methylene stretching bands
whose centres (determined by spectral interpolation and baseline correction) are located at
2924.4 and 2853.7 cm -~. Significant changes in those bands are observed after one hour of
reaction approximately. The intensity of the bands increase and the bands shift toward lower
wavenumber values. The calculated extent of this red-shift is 0.9 cm -1 for the asymmetric
stretching band and 0.4 cm 1 for the symmetric stretching band. Such shifts imply an increase
of the ordering of the surfactant tails notably larger than that observed for the synthesis with
TEOS.
It can be observed that the 5(ot-CH2) band around 1420 cm -1 shifts toward higher
frequency, also suggesting an increase of ordering of the surfactant headgroups. However,
analysis of the corresponding changes of the 8as(CH3-N +) doublet of CTA molecules (bands at
1490 and 1479 cm -~) due to changes in the headgroup packing can not be done in this case
because of the overlapping of the strong 6as(CH3-N +) band due to TMA at 1488 cm -l. In fact,
0.1
0.1
r
24 h~ /
,_ 20 h
o
,-, 1Oh
< 55~
4"5t ~ ~
15 ~
# m.in ~ /
o "nm/~,
.....4 "ninI' "-~"
' ' ' i , | , ,
for an aqueous solution of CTA with the same surfactant concentration and temperature than
those used in this experiment, the surfactant would most probably form spherical micelles
[10]. However, the formation of the hexagonal MCM-41 phase indicates that rod-shaped
micelles must be present in the reaction mixture. Changes in the FTIR spectra of CTA during
the synthesis of MCM-41 support that a change in the micelles shape actually occurs. An
increased ordering of molecular packing is suggested by the FTIR data, as it is expected for a
transition from sphere to rod [8]. According to this interpretation, it can be concluded that in
the aqueous solution of CTA and colloidal silica, the surfactant forms spherical micelles.
Condensation of the negatively charged silicate oligomers, which interact with the cationic
surfactant, would drive a transition of the micelles to rod-shape, allowing the hexagonal
mesophase to form. A similar transition from spheres to rods probably occurs in the synthesis
with TEOS. However, the intensity of the CTA infrared bands is initially very low due to the
low surfactant concentration used in this case. Substantial growth of the mesophase seems to
be required to obtain intense enough infrared bands and, therefore, only the final steps in the
change of micelle shape can be observed.
4. CONCLUSIONS
In the synthesis of MCM-41 using TEOS as the silica source, fast hydrolysis/condensation of
the precursor occur at room temperature. A highly ordered hexagonal mesophase is obtained
in less than one hour. Small changes observed in the infrared bands of the surfactant are
attributed to an increase of the degree of ordering of the surfactant molecules as the CTA-
silicate mesophase grows.
Synthesis of MCM-41 takes place at a much slower rate using colloidal silica. After 24
hours at 75~ the reaction is not complete and a hexagonal mesophase with poor order is
obtained. FTIR data indicate significant changes in the packing of surfactant molecules that
might be attributed to a sphere to rod-shape transition of the surfactant micelles driven by the
interaction with silicate species.
ACKNOWLEDGEMENT
This work has been supported by the Spanish Ministerio de Ciencia y Tecnologia (Plan
Nacional de Investigaci6n Cientifica, Desarrollo e Innovaci6n Tecnol6gica) under projects
MAT2001-5131-E and MAT2003-07769-C02-02.
REFERENCES
[ 1] J. Patarin, B. Lebeau and R. Rana, Curr. Opin. Colloid Interf. Sci., 7 (2002) 107
[2] A.A. Kline, M.E. Mullins, J. Am. Ceram. Soc., 74 (1991) 2559
[3]. A.A. Kline, T.N. Rogeres, M.E. Mullins, B.C. Cornilsen, Lj.M. Sokolov, J. Sol-Gel Sci.
Technol., 2 (1994) 269
[4] D.C. Calabro, E.W. Valyocsik, F.X. Ryan, Microporous Mater., 7 (1996) 243
[5] V.L.Zholobenko, S.M. Holmes, C.S. Cundy, J. Dwyer, Microporous Mater., 11 (1997) 83
[6] S.M. Holmes, V.L. Zholobenkko, A. Thursfield, R.J. Plaisted, C.S. Cundy, J. Dwyer, J. Chem.
Soc., Faraday Trans., 94 (1998) 2025
[7] Q. Cai, W.-Y. Lin, F.-S. Xiao, W.-Q. Pang, X.-H. Chen, B.-S. Zou, Microporous Mesoporous
Mater., 32 (1999) 1
[8] J.G. Weers and D.R. Scheuing, J. Colloid Interf. Sci., 145 (1991) 563
[9] C.J. Brinker and G.W. Scherer, Sol-gel science. The physics and chemistry of sol-gel
processing, Academic Press, San Diego, 1990
[10] J. Frasch, B. Lebeau, M. Soulard and J. Patarin, Langmuir, 16 (2000) 9049
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 121
Department of Chemistry, Moscow State University, Lenin Hills, 119892 Moscow, Russia
1. INTRODUCTION
Microporous molecular sieves, such as zeolites and zeolite-like materials, are among the most
widely used heterogeneous catalysts. However, because of the limited size of their pores (< 1
rim), they cannot be applied for conversion of large molecules. The appearance of mesoporous
molecular sieves was expected to solve this problem. However, as it was revealed by
subsequent studies, their rather low stability in the presence of water vapour and
comparatively low acidity did not allow them to satisfy the requirements for commercial
catalysts and therefore none of these materials has been commercialised as yet. It was thus
very attractive to design the materials, which will combine the advantages of both
mesoporous molecular sieves and zeolites.
Two approaches can be used for the design of such kind of systems. The first approach
is based on the creation of mesopores in the crystals of zeolites by conventional
dealumination procedures, such as steaming, acid and alkaline treatments [1-4], or matrix
assisted synthesis [5]. The second involves building of mesoporous materials from tiny
zeolitic species [4,6-14] or transformation of the preassembled walls of mesoporous materials
such as MCM-41 and SBA-15 into zeolitic structures by post-assembly treatment with zeolite
structure directing template [ 15-17].
The analysis of the literature data on the micro/mesoporous composites suggest that in
most of the cases these materials show enhanced stability and acidity with respect to
mesoporous materials. In some cases, there are also evidences for the enhanced catalytic
activity. However, the information on the catalytic properties of such materials is rather
limited.
In this paper, we present the results on the catalytic properties of micro/mesoporous
composite materials obtained by recrystallization of mordenite. Transalkylation of biphenyl
with para-diisopropylbenzene and cracking of 1,3,5-triisopropylbenzene are selected as
model reactions.
122
2. EXPERIMENTAL
The characteristics of the parent and recrystallized materials presented in Table 1 indicate that
recrystallization of De-MOR have not affected the chemical composition of the sample. On
the contrary, the structure and texture of recrystallized samples were influenced significantly
by the conditions of recrystallization. In particular, the concentration of NaOH solution taken
for zeolite destruction was the most important parameter (Table 1, Figs. 1-4). The effect was
fully reproducible on a large number of experiments.
The results of FTIR study showed that the destruction of starting mordenite in 0.4 M
and 0.8 M solutions did not affect significantly the crystallinity of recrystallized mordenites
(Fig. 1). In the case of the Re-MOR 1, the bands at 570 and 592 cm L corresponding to
123
stretching vibrations became even more intensive and sharper. On the contrary, treatment in
1.6 M solution led to significant decrease of the intensity of these bands and pointed to the
absence of large mordenite crystals.
Table 1
Characteristics of parent and recrystallized mordenites
Samples Si/A1 V, cm3/g Vmic, cm3/g Dmeso, ]k ea , A ao b, lamol./g
t 7/
c' =
I
i40o ~2oo 1ooo soo ;oo 10 20 30 40
Wavenumbers (cm-l) 20
Fig. 1. IR specra of parent mordenite De-MOR Fig. 2. XRD patterns of parent mordenite
(l) and recrystallized samples Re-MOR 1 (2), De-MOR (1) and recrystallized samples
Re-MOR 2 (3), Re-MOR 3 (4). Re-MOR 2 (2), Re-MOR 3 (3).
XRD data obtained in the range of 5<20<50 (Fig. 2) confirmed the results of FTIR
study. On Re-MOR 2, only small halo in the region of 15 - 30 20 due to amorphous material
was observed, while on Re-MOR 3 the peaks corresponding to mordenite phase practically
disappeared and the broad halo gave the main contribution to the XRD pattern. The XRD data
recorded in the region of 5<20<50 did not allow to verify, whether the observation of
amorphous phase was due to formation of mesoporous material. This information was
obtained from XRD patterns obtained at low angles (Fig. 3), pointed that recrystallization leds
124
to formation of ordered mesoporous materials. The group of peaks observed in Fig. 3 can be
indexed as (100), (110), (200) and (210) reflections associated with hexagonal symmetry. The
increase of the concentration of alkali solution in a series of Re-MOR1, Re-MOR2 and Re-
MOR3 samples resulted in the improvement of the quality of mesoporous phase, suggesting
that the decrease of the size of mordenite species obtained during dissolution step simplify
their assembling in ordered mesoporous phase.
The information obtained from the isotherms of N2 adsorption- desorption also
pointed to formation of ordered mesoporous materials. Fig. 4 shows that recrystallized
mordenites exhibit rather sharp steps at p / p o ~ 0.35, corresponding to the existence of uniform
mesopores. The horizontal hysteresis loop after p / p o ~ 0.45 further suggested the existence of
secondary texture mesopores. According to BJH calculation, the size of the mesopores was
about 3.0 nm. The formation of mesopores resulted in significant increase of pore volumes of
the samples upon recrystallization (Table 1). It should be noted that the higher was the
concentration of NaOH solution used for zeolite destruction the higher was the growth in
mesopore volume. On the contrary, the micropore volume decreased with the concentration of
NaOH solution used for zeolite destruction. This suggests that the degree of recrystallization
of mordenite into mesoporous material can be adjusted by variation of the concentration of
NaOH in solution.
800
700 4
, ~ 6oo
500
i 3oo 2
~o
100
[ I I [ [ I [ o . . . . .
0 2 4 6 20 0 0.2 0.4 0.6 0.8 1
ptpo
The textural characteristics of the materials were further assessed on the basis of the
combination of the XRD and adsorption data using the geometrical method described in [ 18].
The results obtained (Table 1) suggest that the increase of the concentration of NaOH solution
in a series of Re-MOR1, Re-MOR2 and Re-MOR3 samples leads to the decrease of wall
thickness of mesoporous material from 24.0 to 9.5/~. This observation is probably due to the
125
decrease of the size of mordenite building units, from which mesoporous materal is
assembled.
To summarize, the analysis of the structure and texture of recrystallized mordenites
suggests that after dissolution step, the fragments of mordenite existing in solution are further
tailored to mesostructure, giving composite micro/mesoporous molecular sieves with the
chemical composition similar to starting material. The degree of recrystallization of mordenite
into micro/mesoporous composite material as well as the structural and textural characteristics
of this material can be controlled by the variation of the concentration of NaOH in solution.
The acidity of micro/mesoporous materials was studied by TPD NH3 (Fig. 5). The
results show that the samples Re-MOR 1 and Re-MOR 2 exhibited strong acidity. On
Re-MOR 1, the amount of strong acid sites was even higher than on parent mordenite. This
suggests that these materials are not simply a mixture of mordenite and mesoporous
composites. On the contrary, the TPD profile for the sample Re-MOR 3 was quite different:
the contribution of weak and strong sites decreased significantly, while the amount of medium
sites remained the same. The total amount of sites on Re-MOR 1 and Re-MOR 2 was similar
to starting moredenite, while for on Re-MOR 3 it decreased by a factor of 3 (Table 1).
0.3
1
~. 0.25 --2
0.2 34
~" 0.1
~ 0.05
0 ~ ~
and meta-DIPB, meta-DIPB being the major product of isomerization. Among all the
products formed, DIPBP were the target products, since they can be further used in the
synthesis of advanced polymer materials with high thermal and mechanical stability or
thermotropic liquid crystal properties.
The parent mordenite showed rather low conversion of p-DIPB and BP (Table 2). The
main reaction pathway over this catalyst was transalkylation of BP and p-DIPB leading to
IPBP. Besides that, the formation of QP occurred in rather large extent. The latter was
probably responsible for very fast catalyst deactivation in the beginning of the reaction over
this catalyst.
Table 2
Transalkylation of biphenyl and p-diisopropylbenzene over re-crystallized mordenites
Catalyst De-MOR Re-MOR1 Re-MOR2 Re-MOR3 MCM-41
Conversion of (%)
BP 24.7 69.1 54.5 15.2 23.6
DIPB 3.7 24.9 22.3 5.4 14.7
Selectivity to BP products (% mol.)
IPBP 37.8 34.7 42.4 43.7 73.4
DIPB 17.9 47.1 39.1 2.5 6.7
C~8+ 5.4 5.1 9.7 12.4 4.2
QP 18.9 4.6 0.7 34.8 2.6
Other 19.9 8.4 8.2 6.6 13.1
Recrystallization of De-MOR into Re-MOR1 and Re-MOR2 led to 2-3 fold increase of
BP conversion and 6-7 fold increase of the conversion of p-DIPB with respect to starting
mordenite. Furthermore, it resulted in the increase of the selectivity towards DIPBPs, while
selectivity to QP decreased tremendously and the stability of the catalytic activity increased.
Thus, composite micro/mesoporous materials Re-MOR1 and Re-MOR2 showed improved
catalytic properties with respect to both microporous and mesoporous materials, again
suggesting that they are not a physical mixture of mordenite and mesoporous material. On the
other hand, deeper recrystallization into Re-MOR3 led to rather low catalytic activity,
comparable with MCM-41 catalyst (Table 2).
The results on the catalytic activity of parent and recrystallized mordenites in 1,3,5-
TIPB cracking are listed in Table 3. The main reaction products were para- and meta-
diisopropylbenzenes (p-DIPB and m-DIPB), cumene, benzene and propylene. Besides
propene, other C2-C4 gases, mainly propane and butanes, were formed in considerable
amounts. Aliphatic products C4+, as well as other aromatics including toluene, ethyllbenzene,
xylenes, butylbenzenes and polysubstituted benzenes were observed in smaller amounts.
The starting De-MOR sample showed rather low conversion due to diffusion limitations
of bulky 1,3,5-TIPB in the pores of mordenite. The main product of cracking over this
catalyst was cumene. Besides cracking, other processes such as hydride transfer, propene
oligomerization, aromatisation and alkylation of aromatics with the fragments formed upon
propene transformation contributed to the overall reaction pathway. Strong acid sites of
mordenite were most probably responsible for these reaction pathways.
Recrystallization of De-MOR into Re-MOR1 led to 2 fold increase of 1,3,5-TIPB
conversion as a result of formation of mesopores favouring transport of 1,3,5-TIPB. In
addition, the contribution of deep cracking into cumene and benzene increases, probably due
127
Table 3
Cracking of 1,3,5-trisiopropilbenzene over recrystallized mordenites
Catalyst De-MOR Re-MOR1 Re-MOR2 Re-MOR3 MCM-41
Conversion of 1,3,5-TIPB (%)
7.1 13.4 15.2 36.8 34.0
Selectivity (% mol.)
C3H6 28.3 38.1 43.7 39.1 35.8
~C2-C4 24.2 19.6 10.6 12.6 3.4
aliphatic C4+ 1.1 1.1 0.5 0.9 0.2
benzene 2.7 4.5 1.9 0.3 -
cumene 18.7 22.2 22.4 18.3 3.0
m-DIPB 6.4 4.3 8.8 17.8 40.1
p-DIPB 1.1 1.1 2.6 5.3 4.0
Other aromatics 17.5 9.1 9.5 5.7 13.5
To summarize, the catalytic data obtained in both catalytic reactions reveal that the
composite micro/mesoporous materials prepared by mordenite recrystallization exibit
improved catalytic properties with respect to both microporous and mesoporous materials. It
should be mentioned, however, that the optimal degree of recrystallization depends on the
catalytic reaction selected. Thus, for transalkylation reactions mild recrystallization leads to
the best catalyst performance. On the contrary, for cracking, deeper recrystallization is
needed.
4. CONCLUSIONS
ACKNOWLEDGMENTS
The financial support by Russian Foundation of Basic Research, Council for the Grants of
Russian President (project NSh-1275.2003.3) and INTAS grant (project 03 51-5286) is gratefully
acknowledged.
128
REFERENCES
alnstitut Europ6en des Membranes (UMR 5635 CNRS), UM II, CC47, Place Eugbne
Bataillon, 34095 Montpellier Cedex 5, France
bUniversity of Colorado, Chemical & Biological Engineering Dept., UCB 424, Boulder, CO
80309, USA
CKauno Technologijos Universitetas, Organines Chemijos Katedra, Radvilenu pl. 19, 50299
Kaunas, Lithuania
The objective of this work is to evaluate the use of microwave (MW) heating to accelerate the
synthesis of silicalite-1 membranes by secondary growth. Silicalite-1 seeds (50-60 nm) were
first prepared by MW-assisted hydrothermal synthesis from fresh sols. The MW seed
suspension was deposited by slip-casting on commercial c~-A1203 supports and zeolite
membranes were obtained after a synthesis time as short as 30-60 min, by MW-assisted
secondary growth at 393-453 K. The membrane thickness varied typically from 200 nm to a
few gin, with glassy like or columnar morphologies respectively, depending on the synthesis
conditions. The very first single gas permeation results revealed a relatively high single gas
permeance for the membranes prepared during 2 hours at 403-453 K, and a n/i-butane
selectivity which tends to reach a maximum (40-50) for the membrane prepared at 433 K.
1. INTRODUCTION
The preparation of zeolite membranes with controlled microstructure, quality, thickness and
composition is desirable for a number of applications such as pervaporation, gas separation and
catalytic reactors. On the other hand, the development of efficient, reproducible and low cost
synthesis methods are still required for industrial development of this new and promising
generation of inorganic membranes. The secondary growth of a layer of zeolite seeds deposited
on a support is currently recognized as one of the most attractive and flexible methods for
orienting the formation of consolidated membranes and films including A, L, Y, ZSM-5, and
silicalite-1 [ 1]. Colloidal zeolites seeds can be prepared by either classical [2] or MW assisted
hydrothermal synthesis [3,4]. Membrane growth from seeds provides better control of the
membrane formation process by separating the crystal nucleation and growth, which results in
short crystallization times [5]. MFI membranes with attractive performance (high selectivity
and fluxes) have recently been obtained by mastering the seed properties and deposition
methods [6,7].
130
Among the recent developments for further shortening zeolite membrane synthesis duration,
microwave (MW) heating appears to be an attractive option. Although MWs have been used
for approx. 15 years for zeolite synthesis, the first investigations devoted to membranes were
only reported a few years ago (A1PO4-5, SAPO-5 and LTA). NaA planar membranes were
recently successfully synthesized by Chinese teams [8-9] and the batchwise method was
extended to the continuous synthesis of NaA and NaY membranes, by H. Kita [10]. SOD
membranes were also derived from this method [ 11-13 ]. MW assisted hydrothermal synthesis is
attractive for limiting the synthesis duration, but also for improving the properties of zeolite
membranes. Indeed chemical attack of the support (usually ot-A1203) during long-term synthesis
alters both the zeolite composition and the support integrity. Furthermore, membrane thickness,
degree of crystallinity and membrane performance can be adjusted for MW assisted synthesis
derived membranes. This presentation reports our recent results on MW derived MFI
membranes obtained by secondary growth of a layer of MW derived seeds on an ot-A1203
support.
2. EXPERIMENTS
The MW oven used in this study was a computer controlled Milestone ETHOS 1600, with
pressure and temperature sensors in one of the Teflon autoclaves. For each sol composition
and autoclave filling rate, the maximum MW power (Pmax 0-1000 W), the maximum
temperature (T) and the experiment duration (t) were fixed. In this work Pma.~(directly related
to the sample heating rate) was fixed at 400 W. The evolutions of temperature, pressure and
power were continuously recorded during each experiment.
3. RESULTS
~,~
* ~i:~ ~-,,~I~i,f! '.~,~ ? ~ ~ ~ #,',~ ~ , ~5 ~ N~,~ ' ii~! . . . . ..........
....... !d;ii~i:L:;:;;~ii
Fig. 1. FESEM observation of the ct-A1203 support covered with an homogeneous layer of MW
derived silicalite-1 seeds. The layer was dried for 6 hours at 428 K.
132
Fig. 2. FESEM observations (surfaces on top al,bl,c],dl and cross-sections in the bottom a2,b2,c2,d2) of
silica|ite-1 membranes prepared by MW assisted secondary growth with different synthesis
conditions: a) 2.5 hours, 453 K; b) 2.5 hours, 433 K; c) 2.5 hours, 413 K; d) 2.5 hours, 393 K.
133
4.5
4
3.5
-+-,150 "J
3
m, + 120 i/
U
2.5
7/"
U /
I
0.5
r-- 9 - :: . . . . lr . . . . . . . . . . . . . . !r . . . . . . . . . . . . . . . . . . ! . . . . . . . . . . ! ...... ' ~ . . . . . . . . f " ................... !
Fig. 3. Influence of the MW-assisted hydrothermal synthesis temperature and duration on the final
silicalite-! membrane thickness.
Very thin membranes are obtained (below 500 nm) at low temperature (below 410 K)
whatever the synthesis duration. The influence of the synthesis time on the membrane
thickness is more effective at high temperature.
A comparison of surface morphology and cross-section of the silicalite-1 membranes
prepared with two different synthesis durations at either 453 K or 413 K is shown in Fig. 4.
At 453 K, the membrane thickness decreases from 4 lam to 1.2 ~tm, respectively when the
synthesis is shorten from 150 to 30 min. The influence of synthesis duration on membrane
thickness is lower at 423 K.
The influence of the initial seed layer thickness and its competitive dissolution during the
secondary growth step, in relation with the synthesis temperature, is under study.
134
. . . . . . . . . . . . . . . . . . . . . . . . . . . ~. +++;,~+>o, +
+~:++;+"j'?:~;;~:+ ?,+::~!,,+:::k.:~+,
'::!....... :+~i+',.:. +::..,::~,~+'~+:+
::i ~.+:'~+
. j+'+;:, ++++++
'+i++/+.++i. .
' ;+ ~. + ,''~ ~ :' "=., . +~' ;+' ".~';T.', ....... : : 7 " '~, '' .. .
Z I :: ,. . . .
..... '.. ,, . . . . . .... ~ ,. ;, . ~ , ,~. ,. .
.,..: , ' . . ; . . , i="+ ' . , :,,., ,+ ,. ,
, ,/ :~+ ~ '~ ,,+, ~,,,,,: ,~ ,,,, +. i ~' ,:~+~ ' ,, .... '.~,+
Fig. 4. FESEM observations (surfaces on top a~,b~,ct,d~ and cross-sections in the bottom a2,b2,c2,d2) of
silicalite-1 membranes prepared by MW assisted secondary growth with different synthesis
conditions: a) 2 hours, 413 K; b) 2 hours, 453 K; c) 30 min, 413 K; d) 30 min, 453 K.
135
1~'~I1 ~/I2 : ' i ,i =~ ~""~ ~;' " ' { ' i~.,~:=~,;,~,.,,?.;~i ~ r,lk~ .......i ~ . ~ ~
.;,~i~"
Fig. 5. FESEM observations of the silicalite-1 derived membranes prepared by secondary growth at
different temperatures, and used for gas permeation tests. Synthesis conditions 9 M 1) 2 hours, 403 K;
M2) 2.5 hours, 423 K; M3) 2 hours, 433 K and M4) 2 hours, 453 K.
The single gas permeation results (Table 1) revealed a relatively high single gas
permeance for all the four membranes. As expected the highest permeance was for the
thinnest membrane M1. On the other hand, the ideal selectivity of n/i-butane is different and
tends to reach a maximum (40-50) for membrane M3. The lowest selectivity factor was
obtained for membrane M4 made at the highest temperature. Since this membrane is very
thick and has a strongly columnar morphology, we could assume that crack formation is more
easily obtained for this morphology during the calcination treatment. This conclusion is
supported by the fact that the/-butane permeance is much higher through M4 than through the
other membranes.
Table 1
Results of single gas permeance for MW derived silicalite-1 membranes prepared at different
temperatures
Membrane permeance, mol/(Pa.s.m 2)
Test ....Membrane Membrane Membrane Membrane
Single Gas Temperature M1 M2 M3 M4
[K] 403K-2h 423K-2.5h 433K-2h 453 K - 2 h
H2 294 3.1.10 -6 2.1.10 -6 1.5.10 -6 1.0.10 -6...
CO2 294 3.4.10 -6 3.0.10 -6 2.7.10 -6 3.4.10 -6
N2 294 2.2.10 -6 1.9.10 -6 1.5.10 -6 4.6.10 .7
n-butane 294 1.1.10 .6 6.1.10 -7 2.2.10 .7 2.4.10 -7
n-butane 473 4.7.10 7 3.5.10 -7 2.4.10 -7 2.7.10 -7
/-butane 473 3.4.10 ~ 2.3.10 -s 5.7.10 -9 9.5.10 s
/-butane 294 5.7.10 ~ 7.1.10 -8 4.1.10 .9 1.7.10 ~
4. C O N C L U S I O N
Good quality silicalite-1 membranes can be obtained very rapidly (30-150 minutes) by MW
assisted secondary growth, from MW derived seeds. When comparing a series of membranes
prepared for 120-150 minutes at different temperatures, the highest single gas permeance and
medium n/i-butane selectivity were obtained with a membrane prepared at the lowest
temperature (120 min at 403 K), although a higher temperature (433 K) was needed to get a
higher n/i-butane selectivity (40-50) with a slightly lower (but still attractive) permeance.
Microwave assisted secondary growth is a promising method for the rapid synthesis of
silicalite-1 membranes with high permeance. We are now working to better characterize the
membrane crystallinity, to test their performance for gas mixtures, at higher temperature and
also in pervaporation mode. The extension of this synthesis method to other zeolite membrane
structures is also being considered.
ACKNOWLEDGEMENTS
The authors thank the French embassy in Vilnius (Lithuania), the CNRS (SC-exl9) and the
U.S. National Science Foundation for supporting for the stays of J. Motuzas and R.D. Noble
at I.E.M. Montpellier and of J. Motuzas at the University of Colorado (NSF grant no. EEC
0437144).
REFERENCES
aFoundation for Research and Technology Hellas, Institute of Chemical Engineering and
High-Temperature Chemical Process, P.O. Box 1414, GR 265 04 Patras, Greece
1. INTRODUCTION
The separation of light olefin/paraffin mixtures is usually carried out by cryogenic distillation,
which requires vast amounts of energy. As a result, an alternative separation process, such as
a membrane based process, could have significant economic impact. Recently published
reviews provide a summary of the work done towards the separation of olefin/paraffin
mixtures using polymeric [1] or carbon [2] membranes. The unique properties of zeolite
crystals, such as molecular sieving ability and selective adsorption, make them promising
membrane materials. Membranes from a large number of zeolite framework structures have
already been synthesized and tested in the separation of gas or liquid mixtures [3-11 ].
Faujasite type membranes have already shown promising results in the separation of a
large number of mixtures such as C02/N2 [12], C02/CH4 [13], saturated/unsaturated
hydrocarbon mixtures [14, 15], and HzO/alcohols [16]. The ability of faujasite membranes to
separate olefin/paraffin mixtures has recently been reported [ 14, 17, 18]. In the present study,
138
the effect of membrane hydrothermal synthesis temperature on the framework type, growth
kinetics and membrane permselective performance is presented.
2. EXPERIMENTAL METHODS
O-rings
. . . . . . . . ]
i i
(Retentate)
~ itl I~
[--- ~l I (Permeate)
~ gas
I chromatograph
Propane (Feed) 1 [,J I(S'
(Sweep gas)
Propylene
Helium ! [ ?t
Zeolite
Helium
III-]
I
film Support
The feed side flow rates of propane and propylene were adjusted using a variable-area
rotameter (Omega), while the flow rate of helium was controlled by a mass flow controller
(Aera FC-7700C). The total pressure on both sides of the membrane was atmospheric. The
composition of the permeate gas was measured using a Shimadzu (GC-14B) gas
chromatograph, equipped with a flame ionization detector. The separation factor was
calculated from the following equation:
( C1/ C2 ) permeate
S.F(1 / 2 ) :
( C 1/ C2) feed
,N
Fig. 2. Top-view and cross-sectional SEM images of zeolite films synthesized at 70~ (for 358hrs),
85~ (for 164hrs), 100~ (for 91hrs) and 115~ (for 35hrs).
The XRD patterns (Fig. 3) of the zeolite membranes and powders synthesized from the
same gel have been used to investigate the effect of hydrothermal synthesis temperature on
the zeolite framework formed. The characteristic peaks of faujasite can be seen in the X-ray
patterns of the membranes synthesized at 70 and 85~ (Fig. 3a). On the other hand, the
patterns of the membranes synthesized at 100 and 115~ reveal the coexistence of two
different zeolite frameworks (FAU & LTA). The X-ray patterns of the zeolite powder show
140
that very poor crystallinity was obtained at 70~ while only zeolite LTA was formed at the
highest synthesis temperature. This observation is in agreement with previously reported
results [19], indicating that with increasing hydrothermal synthesis temperature denser
structures are formed. Furthermore, by comparing the crystal phases of the zeolite membranes
and powders it is clear that the existence of a seeded layer can play a significant role in the
type of zeolite formed.
(a) ('0)
~ 115~
~ ~ . ~ ~ ~ ~ 115~
~F 100~
~ 100oc
F FFF (D
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
2 theta [degrees] 2 theta [degrees]
Fig. 3. XRD patterns of the membranes (a) and the corresponding zeolite powders (b) synthesized at
different hydrothermal synthesis temperatures, (S" support, A" LTA type zeolite, F: FAU-type
zeolite).
The effect of hydrothermal synthesis temperature on the membrane growth rate has also
been examined. As seen in Fig. 4a membrane thickness increases linearly with time until
reaching a maximum. An Arrhenius plot (Fig. 4b) of the zeolite film linear growth rates,
indicates that film growth is activated. This appears surprising because the membranes consist
of two different zeolite frameworks (FAU and LTA). However, the structure of these two
zeolite types, comprised from interconnected sodalite cages, are known to have similar
activation energies of formation [20].
20 0
# -1
15
~'m/ ~Synthesis v -2
~1o 4/ //iv T e m p e r a t u r e
i // ~-3
m
~~ " 100~ -4
O i , I I I ~ I , I ~ I
[23]. Based on the single component adsorption isotherms it can be argued that propane flux
is temperature independent and smaller than propylene's because propane molecules can not
adsorb in the crystal. Furthermore, the decrease of the propylene flux with increasing
temperature can be attributed to the observed decrease of its adsorption capacity.
Wynthesis ,-*
r~
N, T y n t h e s i s - 85~ &
m.o J~
~ 1E-3 .~ .~. =~ 1F~3 ~o
202
O
--~~~fi~o/~ m ,-Ic~
~= 1F_,4 .,-- i--_.__i ~ / 0 ~ /l /
=
1F~
__o~~-____,~, /
= SF 10 "~ = 10 "~
~o ~ 1E-5 --~
(I,)
1 I~:.Jfil , i , 1 , i , i
0 ~ , 1 ,
*~ 1E-6 i i , i , i i i t I , 0 e
211 40 6O 811 100 120 140 ~ 21) 40 (II 8O 111~ 120 140 ~
Temperature [~ Temperature [oq
--~ . ~~ Io--~ 20 O
----I ----A/ o/F1/i/l
S
w ~m. /o11O1-Jm / ~c~
~= 1F,4 ...,.o__oAlJ.i_m~n F.,,I) ~= 1F_,4
u 40"=
= 10 "~ =
~ IF_~ ~ 1F~ 0
2O 40 6O 8O 1~ 120 140 ~
v
40 60 80 100 120 140
Temperature [~
T e m p e r a t u r e ~'q
Fig. 5. Temperature dependence of permeation flux, separation factor (S.F) and ideal selectivity (S)
for membranes synthesized at different temperatures. (u, [] ~ propane, A, A - , propylene, open
symbols: Ppropylene -- 55kPa,/:)propane-----46kPa, Phelium -- 46-55kPa, closed symbols: 55kPa C3H6 / 46kPa
C3H8 mixture)
The maximum separation factor was 13.7 (C3H6 permeation flux - 3.92• 10-4 mol.m-2.sl),
measured using membranes synthesized at 85~ For the same membranes the optimum ideal
selectivity was 28.7, and had been observed at 35~ Surprisingly the membranes synthesized
at 115~ showed the highest ideal selectivity, 58.5 (observed at 35~ and the lowest
separation factor,-1.9, which did not change significantly with permeation temperature. As
mentioned in the previous section, the differences between the permeation behavior of the
membranes synthesized at 115~ and the membranes synthesized at lower temperatures can
be attributed to the existence of a continuous well intergrown LTA crystal layer.
The effect of membrane synthesis temperature on the maximum separation factor and the
corresponding propylene permeation flux is presented in Fig. 6. It is clear that the optimum
separation performance has been observed when membranes synthesized at 85~ were used.
It can also be seen in Fig. 6 that membranes synthesized without the seeding procedure
exhibited a maximum propylene/propane separation factor of-1.5 and a propylene flux on the
order o f - 1 0 -2 mol.m-2.s-]. This result indicates that the presence of seed crystals plays a
significant role in the synthesis of"good quality" membranes.
143
14
12
=
10
o/
oH 85 C 9 100 ~
~" 8
115
6 I 9 85~
4 70 ~ No seeding
g~
2 1
~q~ . . . . . . . . . 1 . . . . . . . I~
Fig. 6. Effect of membrane synthesis temperature on the maximum separation factor and the
corresponding propylene permeation flux.
4. CONCLUSIONS
Zeolite membranes were synthesized on (x-alumina disks from solutions of the same molar
composition under different hydrothermal synthesis temperatures (70-115~ Membrane
thickness increased linearlv with time and ~rowth armeared to be activated. Membranes
synthesized at 70~ and 85~ were faujasite (FAU) while the membranes synthesized at
100~ and 115~ consisted of LTA and FAU crystal phases. The maximum separation factor
of a 55kPa C3H6 / 46kPa C3H8 mixture was 13.7 at 90~ with a propylene flux of 3.92 x 10-4
mol.m-Z.s-~, and was measured using membranes synthesized at 85~ For each membrane
tested, the maximum value of ideal selectivity was higher than the maximum separation factor
indicating that presence of propylene, under certain conditions, enhances propane transport
through the membrane. The value of the maximum ideal selectivity (58.5) was observed at
35~ using membranes synthesized at 115~
ACKNOWLEDGEMENTS
The authors would like to acknowledge financial support form the Ministry of Development,
GSRT, Greece, through funding of "Greece-Poland Joint Research & Technology
Programme" (EHAv.M4.3.6.1-4282-676). The authors would like to acknowledge Prof.
Michae! Tsapatsis and Dr. Zhipping Lai, for assisting acquisition of the X-ray diffraction
patterns.
REFERENCES
[4] J. Caro, M. Noack, P. Kolsch and R. Schafer, Microporous Mesoporous Mater., 38 (2000) 3.
[5] K. Jansen and T. Maschmeyer, Topics Cat., 9 (1999) 113.
[6] Z.P. Lai, G. Bonilla, I. Diaz, J.G. Nery, K. Sujaoti, M.A. Amat, E. Kokkoli, O. Terasaki, R.W.
Thompson, M. Tsapatsis and D.G. Vlachos, Science, 300 (2003) 456.
[7] Y.S. Lin, Sep.Pur.Tech., 25 (2001) 39.
[s] M. Noack, P. Kolsch, R. Schafer, P. Toussaint and J. Caro, Chem. Eng. Technol., 25 (2002)
221.
[91 A. Tavolaro and E. Drioli, Adv. Mater., 11 (1999) 975.
[10] Y. Yan, M.E. Davis and G.R. Gavalas, J. Membr. Sci., 123 (1997) 95.
[ 11 ] C.D. Baertsch, H.H. Funke, J.L. Falkoner and R.D. Noble, J. Phys. Chem., 100 (1996) 7676.
[12] K. Kusakabe, T. Kuroda and S. Morooka, J. Membr. Sci., 148 (1998) 13.
[13] Y. Hasegawa, T. Tanaka, K. Watanabe, B.H. Jeong, K. Kusakabe and S. Morooka, Korean J.
Chem. Eng, 19 (2002) 309.
[14] V. Nikolakis, G. Xomeritakis, A. Abibi, M. Dickson, M. Tsapatsis and D.G. Vlachos, J. Membr.
Sci., 184 (2001) 209.
[15] B.H. Jeong, Y. Hasegawa, K.I. Sotowa, K. Kusakabe and S. Morooka, J. Membr. Sci., 213
(2003) 115.
[16] H. Kita, K. Horii, Y. Ohtoshi, K. Tanaka and K. Ichi, J. Mater. Sci. Lett., 14 (1995) 206.
[17] S. Nair, Z. Lai, V. Nikolakis, G. Bonilla and M. Tsapatsis, Microporous Mesoporous Mater., 48
(2001) 219.
[18] I.G. Giannakopoulos and V. Nikolakis, Ind. Eng. Chem. Res., 44 (2005) 226.
[19] E.J.P. Feijen, J.A. Martens and P.A. Jacobs, in J. Weitkamp, H. Karge, H. Pfeifer and W.
Holderich (Editors), Zeolites and related microporous materials: State of the art 1994, Elsevier
Science B.V., 1994, p. 3.
[20] H. Kacirec and H. Lechert, J. Phys. Chem., 79 (1975) 1589.
[21 ] J.M. van de Graaf, F. Kapteijn and J.A. Moulijn, J. Membr. Sci., 144 (1998) 87.
[22] J.R. Xiao and J. Wei, Chem. Eng. Sci., 47 (1992) 1123.
[23] F.A. Da Silva and A.E. Rodrigues, Ind. Eng. Chem. Res., 38 (1999) 2051.
[24] Y.H. Huang, A.I. Liapis, Y. Xu, O.K. Crosser and J.W. Johnson, Sep. Technol., 5 (1995) 1.
Studies in Surface Science and Catalysis,volume 158
J. 0ejka, N. 2;ilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 145
The overall 3-step process for the manufacturing of SiSiC-zeolite structured cellular
composites was studied. The process consists of a 2-step procedure for the ceramic monolith
fabrication, followed by the functionalisation of the support surface with zeolite coating
(3 rd step). SiSiC ceramic monoliths were prepared by reactive Liquid Silicon Infiltration
(LSI) of carbon preforms from corrugated cardboard monoliths. The zeolite coating process
consisted of a hydrothermal treatment of the SiSiC carrier in an alkaline solution containing
the template and the Al-source, while the Si was provided from the ceramic substrate. MFI-
type (ZSM-5) zeolite crystals were directly grown on the ceramic supports via a partial Si
dissolution (from the SiSiC matrix) and zeolite crystallisation (support self-transformation).
Cellular SiSiC-zeolite monoliths possess bimodal (micro-/macro-) porosity and high
mechanical, chemical and thermal stabilities.
1. INTRODUCTION
glass substrates. Diverse techniques can be employed for the substrate functionalisation with
zeolite coatings: monolayer assembly through chemical linkers, dip- or slurry-coating, in-situ
hydrothermal synthesis, seeding techniques. More recently, a novel alternative technique,
called partial conversion/transformation of reactive supports, has been developed [4,6-9].
With this method, the support participates actively in the zeolite crystallisation process,
supplying part or all of the framework-builders for the zeolite synthesis [6-9]. In this work,
we present the complete process for the manufacturing of structured cellular SiSiC ceramic
supports via LSI (templating of corrugated cardboard monoliths) [6,10] and for the surface
functionalisation (via the support self-transformation technique) of these substrates with
MFI-type zeolite. Such a combined 3-step method has recently been presented for the
fabrication ofbiomorphic (Rattan-derived) SiSiC/zeolite composite monoliths [6].
2. EXPERIMENTAL
represents formally the fraction (- 45 wt %) of solid silicon in the ceramic SiSiC sample used
as reactive support for the zeolite crystallisation. No additional Si-source was used in the
solution. The hydrothermal synthesis was carried out in Teflon-lined stainless steel
autoclaves at 175 ~ (0-192 h for the kinetic studies of the zeolite crystal growth), under
static conditions and autogenous pressure. After the desired crystallisation time, the
specimens were removed from the synthesis mixtures, washed first in boiling water and then
in ethanol, dried overnight at 100 ~ and finally calcined at 600 ~ in air with a - 1.6 ~
i ramp and 10 h of dwelling time.
2.3. Characterisation
X-ray diffraction pattems (XRD) were taken on an X'pert Pro diffractometer (Philips
Analytical), using CuKcx radiation. The crystallinity was measured with the Qm method [7,8]:
(EDX) analyses were performed by using a Scanning Electron Microscope (Quanta 200, FEI)
equipped with an EDX detector INCA Energy 200 (Oxford Instruments Analytical Ltd).
Inductive Coupled Plasma (ICP-OES) analyses of the filtrate solution after synthesis were
performed on a Plasma 400 (Perkin Elmer).
Fig. 1. Digital picture of corrugated cardboard-derived cellular SiSiC monoliths: cardboard template
(left side) and SiSiC monolith (right side) after pyrolysis and reactive LSI of the carbon preform.
SiC
MFI zeolite
I i !
A II _~_~ 4h/ g
5 10 15 20 25 30 35 40 45 50
20 (o)
Fig. 2. The crystallisation kinetics of zeolite (0-192 h) on corrugated cardboard-derived cellular SiSiC
monoliths" XRD patterns (20=5-50 ~ of grinded and calcined SiSiC/zeolite composites.
The time dependency of zeolite crystallisation showed the typical S-type behaviour, with
incubation period o f - 24 h and maximum crystallinity o f - 45 % (Fig. 3).
0,5
0,4
0,3
m
0 0,2
0,1.
24 48 72 96 120144168192
Synthesis time (h)
Fig. 3. The crystallisation kinetics of zeolite (0-192 h) on corrugated cardboard-derived cellular SiSiC
monoliths: QAI(crystallinity) of calcined SiSiC/zeolite composites.
192 h. If compared to the crystallinity curve (Fig. 3) obtained from the XRD measurements
by applying the QAI method, it becomes clear that the weight of the samples increased again
because of the crystallisation start off between 24 and 36 h. In the early stages, the
predominant process was the dissolution of Si from the SiSiC substrate.
2O
ii- -ii
21 ......................--/----------W-e-i-ght-.ga!n- ....:1'
5 4\ , /
._~" 1"'\',, / Weightloss *
-10] ~ (
.15~ . . . . . . . . . . . . . . . .
0 24 48 72 96 120144168192
Synthesis time (h)
Fig. 4. The crystallisation kinetics of zeolite (0-192 h) on corrugated cardboard-derived cellular SiSiC
monoliths: weight change (%) of the SiSiC/zeolite composites after synthesis. The initial weight loss
is to be ascribed to the dissolution of Si from the SiSiC matrix. After a certain induction period the
zeolite crystallisation takes place and leads to an overall weight increase.
ICP-OES elemental analyses (Fig. 5) of the filtrate solutions after synthesis showed the
dissolution kinetics of Si and the Si consumption from the synthesis mixture during the
zeolite incubation and crystallisation periods. Negligible amount of A1 was observed already
after 6 h, indicating the possible complexation or adsorption of the A1 species on the substrate
surface. Almost no excess powder was found in the solution after synthesis. Most likely, the
process of support self-transformation leads to a zeolite crystallisation extremely Iocalised
and confined into a boundary layer around the surface of the substrate. Probably, this is due
to a gradient of species (Si) concentration in the proximity of the SiSiC carrier, caused by the
controlled dissolution process of Si from the ceramic substrate.
A 7 0,5
\__ 0,4
r 0,2
1 0, 1
6 / -._1 o • A, oo.o I
O,L' ~_. . , . . 0,0
0 2'4 4'8 72 120 .4 68192
Synthesis time (h)
Fig. 5. The crystallisation kinetics of zeolite (0-192 h) on cardboard-derived cellular SiSiC monoliths:
Si and A1 concentration in the filtrate (after synthesis) versus to the crystallinity (QA~).
150
100 0-12h
9. 24h
36h
48h
=96
94 96h
192h
93
Tel~Eaure (%)
Fig. 6. The crystallisation kinetics of zeolite (0-192 h) on cardboard-derived cellular SiSiC monoliths:
TGA analyses of the as-synthesised zeolite-ceramic composites.
192 h
A
50-
= 4oi
E
o
30.
'1o
,r 20 48 h
_..)
>o 10~ ~6,h
0-24 h
J.,
0
0,0 012 014 016 018 1,0
Relative pressure P/P0
Fig. 7. The crystallisation kinetics of zeolite (0-192 h) on cardboard-derived cellular SiSiC monoliths:
N2 adsorption-desorption isotherms.
151
Table 1
Properties of corrugated cardboard-derived cellular SiSiC/zeolite composites:
crystallisation kinetics (0-192 h) of zeolite
Fig. 8. SEM pictures of the calcined SiSiC/zeolite composite (after 96 h synthesis): (a) top-view of
the channels, (b and c) cross-section view along the channels.
152
4. C O N C L U S I O N S
We described a complete route (templating LSI for the preparation of cellular ceramic
monoliths and support self-transformation for the functionalisation of the substrate surface
with MFI-type zeolite) to fabricate structured SiSiC-zeolite composites possessing
hierarchical porosity, molecular sieving capabilities and exhibiting high thermo-mechanical
stability. Such materials are expected to find applications in areas such as catalysis
(monolithic reactors, catalytic filters and converters, microreactors) and
adsorption/separation.
ACKNOWLEDGEMENTS
We thank F. Mtiller, P. Widlok and E. Springer for the analytical measurements, A. Rudolph,
S. Kullmann and T. Fey for the material preparation and characterisation and R. Herrmann
for advice.
REFERENCES
The influences of the concentrations of Li+, Zn 2+ and organic cations as well as the type of
organic cations and liquid phase on the preparation of VPI-8 using vapour phase transport
method (VPT) were investigated. The typical sample was characterized by XRD, SEM, FT-IR
and Chemical analysis. There appeared to be an optimal reaction mixture composition for the
best crystallization intensity of VPI-8 using VPT. In addition, the selection of organic cations
is important to the crystallization of VPI-8.
1. INTRODUCTION
Molecular sieves have found applications as catalysts, adsorbents and ion exchanger [1].
Their microporous properties are based mainly on the structure of the tetrahedral framework
building from the TO4 tetrahedral and the possible variation of T atoms (A1, P, Ga, B, etc.).
The preparation of zeolite molecular sieves with substitutions of AI and/or Si by the other tri-
or tetravalent elements in the tetrahedral positions (T) is an active research. In addition to
synthesizing novel zeolites with new chemical compositions, the presence of heteroatoms in
the synthesis mixture plays an important role in the forming of new topologies. For examples,
zeolite ECR-34, which Ga substitutes for A1, is a novel microporous material with 18-rings
channels [2]; zeolite IM-12, a zeolite containing Ge and Si, has 12- and 14-rings channels[3].
However, the frameworks of molecular sieves containing the divalent heteroatoms are rare.
In 1991, the zincosilicate molecular sieves VPI-7 [4] and RUB-17[5] containing 3-ring units
were hydrothermally synthesized from the sol containing zinc. Since the network of alkali
metal cations and water molecular plays an important role in their poor thermal stabilization.
In 1993, the high-silica zincosilicate VPI-8 with good thermal stability was synthesized [6].
The effects of reaction components and mechanism of crystallization of VPI-8 were also
studied under hydrothermal conditions [7].
The vapor phase transport (VPT) method was first reported in 1990 as a new synthetic
technology attempting to reduce the vent of waste liquid and environment pollution. ZSM-5
[8] had been crystallized from an amorphous aluminosilicate gel in the vapor of
154
ethylenediamine, triethylamine and distilled water. The method has also been considered as a
very useful method to synthesis zeolitic membrane on the supports having complicated
structure like honeycombs. The membranes on a porous aluminum support and a
hydrogen-storage composite material were successfully synthesized using the VPT method
[9,10].
In this paper, the effects of the concentrations of Li +, organic cations and Zn 2+ as well as the
type of organic cations and liquid phase on the preparation of VPI-8 using VPT method were
investigated. An optimal reaction mixture composition for the best crystallization intensity of
VPI-8 was found. In addition, the presence of organic cations is necessary to the formation of
VPI-8. The samples were characterized by XRD, SEM and Chemical analysis.
2. EXPERIMENTS
2.1. Synthesis
The reagents used in our experiments were the following: zinc nitrate hexahydrate
(Zn(NO3)2"6H20, _>99wt%, Beijing Chemical Plant), fume silica (99%SIO2, A1203<10ppm,
Shenyang Chemical Plant), tetraethylammonium hydroxide (TEAOH, 25wt%, Beijing
XingFu Fine Chemical Plant), lithium hydroxide monohydrate (LiOH-H20, LiOH _>56.5wt%,
Beijing Chemical Plant), tetramethylammonium hydroxide, TMAOH (>_25wt% aq. solution,
China National Medicine Group Shanghai Corporation), tetraethylammonium bromide,
TEABr (_>98wt%, Beijing Xingfu Special Chemicals Research Department),
tetramethylammonium bromide, TMABr (_98wt%, Beijing Xingfu Special Chemicals
Research Department ), triethylamine (Et3N, 98%, Tianjin Chemical Plant), diethyleneamine
(EDA, 98.5%, Tianjin Chemical Plant), ammonia (NH4OH >__36wt%, Taiyuan Fertilize Plant),
acetone (_>99.5wt%, Beijing Chemical Plant) and distilled water.
In a typical run, firstly, a dry gel was prepared by combining 2g of Zn(NO3)2"6H20 and
3.67g of SiO2 with 10ml of H20 and drying at 150~ Secondly, the reaction mixture was
obtained by combining 12.47g of TEAOH and 1.1g of LiOH with the dry gel. The final molar
compositions of the mixture were as followings: 0.12 L i 2 0 : 0 . 1 ZnO: SIO2- 24.6
H20: 0.35 TEAOH. Additionally, 15 ml distilled water was added in the bottom of an
autoclave with 600 ml volume. The mixture gel was then transferred into autoclave. It was
sealed and its crystallization was carried out under vapor without stirring at 220~ for six
days. The solid product was collected by washing with distilled water and dried in air at
ll0~
2.2. Characterization
X-ray powder diffraction patterns were obtained on a Rigaku D/Max 2500 powder
diffractometer fitted with a fine-focus copper X-ray tube and the operation conditions were
the speed of scanning 20=1~ per minute and the voltage and electric current equal to 100V,
40mA, respectively. The relative crystallization of sample was calculated as following: the
characterized diffraction peaks of the typical sample at 20 = 6.77, 9.58, 20.08, 21.52, 22.29,
23.32, 24.58 and 26.18 were selected, the value of all these characterized peaks intensity were
summed, then the sample with the largest intensity value was considered as 100% relative
155
crystallization and standard sample; the intensity ratio of other sample to standard sample was
the relative crystallization. SEM photograph were recorded on a JEOL, JSM-35C scanning
electron microscope under the operation voltage of 25 KV. The IR spectra were recorded on a
Perkin-Elmer 1730 FT-IR spectrometer. The samples were prepared using the standard KBr
pellet method. The element composition of the product was also obtained by chemical
analyses. The content of silica was analyzed by the gravitational method and that of zinc was
measured by volume method. The IR spectra were recorded on a Perkin-Elmer 2000 FT-IR
spectrometer. The samples were prepared using the standard KBr pellet method.
Table 1
The influences of liquid phase on the synthesized product
Liquid phase H20 NH4OH EDA Et3N Acetone
Phase VPI-8 VPI-8 VPI-8 VPI-8 VPI-8
Relative crystallization 100% 59.3% 96.7% 98.1% 73%
was obtained in the case of TMAOH. But using TEABr or TMABr in the reaction mixture, an
amorphous solid could only be obtained. The presence of organic cations in the reaction
mixture is necessary to synthesis of VPI-8. However, structure-direction-agent used in organic
ammonium salt was disadvantaged to prepare molecular sieve VPI-8. The data were not
similar with the conditions which synthesized molecular sieve VPI-8 in hydrothermal system.
At using TEABr as structure-direction-agent, molecular sieve VPI-8 may be given in
hydrothermal system [11]. Certainly, to synthesize molecular sieve VPI-8 in vapour phase
system was more difficulty than to do this in hydrothermal system.
Table 2
The influences of organic additives on the product
Organic ammonium type TEAOH TEABr TMAOH TMABr No
Phase VPI-8 amorphous VPI-8+Q* amorphous amorphous
Relative crystallization 100% / 63% / /
*: Q is a quartz.
Next, the effect of changing gel compositions was studied by choosing zinc nitrate
hexahydrate, fume silica, tetraethylammonium hydroxide (TEAOH), lithium hydroxide
monohydrate (LiOH'H20) and distilled water as the raw chemicals. The experiments were
crystallization time equal to 6 days and crystallization temperature equal to 220~ Table 3
was the typical data.
Table 3
The influences of gel compositions (mol ratio) on the product
TEAOH Li20 ZnO SiO2 Phase Relative
crystallization
0.35 0.05 0.10 1.0 Amorphous /
0.35 0.08 0.10 1.0 VPI-8 70%
0.35 0.12 0.10 1.0 VPI-8 100%
0.35 0.18 0.10 1.0 VPI-8 92%
0.35 0.25 0.10 1.0 VPI-8+Quartz /
0.35 0.35 0.10 1.0 Quartz /
0.35 0.12 0.01 1.0 vPI-8+Quartz /
0.35 0.12 0.02 1.0 vPI-8 86%
0.35 0.12 0.05 1.0 vPI-8 100%
0.00 0.12 0.10 1.0 Amorphous /
0.20 0.12 0.10 1.0 VPI-8 66.8%
0.60 0.12 0.10 1.0 VPI-8 100%
157
3.2. Characterization
3.2.1. X-ray powder diffraction
Fig. 1 showed the representative XRD pattern of the product. The diffraction peak positions
and relative intensity of the XRD pattern of the sample (Fig.l) were corresponded to the
results of literature [7] showed molecular sieve VPI-8. It may conclude that the as-synthesized
product was the VPI-8 molecular sieve.
1400
1200
1000
800
600
_=
400
200
l . ~t ~ ~ i7--~-,~---i --( . . . .
5 10 15 20 25 30
2 Theta (o)
3.2.2. Morphology
SEM method was used to observe the morphology of typical sample. Fig.2 showed the
SEM photo of as-synthesizes sample. Obviously, the crystals were irregularly plate-like
material. The crystals size of molecular sieve VPI-8 obtained from VPT method was smaller
and more irregular than that from hydrothermal system [ 11].
Table 4
The chemical analysis result of the product synthesized by different method
Zn/Si tool ratio Zn/Si tool ratio in product synthesized by different method
in start gel VPT hydrothermal method
0.05 0.048 0.041
0.1 0.093 0.091
0.2 0.017 0.015
159
r
8O
Q;
60
e~ VPI-8
E 40
~ 2o
0 , I ~ I , I , I , I ~ I , I
4. CONCLUSION
Zincosilicate molecular sieve VPI-8 was successfully synthesized using VPT method in the
vapor of different liquid phase. There appears an optimal composition for reaction mixture to
obtain VPI-8 with the best crystallization intensity. The suitable gel molar compositions for
synthesizing molecular sieve VPI-8 were as follows: 0.12 Li20:0.1 ZnO: SiO2:24.6 H20:
0.35 TEAOH and the effective vapour was water vapour. The organic structure-direct-agent
played an important role in the crystallization of VPI-8. The Zn/Si molar ratio of VPI-8
framework changes regularly with that of reaction mixture.
ACKNOWLEDGEMENTS
The authors are grateful to the National Science Foundation of China (Grants 20376052) for
financial support.
160
REFERENCE
Fly ash was converted to zeolite by a hydrothermal method using NaOH and KOH solutions
under various conditions. The fly ash and derived zeolites were tested for removal of dyes
from wastewater. It is found that basic dyes can be effectively adsorbed on fly ash and the
modified forms. Modification of fly ash with basic solutions can significantly enhance the
adsorption capacity depending on the treatment temperature and base concentration, not the
type of bases. The adsorption capacity increases with the increasing adsorption temperature
and pH of the solution. The sorption capacities of FA-NaOH can reach 2 x 104 and 3 x 104
mol/g for crystal violet and methylene blue, respectively. The adsorption isotherm can be
described by a two-site Langmuir isotherm equation.
1. INTRODUCTION
Considering both volume discharged and effluent composition, the wastewater generated by
the textile industry is rated as the most polluting among all industrial sectors. This challenge
has prompted intensive research in new advanced treatment technologies, some of which are
currently making their way to full-scale installations. These comprise polishing treatments
such as filtration, chemical oxidation and specialised flocculation techniques and pre-
treatment steps including anaerobic digestion, fixed-film bioreactors, Fenton's reagent
oxidation, electrolysis, or foam flotation [ 1]. Despite the fact that effective decolorisation has
often been shown to be feasible via sorption onto activated carbon, full-scale use of this
technology has apparently not been reported yet, a likely consequence of high capital and
regeneration costs of the activated carbon.
At present, there is a growing interest in using low cost materials for adsorption of dyes.
In .the past a few years, several investigations have been reported using industrial or
agricultural wastes as adsorbents and the data show that those wastes exhibit more or less
adsorption capacity [2-7]. Fly ash is a by-product generated during the combustion of coal in
the electricity generation process. Disposal of fly ash has become an increasing economic and
environmental burden. In the past decade, efforts are being made to explore other applications
such as transformation into zeolite, potassium fertiliser and flue gas absorbent [8], in addition
to civil engineering for cement production. The production of zeolites using coal fly ash as a
resource constitutes one important issue of waste management and many techniques have
been investigated [9]. We have investigated the adsorption of dyes using fly ash as a
adsorbent and found that fly ash could be a cheaper adsorbent while still having higher
adsorption capacity [ 10]. Conversion of fly ash into zeolite will make porous materials with
162
higher capacity for adsorption. In this paper, we report our investigation on hydrothermal
conversion of fly ash into zeolite and their adsorption capacity of dyes in aqueous solution.
2. E X P E R I M E N T A L
Table 1
Propertie s of fly ash adsorbents.
Sample SBET(m2/g) pH Phases
FA 15.6 4.4 quartz, mullite, with minor hematite and magnetite
FA-NaOH-RT 20.2 8.8 quartz, mullite, with minor hematite and magnetite
FA-NaOH-60 35.0 9.9 quartz, mullite, with minor hematite and magnetite
FA-NaOH-100 52.4 ).3
10. quartz, mullite, P1 zeolite
seen that the surface area will increase with the increasing temperature when fly ash is treated
with NaOH. The pH of solid slurry is also increased with the treatment temperature, due to
the changes of surface functional groups.
2.0e-5
FA-NaOH-100
30x103 1.5e-5 (~ o o A
.,,-.,
:5 FA-NaOH-60
03 3
*.-.- L | ,
: ~ 2 0 x l 03 E1.0e-5 /k
._
if)
O
6e-5
2"5E-05 ]1 I
5e-5 ,.. OMB
o 2.05-05 -1]IIcv ]
4e-5
o
E 3e-5
0 2e-5
O
E!
1M NaOH
3M NaOH
,
1.5E-05
1.0E-05 L ,
le-5
Z~ 5M NaOH
~ 5.o5-o6
Two-site Langmuir model
0.0E+00
0 le-5 2e-5 3e-5 4e-5 5e-5
FA FA-RT FA-60 FA-100
Ce (M)
Fig. 3. Effect of NaOH concentration on FA Fig. 4. Adsorption o f M B and C V on various fly
adsorption capacity of methylene blue. ash samples at 30 ~
adsorption capacity for methylene blue than crystal violet, which is due to the dye chemical
structure. The molecular size of crystal violet is much larger than methylene blue prohibited it
penetrating into the small pores of fly ash.
2.0e-5
.~ 2.0e-5
o Crystal violet
"~ 1.8e-5 [] Methylene blue
E 1.5e-5 O
v 1.6e-5
"o
.~
I,_ 1.4e-5
O ~1.0e-5
"(3 1.2e-5
0 0 Experiment
"E 1.0e-5 I One-site Langmuir model
o 8.0e-6 5.0e-6 | --- R-P model
E I . . . . Two-site Langmuir model
< 6.0e-6 . . . . . Freudlich model
2 4 6 8 10 0.0
0.0 5.0e-6 1.0e-5 1.5e-5 2.0e-5
pH
Ce (M)
Fig. 5. Effect of solution pH on adsorption. Fig. 6. Comparison of adsorption isotherms
165
3. 2. 5. Adsorption i s o t h e r m s
The equilibrium adsorption isotherm is of importance in the design of adsorption systems.
For solid-liquid systems, several isotherms equations are available. In this work, we select
three important isotherms, the Langmuir, Freundlich, and Redlich-Peterson isotherms.
The Langmuir isotherm takes an assumption that the adsorption occurs at specific
homogeneous sites within the adsorbent. The generalised Langmuir isotherm can be written in
the form.
n
aibiCe
Qe = ~ (1)
,:1 1 + biG e
Qe is the adsorbed amount of the dye, Ce is the equilibrium concentration of the dye in
solution, ai and bi are the parameters on the Langmuir isotherm. In most cases, the single- or
two-site models give a satisfactory fit to experimental data.
The Freundlich isotherm is an empirical equation employed to describe heterogeneous
systems. The Freundlich equation is expressed as
where K and n are Freundlich adsorption isotherm constants, being indicative of the extent of
the adsorption and the degree of non-linearity between solution concentration and adsorption,
respectively.
The Redlich-Peterson isotherm model combines the elements from both the Langmuir
and Freundlich equations, and the mechanism of adsorption is a hybrid one and does not
follow ideal monolayer adsorption. The Redlich-Peterson equation is widely used as a
compromise between Langmuir and Freundlich systems. It is expressed as
Qe = ~ K C e (3)
1 + aCe p
K and ot are the Redlich-Peterson isotherm constants and [3 is the exponent, which lies
between 1 and 0.
Table 2
Comparison of adsorption isotherm models.
Model Parameters R2
Fig. 6 shows a comparison of adsorption isotherms for curve fitting of the experimental
results with above three adsorption isotherms. The model parameters from all isotherms
obtained from nonlinear regression are presented in Table 2. As seen that the Redlich-
Peterson model is better than the one-site Langmuir isotherm and Freudlich model in
simulation of the adsorption isotherm and the two-site Langmuir model will be the best (as
evident from correlation coefficients). This suggests that some heterogeneity in the surface or
pores of the fly ash will play a significant role in dye adsorption.
3.0e-5
2.5e-5
~ A u 0
0 ~ Jd"' 2.5e-5
~,2.0e-5
o
E 1.5e-5
--•2.0e-5
E
"~'1.5e-5
o 1.0e-5
0
O FA 1.0e-5 %
[] FA-NaOH-RT
5.0e-6 zx FA-NaOH-60
5.0e-6
O Fa-NaOH-100
Two-site Langmuir model
0.0 {~- 0.0
0.0 1.0e-5 2.0e-5 3.0e-5 4.0e-5 5.0e-5 0 1e-5 2e-5 3e-5
Ce (M) Ce (M)
(A)
2.5e-5
O A
1.5e-5 2.0e-5
1.5e-5
o 1.0e-5 5
E E
(o (~ 1.0e-5
o O FA
5.0e-6 [3 FA-NaOH-RT
zx FA-NaOH-60 5.0e-6
O Fa-NaOH-100
Two-site Langmuir model
0.0 ~ o.o ~ . . . .
0.0 5.0e-6 1.0e-5 1.5e-5 2.0e-5 0.0 5.0e-6 1.0e-5 1.5e-5 2.0e-5
Ce (M)
Ce (M) (B)
Fig. 7. Adsorption isotherms of MB and CV on fly ash at 30 and 40 ~ (A) methylene blue (B) crystal
violet
167
Table 3
Thermodynamic Parameters for the Adsorption of methylene blue and crystal violet on fly ash
samples. . ...........
Sample Dye AG~ (kJ/mol) AH~ (kJ/mol) AS~ (J/mol K)
30 ~ 40 ~
FA -38.8 -40.1 0 132.3
FA-NaOH-RT MB -33.4 -36.6 -63.0 -93.8
FA-NaOH-60 -33.2 -38.1 -114.9 -265.8
FA-NaOH- 100 -37.4 -35.1 107.6 482.3
FA -32.8 -38.1 -129.0 -551.4
FA-NaOH-RT CV -34.1 -36.6 -61.3 -87.1
FA-NaOH-60 -33.5 -29.4 158.3 636.4
FA-NaOH- 100 -34.3 -35.1 10.9 153.0
which is presented in Fig. 7. One can see that the adsorption shows a complex behaviour with
the adsorption temperature. In some cases, the adsorption increases as the temperature is
increasing, but in other cases, the adsorption decreases with the increasing temperature.
Thermodynamic parameters, i.e., free energy (AG~ enthalpy (AH~ and entropy (AS~
changes were also calculated using Eqs. 4-6 [7] and are given in Table 3.
AG O= - R T l n K 1 (4)
( T2TI ) K2
AH ~ --R T2 ,]el lnKl (5)
AH ~ - A G O
AS ~ = (6)
T
where K1 and Kz are the Langmuir constants at TI = 30 and T2 = 40 ~ The negative values of
AG~ indicate the feasibility and spontaneous nature of methylene blue and crystal violet
adsorption on fly ash. The change in enthalpy (AH~ for FA-NaOH-100 was found to be
positive (Table 3). The positive values confirm the endothermic nature of adsorption. The
positive values of the entropy change show the increased randomness at the solid/solution
interface with some structural changes in the adsorbate and adsorbent and an affinity of the
adsorbent toward methylene blue [7]. For FA-NaOH-RT, adsorption enthalpy of methylene
blue and crystal violet show negative value, indicating the exothermic characteristic. Other
two adsorbents exhibit a mixed thermodynamic behaviour.
4. CONCLUSION
Hydrothermal treatment of fly ash by basic solution will change fly ash into effective
adsorbents for dye removal from aqueous solution. High temperature treatment will result in
the increase of surface area and formation of zeolite (NaP 1). These adsorbents will generally
exhibit higher adsorption capacity but change with pH and temperature. A two-site Langmuir
isotherm equation can best fit the experiments.
168
REFERENCES
[1] P.C. Vandevivere, R. Bianchi, W. Verstraete, J. Chem. Technol. Biot. 72 (1998) 289.
[2] Z. A1-Qodah, Water. Res. 34 (2000) 4295.
[3] G. Annadurai, R.S. Juang, D.J. Lee, J. Hazard. Mater. 92 (2002) 263.
[4] V.K. Gupta, D. Mohan, S. Sharma, M. Sharma, Separ. Sci. Technol. 35 (2000) 2097.
[5] Y.S. Ho, G. McKay, Chem. Eng. J 70 (1998) 115.
[6] P. Janos, H. Buchtova, M. Ryznarova, Water. Res. 37 (2003) 4938.
[7] D. Mohan, K.P. Singh, G. Singh, K. Kumar, Ind. Eng. Chem. Res 41. (2002) 3688.
is] R. Kikuchi, Resources, Conserv. Recycl.. 29(1999) 333.
[9] X. Querol, N. Moreno, J.C. Umana, A. Alastuey, E. Hemandez, A. Lopez-Soler, F. Plana,
Intern J. Coal Geol. 50 (2002) 413.
[lo] S. Wang, Y. Boyjoo, A. Choueib, J. Zhu, Water. Res. 39 (2005) 129.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka, N. Zilkov~and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 169
Laboratoire des Mat6riaux, Surfaces et Proc6d6s pour la Catalyse (LMSPC, UMR 7515),
ECPM-ULP, 25, rue Becquerel, F-67087 Strasbourg Cedex 2, France.
E-mail: lacroixm@ecpm.u-strasbg.fr. Part of the ELCASS (European Laboratory of Catalysis
and Surface Science)
Y zeolite has been successfully synthesised inside multi-walled carbon nanotubes (MWNT)
via a two-step procedure. Zeolite nanocrystals are structured in a nanowire 1D-filamentous
morphology. This led to a peculiar crystallographic arrangement, thus to a anomalous
behaviour in a typical Friedel-Crafts reaction.
1. INTRODUCTION
From an industrial point of view, zeolites are important in many processes, ranging from
adsorption, ion-exchange to catalysis. With the knowledge and experience gained from
petrochemical applications, rich possibilities have in turn become available for the intensional
design of zeolites as catalysts molecularly and microscopically tailored to fuit a desired
application [1]. Over recent decades, there has been a growing interest by making zeolite
crystals grow directly on macroscopically pre-shaped supports [2,3] to gain hydrodynamic
advantages. It is then evident that a fine-tuning of the size and also the shape of zeolite
crystals appears crucial for the activity and selectivity of many catalytic reactions [4]. The
aim of this work was to design FAU-type zeolites at the microscopic scale: via modification
of the shape of the crystals by using multi-walled carbon nanotubes (MWNT) as exo-template
(inner Q = 80 nm). One can hope that the filamentous morphology of the Y-zeolite catalyst
can heve an influence on the distribution of the products in a given catalytic reaction.
2. EXPERIMENTAL
groups on the external surface of the MWNT. The zeolite synthesis was done following a
two-step procedure. First a seeding gel was prepared in a PTFE sealed flask and then
vigorously stirred for 24h. The constituants concentration of the gel (mole) were as follow :
NaOH : NaA102 : Na2SiO3 : H20 = 4 : 1 : 3,1 : 43,2. Afterwards, the gel was diluted in
distilled water to disperse the seeds in a volume corresponding to the volume adsorbable by
the carbon nanotubes (about 6 mL of liquid per gram of MWNT). The diluted seeding gel was
then added to the nanotubes under vigorous stirring until complete absorption. The solid was
then smoothly dried for 2h in a oven at 100~ to remove the water and keep the seeds inside
the nanotubes. Secondly, the synthesis gel was prepared according to Bell et al. [6] with mole
ratios as follow : NaOH : NaA102 : NazSiO3 : H 2 0 = 4 : 1 : 3,1 : 43,2. The gel was added to
the dried carbon nanotubes containing the seeds of Y zeolite. The synthesis was performed at
90~ for 24h in a closed PTFE sealed autoclave. After drying, the zeolite was exchanged
three-times at 80~ with a 1M NHaC1 solution and calcined at 550~ for 1Oh in order to
remove ammonia to get the acidic H-form.
2.2. Characterisations
The composite zeolite / MWNT was characterised by XRD using a Siemens D-5000
diffractometer equipped with a copper cathode (Z=1.5406 A). The specific surface areas and
the porosity of the samples were measured by nitrogen adsorption-desorption at 77K with a
Coulter-Beckman SA 3100 sorptometer. The SEM images were recorded on a JEOL F-6700
FEG working under 3 kV. High-resolution transmission electron microscopy (HRTEM)
measurements were performed with a TOPCON 002B microscope working under 200 kV.
The same zeolite was also synthesised without the carbon nanotubes to compare the structures
and the catalytic properties of both zeolites.
3. R E S U L T S AND DISCUSSION
I -"--Y / MWNT
10 15 20 25 30 35 40 45
20 (~
Table 1
Conversion after 120 h (C120h) and selectivity (S) towards the ortho and para isomers
Bulk Y 60% 85 % 15 %
4. CONCLUSION
REFERENCES
1. INTRODUCTION
oxidation catalyst [10]. The H20 affinity of Ti-Beta zeolites changes significantly
depending on the preparation methods and their hydrophobic-hydrophilic properties can
modify not only the catalytic properties but also the photochemical process in the zeolite
pores.
In this study, we have used the two types of Ti-Beta zeolites synthesized under
different conditions as photocatalysts for the liquid phase degradation of 2-propanol diluted
in water to form CO 2 with H20 and the effects of the hydrophobic-hydrophilic properties of
zeolites on the photocatalytic reactivity have been investigated.
2. EXPERIMENTAL
2.3. Characterization
The diffuse reflectance absorption spectra were recorded with a Shimadzu
UV-2200A spectrometer at 295 K. The photoluminescence spectra were measured at 77 K
using a Shimadzu RF-5000 spectrophotofluorometer. The XAFS spectra (XANES and
EXAFS) were measured at the BL-9A facility [11] of the Photon Factory at the National
Laboratory for High-Energy Physics, Tsukuba. A Si(lll) double crystal was used to
monochromatize the X-rays from the 2.5 GeV electron storage ring. The Ti K-edge
absorption spectra were recorded in the fluorescence mode at 295 K. In a typical
experiment, 50 mg of the sample pressed in to a disc was loaded into the in situ cell having
the Kapton windows. The normalized spectra were obtained by a procedure described in
previous literature [12] and Fourier transformation was performed on k3-weighted EXAFS
oscillations in the range of 3-10 ~-1. The preedge peaks in the XANES regions were
normalized for atomic absorption, based on the average absorption coefficient of the spectral
region from 5050-5200 eV [13,14]. The curve-fitting of the EXAFS data was carried out by
employing the iterative nonlinear least-squares method and the empirical backscattering
parameter sets extracted from the shell features of titanium compounds.
From the results obtained by the curve-fitting analysis of the EXAFS spectra, it was found
that the Ti-Beta zeolites consist of 4-coordinated titanium ions with an atomic distance of
1.84 A for Ti-Beta(OH) and 1.83 A for Ti-Beta(F) as shown in Fig. 1. These XANES and
FT-EXAFS investigations indicate that the dispersion level of the titanium oxide species in
the present Ti-Beta zeolites is high and these two zeolites have the same coordination
geometry as the tetrahedral titanium oxide species included within the zeolite framework.
15 t
/ /I CN:3.9 I :5 I ~A,a ~ Ti-Beta(OH)
1 g ..............
pre~dge "~ - T i- B eta (F )
8 :~ 5-coo
01 , i
The change in the coordination geometry of Ti atom reflects very sensitively on the
intensity and position ofpreedge peak in XANES region at Ti K-edge [13,14]. Fig. 3 shows
the, XANES spectra of Ti-Beta(OH) and Ti-Beta(F) zeolites and the effect of the addition of
H20 molecules on their preedge peaks. The intensity and position of the preedge peaks of
zeolites under vacuum condition clearly indicate the presence of the tetrahedrally coordinated
titanium oxide species in these zeolites under vacuum condition. As shown in Fig. 2, the
addition of H20 molecules onto the Ti-Beta zeolites leads to the efficient decrease in the peak
intensity and the shift to the higher energy in the peak position, its extent depending on the
amount of added H20. Such efficient changes in the peak intensity and position with the
H20 addition suggest not only that tetrahedrally coordinated titanium oxide species locate at
positions accessible to the added H20 molecules but also that added H20 molecules interact
179
directly with the titanium oxide species changing its coordination geometry. The change in
the intensity and position of preedge with the addition of H20 molecules also shown in Fig. 2.
From the changes of these values with the H20 addition, it is clearly found that the
coordination number of titanium oxide species increases from its original 4-coordination to
5-coordination and finally to 6-coordination. Although, the peak intensity decreases and the
peak position shifts to the higher energy by the addition of H20 molecules on Ti-Beta(F), as
shown in Fig. 2, these values are not included in the region for 6-coordinated standard
titanium compounds reported by Sankar et al. This deviation indicates that strength of the
additional bondings between Ti atom and added H20 molecules to form 6-coordination in
Ti-Beta(F) is weak in comparison with the bondings between Ti atom and four O atoms of
zeolite framework in the original tetrahedrally coordinated unit.
As shown in Figs. 2 and 3, the changes in the peak intensity and position with the
H20 addition are more remarkable on Ti-Beta(OH) than Ti-Beta(F). This difference
indicates that the opportunity to have the interaction between the added H20 molecules and
titanium oxide species in the zeolite framework is the higher in the pore of hydrophilic
Ti-Beta(OH) than the hydrophobic Ti-Beta(F).
a) Preedge
Ti-Beta(OH)
F~ Ti-Beta(F)
o 4965 4970 4975
A21
o~ 4
"O b)
._N
m
E
E
c) 3
z
"5 2
<
Q
Energy / eV
Fig. 3. The effect of the addition of H20 Fig. 4. Adsorption isotherms for H20
molecules on the preedge peak intensity of molecules at 298 K on the Ti-Beta(OH)
the XANES spectra. (a) Ti-Beta(OH) and and Ti-Beta(F).
(b) Ti-Beta(F). The amount of added H20
molecules; 0, 1.4, 3.0, 4.6 mmol/g-cat (from
top to bottom).
180
Because Ti-beta(OH) and Ti-Beta(F) zeolites have the same pore structural geometry, the
present results clearly indicate that the hydrophobic-hydrophilic properties of zeolite affect
on the accessibility and the interaction between photocatalytic active sites (tetrahedrally
coordinated titanium oxide species) and reactant gasses (H20 molecules).
The H20 affinity of Ti-Beta zeolites can be evaluated by monitoring the H20
adsorption isotherms. Fig. 4 shows the H20 adsorption isotherms on the Ti-beta zeolites.
These results indicate that the Ti-Beta(OH) and Ti-Beta(F) zeolites synthesized using a
typical OH- ion and F- ion as the anion of the SDA exhibit hydrophilic and hydrophobic
properties, respectively. The large molecule of SDA, N'-dibenzyl-4,4'-trimethylenebis
(N-methylpiperidinium) dihydroxide, for the synthesis of Ti-Beta(OH) can be removed only
by calcination leaving many defect sites in the zeolite pore [9]. On the other hand, it has
been reported that Ti-Beta free of detect sites can be prepared by using TEAF as SDA [ 10].
This difference in the numbers of defect sites between these Ti-Beta zeolites probably
generates the hydrophilic and hydrophobic properties, respectively.
OH O
i H20 II
CH3-CH-CH3 ~ CH3-C-CH3 ~ CO2 + H20
O2
I
m Adsorption
100[ i ~ 100 o~ "7. 0.4
-r- 80 80 O0
" [i;
O PrOH
1 7=
=
ooF ,,f c> 1 oo- o 0.2
increased as the intermediate, finally 2-propanol and acetone were degradated into CO2 and
H20. Fig. 6 shows the adsorption properties and photocatalytic reactivities of Ti-Beta
zeolites and TiO2 powder (P-25) photocatalysts for the degradation of 2-propanol in water.
Ti-Beta zeolites indicate the higher ability for 2-propanol adsorption and higher efficiency for
the photocatalytic degradation than commercial TiO2 powder (P-25) photocatalyst. Among
the Ti-Beta zeolites, the amount of adsorbed 2-propanol on Ti-Beta(F) is higher than that of
Ti-Beta (OH), indicating that efficient adsorption of organic compounds in the aqueous
solutions can be realized by fluoride-modified hydrophobic zeolite. Because the Ti-Beta(F)
exhibited the photocatalytic reactivity higher than Ti-Beta(OH), it can be also found that the
more highly adsorption, the more efficiently photocatalytic reactivities were promoted. The
hydrophobic surface of Ti-Beta(F) zeolite was suitable for the adsorption of organic
compounds and the larger amount of adsorbing organic compounds on Ti-Beta(F) zeolite led
to the higher reactivity for photocatalytic degradation.
4. CONCLUSION
Two kinds of Ti-Beta zeolites which can exhibit hydrophilic and hydrophobic properties
respectively has been synthesized using a typical OH- ion and F- ion as the anion of the
structure-directing agent. Theses Ti-containing zeolites are good candidates as efficient
photocatalysts. Especially, Ti-Beta(F) exhibited a higher reactivity for the degradation of
2-propanol as compared to Ti-Beta(OH) due to the higher affinity of the 2-propanol
molecules for adsorption in aqueous solutions, which was induced by the fluoride-modified
hydrophobic surface. The present results indicate that the combination of the porous
adsorbents with fluoride-modified hydrophobic surface and titanium oxide photocatalyst is a
good candidate as the efficient system for the degradation of organic pollutants diluted in
water.
ACKNOWLEDGMENT
This work was supported by the Grant-in-Aid Scientific Research from the Ministry of
Education, Science, Culture, and Sports of Japan (Grants 16656251). This work is also partly
performed under the project of collaborative research at the Joining and Welding Research
Institute (JWRI) of Osaka University with helpful advice from Prof. H. Fujii. The X-ray
adsorption experiments were performed at the Photon Factory of KEK (2003G251) with
helpful advice from Prof. M. Nomura.
REFERENCES
[ 1] C. Minero, G. Mariella, V. Maurino, and E. Pelizzetti, Langmuir, 16 (2000) 8964.
[2] H. Yamashita, S. Kawasaki, Y. Ichihashi, M. Harada, M. Anpo, G. Stewart, M.A. Fox, C. Louis
and M. Che, J. Phys. Chem. B, 102 (1998) 5870.
182
[3] H. Yamashita, M. Honda, M. Harada, Y. Ichihashi, M. Anpo and Y. Hatano, J. Phys. Chem. B,
102 (1998) 10707.
[4] H. Yamashita, Y. Ichihashi, M. Harada, G. Stewart, M.A. Fox and M. Anpo, J. Cata|., 158
(1996) 97.
[5] H. Yamashita, Y. Ichihashi, M. Anpo, C. Louis and M. Che, J. Phys. Chem., 100 (1996) 16041.
[6] H.Yamashita and M. Anpo, Curr. Opin. Solid St. Mater. Sci., 7 (2004) 471.
[7] B. Notari, Adv. Catal., 41 (1996) 253.
[8] T. Tatsumi and N. Jappar, J. Phys. Chem., B, 102 (1998) 7126.
[9] T. Takewaki, S. J. Hwang, H. Yamashita and M.E. Davis, Microporous Mesoporous Mater. 32
(1999) 265.
[10] M.A. Camblor, A. Corma, P. Esteve, M. Martines and S. Valencia, Chem. Commun., (1997)
795.
[ 11] M. Nomura and A. Koyama, J. Synchrotron Radiat., 6 (1999) 182.
[12] H. Yamashita, M. Matsuoka, K. Tsuji, Y. Shioya and M. Anpo., J. Phys. Chem., 100(1996) 397.
[ 13] J.M. Thomas and G. Sankar, Acc. Chem. Res., 34 (2001) 571.
[14] G. Sankar and J. M. Thomas, Top. Catal., 8 (1999) 1.
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 183
1. INTRODUCTION
There are still strong driving forces for developing new zeolites with peculiar structural
features, because these microporous solids find main uses as catalysts, adsorbents, and ion
exchangers in a large number of technologically important chemical processes. Over the last
several decades, in fact, numerous zeolite structures have been synthesized hydrothermally in
the presence of the inorganic and/or organic additives commonly known as structure-directing
agents (SDAs) [ 1].
Recently, we have been involved in one research program to investigate the structure-
directing ability of linear diquarternary alkylammonium ions with aliphatic and/or cyclic
moieties in zeolite syntheses. A series of results from our synthetic program has revealed that
the phase selectivity of the crystallization in the presence of these organic species is sensitive
not only to the length of the polymethylene bridging unit, but also the nature of the groups on
the ammonium ion [2-10]. For example, we found that the (C2Hs)3N+(CH2)sN+(C2Hs)3 ion
(N,N'-bis-triethylpentanediyldiammonium, Et6-diquat-5) can direct the synthesis of at least
four different zeolite structures (i.e., P1 (GIS), SSZ-16 (AFX), SUZ-4, and ZSM-57 (MFS)),
depending on the oxide composition of synthesis mixtures [2-5]. When ZSM-5 (MFI) and
mordenite (MOR) that are known to crystallize without the aid of organic SDAs are not
considered as a result of structure direction exerted by the ammonium cation employed,
however, neither of the zeolites prepared with Et6-diquat-5 possesses the same framework
topology as that of any of the materials that can be synthesized in the presence of a slightly
shorter (CH3)3N+(CH2)5N+(CH3)3 ion (N,N'-bis-trimethylpentanediyldiammonium, Me6-
diquat-5) as an organic SDA [6,7,11 ].
184
2. E X P E R I M E N T A L
2.1. Synthesis
The Me4Etz-diquat-4 and Et4Me2-diquat-4 ions were prepared by refluxing 1,4-
dibromobutane (99%, Aldrich) with an excess of asymmetric N,N-dimethylethylamine (99%,
Aldrich) and N,N-diethylmethylamine (98%, Aldrich) in tetrahydrofuran (THF) as a solvent
overnight, respectively. The excess amine was removed by extraction with THF and
recrystallizations were performed in methanol-diethyl ether mixtures. The other two sets of
divalent cations (Me4Et2-diquat-n and Et4Me2-diquat-n with n - 5 and 6) were synthesized
using the corresponding dibromoalkanes instead of 1,4-dibromopropane, respectively, with
procedures similar to the preparation of Me4Etz-diquat-4 or Et4Me2-diquat-4. After
purification, the formation and purity of all these diquaternary ammonium salts were
confirmed by lH and 13CNMR.
The reagents used for zeolite syntheses included NaOH (50% aqueous solution, Aldrich),
diquaternary cations prepared here, Al(NO3)3-9H20 (98%, Junsei), fumed silica (Aerosil 200,
Degussa), and deionized water. The final composition of the synthesis mixture was
4.5R'xNa20"yA1203"30SiO2.1200H20, where R is the organic SDA, x is varied between 5.0 <
x < 17.0, and y is varied between 0 < y < 1.0. When necessary, NaOH was replaced by the
equivalent amount of KOH (45% aqueous solution, Aldrich). After being stirred at room
temperature for 24 h, the synthesis mixture was charged into Teflon-lined 23-mL autoclaves
and heated to 160 ~ under rotation (100 rpm) for 5-14 days. The solid products were
recovered by filtration, washed repeatedly with water, and then dried overnight at room
temperature.
2.2. Characterization
The phase purity and crystallinity were determined by powder X-ray diffraction (XRD)
using a Rigaku Rigaku 2500H diffractometer with Cu Ks radiation. The C, H, and N contents
of selected samples were analyzed by using a Carlo Erba 1106 elemental organic analyzer.
Thermogravimetric analyses (TGA) were performed in air on a TA Instruments SDT 2960
thermal analyzer, where the exothermic weight losses related to the combustion of organic
SDA were further confirmed by differential thermal analyses (DTA) using the same analyzer.
Crystal morphology and size were determined by a JEOL JSM-6300 scanning electron
microscope (SEM).
The ~H-13C CP MAS NMR spectra at a spinning rate of 4.5 kHz were measured on a
Bruker Avance 500 spectrometer at a 13C frequency of 125.767 MHz with a rt/2 rad pulse
185
length of 5.0 Its, a contact time of 1 ms, and a recycle delay of 5 s. Approximately, 4000 pulse
transients were accumulated. The ~3C chemical shifts are referenced to TMS. The Raman
spectra in the 200-3200 crn-~ region were recorded on a Bruker RFA 106/S FT-Raman
spectrometer equipped with an Nd:YAG laser operating at 1064 nm. The samples were
exposed to a laser power of 60-80 mW at the spectral resolution of 4 cm -~. Typically, 1024
scans were accumulated for obtaining the Raman spectra.
Due to their high flexibility, linear diquaternary ammonium cations can undergo a large
number of conformations at zeolite crystallization conditions, depending on the nature and
extent of interactions with the surrounding inorganic species in synthesis mixtures. This
implies that the role of inorganic variables acting as phase selectivity factors may be more
important in syntheses using flexible, charged organic molecules than in those involving
relatively large polycyclic SDAs with a high degree of rigidity. We have started our zeolite
syntheses by explore the influence of the A1 content in synthesis mixtures on the phase
selectivity of the crystallization in the presence of each of Me4Etz-diquat-n and Et4Me2-
diquat-n ions with n = 4-6. A summary of products from a set of synthesis experiments in
which the SIO2/A1203 ratio in the gel is varied from 30 to oowhile fixing the NaOH/SiO2 ratio
to 0.33 is given in Table 1. In each case, the zeolitic phases listed were the only ones obtained
in repeated trials. When the initial SIO2/A1203 ratio in the synthesis mixture is in the range 30-
120, many of the organic SDAs studied tend to stabilize the one-dimensional, 12-ring zeolite
ZSM-12 (MWW). As seen in Table 1, however, some degree of diversity in the zeolitic
product is observed from syntheses carried out in Al-free media. For example, ZSM-48, a
one-dimensional 10-ring zeolite, is the product obtained from a pure-silica gel in the presence
of either of three Me4Etz-diquat-n ions with n = 4-6.
The most interesting observation obtained from Table 1 is that the synthesis using
Et4Mez-diquat-4 under the conditions described above yielded MCM-47, a layered silicate
composed of ferrierite (FER) layers [12]. To our knowledge, the 1,4-bis(N-methyl-
pyrrolidinium)butane (1,4-MPB), where the shape of the ammonium group is substantially
different from that on Et4Mez-diquat-4, has thus far been known as the only organic SDA
leading to the crystallization of this layered silicate [13]. It should be noted here that
replacement of NaOH by the equivalent amount of KOH under otherwise identical conditions
Table 1
Syntheses from gel composition 4.5R'xNazO'yAI203"30SiO 2' 1200H20 a'b
x - 5.0 x= 11.0
Organic SDA SIO2/A1203 ratio in the gel SIO2/A1203 ratio in the gel
30 60 120 oo 30 60 120 OO
Table 2
Syntheses from gel composition 4.5R'xNa20"0.5A1203"30SiO2" 1200H20 a'b
NaOH/SiO2 ratio in the gel
Organic SDA
0.33 0.47 0.60 0.73 0.87 1.00 1.13
Me4Et2-diquat-4 A A MOR MOR MOR MOR MOR
Me4Et2-diquat-5 A D+MOR D+(MOR) MOR+D MOR+D MOR MOR
Me4Et2-diquat-6 ZSM-12 A A A MOR A ANAc
Et4Mez-diquat-4 A ZSM-12 A A MOR MOR + L MOR
Et4Mez-diquat-5 ZSM-12 ZSM-12 A A MOR SSZ-16+ ANAc +
P1 SSZ-16
Et4Mez-diquat-6 ' ZSM-12 A MOR+(L) A MOR MOR + L SSZ-16
aR is organic SDA and x is varied between 5.0 _<y < 17.0. bThe crystallization conditions and
product notations are the same as those in Table 1. CAnalcime.
also produced MCM-47, which is in contrast to the trend found in the 1,4-MPB-mediated
synthesis of the same material [14]. While the synthesis procedure for MCM-47 with Et4Mez-
diquat-4 described above proved to be highly reproducible, on the other hand, the NaOH/SiO2
ratio yielding its pure form was found to be narrow. The synthetic results in Table 1 reveal
that quartz instead of MCM-47 is the phase obtained in the presence of this diquaternary
cation, when the NaOH/SiO2 ratio in the Al-free gel is increased to 0.73 while keeping all
other synthetic parameters constant. Such a change in product is also observed for the Me4Et2-
diquat-n ions with n = 4-6, since the synthesis at NaOH/SiO2 = 0.73 does not lead to the
formation of any zeolitic phase.
To more closely examine the effects of the NaOH content on the final product in
synthesis mixtures containing each of six different organic SDAs, we have varied NaOH/SiO2
ratio in the gel from 0.33 to 1.13 and performed zeolite syntheses under conditions where the
SIO2/A1203 ratio is fixed to 60. Our previous studies on zeolite syntheses using flexible
diquarternary alkylammonium ions have shown that a considerable diversity in the phase
selectivity of the crystallization is frequently observed for this particular SIO2/A1203 ratio [5-
9]. Neither of the organic SDAs studied here gives zeolitic products in their pure form, except
ZSM-12 and mordenite, in the NaOH/SiO2 ratio range 0.33-1.00. As seen in Table 2, however,
zeolite SSZ-16 was found to crystallize in the presence of Et4Mez-diquat-6 with increasing
NaOH/SiO2 ratio in the gel to 1.13. SSZ-16, first reported by Zones in 1985 [15], is
isostructural to the silicoaluminophosphate molecular sieve SAPO-56 (AFX) with a very
large cavity referred as the aft-cage [16-17]. Since this small-pore zeolite can also be
synthesized with several other diquarternary alkylammonum ions, which are formed of two
bulky organic groups with different sizes (e.g., triethylammonium, 1,4-diazabicyclo[2.2.2]-
octane, etc.) connected by the polymethylene bridging unit [5,15,16], it appears that Et4Mez-
diquat-6 may play a generally beneficial but structurally non-specific space-filling role in the
crystallization of SSZ-16. Among the doubly charged ammonium ions used in zeolite
synthesis thus far, on the other hand, Et6-diquat-5 is unique in that it is the only known SDA
leading to the crystallization of ZSM-57 with the intersecting 10- and 8-ring pore system [2-5].
The synthesis results in Table 2 reveal that Et4Me2-diquat-5, the cation most closely similar to
Et6-diquat-5, did not yield this medium-pore zeolite even at the optimized gel composition
(SIO2/A1203 = 60 and NaOH/SiO2 = 0.60) for its formation [2,5]. This suggests that the phase
selectivity of the crystallization in the presence of linear diquaternary alkylammonium ions
can differ according to both the bulkiness and length of the groups on the ammonium ion,
despite their highly flexible nature which allows them to undergo a large number of
conformational variations under the zeolite synthesis conditions.
187
(b)
(a)
i i I !
10 20 30 40
2 theta (degree)
Fig. 1. Powder XRD patterns of the (a) as-made and (b) calcined forms of MCM-47 prepared with
Et4Mez-diquat-4 and Na + ions, and (c) SSZ-16 prepared with Et4Mez-diquat-6. Detailed synthesis
conditions for these two materials are given in Table 1 and Table 2, respectively.
Fig. 1 shows the powder XRD patterns of the as-made form of MCM-47 and SSZ-16
prepared here. The positions and relative intensities of all the X-ray peaks are in good
agreement with those reported in the literature [12,16]. Calcination of as-made MCM-47 in
air at 550 ~ led to a notable peak broadening, as well as a decrease in d spacing for the (020)
reflection, accompanied by the appearance of an amorphous background. These changes are
consistent with those previously reported by Burton et al. [12]. The SEM photographs of two
MCM-47 samples obtained from Na +- and K+-containing pure-silica gels in the presence of
Et4Mez-diquat-4, respectively, are given in Fig. 2. The sample prepared using Na + as an alkali
metal appears as cuboids of c a . 1 #m. However, plate-like crystallites of approximately 1 #m
in length and 0.1 #m in thickness are observed for the sample prepared with K +. Elemental
analysis has revealed that the total organic contents in these two MCM-47 materials are 18.6
(C, 13.44; H, 2.90; N, 2.26) and 18.9 wt% (C, 13.64; H, 2.92; N, 2.35), giving numbers of
0.98 and 0.97 organic SDA molecules per half unit cell (18 Si atoms), respectively. These
values are essentially identical with 1 SDA/18 Si atoms expected from the proposed model
[12]. The same result can also be obtained from the TGA/DTA curves of both MCM-47
~!!!iiiiiiiiiii~
84 ~ii~iiiii~iiii,
84184184
ili
.......
~!!i~'!ii!~i~ i~
i~iiii~il
84iili, ii!iiii~
j
Fig. 2. SEM photographs of MCM-47 materials prepared by the combined use of Et4Me2-diquat-4 and
(a) Na+ and (b) K+ ions.
188
2 4~5
~ N §
4 5
(a) ~] 3
1 ]
i
80 6'0"4'0"2'0 ()
13C 6 (ppm)
Fig. 3. ~3C NMR spectra of Et4Me2-diquat-4, showing the assignment of each resonance: ~3C NMR of
(a) its bromide salt in D20 solution and 1H-13CCP MAS NMR of as-made (b) ZSM-12, (c) mordenite,
and (d) MCM-47 with Et4Mez-diquat-4 occluded in their pores.
materials, since their overall organic contents (18.8 and 18.9 wt%, respectively), calculated
from the large, exothermic weight loss appearing in the temperature region 250-700 ~
assignable to the decomposition of the occluded Et4Me2-diquat-4, are very close to the values
determined from elemental analysis.
Fig. 3 shows the 1H-~3C CP MAS NMR spectra of as-made ZSM-12, mordenite, and
MCM-47, all of which were prepared using the same organic SDA, i.e., Et4Mez-diquat-4,
together with the liquid ~3C NMR spectrum of a D20 solution of the bromide salt of this
diquaternary cation. The assignments of the observed resonances are also indicated in Fig. 3.
It is clear that Et4Me2-diquat-4 not only remains intact upon its occlusion into the pores of
zeolites prepared here, but also experiences the chemical environment that is notably different
in each of these three materials. When Et4Mez-diquat-4 is located inside the channels of ZSM-
12, for example, the resonance of the second methylene carbons in its central chain splits into
two peaks at 20.1 and 22.2 ppm, unlike that of the corresponding carbon resonances of free
Et4Me2-diquat-4. A similar trend is also observed for the case of the other two different types
of methylene carbons adjacent to the nitrogen, due to the appearance of four resonances in the
chemical shift region 55-65 ppm. These results suggest that two different degrees of
geometric constraints and van der Waals interactions with the zeolite framework are imposed
on all the methylene carbons of the occluded organic SDA. As seen in Fig. 3, on the other
hand, as-made mordenite shows signs of three components in the chemical shift region 60-65
ppm, despite the fact that two different types of methylene carbons adjacent to the nitrogen
exist in Et4Me2-diquat-4. In contrast, the ~H-~3C CP MAS spectrum of the organic in as-made
MCM-47 shows only one broad methylene carbon resonance at 62.4 ppm that is shifted
slightly downfield relative to that of free Et4Mez-diquat-4. This layered silicate also gives
189
N "
Fig. 4. Raman spectra in the 200-3200 cm~ region of (a) 0.5 M aqueous solution of Et4Mez-diquat-4
dibromide and as-made (b) ZSM-12, (c) mordenite, and (d) MCM-47.
only one resonance around 20 ppm assigned to the second methylene carbons in its central
chain, unlike as-made ZSM-12 or mordenite. Therefore, it is clear from Fig. 3 that the
Et4Me2-diquat-4 molecules in ZSM-12, mordenite, and MCM-47 adopt conformations
distinctly different from one another, which is mainly due to the difference in geometrical
restrictions imposed by the particular structure of each material.
Fig. 4 shows the Raman spectra in the 200-3200 cm -t region of ZSM-12, mordenite, and
MCM-47 containing Et4Mez-diquat-4 as a guest molecule. The Raman spectrum of Et4Mez-
diquat-4 in aqueous solution is also compared in Fig. 4. In parallel with the ~H-13C CP MAS
NMR data given above, Raman spectroscopy reveals that Et4Mez-diquat-4 is occluded intact
in each of these three materials. When compared with the Raman data of various tetraalkyl-
ammonium salts in the literature [ 18], the sharp band appearing at 715 cm l in the spectrum of
Et4Mez-diquat-4 in aqueous solution can be tentatively assigned to the symmetric C-N
stretching mode. Both the as-made ZSM-12 and MCM-47 materials give this vibration at 711
cm -~ that is essentially unchanged relative to free Et4Me2-diquat-4. As seen in Fig. 4,
however, the organic in mordenite exhibits it at 703 cm ~ The effect of the confined space
around the Et4Mez-diquat-4 can also be evidenced by differences in the band position and
intensity of the C-C stretching mode at 950-1150 cm -1 and the CH2 twisting and rocking
modes at 1175-1300 cm -~, as well as by the blue shift of the C-H stretching bands in the
region 3100-2800 cm 1, with respect to Et4Mez-diquat-4 in aqueous solution. These results
again confirm that the conformations of the organic molecules occluded in ZSM-12,
mordenite, and MCM-47 are distinctly different from one another.
4. CONCLUSIONS
A novel synthesis route to the layered silicate MCM-47, which includes the combined use of
Et4Me2-diquat-4 and Na + or K + ions, is presented. The overall results of this study
demonstrate that the phase selectivity of zeolite syntheses in the presence of flexible Me4Et2-
diquat-n and Et4Mez-diquat-n ions with n = 4-6 is sensitive t~ the bulkiness and length of the
190
groups on the ammonium ion, as well as the concentrations of inorganic components in the
synthesis mixture. Analysis of the ~H-~3C CP MAS NMR and Raman spectra obtained from
the EtaMez-diquat-4 molecules occluded in ZSM-12, mordenite and MCM-47 reveals that the
host-guest interactions occurring within the respective materials are considerably different
from one another, yielding differences in the conformation of the occluded organic molecules.
ACKNOWLEDGMENT
Funding for this work was provided by the Korea Science and Engineering Foundation under
the Grant No. R02-2003-000-10087-0 to S.B.H.
REFERENCES
ZSM-5 type zeolites containing alkaline earth metals (M-HZSM-5) were prepared under
various synthesis conditions and their hydrothermal stability was investigated. It was found
that the average crystal size of M-HZSM-5 can be easily controlled by addition of H3BO3
with no reduction of the amount of alkaline earth metal occluded. It was also found that the
hydrothermal stability of M-HZSM-5 evaluated by steaming at 600 ~ depends upon not the
crystal size but the amount of the alkaline earth metal occluded. Ca-, Sr- and Ba-HZSM-5
showed the higher hydrothermal stability, and Ca- and Sr-HZSM-5 exhibited the high
selectivity to light olefins in methanol conversion even after steaming at 600 ~ for 7 days.
1. INTRODUCTION
Deactivation of zeolite catalyst for methanol conversion to hydrocarbons such as light olefins
and gasoline under the high reaction temperature is attributed to (1) poisoning of outer surface
by coke deposition, accompanied by the restricted access of reactant to inner acidic sites by
pore blocking, and (2) degradation by dealumination with steam as a by-product [1,2]. As
coke deposits in the catalyst are easily burned off under a stream of air, the deactivation by
coke deposition is not so serious. In order to expand the catalyst life, therefore, dealumination
of zeolite catalyst must be depressed. From such a viewpoint, we have studied the
dealumination behavior of several zeolites by thermal and/or steam treatment [3-5]. It was
found that the dealumination of ZSM-5 type zeolite via hydrolysis of Si-O-A1 bond in the
framework is catalyzed by an acid (proton), which moves freely in the zeolitic pores. We have
also reported that the release of tetrahedrally coordinated aluminums from the ZSM-5 zeolite
framework is depressed by adding alkaline earth metals [6-8]. However, a detail
dealumination behavior of ZSM-5 type zeolites containing alkaline earth metals has not been
clarified. In this paper, therefore, various ZSM-5 type zeolites with different alkaline earth
metal and crystal size were prepared and their dealumination behavior was investigated in
detail.
2. EXPERIMENTAL
metal acetate and sodium hydroxide in deionized water. The hydrogel composition was:
SIO2/A1203-200, NaOH/SiO2=0.1-0.4, TPABr/SiO2--0.1, H20/SIO2=40, SIO2/B203=2.5 -c~,
and M/Si=0.005-0.025. The resultant hydrogel was transferred into a stainless-steal autoclave
and stirred at 160 ~ under autogenous pressure for 16-72 h. The precipitated crystals
obtained were washed with deionized water, dried at 120 ~ for one night, and calcined at
500 ~ for 20 h to remove the organic cations occluded into zeolite framework. The zeolite
was protonated in a 0.6 M HC1 solution at 80 ~ for 6 h, and calcined in air at 500 ~ for 8 h.
2.2. Steaming
Steaming was carried out using an atmospheric pressure flow system at 600 ~ 1 g of
zeolite (12-24 mesh) was charged into a quartz-tube reactor. The 50 % water vapor gas was
generated by nitrogen. The total gas flow rate (nitrogen + water vapor) was kept constant
(0.531 tool/h). 100% steam was also used to clarify an effect of water vapor pressure on the
dealumination behavior.
2.3. Characterization
Characterization of M-HZSM-5 before and after steaming was carried out by means of
X-ray diffraction (XRD, Rigaku RINT-2000), scanning electron microscopy (SEM, Hitachi
S-4000), FT-IR (JEOL JIR-7000), N2 adsorption (Bel Japan, Belsorp-28SA), X-ray
fluorescence (XRF, Philips PW-2000), inductively coupled plasma optical emission
spectroscopy (ICP, Seiko SPS-7700) and 27A1 MAS NMR (Varian VXP-4000). Prior to the
27A1 MAS NMR measurements, the sample was moisture-equilibrated over a saturated
solution of NH4C1 for 24 h.
groups are responsible for the cation exchange regardless of a l u m i n u m content [9,10]. The
B E T surface area measured by N2 adsorption was larger than 300 m2/g. Fig. 2 shows the
typical S E M images of various M - H Z S M - 5 . It was found that the average crystal size of M-
H Z S M - 5 could be easily controlled by varying the amount of H3BO3 added and the alkalinity
in the starting gel (NaOH/SiO2 ratio). In the case of H3BO3 addition, the reduction of alkaline
earth metal content occluded in the M - H Z S M - 5 (M/AI2 ratio) was hardly observed (Table 1).
Table 1
Characteristics of various M - H Z S M - 5 prepared.
Sample Zeolite Starting synthesis gel Product
Average BET surface
no. SiO2 NaOH M/Si Si/AI2 M/AI2
crystal size area
....../B203 /Si02 ...................(>m) (mZ/g)
1 HZSM-5 ~ 0.1 - 190 - 9.5 352
2 HZSM-5 40 0.1 - 199 - 4.5 _348_ .......
3 Ca-HZSM-5 ~ 0.1 0.025 200 1.73 8 318
4 Ca-HZSM-5 80 0.1 0.025 207 2.56 5 332
5 Ca-HZSM-5 30 0.1 0.025 197 2.56 3 334
6 Ca-HZSM-5 10 0.1 0.025 204 2.08 1.5 322
7 Ca-HZSM-5 e~ 0.3 0.025 167 0.40 2 347
8 Ca-HZSM-5 ~ 0.4 0.025 133 0.34 1 345
9 Ca-HZSM-5 80 0.1 0.015 199 1.58 5.5 -
10 Ca-HZSM-5 80 0.1 0.005 201 0.89 6 -
11 Sr-HZSM-5 ~ 0.1 0.025 195 2.51 7 325
12 Sr-HZSM-5 40 0.1 0.025 205 2.87 4.5 335
13 Sr-HZSM-5 20 0.1 0.025 205 2.51 3.5 -
14 Sr-HZSM-5 ~ 0.3 0.025 177 0.62 3 358
15 Sr-HZSM-5 40 0.1 0.015 206 2.07 4.5 -
16 Sr-HZSM-5 40 0.1 0.005 200 0.90 4 -
17 Ba-HZSM-5 ~ 0.1 0.025 198 1.07 9 327
18 Ba-HZSM-5 30 0.1 0.025 205 2.77 4 -
19 Ba-HZSM-5 c~ 0.3 0.025 166 0.34 3 355
Synthesis conditions 9 SIO2/A1203=200, TPABr/SiO2=0.1, H20/SIO2=40, Temp. = 160 ~ Time= 18 h.
Ca-HZSM-5
6"m
Sr-HZSM-5
Ba-HZSM-5
~~~[~ii~6 ~m
However, a significant decrease in the M/A12 ratio was observed for the zeolites synthesized
with the higher NaOH/SiO2 ratios. The SIO2/B203 ratio in the M-HZSM-5 (ca.400) was
considerably smaller as compared with that in the starting gel. It is well recognized that B
atoms in the zeolite framework are easily released through the thermal and acid treatment.
To clarify the chemical state of aluminums in the M-HZSM-5, the 27A1 MAS NMR
spectra were measured. Fig. 3(A) shows the typical 27A1 MAS NMR spectra of various M-
HZSM-5. Only a sharp peak at ca. 54 ppm was observed in all spectra, which is a
characteristic resonance of tetrahedrally coordinated framework aluminums [11]. No peak
assigned to non-framework aluminums (extraframework aluminums) was observed around 0
ppm. Therefore, it was clearly revealed that all of aluminums in the M-HZSM-5 are present in
the zeolitic framework.
(B)
(A) ~ (d)
(d)
(c) (c)
rJ3
(b) (b)
Fig. 3. 27A1MAS NMR spectra of various M-HZSM-5 (A) before and (B) after steaming at 600 ~
for 12h.
Sample no. (a) 1, (b) 3, (c) 11, (d) 17 ~ 100
\
3.2. Hydrotherma! stability of M-ItZSM-5 ~ ~, 80
At first, M-HZSM-5 was steamed at 600 ~ ~
~ under water vapor pressure of 50 kPa in ~ x 60
order to elucidate their dealumination ~ ~
~ o
behavior. The degree of dealumination was N N
evaluated from a difference in the 54 ppm ~ ~ 40
peak intensity in the 27A1 MAS NMR spectra .o ~
of samples before and after steaming. Fig. ~ =
3(B) shows the 27A1 MAS NMR spectra of ~ ~ 20
various M-HZSM-5 after steaming for 12 h. ~ .~
As compared with the 27A1 MAS NMR ~ ~N 0 , I ~ I i I i I i I i I , I
1 2 3 4 5 6 7
spectrum of HZSM-5 before steaming (Fig.
Time of steaming (day)
3(A)-(a)), a rapid decrease in the intensity of
the peak at 54 ppm was observed for the Fig. 4. Effect of crystal size on dealumination
steamed HZSM-5. On the other hand, only a of Ca-HZSM-5.
Sample no. @ :4, I :5, A "6
195
slight decrease in the peak intensity was observed for Ca-, Sr- and Ba-HZSM-5 after
steaming. This strongly suggests the great stability of these zeolites for steaming. For all
steamed zeolites, no peak was observed around 0 ppm. This phenomenon can be explained as
follows. The non-framework aluminium is not uniform and exists in different states, resulting
in broadening the 27A1NMR signal and therefore leading to some 27A1 NMR invisible peaks
[12, 13].
To get further information concerning the dealumination behavior of M-HZSM-5, the
effects of the crystal size and the amount of alkaline earth metal occluded on the
dealumination were investigated. Fig. 4 shows the time dependence of the intensity ratio of 54
ppm of the steamed Ca-HZSM-5 to unsteamed zeolite. Although the framework aluminums
decreased with the steaming time, there was no difference in the degree of the reduction of the
amount of framework aluminums. Fig. 5 shows the effect of the Ca content (Ca/A12 ratio) on
the dealumination of Ca-HZSM-5. It
100
became clear that the degree of ~D
the ZSM-5 zeolite framework is considerably depressed by alkaline earth metal occluded in
the M-HZSM-5 to a different degree.
the steaming time. On the other hand, in the case of Ca-HZSM-5 (Fig.10(B)), although the
intensity of the peak at 3605 cm -~ decreased in a similar manner, the intensities of the peaks at
3568 and 3500 cm -~ hardly changed even after steaming for 7 days. In the FT-IR spectra of
Sr-HZSM-5 (Fig.10(C)), a clear trend was not observed.
~- I ,., .. I I
4000 3500 3000 4000 3500 3000
Wavenumber (cml) Wavenumber (cm 1)
Fig. 9. FT-IR spectra of (A) HZSM-5 and (B) Ca-HZSM-5
(a) before and after (b) Na + ion exchange and (c) pyridine adsorption.
(A)3740
03)
3740 3658
~ (c)
I 3605 I / 3605
' 0 U--""-L~ 0
Table 2
Effect of steaming on catalytic activity of M-HZSM-5 in methanol conversion.
Zeolite Steaming Effective Product distribution (C-%)
(day) cony. a C2H4 C3H6 C4H8 CsHto CH4 C2- BTX C6 + CO +
(%) C5 b CO2
HZSM-5 0 100 18.4 25.9 8.7 1.5 12.3 5.5 20.1 7.5 0.1
HZSM-5 1 100 13.9 44.6 19.0 6.0 1.6 2.8 4.4 7.7 0
HZSM-5 3 100 10.5 40.1 1 9 . 3 12.0 1.7 2.2 2.1 12.1 0
. .HZSM-5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7. . . . . . . . . . . . . . . .67.7
. . . . . . . . . . . . . . . .5.9
. . . . . . . . . .27.5
............1 . . .7. ... .3. . . . . .13.2
. . . . . . . . . . . . . 3.5
. . . . . . . 2.6 1.8 28.2 0
Ca-HZSM-5 0 100 16.4 47.2 18.4 3.5 1.5 1.5 6.1 5.3 0.1
Ca-HZSM-5 1 100 14.7 46.2 18.9 6.0 1.4 0.9 4.8 7.0 0.1
Ca-HZSM-5 3 100 11.2 44.7 19.3 8.9 1.3 1.0 4.3 9.2 0.1
. .Ca-HZSM-5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7. . . . . . . . . . . . . . . . 100
. . . . . . . . . . . . . . . .6.5
. . . . . . . . . .39.2
............1 . . .9. . .. .0. . . . . .14.8
. . . . . . . . . . . . . .1.7
...... 0.7 2.6 15.5 0
Sr-HZSM-5 0 100 9.0 41.6 20.0 11.3 6.8 1.2 2.1 7.8 0.2
Sr-HZSM-5 3 100 9.9 42.5 19.6 10.4 2.0 0.9 4.3 10.2 0.2
. .Sr-HZSM-5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7. . . . . . . . . . . . . . . . 100
. . . . . . . . . . . . . . . .7.9
. . . . . . . . . .39.8
. . . . . . . . . . . . 19.6
. . . . . . . . . . . . .13.9
. . . . . . . . . . . . . .1.9
...... 0.7 3.3 12.8 0.1
Ba-HZSM-5 0 37.0 0.9 2.3 0.4 0.7 26.8 0.2 0 0 68.7
Ba-HZSM-5 3 22.5 1.8 9.5 7.6 6.1 37.6 0.5 0.2 16.3 20.3
Ba-HZSM-5 7 14.1 0.8 4.9 4.1 1.2 41.4 0 0 8.7 38.9
aEffective conversion= 100-(methanol + dimethyl ether), bc 2 to C5 paraffins
4. C O N C L U S I O N S
From all of above results, it was found that the crystal size of M-HZSM-5 can be easily
controlled by addition of H3BO3 and the dealumination by steaming is greatly depressed by
adding alkaline earth metal. The high hydrothermal stability of M-HZSM-5 was also
confirmed from methanol conversion on the steamed zeolites. Ca- and Sr-HZSM-5 showed
the high selectivity to light olefins even after steaming at 600 ~ for 7days.
REFERENCES
[1] W.W. Kaeding and S.A. Butter, J. Catal., 61 (1980) 155.
[2] C.D. Chang, C.T.-W. Chu and R.F. Socha, J. Catal., 86 (1984) 289.
[3] T. Sano, H. Ikeya, T.Kasuno, Z.B. Wang, Y. Kawakami and K. Soga, Zeolites, 19 (1997) 80.
[4] Y. Oumi, R. Mizuno, K. Azuma, S. Nawata, T. Fukushima, T. Uozumi and T. Sano,
Microporous Mesoporous Mater., 49 (2001) 103.
[s] Y. Oumi, S. Nemoto, S. Nawata, T. Fukushima, T. Teranishi and T. Sano, Mater. Chem. Phys.,
78 (2002) 551.
[6] K. Fujisawa, T. Sano, K. Suzuki, H. Okado, K. Kawamura, Y. Kohtoku, S. Shin, H. Hagiwara
and H. Takaya, Bull. Chem. Soc. Jpn., 60 (1987) 791.
[7] T. Sano, T. Murakami, K. Suzuki, S. Ikai, H. Okado, K. Kawamura, H. Hagiwara and H.
Takaya, Appl. Catal., 33 (1987) 209.
[8] H. Okado, H. Shoji, T. Sano, S. Ikai, H. Hagiwara and H.Takaya, Appl. Catal., 41 (1988) 121.
[9] S.G. Fegan and B.M. Lowe, J. Chem. Soc., Chem. Commun., (1984) 437.
[10] G.L. Woolery, L.B. Alemany, R.M. Dessau and A.W. Chester, Zeolites, 6 (1986) 14.
[11] C.A. Fyfe, J.M. Thomas, J. Klinowski and G.C. Gobbi, Angew. Chem. Int. Ed. Engl., 22
(1983) 259.
[12] E. Loeffier, U. Lohse, Ch. Peuker, G. Oehlmann, L.M. Kustov, V.L. Zholobenko and V.B.
Kazansky, Zeolites, 10 (1990) 266.
[13] E. Brunner, H. Ernst, D. Freude, T. Frohlich, M Hunger and H. Pfeifer, J.Catal., 127 (1991) 34.
[141 P.A. Jacobs and R.van Ballmoos, J. Phys. Chem., 86 (1982) 3050.
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 199
The binary porous materials containing nitrogen (imido-, nitrido- groups) as a bridging anions
between the framework atoms such as Si, A1, P, Sn, Zn, B have been prepared mostly by joint
ammonolysis of chlorides of respective elements. The products comprising two different
framework elements (e.g. Si/A1, A1/P) were also obtained by separate ammonolysis of
substrates followed by mixing of resulting amides or imides. The products were always
amorphous, but after thermal removal of NH4C1 they showed high surface area and some
ordering of mesoporous system. The attempts to crystallize them in liquid ammonia or other
aprotic solvents with contribution of alkalis (e.g. NaNH2) usually did not result in products of
distinct crystallinity, moreover their porosity was always drastically decreased. Some of the
samples (particularly binary Si/A1, A1/P systems) showed a noticeable catalytic activity in
propan-2-ol decomposition, which reflected the presence of acid sites.
1. INTRODUCTION
Zeolites and zeolite-like materials are as a matter of fact the combined oxide compounds (e.g.
xNazOxAlzO3ySiOzzH20). Very common and broadly applied isomorphous substitutions of
the T atoms in zeolites allow to prepare a great variety of the molecular sieves containing
different elements (or rather their respective oxides). The introduction of different elements
into the framework positions affects the physicochemical properties of obtained materials and
sometimes results in formation of novel structures. Another challenging problem in synthesis
of the molecular sieves was a substitution of oxide anions between the framework atoms for
another elements or groups.
Substitution of oxides for sulfides resulted in numerous crystalline open framework structures
(thiozeolites) [ 1]. The sulfide containing structures appeared, however, always less stable than
the conventional oxide molecular sieves.
~8! ,if
H H I
/\ /\ /\ /\ /\ /\
H H
/\ /\ /\ /\ /\ /\
Fig. 1. Some of potential structure segments in the nitrogen containing materials.
200
Another isoelectric anion which could substitute the bridging oxide is imide (-NH-). Also
nitride (-N =) can be considered as a bridging component of the framework (Fig. 1). The
nitrides of some elements are known as very stable materials applied as ceramics, insulators
or semiconductors. The attempts to prepare the nitrogen containing porous materials have
been reported since several decades. Schnick obtained several nitridophosphate- or
oxonitridophosphate-sodalites as well as nitridosilicates by means of high temperature solid
state reactions [2-5]. Several authors [6-8] presented a possibility to introduce nitrogen
functional groups into crystalline structures of zeolites or zeolite-type materials by their
thermal treatment with ammonia. The introduced amide or imide groups were considered as
basic catalytic sites, active for Knoevenagel condensation. Another approach comprised
preparing the porous imides, nitrides or oxynitrides by means of ammonolysis of silicon
halides (mostly SIC14) with ammonia or amines. The obtained products showed an ordered
micro- or mesoporosity and could be applied as a support for basic catalysts after
impregnation with metallic potassium [9-13].
The results presented here comprise our preliminary attempts to prepare and
characterize the nitrogen bearing, binary materials, containing such potential T atoms as Si,
A1, P, Sn, Zn, B. The general idea of synthesis was to mimics the conventional hydrothermal
synthesis of zeolites [14]. The amides or imides played a role of oxides (SiO2, A1203), NaNH2
was considered the NaOH analog. The latter amides can form alkaline amido or imido
aluminates and silicates. Subsequently they can yield the aluminosilicate analogs. The
compositions of other elements could follow a similar reaction route. Liquid ammonia was a
principal crystallization medium instead of water applied in hydrothermal process. Liquid
ammonia was already applied as alternative medium for crystallization of zeolites [15].
Another aprotic solvents of relatively high dielectric constant and dipole moment were also
applied. The potential structure directing agents were admitted to some crystallization
mixtures.
2. E X P E R I M E N T A L
The chlorides of silicon, aluminum, tin, zinc and phosphorus were used as main sources of
potential framework atoms. The metallic powder of Zn, A1 as well as A1N were also used in
some experiments. Usually the first step of preparation comprised an ammonolysis of
respective chlorides with gaseous ammonia. To prevent the explosive course of reactions
(particularly with SIC14) the solvents such as chlorophorm, nitrobenzene, acetonitrile were
employed. In many cases two mixed substrates (e.g Si/A1, P/Si, P/A1) were jointly treated
with ammonia. The separate ammonolysis was applied too in some cases and the resulting
products were mixed afterwards. The ammonolysis products were always heated (300 -
500~ in nitrogen stream in order to remove the ammonium chloride formed upon
ammonolysis. Another way of preparing amides consisted in reaction of metals with
ammoniacal solutions of NaNH2 or metallic sodium. The latter alkaline reagents were also
applied to form the crystallization mixture from the binary amides or imides. Sometimes urea,
amines or alkylammonium salts were used as potential template agents. Ammonia was a
principal crystallization medium, although some other aprotic solvents such as formamide,
DMF, DMSO, nitrobenzene (NB), acetonitrile (AN) were employed. The crystallization was
conducted in steel autoclaves for various period of time (several hours - several weeks).
Using of liquid ammonia limited the temperature of the process. Therefore, many experiments
were carried out at room or only slightly elevated temperature (80~ The other solvents
allowed to heat safely the mixtures at above 100~ The obtained products were characterized
201
~iNH 2 AI,Si
E E
8 8
!- ,Si,Na(NH) a
...
. ,
1600 14;0 ' 12;0 ' 10~30 ' 8()0 ' 6()0 " 4i
400 1600 ' 1400 ' 12~)0 ' 1000 ' 8(~0 ' 600 400
Wavenumber [cm 1] Wavenumber [crn 1]
Fig. 2. Typical IR spectra of Si(NH)2 and the Fig. 3. IR spectra of step-product (Si/AI=30)
products ofjoint ammonolysis of SIC14 and and of the crystallization (with NaNH2) product
A1CI3 (Si/Al = 2 or 50) followed by calcination
in nitrogen at 350~
202
In most cases the crystallization procedure did not lead to crystalline products. However, the
XRD patterns (Fig. 4) of some samples indicate the presence of the crystalline phase. The
thermal treatment of the crystalline product Na/A1/Si=I at 500~ does not bring about any
considerable changes in the structure, whereas the heating at 900~ causes a transformation,
which probably results from deammination of imide and formation of nitride.
drastically. The samples indicating a contribution of crystalline phase always show very low
surface area (e.g. Na/A1/Si-1 -2 mZ/g ).
1000
900 3,025 _ I //
800 o,o.o/\ I //
700 0o,o ts t I ~ /
600
0o,0\ I / /
o~ ooo~ ~ _ _ I //
~'E 500
o ooo0 . . . . . .--1 / /
4OO
300
2O0
100
0,0 o12 o14 o:6 o18 11o
P/Po
Fig. 6. Typical ( sample Si/A1 =30) nitrogen Fig. 7. Typical TEM photograph adsorption
isotherm and pore distribution (sample Si/A1-2)
The fragments of the samples show ordered pore system, which is well seen in the TEM
pictures, although the ordered channels are not visible in entire volume of the samples.
Si/B=6400~ 1600 ]
14001
12001 o,oot/ \ //
~8oo1o,oo,
1000 I lv //
oE 600 t
o,ooo; ~6oA ~
~ 400t
20001 f
9 | 9 , , | 9 ! 9 | , ,
O
r-
P,Sn
E
r-
E
I.-
- - ~
., 9 i ' '
1600 ' 14;0 " 12;0 ' 10;0 _8,0o 6;0 400
Wavenumber [cm- ]
The syntheses of binary products with phosphorus and other elements (A1, Sn) were
conducted similarly as the described above. In the case of the A1/P samples only equimolar
ratio of A1 and P was applied. The products showed also considerable high surface area (560
mZ/g). In some cases (Sn, Zn) the syntheses comprised additional treatment with sodium
amide in liquid ammonia solution, but the resulted products were amorphous and their surface
areas were rather small.
propene, which rather indicates the presence of acid sites. The testing of active samples for
cumene cracking showed none or negligible activity, which suggests rather weak acid
strength. The silicon diamide was inactive, but an introduction of A1 increased the activity
very substantially. The further transformation of the A1/Si semi-products with NaNH2
(crystallization) affected the porosity and even more practically eliminated a catalytic activity.
Considerable activity is also seen in the case of sample P/A1. Most of the other samples
appeared practically inactive.
./t ............................................................................................................................................................................ ]
Fig. 11. Catalytic activity of
50-( i m i typical samples for dehydration
of propan-2-ol at 230~
!1
9.~ ~ - ~ ~ !
~
"~o-/3>
t- i I i I I I
ii
o
O -,._
15 / 1 !1 II II !
~o_~ II II ii !
The presented results illustrate preliminary series of experiments, which require further, more
systematic elaboration of the most promising findings. Properties of the most representative
samples are compiled in the table below:
Table 1
Typical samples obtained in syntheses
Framework Substrates Synthesis Properties
elements
Si SiCI4 SIC14 treated with NH3, calc. at 500~ Amorph. surf. area.
1070m2/g, cat. inact.
Si, AI (50/1) SIC14,A1C13 Joint NH3 treatment of both substrates, calc. Amorph. 750 m2/g,
at 500~ cat. active
Si, A1 (2/1) SIC14,A1CI3 Joint NH3 treatment, calc. at 500~ Amorph. 560 m2/g,
cat. active
Si, P (1/1) SIC14,PCIs Joint NH3 treatment, calc. at 500~ Amorph. 370 mZ/g,
cat. active
Si,,P (2/1) SIC14,PCI5 Joint NH3 treatment in NB, crystallization in Cryst. 30 mZ/g, cat.
NB, 180~ 4 days, calc 500~ active
AI, P (1/1) A1CI3,PCI5 Joint NH3 treatment, calc. at 500~ Amorph. 560 mZ/g,
cat. active
Zn, Sn (1/3) Zn, Na, Reaction of Zn with Na dissolv, in NH3 + Cryst. 10 mZ/g, cat.
SnCl4 SnCI4, cryst, at r.t. in NH3 for 3 days active
Si, B (1/6) SIC14,BC13 Joint NH3 treatment in acetonitrile, calc. at Amorph. 760 mZ/g,
400~ cat. inact.
206
4. CONCLUSIONS
REFERENCES
[ 1] R.L.Bedard, S.T. Wilson, L.D. Vail, J.M. Bennet and E.M Flanigen, Stud. Surf. Sci. Catal.,
49A (1989) 375.
[2] W. Schnick and U. Berger, Angew.Chem. Int. Ed., 30 (1991) 830.
[3] W. Schnick, and J. LiJckow, Angew.Chem. Int. Ed., 31 (1992) 213.
[4] W. Schnick, Stud. Surf. Sci. Catal. 84 (1994) 2221.
[5] W. Schnick and H. Huppertz, Chem.Eur. J., 3 (1997) 679.
[6] S. Ernst, M. Hartmann, S. Sauerbeck, T. Bongers, Appl. Catal. A 200 (2000) 117.
[7] P. Grange, P. Bastians, R. Conanec, R. Marchand, Y. Laurent, Appl. Catal. A 114 (1994) L 191.
[8] A. Massion, J.A. Odrizola, Ph. Bastians, R. Conanec, R. Marchand, Y. Laurent, P. Grange,
Appl. Catal. A 137 (1994) 9.
[9] P.W. Lednor and R. de Ruiter, Chem. Commun., 1991,1625.
[10] S. Kaskel, D. Farrusseng, K. Schlichte, Chem. Commun. (2000) 2481.
[ 11] S. Kaskel, and K. Schlichte, J. Catal., 201 (2001) 270.
[12] S. Kaskel, M. Khanna, B. Zibrowius, H-W. Schmidt, J. Crystal Growth, 261 (2004) 99.
[ 13] D. Farrusseng, K. Schlichte, B. Spliethoff, A. Wingen, S. Kaskel, J. Bradley, and F. Schtith
Angew.Chem. Int. Ed. Engl.,40 (2001) 4204.
[ 14] R. Kocifiski, J. Migdat, K. Held, S. Kowalak, E. Szymkowiak, Proc. IX Polish Zeolite Forum,
Wysowa, 2002, 98.
[15] D.M. Millar, J.M. Garces, 12th International Zeolite Conference, Baltimore, 1999 (Material
Research Society), 1535.
[ 16] F. Cheng, B. Toury, F. Lefevre and J.S. Bradley, Chem. Commun. (2003) 242.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 207
The zincosilicalites (MFI) with Si/Zn - 100 have been prepared from sodium silicate and
sodium zincate. Various acids were used for reducing the alkalinity of the starting mixture (to
the pH value of 11) in order to form the initial gel. The anions introduced with acids affected
the crystallization as well as the properties of the resulting zincosilicates. The samples
prepared with phosphates and sulfates showed the highest crystallinity, whereas the presence
of HC1 or acidic acid hindered forming a zeolite structure. The samples modified with various
cations or with dispersed platinum showed a considerable catalytic activity in following
reactions: propan-2-ol decomposition, cumene cracking, 1-butene isomerization, propan-2-ol
oxidation, thiophene hydrodesulfurization. The samples synthesized in the presence of
phosphates showed always the highest activity.
1. INTRODUCTION
The isomorphous substitution of the framework atoms in zeolites allows to prepare the novel
zeolite-like materials with properties depending on nature of incorporated atoms [1]. The
zincosilicate molecular sieves form several structures. Most of them is analogous to those of
aluminosilicate zeolites. The novel structures without any known zeolite analogs have been
obtained too. Our earlier study [2-5] have shown that MFI structure can be attained by
zincosilicates with the Zn/Si ratio as high as 0.1. The higher zinc loading in the initial mixture
always resulted in forming the amorphous products or in dense phases (e.g. quartz). The
crystallinity and properties of the products depended on sources of zinc and of silicon [3,4].
Recently, we have found [4] that sodium zincate is an efficient reagent for crystallization of
zincosilicate MFI. The preparation procedure reminded a conventional synthesis of zeolites
that always involves a forming of the starting gel by mixing the solutions of sodium silicate
and sodium aluminate. In the case of zincosilicate the highly alkaline solutions of zincate and
silicate do not form any gel after mixing. However reducing of the pH of the mixture to the
level 11 by means of acids causes an immediate gelling. The pH value of 11 is recommended
for the mixture used for crystallization of metallosilicate MFI [6].
We applied several different acids for adjusting the pH of the starting zincosilicate gel.
It has been known that the presence of various anions in the crystallization mixture (such as
OH-, F-, PO43-) affects very much the course of crystallization [7,8].
The main aim of the study was to compare the structure and properties of the resulting
products regarding the influence of anions present in the zincosilicate gel. The series was
restricted to the samples with the even Zn/Si ratio of 0.01. The catalytic activity of the
samples were examined in several reactions requiring the acid and red-ox sites (propan-2-ol
208
2. E X P E R I M E N T A L
The initial gels were formed from solutions of water glass (supplied by Prayon-Rupel,
Belgium), sodium zincate and tetrapropylammonium bromide (TPABr) (Lancaster). Sodium
zincate was prepared by dissolving ZnO (P.O.Ch., Poland) in concentrated NaOH. Certain
amounts of different acids (H3PO4, H2804, H2C204, HF, HNO3, HC1, CH3COOH) were used
to adjust the pH of the initial gel to the value of 11. The gel was aged at room temperature for
an hour. Then the hydrothermal crystallization was carried out in Teflon lined autoclaves at
170~ The crystallization time was the same (46 hours) for all the samples. The resulting
products were washed with water and dried. Then the template was removed by calcination in
air at 450~ as long as the constant mass was attained (-50 hours). The products were
characterized by means of standard methods such as XRD (TUR M-62), BET (ASAP 2010,
nitrogen adsorption), AES-ICP (ARL 3410), IR (Bruker FTIR Vector 22, KBr pellets or self-
supported wafers with a vacuum cell, pyridine was adsorbed at room temperature), SEM
(Philips SEM 515), 298i MAS NMR (Bruker AMX 300 WB), TG/DTA (SETARAM
SETSYS 12, air atmosphere).
The samples were modified with cations (Cu 2+, Zn 2+, A13+, Ce 3+, H +, Ca 2+, Na +) by means of
conventional ion-exchange procedure. The hydrogen forms were prepared by thermal
treatment of ammonium derivative. The samples (prepared with H3PO4) used as catalysts for
HDS of thiophene were prepared by impregnation of zincosilicate MFI with HzPtC16 aqueous
solution. The samples (5 wt.% Pt) were then dried and calcined in air at 400~ for 4 hours.
The catalytic tests for cumene cracking, propan-2-ol decomposition were conducted in a pulse
micro-reactor attached to the gas chromatograph, equipped with TCD. The catalyst samples
(powder 0.015g) were activated prior to reaction at 400~ (propan-2-ol decomposition) or
450~ (cumene cracking) in helium stream for 30min. The propan-2-ol decomposition was
carried out at 230~ and cumene cracking at 350~ respectively. The pulses of 1#1 were
injected. The oxidation of propan-2-ol with H202 as oxidant was conducted in a batch system.
The molar proportions of alcohol/oxidant were 1/0.23, respectively. The reagent mixture
(1.35cm 3) and 0.05g of the catalyst powder were maintained in a glass ampoules. The samples
were held at 75~ for 20 hours. After cooling the reaction mixture the products were analyzed
by means of gas chromatograph equipped with TCD. Isomerization of 1-butene was carried
out in closed-circulation system at 100~ The catalysts (0.05g) were evacuated at 400~
under vacuum for 2 hours before the reaction. The reaction products were analyzed by gas
chromatograph (TCD) connected to the closed-circulation system. HDS reaction of thiophene
was carried out in the conventional fixed-bed flow reactor at 350~ under 1 atm using 0.1 g of
catalyst. The catalyst was pretreated in He stream at 400~ for 1 hour and then was reduced
by H2 at 400~ for 1 hour. The reaction conditions are as follows: Hz/thiophene = 30mol/mol;
W/F= 37.9 g.h/mol. The reaction products were analyzed by gas chromatograph (FID).
The properties of the resulted products are listed in the Table 1. As indicated XRD results
(Fig. 1) the presence of phosphoric, oxalic or sulfuric acids in the starting gel allowed to
obtain pure zincosilicates MFI with high crystallinity. There are no reflections that could be
attributed to ZnO admixture. The samples obtained with contribution of HF and HNO3
209
showed lower crystallinity. The latter show also some admixture of quartz. The high
crystallinity of the product of phosphate assisted synthesis is consistent with observation of
Kumar [7] who emphasized the promoting effect of these anions in crystallization process.
The sulfates seem to facilitate this process too. The HC1 and acetic acids used in mixtures
resulted in very hard, packed materials, which, however, completely dissolved in water on
washing.
Table 1.
Properties of obtained samples
Acid applied H3PO4 H2C204 H2SO4 HF HNO3 HC1 CH3COOH
Unit cell volume 5325.13 5198.97 5389.51 5405.89 5225.63 - -
Zn/Si in products 0.0143 0.0121 0.0118 0.0116 - - -
BET surf. area [m2/~] 326.2 - 313.3 - ~ - -
- not determined
The calcination at 450~ used to remove the template did not bring any noticeable changes in
crystallinity of the samples. Only transformation of the orthorhombic to monoclinic symmetry
was noticed. The calcination at higher temperature (900~ caused the recrystallization to
crystobalite (Fig. 2).
Q
Acid
HNO3
I calcinedHat3P900
40oC
i I HF
4 14 24 34
4 14 2 Th2~a 34
2 Theta
Fig. 1. X R D patterns o f the as synthesized Fig. 2. XRD profiles of sample calcined at
samples in the presence of different anions. indicated temperatures.
oxidation products. The effects mentioned above, recorded for silicalite-1 always take place at
markedly lower temperatures (i.e. 380~ [9]. It can suggest a stronger bonding of
alkylammonium cations to zincosilicalite MFI inner surface due to negative charge of the
framework (Fig. 3).
The unit cell volume of the samples depends on the type of anions present in the synthesis
mixture during the crystallization (Table 1), although the differences are relatively small.
Only for the sample obtained in the presence of oxalic anion this value differs significantly. It
can suggest a lower zinc content in the framework positions. The chemical analysis of zinc
indicated the influence of the accompanying anions on the amount of zinc in the products.
The highest zinc content is in the sample prepared with phosphorous acid, the lowest was
found in the sample obtained with hydrofluoric acid. The results of elemental analysis are not
very much consistent with the values of the unit cell volumes. The sample crystallized in the
presence of F shows the lowest Zn loading, whereas its unit cell volume is the largest among
the studied samples. This suggests that some zinc atoms in Zn-rich samples occupy the extra-
framework positions.
The morphology and size of crystallites are affected by the anions employed in syntheses
(Fig. 4). In the case of samples prepared with oxalic and sulfuric acids the resulting samples
form the circular crystallites (-7 #m in diameter). In the case of earlier sample some
admixture of small irregular in shape particles is seen. The sample obtained in the presence of
phosphoric acid consists of particles with uniform circular shape, but they are evidently
composed of small rectangular crystallites. Smaller magnification of picture shows that the
samples obtained in the presence of sulfates and phosphates form separated particles, whereas
the sample obtained with oxalic anions form larger agglomerates. Crystallites of the sample
prepared with HF show considerably different morphology. We can observe two types of
particles. One of them show poor crystallinity and irregular shape. These particles stuck
together and form the bigger aggregates. The second type of crystallites exhibits a needle
shape.
The BET surface area measured for the samples crystallized in the presence of H3PO4 and
H2SO4 showed very similar values, regardless of the kind of anions accompanying the
synthesis (Table 1). The IR spectra (KBr) did not bring any distinct evidence on the presence
of zinc in the framework. We can only observe very weak band a t - 960 cm -~ that can reflect
the Zn-O bond vibration (Fig. 5).
298i MAS NMR analysis showed the presence of two signals a t - 1 1 5 a n d - 1 0 5 ppm [10].
Signal a t - 1 1 5 ppm is assigned to silicon atoms surrounded by four another silicon atoms.
The other signal can be attributed to silica atoms combined with three silicons and one zinc
atom Si(3Si, 1Zn) [ 11, 12]. The intensity of this signal decreased after calcination. It suggests
the migration of the part of zinc atoms to the extra-framework positions on calcination.
211
............................: ...........................................................................................................................................................................................................................
H A! Ce Zn Cu Ca Na
form
100 ~ Fig.7.
El a c e t o n e Reaction selectivity to acetone
I propene and propene over different
8o cation modifications of the
1 sample obtained in the
60~ presence of phosphorous
anion.
9
r~
~, 40 ,
20 / I
H Ai Ce Zn Cu Ca Na
form
212
The cations introduced into the studied zincosilicate MFI by means of ion-exchange affected
considerably their activity and selectivity. The aluminum and protonic forms showed the
highest activity, whereas the sodium and calcium forms were practically inactive (ca. 2%).
The dehydration was a predominant route of propan-2-ol decomposition over the catalysts
modificated with H +, Cu 2+, Zn 2+ and Ce 3+, although some amount of acetone was recorded
too. In the case of the samples modified with A13+ the only product was propene. The
dehydrogenation towards acetone was prevailing process in the case of calcium and sodium
forms used as catalysts (Fig. 6, 7).
Another reaction involving acid sites applied in our study was 1-butene isomerization (Fig. 8).
I ~___d [] Zn form
40 22VI ] aICeform 1 I
~ 1 0 ~ - - ~ [] AI form i
30
~" 20-
4 i I
lll _ . I..,.d I
z i
0
H3PO 4 HzCzO 4 HNO 3 HzSO 4 HF acid
H Zn Cu Na fo'rm
Fig. 8. Conversion of 1-butene isomerization Fig. 9. Activity of the indicated samples for
for samples obtained by using PO4 3 anion. cumene cracking.
The sample prepared with phosphates showed a remarkable activity, which depended very
much on kind of cation introduced into the zincosilicate. The H form was the most active,
while the sodium form was completely inactive. The cis- and trans-2-butenes were the
reaction products. The contribution of both isomers was similar, although the content of trans
isomer was always a little prevailing.
The cumene reaction always led to propene and benzene, which suggests the presence of
strong protonic acid sites. The highest activity showed the samples obtained in the presence of
sulphuric acid, whereas the samples crystallized with fluoride anions indicated the lowest
activity. Similarly as in the case of propan-2-ol decomposition, the modification with cations
such as H, A1, Ce and Zn resulted in a noticeable activity, while the sodium and calcium
forms appeared totally inactive (Fig. 9).
The spectral analysis of adsorbed pyridine did not provide an unambiguous evidence of the
presence of protonic acid sites. The spectra indicate only the bands at-1491 cm -~ and 1450
-1
cm , which suggests only the presence of the Lewis acid sites (Fig. 10). The spectral data are
not quite consistent with the results of cumene cracking, which involves a protonic acidity.
Some evidence on the presence of protonic centers in the zincosilicalites MFI gave
temperature programmed hydrogen evolution analysis [10]. We observed the evolution of
hydrogen gas as a result of the reaction the hydrogen form of samples with metallic zinc.
The obtained samples showed some activity in the propan-2-ol oxidation. The series prepared
with phosphates was distinctly more active than the others. It is surprising that modification
with Cu cations did not show significant activity (Fig. 11).
213
1614
1452 20
//l ...........................................................................................................................................................................................................................
1491 18
des 400~
16
~14
///des 300~ 12
"~ 10
, _ . / / d e s 200~ 8
r~
~ 6
<
~ ~ d e s 100~
, ,
H3PO4 H2C204 H2SO4 HF acid
50
9 5wt%Pt/H form
40
9 5wt%Pt/Zn form
9 5wt%Pt/Cu form
J
= 30
O
r~
20
~ " 5wt%Pt/Na form
9 a 9 9 9
10
~ ~ Q K ~ ' ~ * ~ ~ ~, " *- ~- 9 __ .. ! - --
0 | I i i ....... 1,. I i t | I
0 1 2 3 4 5
Time on stream (hour)
Fig. 12. Hydrodesulfurization of thiophene at 350~ Conversion versus time on stream
Table 2.
Products distribution in the HDS of thiophene (after 2 hours) over Pt supported on the
indicated cation modifications
Conversion Composition of h~cdrocarbon [%]
Catalyst [%] <C3 i-C4 n-C4 1-C4,+ i-C4, t-C4, c-C4, 1,3-C4,, >C5
Pt/H 23.0 3.3 0.9 41.5 16.4 21.5 16.3 0.0 0.1
Pt/Zn 20.9 1.2 0.3 51.3 11.9 20.2 15.1 0.0 0.0
Pt/Cu 5.7 10.8 0.5 24.0 18.8 26.4 19.3 0.2 0.0
,
Pt/Na 6.0 4.0 0.0 6.6 3.7 49.6 36.1 0.0 0.0
'- stands for olefins, "- stands for dienes
214
The cation modified samples containing platinum showed a considerable activity for HDS of
thiophene. The activity depended on kind of cation introduced. The Pt catalysts supported on
H and Zn modified samples indicated relatively high activity (over 20% of thiophene
conversion), whereas the samples based on Cu and Na showed four times lower activity. It is
clear that acidity of support facilitates the activity of platinum. Although the samples were
markedly less active than the conventional CoMo/AI203 catalyst, they showed considerably
higher contribution of paraffins in products (more than two times for the H- and Zn- modified
zeolite support). They also indicate relatively good stability in time. The C4 hydrocarbons
(Table 2) are the main reaction products (over 90%). Only in the case of Cu modified sample
the contribution of light hydrocarbons is more pronounced (-10%).
4. CONCLUSIONS
The zincosilicate MFI can be obtained from sodium zincate analogously as aluminosilicate
zeolites. The nature of the acids used for adjusting the pH of the initial gel affects the
crystallization process and the properties (including catalytic) of resulting samples. The
phosphates and sulfates acted as crystallization promoters, whereas HC1 and CH3COOH
totally disturbed a crystallization and no zeolite phase is obtained.
The obtained samples showed considerable activity in reactions requiring the acid sites
(propan-2-ol decomposition, 1-butene isomerization, cumene cracking) and red-ox centers
(e.g. propan-2-ol oxidation). The samples prepared with phosphates were always the most
catalytically active (only in cumene cracking the samples prepared with sulfates prevail).
Cations introduced into the zincosilicate MFI affected considerably the catalytic activity and
selectivity. The proton and polyvalent cations introduced similarly as in zeolites provide the
acid sites, which is reflected in higher activity. The acidity of the modified samples influences
also the activity of the Pt treated catalysts in the HDS reaction of thiophene.
REFERENCES
The ultramarine analogs with different coloration and different structure can be obtained by
means of thermal treatment of synthetic zeolite A modified with various cations mixed with
elemental sulfur and alkalis. The post-synthesis ion-exchange treatment of sodium forms of
ultramarine analogs can also modified to some extent their properties, although the results are
less spectacular than that attained upon the direct synthesis with various cations. The
ultramarine analogs with SOD structure are less susceptible to cation modification than the
colored products with retained LTA structure.
1. INTRODUCTION
the ultramarine analogs prepared from zeolite exclusively with sodium cations. The
ultramarine analogs with retained original LTA structure as well as the samples showing SOD
structure or that of nepheline hydrate II were treated with aqueous solutions of various cations
or were treated with respective salts at elevated temperature (solid state ion-exchange). The
commercial ultramarine was modified with the above methods for comparison.
2. EXPERIMENTAL
Zeolite NaA (donated by Atofina, Poland) was used as starting material for syntheses of
ultramarine analogs. The elemental sulfur (P.O.Ch. Poland) was employed as radical
precursor in all experiments and its content in the initial mixture always made 40 wt.% of
zeolite. The parent sodium zeolite A was modified with cations such as K +, Li +, NH4+, Ca 2+
by means of conventional procedure with aqueous solutions of respective salts (KC1, LiC1,
NH4C1, CaCI2). The 0.1M solutions were used for the cation modifications. The procedure
was repeated three times in order to attain a high exchange degree. NazCO3 was the principal
alkali source. Another alkalis applied were K2CO3, Li2CO3, CaO. The preparation procedure
comprised a mixing and grinding of zeolite with sulfur and alkalis. The alkalis content
(expressed as Na2/S or Me2/S if other cations applied) varied in the range of 0 - 1. The initial
mixtures were maintained in covered ceramic crucibles and heated for 2 hours in the furnace
in the range 5 0 0 - 800~ The samples were inserted into hot furnace. The products were
cooled down in desiccator and then washed with water and dried.
Another series of experiments comprised a post-synthesis modification of the
ultramarine analogs with various cations. Usually 3 g of the sample was treated at room
temperature with 30 ml of 0.1 M solutions of adequate salt. The procedure was repeated three
times with fresh solutions after each 2 hour period. The same experiment with commercial
ultramarine (provided by Prayon-Rupel) was conducted for comparison. The solid state ion-
exchange was carried out at elevated temperature with the salts of chosen cations. The
modified samples were washes with water and dried.
The characterization comprised the following methods: XRD (TUR-62M), UV-vis
(Varian-Cary 100), FTIR (Bruker-Vector 22), ESR (Radiopan), SEM (Philips-SEM 515)
Properties of the samples prepared from cation modified zeolites A and sodium carbonate (as
alkali source) are given in Table 2. The samples prepared from the mixtures of zeolites and
sulfur without any additional alkalis differ very much from the samples obtained from
unmodified NaA zeolites. Contrary to the latter, the products containing other alkali or alkali
earth cations are colorless. The sample based on NH4A zeolite undergoes total amorphization.
The amorphization is also noticed in the case of the products obtained from the Cu and Zn
modified zeolites. The brown color of the sample prepared from CuA results from generation
of copper sulfide. The admittance of some alkali (e.g. Na2/S = 0.2) results in coloration of the
samples, but the colors are different than that of the sample obtained from unmodified zeolite.
The structure of the products depends markedly on the cation introduced. For instance, LiA
leads to forming of SOD already under low alkalinity, whereas the presence of other cations
under study requires higher alkalinity to attain this structure. The differences in the properties
of the products obtained from modified and unmodified zeolites become less distinct for the
samples obtained from the more alkaline mixtures (e.g. Na2/S = 1.0). All the samples indicate
the SOD structure (although in some cases the impurities of unknown structures are
217
noticeable). The intensity of the color is rather low. The shades of blue and green are
prevailing. Only the sample modified with Cu remains brown. The declining influence of the
cations introduced into zeolites on the properties of the products with growing alkalinity
results from raising contribution of sodium (from Na2CO3) in the mixture. The samples
prepared with zeolite NaA contain substantial content of sodium even if any NazCO3 is not
added. The sample denoted as NaJS=0 contains as a matter of facts the amount of Na that
corresponds to real Na2/S=0.23. The ion-exchange degree in modified samples was always
-80%. For instance, the KA sample mixed with sulfur (Na2/S=0) contains some remnant of
Na and the N a / K ratio is 0.3. In the samples containing Na2CO3 the contribution of sodium
becomes prevailing (Table 1).
Table 1
Molar proportions of the mixtures based on zeolite KA.
Na2/S 0 0.2 0.4 0.6 0.8 1
Real Na2/S 0.05 0.25 0.5 0.7 0.9 1.1
Na/K 0.3 1.5 3 4 5 6.5
The differences in the products coloration are reflected in the UV-vis spectra (Fig. 1). In the
case of samples based on alkaline cation modified zeolites the presence of sulfur radical is
evident in ESR spectra (Fig. 2). The spectrum of the samples containing introduced Cu (or
Zn) show only the signal resulting from metal cation, whereas the signals typical for the sulfur
radicals are not seen.
1 36o !
NaA
Na2/S=I
..__..._--------
~ LiA
~
Naz/S=I
g=2,029 ~ f !
CuA ~
Na2/S=I
I I I I I I
200 300 400 500 600 700 800 900 300 305 310 315 320 325 330
Wavelength (nm) Field [mT]
Fig. 1. UV-vis spectra of selected samples Fig. 2. EPR spectra of the samples obtained
obtained from zeolites A modified with from zeolites A modified with indicated
indicated cations (Na2/S= 1). cations (Na2/S= 1).
218
The products of thermal syntheses at 500~ (Table 3) also depend on nature of cations
introduced into zeolites, although the thermal structure transformations are less pronounced
than those at 800~ (Fig. 3,4). For example, the NH4A mixed with sulfur (Na2/S=0) retains
the LTA structure after heating, while it undergoes an amorphization at 800~
The significant influence of heating temperature is conspicuous for the LiA based samples. It
has not attained the SOD structure even at high alkalinity (Na2/S-1), while at 800~ the
transformation to SOD was noticeable already after adding small amount (Na2/S=0.2) of
sodium carbonate.
Table 2
Color and structure of the products obtained at 800~ from the mixtures of cation modified
zeolites A, elemental sulfur and NazCO3.
Na2/S ' 1.0 0.8 ........ 0.6 0.4 0.2 " 0
NaA Light blue Blue Turquoise Green Green Yellow-green
LiA Light green Green, turq. Turquoise Blue Light blue White
KA Palett~.-gray L. turq.-gray Blue Blue Light yellow White
NH4A Light blue Blue Blue L. turquoise Yellow-green White
CaA L. turquoise L. turquoise L. turquoise Light green Light green White
CuA Brown Brown Brown Brown Brown Brown
ZnA Blue Blue
,,
Blue White White White
The influence of the cations on properties of the products is much more pronounced in
the syntheses involving the cation modified zeolites A and the alkalis containing the same
cations (Tables 4-6). The potassium containing mixtures lead to the products of rather low
color intensity. The yellow coloration prevails in most of samples. It is interesting to
underline that none of the product attained the structure of SOD after the thermal procedure.
219
Table 3
Color and structure of the products obtained at 500~ from the mixtures of cation modified
zeolites A, elemental sulfur and NazCO3.
Na2/S ratio 1.0 0.8 0.6 0.4 0.2 0
NaA Light green Green Green Dark green Blue White
LiA Light green Light green Turq.-green Green Green White
KA L. turq.- gray L. turq.-gray L. turq.-gray L. turq. Blue White
NH4A Light blue Blue Turquoise Green L. turq. White
CaA L. turq. L. turq. Blue Blue Blue White
The products obtained from the mixtures of high alkalinity (i.e. high potassium content) show
a considerable contribution of kaliophilite in the products. The nepheline hydrate II is another
structure detected. The high contribution of yellow chromophore (i.e. $2-, g-2.002) in the
potassium modified samples can result from larger size of K + cations than Na + ions (Fig. 5,6).
The bigger cations occupy considerable part of the B-cage, thus the smaller $2- anion-radicals
can be favored better than $3 in accommodation inside the sodalite cages.
NaA +
[ / 600
Na2CO3
An \ NaA+ r
J KA+
K2CO3
LiA +
Li2C..____
O3
g I 400 600
~ LiA + ~ ~ g=2,002
g=2,029
I I I I I I I I I
200 300 400 500 600 700 800 900 300 305 310 315 320 325 330
Wavelength (nm) Field [mT]
Fig. 5. UV-vis spectra of selected samples Fig. 6. EPR spectra of selected samples
obtained at 800~ with Me~/S=0,4. obtained at 800~ with MeJS=0,4.
Table 4
Properties of the products obtained from KA, sulfur and K2CO3.
K2/S ratio 1.0 0.8 0.6 0.4 0.2 0
500~ Pale yellow Pale yellow Pale yellow Pale yellow Pale green White
Structure N+K N+K N+K N LTA LTA
800~ Pale yellow L. yellow L. yellow Yellow Yellow Pale yellow
Structure N+K N+K N LTA + N LTA LTA
N-nepheline hydrate II, K- kaliophilite
220
The samples prepared from LiA with lithium carbonate as alkali source show a green
coloration at 500~ and blue at 800~ The blue coloration (800~ reflect the predominant
contribution of $3- (g=2.029), which can result from small diameter of Li + cations and
subsequently larger room for bulkier sulfur anion radicals. The green coloration of the
samples prepared at 500~ can result from considerable content of elemental sulfur in the
products, that acts as yellow chromophore (Fig. 5,6).
The presence of lithium affects very much the structure transformation of zeolite. Contrary to
the mixtures containing only sodium cations the structure of SOD is not formed. The LTA
structure is retained (with admixture of nepheline hydrate II) upon heating at 500~ The
nepheline hydrate is prevailing structure of the products obtained at 800~ Only in the
samples of the lowest and the highest alkalinity show structure of carnegiete or unknown
structure, respectively. No sodalite is formed at 800~ either.
Table 5
properties of the samples obtained from LiA and Li2CO3.
Li2/S ratio 1.0 0.8 0.6 0.4 0.2 0
500~ L. green,-gray L. green- gray L. green- gray L. green L. green White
Structure LTA + N LTA + N LTA + N LTA + N LTA LTA
800~ White Light blue Blue Blue Blue White
Structure U N+U N N N C
N- nepheline hydrate II, C-carnegieite, U-unknown
The calcium modified zeolites mixed with sulfur and alkalized by CaO do not lead to any
colored products. The structure of zeolites does not undergo any transformations at 500~
(only some crystalline impurities are noticeable in the sample of highest CaO content). The
LTA structure is also retained at 800~ Only the sample without any CaO undergoes
amorphization and some impurities are seen in the samples of high alkalinity (Tab. 6).
Table 6
Prope.rties of the products obtained from CaA, sulfur and CaO.
Ca/S ratio 1.0 0.8 0.6 0.4 0.2 0
500~ White White White White White White
Structure LTA + U LTA LTA LTA LTA LTA
800~ White White White White White White
Structure LTA + U LTA + U LTA + N LTA LTA Amorph.
It is likely that divalent cations prevent a forming of sulfur anion-radicals. One could consider
that negative charge of the anion radical could be compensate by one positive charge of Ca +2
cation, whereas the other one could be bond to the framework anion (Fig. 7A). It seems,
however, that probably both positive charges are compensated by sulfur anion-radicals. The
close vicinity of two radicals combined with calcium results in their recombination (Fig. 7B).
The above results indicate a significant influence of the cations employed for thermal
syntheses from zeolite A. Both the coloration and the structure of resulting products depend
markedly on nature and concentration of the cations.
221
S~ S--S ,,,S~
A \
/
S Ca
/ \
S ....S -S"
S"
o O\ca/O\ 2+
o
\s, /
/\ /\ /\
si/
c.
/ S--S~s
I
Q QQ 0 Q Q
/s
S ,S"
The color of ultramarine can be modified by means of post-synthesis treatment and the
industrial production comprises such a modification in order to prepare the pink and violet
ultramarine. The SOD structure of ultramarine is not suitable for cation-modification. The
colored products prepared from zeolites can attain the SOD structure, nevertheless, they can
also show the more opened structures such as nepheline hydrate II or of original zeolite
(LTA). The latter structure are much more susceptible to a cation modification and the
influence of the introduced cations on the product properties can be more conspicuous.
The ion-exchange modification of the commercial ultramarine with most of applied cations
causes only very minor changes in coloration of the pigment. Only the treatment with
Fe(NO3)3 brings about some gray shade of the products (Table 7).
Table 7
Results of post synthesis ion-exchange process with aqueous solutions, of respective salts.
Cation Parent Li K Zn Co Cu Ca Fe NH4
Na
Color Blue Blue Blue Blue Blue Blue Blue Blue, Blue
Comm. gray
Structure SOD SOD SOD SOD SOD SOD + SOD Poor Poor
U SOD SOD
Color Olive Olive Olive Olive Olive Olive Olive Brown Olive
green green green green gray gray green green
Structure LTA LTA LTA LTA LTA LTA LTA Poor LTA
LTA
Color Turq. Turq. Turq. Turq. Turq. Green- L. turq. Brown
Turq.
gray gray
Structure N N N N N N N Amorph. N
The contact with acidic solution of the ferrous salt (pH-3) results in a noticeable evolution of
HzS, which evidently suggests some deterioration of ultramarine, although the XRD still
exhibits the SOD structure. The brown color of the suspension above the layer of modified
ultramarine indicates that iron sulfide is formed. The results of the modification of the SOD
ultramarine obtained from zeolite A (not indicated in the Table) were very much alike those
of commercial ultramarine. Neither the coloration nor the structure of the colored material
with LTA structure were considerable affected by the contact with alkaline cations. The
transition cations caused more pronounced color changes, but the LTA structure was still
retained. The modification with iron resulted in brown coloration and retained LTA structure.
The structure of nepheline hydrate was not affected markedly by the cation treatment. Only
the iron cation caused an amorphization. The color of the sample was drastically changed
222
after treatment with iron and somewhat with copper. The products modified with the other
cations did not differ noticeably from the parent sample.
The solid state modification affects the properties of ultramarine noticeably when transition
metal cations were used. The SOD structure was considerably affected by treatment with Co,
Cu, Fe. The crystallinity of the latter samples declined and some unknown phase was formed.
The modification of ultramarine analogs of LTA or nepheline hydrate structures with alkaline
cations did not cause any distinct color changes, neither the structure transformations. The
transition metal cations resulted in more substantial color changes and they caused a
noticeable amorphization of the samples.
4. CONCLUSIONS
The presented results show that the color and structure of the ultramarine analogs prepared
from zeolites A can be changed substantially by means of modification of the initial mixture
composition with various cations. The cation modification of the parent zeolite affects the
product properties very considerably in the case of the lowest alkalinity of the initial mixture
(i.e. containing only zeolite and sulfur). Such low alkaline mixtures form the colored
ultramarine analogs only in the case of the unmodified zeolite (i.e. the sodium form). The
coloration of Cu containing mixture resulted rather from generation of CuS. The samples
prepared from alkaline cation modified zeolite show the retained LTA structure, while those
containing Zn, Cu and NH4+ lose their crystallinity. An increase in alkalinity of the starting
mixtures by means of admitted NazCO3 diminishes the influence of cations introduced into
zeolites, because the contribution of Na cations become predominant. The presence of Cu
cations allows to form the SOD structure, but the sulfur radicals are not generated. The
mixtures containing cation modified zeolites, sulfur and alkalis other than sodium carbonate
(i.e. Li2CO3, KzCO3, CaO) resulted in products much different than the mixtures containing
only sodium cations. The size of alkali cations influence the color of the products. The bigger
potassium cations seem to favors forming smaller $ 2 (yellow) radicals, whereas smaller
lithium cations facilitates generation of the bigger $ 3 radicals. The divalent calcium cations
probably prevent generation of sulfur radicals and they probably facilitate their recombination
by forming the ordinary sulfides. It is interesting to emphasize that the SOD structure was
attained only when the thermally treated mixture contained sodium cations. The role of
sodium seems crucial in the thermal recrystallization towards sodalite.
The post-synthesis cation modification of ultramarine and its analogs obtained from
zeolites is usually less efficient then the direct synthesis with contribution of chosen cations.
The cation exchange with alkali metal salts does not affect markedly neither the commercial
ultramarine nor the colored sodalite obtained from zeolite A. The transition metal salts affect
the coloration and to some extent also the structure of the products. It results from partial
deterioration of the original structure and forming the respective sulfides. The solid state
modification seems to be more efficient procedure than the solution treatment.
REFERENCES
aNational Institute of Advanced Industrial Science and Technology, AIST Tohoku, Sendai,
983-8551, Japan
bSchool of Materials Science, Japan Advanced Institute of Science and Technology, Nomi,
923-1292, Japan
New high-silica zeolite CDS-1 (Cylindrically Double Saw-Edged zeolite; Framework Type
Code (FTC) 9 CDO) was prepared by dehydration-condensation from novel layered silicate
PLS-1 (Pentagonal-cylinder Layered Silicate) intercalating tetramethylammonium hydroxide
between silicate layers. Synthesis conditions of PLS-1 was optimized by using SiO2-
TMAOH-1,4-dioxane-H20 system without any alkali metal ions.
1. INTRODUCTION
Three-dimensional layered compounds have been interested for example AMH-3[1] and
RUB-18[2], which have zeolite-like topology. A number of layered silicates, which can be
transformed into zeolites, are reported e.g. PREFER[3] and EBR-1 [4]. Recently, the synthesis
of zeolites by the structure conversion method using novel layered silicates with new type
topology have been reported e.g. from ERB-1 to MWW type, RUB-18 to RWR[5] type,
Nu-6(1) to novel zeolite Nu-6(2)[6] and PREFER to Ferrierite (FER type zeolites), respectively.
In these examples, newly micropores are constructed in an interlayer by a topotactic structural
conversion. This method has some possibilities for synthesis of zeolite with novel framework
topology by various combinations among precursor materials. This naturally leads to the
realization of a molecular engineering that can design zeolite frameworks by assembling the
precursors in dehydration-condensation or other modifications.
We have been investigated that topotactic conversion using layered materials as precursor.
However, there has been no perfect example by this kind of molecular manipulation so far,
although there are some syntheses intending it till fairly recently. For example, we have
clarified the transformation of layered silicates, kanemite and fl-HLS, to zeolite MFI and SOD,
respectively[7-9]. Recently, we reported the synthesis condition and characterization of a
thermostable zeolite with novel framework CDS-I[10] [FTC: CDO] obtained by a molecular
manipulation, or a topotactic dehydration-condensation between the layers of a new silicate
PLS-I[10] with layer structure consisting of face sharing of 5-MR cylinders in detail.
Framework of CDS-1 is very similar to that of FER, in other word, only a stacking period of
5-MR cylinder layers is different. PLS-1 can be synthesized with the SiO2-TMAOH-
1,4-dioxan-KOH- H20 system. However, the crystallization conditions of PLS-1 were very
narrow i.e. the concentration of KOH strongly affects the phase of the obtained products even
though the amount of K § ions in starting gel is very small. Moreover, it is not easy to estimate
the position of K § ions as the position of K § ions estimated by the structural analysis was too
224
close to T site in the CDS-1 framework and an adsorbed water molecule can substitute for K §
ion.
In this work, we succeed the synthesis of PLS-1 without the presence of K + ion and
optimization of topotactic structural conversion by a dehydration-condensation. Crystal
structures of the PLS- 1, mPLS- 1(modified layered silicate PLS- 1) and CDS- 1 zeolite after
condensation are also investigated.
2. E X P E R I M E N T A L
2.1. Synthesis
Layered silicate PLS-1 without K content was crystallized by the hydrothermal synthesis.
Synthesis conditions are following; 10.0 g of silica powder (Cab-O-Sil M5) (167 mmol), 13.6 g
of 26 wt % TMAOH (38.8 mmol), and 50.0 g of 1,4-dioxane were suspended in 34.3g of
distilled water consecutively. The mixture gel was vigorously stirred for six hour at room
temperature by a magnetic stirrer, and heated for 7 days with seed crystal or 14 days without
one, in a stable PTFE-lined stainless-steel autoclave at 423 K under autogenous pressure. The
crystalline product was filtered and washed with acetone and distilled water, then dried at 343
K in a dried oven for 12 h. Prior to convert PLS-1 to CDS-1 by dehydration-condensation,
PLS-1 was washed and protonated with 0.1 N NH4NO3 or 0.1 N HC1 aqueous solution at 60~
for 4 days (mPLS- 1). The mPLS- 1 was heated from 673 K to 1173 K for 100 K per hour under
evacuation ranging in the degree of vacuum from 10-i to 10-6 Pa and zeolite CDS-1 was finally
obtained.
2.2. Characterization
Obtained products were identified by XRD (MacScience M2 IX) and characterized by
TG-DTA (MacScience TG-DTA2000), ICP-AES analysis (Seiko instruments SPS-1500R), Ar
gas adsorption (Quantachrome AutoSorb-lMP). In order to obtaining information of local
coordinates of framework, 29Si MAS NMR spectra were measured on a Bruker AMX-500
recorded at room temperature at 99.362 MHz and 90 ~ pulses. For quantitative spectra the
HPDEC pulse sequence was used. The recycle delay was 300 s for mPLS- 1 and 30 s for CDS- 1,
respectively, with sample rotor spinning at 6 kHz. 29Si CP/MAS NMR spectra ofmPLS-1 were
also measured on a Bruker AVANCE400WB at room temperature, with sample rotor spinning
at 7 kHz. High-resolution transmission electron microscope(TEM) images of CDS-1 were
obtained on a JEOL JEM-4000EX operated at 400 kV.
._...._ 1073K
923K
L~ 848K
773K
698K
623K
5 10 15 20 25 30 35 40
20 /~
Fig. 1. XRD patterns of as-made PLS-1, mPLS-1 and condensed mPLS-1 at each heating temperature.
Arrows indicate impurity peaks.
Fig. 2 shows 29Si MAS NMR spectra of PLS-1, mPLS-1 and CDS-1. Three resonance peaks
characteristic of two HOSi(OSi)3 (Q3) and Si(OSi)4 (Q4) local coordination around-102 ppm,
-106 and-114 ppm, respectively, are clearly observed. On the other hand, CDS-1 exhibits a
broad peak around-113 ppm and small shoulder around-103 ppm, which is practically due to
Q4 structures and possibly contain Q3 slightly. This indicates that during the conversion of
mPLS-1 to CDS-1, silicate layers with open-framework structure lose their terminal silanol
groups by the dehydration-condensation to turn to a closed tetrahedral (SiO4)n network
226
structures. Small amount of Q3 structure is arisen from two reasons; one is residual terminal
silanols caused by imperfect dehydration-condensation and another is contribution of terminal
silanols on outer surface of crystalline.
From 1H-29Si cross polarization (CP) MAS NMR spectra in Fig. 2, the intensity of the
signal of the Q3 i n c l u d i n g - O H group is enhanced for short contact times, supporting the
interpretation of the spectrum as characteristic for a layered structure. Furthermore, intensities
of the Q3 peaks are greatly enhanced in mPLS-1 with narrow peak width. This fact indicates
that some proton exchanges were occurred on a surface of silicate layer.
Q4 t-112.77)
I CDS-1 mPLS-1
Q3 (-106.39) i
Q4 (-114.27)
PLS-1
_g
PLS-1
-80 -90 -100 -110 -120 -130 -140 -80 -90 ' .1~0 ' .fi0 ' -1~o ,13o -i40
Chemical shiht / p p m Chemical shiht / p p m
Fig. 2. 298i MAS NMR spectra (left) of PLS-1 and CDS-1 and 298i CP/MAS NMR spectra (right) of
PLS- 1 and mPLS- 1, respectively.
From Ar gas adsorption isotherm curves, CDS-1 is much higher in the volume of Ar
adsorbed than mPLS-1. The multipoint Brunauer-Emmett-Teller (BET) surface area and the
Saito-Foley cumulative pore volume for CDS-1 were calculated to be 338 m 2 g-1 and 0.14 ml
g -1 from the argon adsorption isotherm curve, respectively. These values are somewhat larger
than those values of original CDS-1 converted from PLS-l(with K + ion). In addition, the
minimum in the pore size distribution calculated from the argon adsorption isotherm data using
non-local density functional theory[14] was 0.488 nm, which is in good agreement with the
pore diameter of general pore size of 8-MR.
On the other hand, refined lattice parameters for CDS-1 were a = 1.8401(2) nm, b =
1.3989(14) nm, c = 0.7385(5) nm based on space group of Pnma. Structural conversion from
mPLS-1 to CDS-1 occurred at 693 K which is consistent with the combustion temperature that
of TMAOH by the thermogravimetric analysis. Crystallinity of CDS-1 is moderate, therefore,
crystal structure could not solve uniquely from the XRD data. We attempt building an initial
structure model using a molecular dynamics (MD) simulation based on the framework structure
of PLS-1. Detailed of the MD calculations were described in ref. 10.
?MAOH
Fig. 3. Structural viewing of layered silicate PLS-1 along [010] direction (left) and its enlarged
illustration along [001] direction (right).
8-MR
"~'b
0.31 nm ~~
a
8-MR
Fig. 4. Structural viewing of novel layered silicate CDS-1 (left) and effective pore size of two 8-MR
along [001] and [010] directions (right).
228
Fig. 4 shows structural drawing for the CDS-1 and its two 8-MR windows after the
Rietveld refinement. The pore openings of CDS-1 are two 8-MR with dimensions of 0.46 x
0.27 nm along [001] and 0.47 x 0.31 nm along [010]. Finally, R-factors were sufficiently low,
Rwp = 4.70 %, RI = 1.21%, RF = 0.99 % and Re = 2.72 % for PLS-1, and Rwp = 1.86 %, R1 =
0.75 %, RF = 1.02 % and Re = 1.27 % for CDS-1, respectively. Average bond distances, l, and
bond angle, ~b, were l(Si-O) = 0.159 nm for mPLS- 1 and 0.158 nm for CDS- 1 ~ O - S i - O ) --
109.41 ~ for PLS-1 and 109.42 ~ for CDS-1, respectively. Obtained values are consistent with
those of other high-silica zeolite. Difference plots of both mPLS-1 and CDS-1 obtained by the
Rietveld refinements were illustrated in Fig. 5.
IO0000F
I
80000~
15000
60000 i
;~10000
5001
200001-
Fig. 5. Observed, calculated, and difference patterns for Rietveld of the XRD data for mPLS-1 (a) and
the XRD data for CDS-I(b) converted from mPLS-1.
High-resolution TEM image along [010] direction for original CDS-1 shows clearly
micropores in Fig. 6. This image indicates that silicate layers were shifted ca. 1/3 c at a time and
condensed each other. Structure model of CDS-1 was interpreted as a monoclinic system
because of an sharp angle of the crystal edge was ca. 98 ~ as shown in Fig. 6, which corresponds
with an initial estimated model from the MD simulation [ 10].
: r
,,
Fig. 6. HRTEM image of zeolite CDS-1 along [010] direction (left) and structural drawing of
monoclinic CDS-1 (right) along same direction.
229
•• e.g. ITQ-6
b/2 shift
Dehydration-condensation
Fig. 7. Structural similarity between PLS-1 and PREFER, and between CDS-1 and Ferrierite,
respectively. Four materials have a nano-part composed by a pentagonal-cylinder layer in common.
The pentagonal-cylinder layer can be preparation by delamination of PREFER.
230
4. C O N C L U S I O N S
A layered silicate PLS-1 without K + ion was successfully synthesized by increasing the amount
of TMAOH of ca. 25 mmol instead of the same quantity o f K + ions in the mixture gel. Terminal
silanols ofPLS- 1 are increased by the acid treatment. Narrower CP/MAS spectra suggested that
local atomic ordering of framework has highly symmetry. Then, the structural conversion from
mPLS-1 to CDS-1 was hereby stabilized without degrading crystallinity in the
dehydration-condensation process with a rapid programming rate. The synthesis of high-silica
zeolite CDS-1 can be optimized by the modification of PLS-1 with acid treatment. We have
investigated the preparation and optimization of the topotactic conversion from such a
precursor materials to zeolite material. Several application of CDS-1 is also now in progress,
e.g. preparation of a zeolite membrane, catalysis and various metal atom substitution.
Acknowledgments: This work is financially supported from New Energy and Industrial
Technology Development Organization of Japan (NEDO). TEM observation was supported the
"Nanotechnology Support Project" of the Ministry of Education, Culture, Sports, Science and
Technology (MEXT), Japan
Additional information: Recently, some new compounds that have almost same structure of
CDS-1/PLS-1 were reported, e.g. MCM-65[16], UZM-25[17] and ERS-12[18]. Furthermore,
very recently Wang et al. reported new framework type of RRO i.e. zeolite RUB-41 prepared
by calcination of unknown layered silicate RUB-39[19].
REFERENCES
[ 1] H. Jeong, S Nair, T. Vogt, L. C Dickinson, M. Tsapatsis, Nat. Mater. 2 (2003) 53.
[2] S. Vortmann, J. Rius, S. Siegmann, H. Gies, J. Phys. Chem. B, 101 (1997) 1292.
[3] L. Schreyeck et al., Microporous Mater., 6 (1996) 259.
[4] R. Millini et al., Microporous Mater., 4 (1995) 221.
[5] B. Marler, N. Str6ter, H. Gies, Recent Research Reports, 14th Int. Zeol. Conf. (2004) pp. 15-16.
[6] S. Zanardi, A. Alberti, G. Cruciani, A. Corma, V. Forns, and M. Brunelli, Angew. Chem. Int. Ed.,
43 (2004) 4993.
[7] M. Salou, Y. Kiyozumi, F. Mizukami, P. Nair, K. Maeda and S. Niwa, J. Mater. Chem., 8 (1998)
2125.
[8] T. Ikeda, Y. Akiyama, F. Izumi, Y. Kiyozumi, F. Mizukami and T. Kodaira, Chem. Mater., 13
(2001) 1286.
[9] Y. Kiyozumi, F. Mizukami, A. Akiyama, T. Ikeda nad T. Nishide, Stud. Surf. Sci. Catal., 135H
(2001) 191.
[ 10] T. Ikeda, Y. Oumi, Y. Akiyama. A. Kawai and F. Mizukami, Angew. Chem. Int. Ed., 43, (2004) 4892.
[11 ] F. Izumi, S. Kumazawa, T. Ikeda, T. Ida, Powder Diffraction (Ed.; S. P. Sen Gupta) Allied Publ.,
New Delhi, pp. 24-36 (1998).
[12] F. Izumi and T. Ikeda, Mater. Sci. Forum, 321-324 (2000) 198.
[ 13] F. Izumi and R. A. Dilanian, Recent Research Developments in Physics, Vol. 3, Part II, Transworld
Research Network, Trivandrum, 2002, pp. 699-726.
[14] R. Evans, J. Phys. Condens. Matter., 2 (1990) 8989.
[15] A. Corma, U. Diaz, M. E. Domine and V. Forn6s, J. Am. Chem. Soc., 122 (2000) 2084.
[16] D. L. Dorset and G. J. Kennedy, J. Phys. Chem. B, 108 (2004) 15216.
[ 17] R. W. Broach, see framework code CDO on http://www.iza-structure.org/databases/.
[ 18] R. Millini, L. C. Carluccio, A. Carati, G. Bellussi, C. Perego, G. Cruciani, S. Zanardi, Microporous
Mesoporous Mater. 74 (2004) 59.
[19] Y. X. Wang, H. Gies, B. Marler and U. Mfiller, Chem. Mater., 17 (2005) 43.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 231
Large zeolite SOD single crystals, of the largest size about 70 •215 m, were synthesized
using a near non-aqueous method. The characterizations of zeolite SOD and synthesis
conditions were explained. Their mechanical properties have been measured by
Instrumented-nano indentation testing that can accurately measure the mechanical properties
of small volumes of materials. The load-displacement curve, hardness-displacement curve and
modulus-displacement curve of zeolite SOD were obtained. The values of the hardness and
the elastic modulus were about 6GPa and 40GPa, respectively. The mechanical properties of
zeolite SOD were compared with SiO2 standard sample.
1. INTRODUCTION
Zeolites are crystalline microporous materials that are attracting great attention owing to their
applications in high-performance catalysis, ion exchange, gas separation and membrane
separation processes of chemistry industry [1]. They are also be utilized in nuclear waste
disposal, chemical sensing, pollution abatement and media for hydrogen storage [1] [2]. In
addition, zeolites are often used as novel optical electronic and magnetic materials [3].
Zeolites have some unusual physical properties. Woodcock et al. [4] have found the
negative thermal expansion in the siliceous zeolites chabazite and ITQ-4. Dong et al. [5] have
reported that zeolite faujasite (FAU) may adsorb microwave energy effectively.
Although many properties of zeolites have been thoroughly investigated, very little study
has been performed systematically on their single-crystal mechanical properties, primarily
because of the problem of being difficult to obtain zeolites samples of larger size single
crystals (100 ~tm or larger). Furthermore, standard mechanical testing procedures can not be
applied to measure their mechanical properties for very small size samples (5mm or less). It is
worth knowing the mechanical properties of zeolites well for the strength design in their
comprehensive applications. For example, catalysis can be re-used many times. In many
applications continuous uses decrease catalytic activity because of degradation of the crystal
structure. In addition, heterogeneous catalysts are exposed to strong compression and shear
232
stresses during the formation of the final catalyst pellet, either through size reduction
operations, palletizing or during the extrusion of the final product [6]. Zeolites are subjected
to ball milling as a means of selectivity enhancement for base catalyzed reactions [7].
Z. Wang et al. [8] researched mechanical properties of zeolite single crystal ZSM-5 (MFI)
in 2002. They measured elastic modulus of zeolite ZSM-5 of average size of 500 microns
using a micro-deformation tester of their own. The materials were seen to have an average
elastic modulus of 4GPa.
Nanoindentation test is the most commonly applied means of testing the mechanical
properties of very small volumes of materials. It employs high resolution sensors and
actuators to continuously control and monitor the loads and displacements on an indenter as it
is driven into and withdrawn from a material within the scope of the respective
micro-Newtons and nanometer. One of the great advantages of the technique is that hardness
and elastic modulus can be determined by analyses of the indentation load-displacement data.
The instrument has a very high precision in testing respect. It can be used as long as the
dimensions of investigated materials are no less than 20 microns. Shapes of the samples are
unlimited. Many materials such as aluminum, tungsten quartz, soda-lime glass, fused silica,
sapphire etc have been measured by means of the indentation experiment [9]. However the
technique has not been applied in the determining mechanical properties of zeolites yet.
Zeolite sodalite (SOD) is microporous tectosilicates of the general composition Na8
[A1SiO4] 6. It contains a three- dimensional network of 6-MR pores (2.6 • 2.6A) intersecting
channels. The sodalite structure exhibits a framework of cubic symmetry formed by a strictly
alternating connection of corner-sharing SiO4 and A104tetrahedra. This linkage gives rise to
the space filling array of truncated octahedral cages, called [3-cages in zeolite chemistry [ 10].
Zeolite SOD is a good optical, electronic and magnetic material. For example, it has
photochromic and cathodochromic properties at vacant halide sites in the center of the cage
[ 11 ]. Zeolite SOD may be applied in host matrices for quantum dot materials in future [ 11 ].
In this paper, large zeolite SOD single crystals were prepared by near non-aqueous
method and XRD, SEM were used to characterize their properties. Synthesis conditions of
zeolite SOD were studied. The mechanical properties of zeolite SOD were measured using
Nano Indenter XP produced by MTS company in American. The hardness and elastic
modulus were gained.
2. EXPERIMENT METHODS
measuring the phase of samples. A scanning electron microscope (JEOL jSM35C) was used
for observing the morphology and size of zeolites SOD single crystals. FT-IR (Perkin-Elmer
1700) recorded the framework vibration of typical samples.
Table 1
Different zeolite SOD crystalline sizes with different HF/Si
Phase HF/Si Size (/xm)
SOD 0.33 60x60x60
SOD 1.33 70x70x70
SOD 2 35x35x35
SOD 2.67 20•215
234
10000
It)
7500
0
>,, 5000
0O
t-
(1) 2500
t-
J[ t _
10 20 30 4'0
2 Theta(degree)
Fig. 1. As-synthesized samples of zeolite SOD single crystals
were smaller than those of A. Kuperman, because HF had more water in synthesis system.
The reason was that increasing the amounts of nuclear and faster dissolution velocity of SiO2.
But the samples can still meet the need of testing mechanical properties of zeolite SOD using
nanoindentation. Different zeolite SOD crystalline sizes were listed in the Table 1 with
different HF/Si.
.......==?~=~.~ ...~..i~-=?~,~,
~
Fig. 2. The image and framework structure of typical sample zeolite SOD, (a) The image of zeolite
SOD (b) The framework structure of zeolite SOD
235
100
150
8O
120
~" 60
E ~" 90
-o 40 E
O
~" 60
" 20 O
30
J .
0
0 300 600' 900 1200 00 360 6 6 0 - 9 6 0 i2'00
Displacement(nm) Displacement(nm)
Fig. 3. Load-displacement curves of zeolites SOD and standard sample SiO2, (a) Load-displacement
curve of zeolite SOD, (b) Load-displacement curve of SiO2
236
40 120
100
"~'30
n n 80
(.9 (.9
"~20 --~ 60
i-"
"5
"O
40
~L_ 10
o 20
"r"
0 0
0 200 400 600 800 1000 0 2()0 4 0 0 600 8()0 10'00
Displacement(nm) Displacement(nm)
modulus. With CSM technique employed [14,16], the stiffness was measured continuously
during the indentation, i.e., at every point that a load-displacement data pair was taken and the
stiffness was directly measured.
Fig 4(a) and (b) showed hardness and modulus values of zeolite SOD sample with the
increasing indentation depth, respectively. In Fig. 4 (a), the hardness was 36 GPa when depth
was 0. On the other hand, hardness was the greatest when the Berkovich indenter was just
contact the sample. As soon as the indenter sank into the sample, hardness suddenly
descended to 2GPa. With the indenter was driven into the sample, it gradually added to 6GPa
until indentation depth was 100nm. After this, its value kept constant of about 6GPa. The
most common explanation was due to the surfaces of the samples have been hardened by the
25 250
~2o ... 200
a.
(.915 n
~t) '
(.9 150
r-- = 100
"O
~- 5
I
o 50
0
0
260 460 6()0 860 10'00 260 400 600 8()0 10'00
Displacement(nm)
Displacement(nm)
polishing process when the sample was prepared. The Berkovich indenter did not measure its
real hardness until indentation depth was 100nm. Fig 5(a) shows hardness experiment data of
SiO2 standard sample. It is clear that hardness values do not change when indentation depths
increase. It keeps constant of 10GPa.
As is shown in Fig 4(b), elastic modulus of zeolite SOD descented gradually from about
45GPa to 40GPa when displacements from 0nm to 600nm. It can be seen that the elastic
modulus kept constant of about 40GPa after 600nm. There was a resemblance reason between
the hardness and elastic modulus. The Berkovich indenter can measure their real elastic
modulus in certain values of displacement. To examine carefully modulus-displacement curve
of SiO2 (see Fig. 5(b)), it is found that elastic modulus of SiO2 is also invariable and its value
is 100GPa. Its value is larger than zeolite SOD'.
Besides zeolite SOD, the mechanical properties of zeolite FER single crystals were also
measured after they were synthesized From load-displacement curve of zeolite FER, the
final depths after unloading were 770nm, but for zeolite SOD it was 400nm (see Fig.3 a). It
indicates that zeolite SOD has larger elastic recovery than zeolite FER. In addition, the
hardness and elastic modulus values of zeolite FER are about 1GPa and 10GPa, respectively.
They are all smaller than zeolite SOD'.
The main reason that the three kinds of materials show different hardness may be
analyzed from their crystal framework structure. There are micro-pores in single crystals of
FER and SOD. But FER's pore (4.3x5.5A)is larger than SOD's(2.6x2.6A).Contrarily, SiO2
was dense structure. A material with larger pores may be softer. A sample with dense structure
may be harder. This may be the reason why hardness of FER is the smallest and hardness of
SiO2 is the largest in the three samples. In general, a harder material has larger elastic
modulus. The reasons of SOD with better elastic property and FER with better plastic
property need further research from crystal framework structure or in other aspect.
4. CONCLUSIONS
In some chemical applications large zeolite single crystals are required. Using a near
non-aqueous synthetic method, large zeolite SOD single crystals with the largest dimension of
70x70x70#m were prepared and their characterizations were mentioned. Nanoindentation
tests are perhaps the most commonly applied means of testing the mechanical properties of
materials in very small dimensions.
The mechanical properties of large zeolite SOD single crystals were measured using
universal instrument--MTS Nano Indenter XE Load-displacement curve,
hardness-displacement curve and modulus-displacement curve of zeolite SOD have been got.
The hardness and elastic modulus values are 6GPa, 40GPa, respectively. As mentioned in
analyses of mechanical properties, the hardness and elastic modulus of zeolite SOD are larger
than zeolite FER'. So the crystal structure of zeolite SOD is faster than zeolite FER'. Zeolite
SOD keeps better chemical properties than zeolite in chemical applications.
Nanoindentation technique is a potential method researching mechanical properties of
zeolites and expects to fill in basic mechanical databank of more zeolites.
238
REFERENCES
[1] E.C. Moloy, L. R Davila, J. F. Shackelford and A. Navrotsky, Microporous Mesoporous Mater.,
54 (2002) 1.
[2] J.Weitkamp, M.Fritz and S.Ernst, Int. J. Hydrogen Energy., 20 (1995) 967.
[3] G..A.Ozin, A.Kuperman and A. Stein, Angew. Chem. Int. Ed. Engl., 28 (1989) 359.
[4] D.A. Woodcock and Philip Lightfoot, Chem. Mater., 11 (1999) 2508.
[5] J.X. Dong, L. L. Xie, H. Xu, F. Wu and J. Hao, Chin. Chem. Lett., 13 (2002) 91.
[6] Z.Wang, R.F.Lobo and J.Lambros, Microporous Mesoporous Mater., 57 (2003)1.
[7] J.H. Xie and S. Kaliaguine, Appl. Catal A,148 (1997) 415.
[8] Z.Wang, J.Lambros and R.F.Lobo, J. Mater. Sci., 37 (2002) 2491.
[9] W.C. Oliver and G. M. Pharr, J.Mater. Res. 7 (1992) 1564.
[10] J.-Ch. Buhl, Th. M. Gesing, I. Kerkamm and Ch. Gurris, Microporous Mesoporous Mater., 65
(2003) 145.
[11] Z. Li, M.V. NevittandS. Ghose, Appl. Phys. Lett., 55 (1989) 1730.
[12] A. Kuperman, S. Nadimi, S. Oliver, G. A. Ozin, J.M. Garc6s and M. M. Olken, Nature., 365
(1993) 239.
[ 13] M.M.J.Treacy and J.B.Higgins (eds.), Collection of Simulated XRD Powder Patterns for Zeolite,
Elsevier, Amsterdam, 2001
[14] G.M. Pharr., W. C. Oliver and F. R. Brotzen, J. Mater. Res., 7 (1992) 613.
[15] T.H. Zhang andY. M. Yang, Chin. Adv. Mech., 32 (2002) 349.
[ 16] J.B. Pethica and W. C. Oliver, Phys. Sci., 19 (1987) 61.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 239
aState Key Laboratory of Coal Conversion, Institute, Chinese Academy of Sciences, Taiyuan
030001, Shanxi, P. R. China
Nonasil single crystals (200•215 with perfect morphology have been successfully
synthesized from HF-SiO2-N,N-dimethylpiperidine chloride-EtaN system. Compared with
other methods, NON prepared in this system has larger size and more regular morphology.
The NON single crystals were characterized by XRD, SEM, FT-IR, 29Si NMR spectroscopy.
The factors that influence the morphology of NON crystals during the crystallization have
been studied. The experiments confirmed that N,N-dimethylpiperidine Chloride was a
template for preparing nonasil in a near non-aqueous system.
1. I N T R O D U C T I O N
The synthesis of large zeolite single crystals is of extensive interest for a large number of
requirements, including single-crystal structure analysis, studies of zeolite crystal growth
mechanism, studies of adsorption and diffusion, the determination of anisotropic electrical,
magnetic or optical properties [1-3]. Since zeolites are metastable phases formed in
hydrothermal systems, small fine-powder and small aggregated crystallites are usually formed.
Therefore, efforts have been made to develop synthesis routes for large single crystals. The F-
synthesis route has been widely used in the formation of large single crystals. Addition of F
favors mineralization and induces crystallization in neutral systems instead of the traditional
basic system. The hydrothermal synthesis of microporous silica phases using fluoride anions
instead of hydroxide anions as mineralizers has been known since the pioneering work by
Flanigen and Patton [4] and extensively used by others during the last two decades [5-6].
Zeolites are normally crystallized from aqueous system. Bibby and Dale [7] first reported
the synthesis of silica-SOD using an organic solvent, which opened a new route for the
synthesis of zeolites and zeolites-related materials, especially for the formation of large single
crystals. Using alcohol as the solvent, some aluminosilicate and all-silicate zeolite large single
crystals can be synthesized [8].
240
2. EXPERIMENTAL
The reactants were fumed silicon (A1203<10 ppm), hydrofluoric acid (HF>40%) and
N,N-dimethylpiperidine Chloride (>98%). The solvents were triethylamine (Et3N, >98%),
tert-butylamine (t-ButNH2), n-dipropylamine (Pr2NH, >98%) and dibutylamine
(But2NH,>98%). Certain amounts of reactants and solvent were mixed and stirred
homogeneously in a Teflon-lined stainless-steel autoclave. The autoclave was then sealed, put
into an oven and heated at 150-220 ~ for 10-30 days under autogeneous pressure. The
product was filtered, washed and dried at 90 ~
The powder XRD data were collected in air by a RigaKu D/max 2500 using CuKa
radiation, electric current of 100 mA and voltage of 40 kV. SEM photos were taken by a JEOL
JSM-35C. The samples were coated with gold. The IR spectra were recorded on a
Perkin-Elmer 1730 FT-IR spectrometer. The samples were prepared using the standard KBr
pellet method. Solid-state NMR experiments were carried out using a Varian Infinity-plus 400
spectrometer operating at a magnetic field strength of 9.4 T. Si chemical shifts were
referenced to tetramethylsilane (TMS).
on the position and relative intensity of diffraction peaks. This confirmed that the sample
prepared was pure NON phase.
4000
~. 3000
9~ 2000
1000
0
5 10 15 20 25 30
2 Theta/degree
Table 1
Typical synthesis conditions and product morphology
In the experiments, the synthesis conditions of zeolite NON were investigated through
changing the chemical composition in reactive system. The typical results were listed in
Table 1. First, there was amorphous solid in given product without adding hydrofluoric acid.
The given sample was zeolite NON with suitable HF in SiO2- N,N-dimethylpiperidine
242
chloride-Et3N system. Zeolite NON may be obtained in F-/Si=1.33-4.00. This showed that F-
anion was an important chemicals in synthesis zeolite NON. Next, N,N-dimethylpiperidine
chloride was another important chemicals in synthesis zeolite NON. Zeolite NON may be
obtained in N,N-dimethylpiperidine chloride/Si=0.67-4.0. But the product was amorphous
solid without N,N-dimethylpiperidine chloride. According to experimental data and
reference's work [ 1], the authors believed that N,N-dimethylpiperidine chloride played a role
of structure-direction agent in synthesis zeolite NON. Triethylamine served only as a solvent.
The best experimental temperature was 453 K.
Table 2
The effect of the solvents
Solvents Template Products Crystal size
Et3N Yes NON a 200x 170x25/xm
Et3N No amorphous
t-ButNH2 Yes NON a 100x 100x 10/xm
t-ButNH2 No amorphous
ButzNH Yes NON a 100x 100x32#m
ButzNH No amorphous
PrzNH Yes NON a 150x 150x27#m
PrzNH No amorphous
Crystallization temperature 453 K, l0 days.
R: N,N-dimethylpiperidine chloride; a: single crystal
the samples synthesized were also zeolite NON when tert-butylamine, n-dipropylamine, and
n-dibutylamine replaced triethylamine. Then, the products were all amorphous silica without
employing N,N-dimethylpiperidine chloride as template,. It proved that these organic amine
solvents had not structure-directing roles. The role of them was only the solvent.
SEM technique was used in observing synthesized samples. From the photos in Fig. 2, it
was found that the type of solvent had obvious effect on the morphology of NON. Obviously,
using other organic amine instead of Et3N, the crystal sizes of NON single crystals were
different in HF-SiOz-N,N-dimethylpiperidine chloride-Et3N system. In these solvents, Et3N
was the best solvent to form large and regular single crystals (200x 170x25/~m). The solvent
of the sample (b) was tert-butylamine, the morphology was same with sample (a), but the
crystal size was 100x 100x 10/xm, it was only nearly half of the sample (a). The solvent of the
sample (c) was n-dibutylamine, the crystal size was 100x 100x32/xm, it was smaller than the
243
sample (a), but it was thicker. The solvent of the sample (d) was n-dipropylamine, the facade
is not regular and glazed.
From the SEM photographs we can also observe the morphology differences with different
solvents, between amine and water. From the SEM photographs shown in Fig. 2, the sample
(a) synthesized in the system using Et3N as solvent is perfect large single crystal. It is 25 #m
thicknesses of the single crystal and the width is about 200#mx 170#m. The facade is much
more regular and glazed than other NON single crystal [ 16].
(a) (b)
(c) (d)
Fig. 2. SEM photographs of NON crystals (a) solvent: Et3N; (b) solvent: t-ButNH2; (c) solvent:
But2NH; (d) solvent: Pr2NH
244
?0-
a~ 6o
o 5o
30
2O
i0 , . . . . ,
1200 1100 1000 900 800 700 600 500 400
Wavenumber (era-1 )
Fig. 3. FT-IR pattern of typical samples
-112.5
/~-I15.4
6.5
-I0
8.1
4. CONCLUSION
NON single crystals with better morphology and large size can be synthesized in the
HF-SiO2- Et3N system containing N,N-dimethylpiperidine Chloride, which acts as structure
directing agent. F- and organic amine solvent provided a feasible medium together, without
any one of them, the perfect crystals can not be formed. From our experiment results, the
choice of organic amine solvent is also very important, it directly affects the morphology of
single crystals.
ACKNOWLEDGEMENTS
The authors are grateful to the National Science Foundation of China (Grant No. 20373047)
and the Science Foundation of Shanxi (Grant No. 20041024) for financial support.
REFERENCES
[4] E.M. Flanigen and R.L. Patton, US Patent No. 4 073 865 (1978).
[5] H. Kessler, J. Patarin and C. Schott-Darie, Stud. Surf. Sci. Catal., 85 (1994) 75.
[6] M.A. Camblor, L.A.Villaescusa and M.J. Diaz-Cabafias, Topics in Catalysis, 9 (1999) 59.
[7] D.M. Bibby and M.P. Dale, Nature, 317 (1985) 157.
[8] S. Li, C. Liu and R. Xu, J. Chem. Soc., Chem. Commun., 17 (1993) 1645.
[9] F. Liebau, H. Gies, R.P. Gunawardane and B. Marler, Zeolites, 6 (1986) 373.
[10] B. Marler, H. Gies and F.Z. Liebau, Kristallogr, 170 (1985) 127.
[11 ] Y. Long, W. Zhong, and X. Shen, J. Inclusion Phenomena, 4 (1986) 121.
[12] H. Gies and B. Marler, Zeolites, 12 (1992)42.
[13] W. Ernest and Y. Valyocsick, US Patent No. 4 568 654 (1986).
[14] A. Kuperman, S. Nadimi, S. Oliver, G.A. Ozin, J.M. Garc6s and M.M. Olken, Nature, 365 (1993)
239.
[ 15] M.M.J. Treacy and J.B. Higgins, Collection of Simulated XRD Powder Patterns for Zeolites,
Elsevier (2001 ) 257.
[ 16] B. Marler and H. Gies, Zeolites, 15 (1995) 517.
[17] H. Koller, R.F. Lobo, S.L. Burkett and M.E. Davis, J. Phys. Chem., 99 (1995) 12588.
[18] E.M. Flanigen, H. Khatami, H.A. Seymenski, in Adv. Chemistry Series 101, E.M. Flanigen and
L.B. Sand (eds.), American Chemical Society. Washington, D. C. (1971) 201.
[19] J.M. Ch6zeu, L. Delmotte, J.L. Guth and M. Soulard, Zeolites, 9 (1989) 78.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka, N. Zilkov4 and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 247
E.R. Cooper, C.D. Andrews, P.S. Wheatley, P.B. Webb, P. Wormald and R.E. Morris
Here we report a new method of preparing aluminophosphate zeolite analogues using ionic
liquids [ 1] as both solvent and template [2]. An imidazolium-based ionic liquid leads to four
zeotype frameworks under different conditions. One material has a novel structure type with
unusual structural characteristics that can be traced back to the solvent chemistry. Because of
the vanishingly low vapour pressure of ionic liquids this synthesis takes place at ambient
pressure, eliminating any safety concerns associated with high hydrothermal pressures. The
ionic liquid can also be recycled for further use.
1. I N T R O D U C T I O N
The most commonly used route to molecular sieve synthesis is the hydrothermal method [3].
This involves the mixing of the reagents in water, which acts as the solvent, then heating in a
sealed autoclave at autogenous pressure for a specific time. The water does however appear
to have other roles in the synthesis. It can act as a space filler in the porous lattice, it
enhances the reactivity and lowers the viscosity of the mixture. So the water is not only a
solvent but a reactant and a catalyst in T-O-T bond formation [4].
An ionic liquid is commonly defined as a liquid at low temperatures (< 100~ which consists
of only ions [ 1]. Some of the important properties of ionic liquids which have made them of
such industrial importance and also of interest to us as solvents and templates in zeolite
synthesis are [5]:
1. excellent solvating properties for a wide range of organic, inorganic and polymeric
materials
2. no measurable vapour pressure
3. non-flammability
4. high thermal stability
For the purpose of this work we have use a broader definition of an ionic liquid as any salt
that melts below the temperatures used in the synthesis of zeolites (typically 150-200~
248
The aim of this work was to synthesis zeolite analogues, primarily aluminophosphates using
ionic liquids as both the organic template and the solvent, hence eliminating the effects in the
reaction from the water. This however is not a simple task as most ionic liquids are
hygroscopic and can absorb significant amounts of water from the atmosphere [6]. Even after
a moderate drying process, water can still be present [7]. At low concentrations of water the
molecules are isolated or exist in small independent clusters hence a small amount of water
will not disrupt the ionic-liquid-framework interactions [8]. There is however still the
possibility that the low concentration of water could act as a catalyst in T-O-T bond
formation.
2. E X P E R I M E N T A L
In the case of the SIZ-4 synthesis the AI[OCH(CH3)2]3, H3PO4 and HF were added to the
Teflon-liner. This was then heated to 50~ on a hotplate for 2 hours to remove the water and
any isopropanol formed during the initial reaction. This was confirmed by following the mass
of the mixture until no more mass loss was observed. The IL was then added and the
synthesis proceeded as normal. After cooling the autoclave to room temperature the product
was suspended in distilled water, filtered by suction and washed with acetone. The products
were all white, crystalline solids.
Table 1
Synthesis details and conditions for the preparation of materials [2].
Mass of reagents (g) (molar ratio of reagents)
Product Al(OiPr) 3 H3PO4 HF H20 IL Temp Time
(~ (hrs)
SIZ-1 0.1018 0.1732 0.00 4.05 150 66
(1.0) (3.0) (0.0) * (2.9) (43)
SIZ-3 0.1013 0.1732 0.015 4.07 150 68
(1.0) (3.0) (0.73) * (3.8) (43)
SIZ-4 O. 1054 O. 1772 0.015 3.82 150 68
(1.0) (3.0) (0.70) * (0.0) (39)
SIZ-5 0.0465 0.0858 0.00 0.495 2.03 150 19
(1.0) (3.1) (0.0) (116) (44)
* No extra water added in these preparations. Small amounts of water present come from the aqueous
HF and H3PO4solutions and the ionic liquid.
249
3. RESULTS
SlZ-1
SIZ-4 ~
i i i i I I
1-ethyl-3-methylimidazolium bromide was used as both solvent and template to make the four
structures illustrated in Fig. 2. The presence of the intact 1-methyl-3-ethylimidazolium
templates was confirmed by ~3C NMR, the results shown in Fig. 1.
SIZ-1 (St Andrews Ionic Liquid Zeolite-I) is a novel structure type consisting of double six
rings joined to form layers that are linked into a three-dimensional framework by units
containing single four rings. The unusual feature of the structure of SIZ-1 is that the AI:P
ratio is not 1:1. The formula of the material is A18(PO4)10H3.3C6HllN2 but the A1-O-P
alternation is maintained. The framework is therefore interrupted, with some unusual
intraframework hydrogen bonding. The negative charge present on the framework (caused by
the existence of terminal P-O bonds) balances the charge on the 1-methyl-3-ethylimidazolium
templates that are present in the pores. The overall structure of SIZ-1 shows a two-
dimensional channel system parallel to the a- and b- crystallographic axes.
SIZ-3 has the same topology as that of A1PO-11 (AEL) [10] The A1PO-11 framework
structure is related to that of the A1PO-5 framework, but one set of four rings in the ALPO-5
framework has been removed. A1PO-11 exhibits a unidimensional channel system with 10-
membered ring pores. These pores are elliptical in shape and each pore is isolated from the
adjacent pore by a wall composed of linked 4- and 6-rings of alternate aluminium and
phosphorous tetrahedral. The fluoride atom is observed in an A104F unit, which forms an A1-
F-A1 linkage.
SIZ-4 has the same framework topology as the triclinic form of chabazite (CHA) [ 11 ]. The
SIZ-4 CHA framework consists of double six-membered rings of altemating aluminium and
phosphorous tetrahedral. These are linked through four-membered rings to produce a three
dimensional pore structure with eight-membered ring windows. Each six ring of the D6Rs
contains one octahedral aluminium that connects to another octahedral aluminium of another
D6R through two bridging fluorine atoms.
251
SIZ-1 SIZ-3
jBr-
OP
OA1
~F
~o
SIZ-4 SIZ-5
Fig. 2. 1-ethyl-3-methylimidazlolium bromide can be used as both template and solvent to prepare
SIZ-1, SIZ-3, SIZ-4 and SIZ-5. SIZ-3 and SIZ-4 are prepared in the presence of fluorine and SIZ-5 in
the presence of excess water [3].
SIZ-5 has the same topology as zeolite A1PO-41 (AFO) [12] with orthorhombic lattice
parameters 9.76 x 25.61 x 8.32 A. The A1PO-41 structure has unidimensional 10-ring pores
parallel to the c-axis direction. This is the same as in the SIZ-3 framework structure. The
layer perpendicular to the c direction is described as a sequence of 4.62.4.10-rings connected
by narsarsukite-type UDUD chains (linked 4-rings with sequence of tetrahedron interlayer
directiveness up-down-up-down). The 10-ring pore in the A1PO-41 framework is not strictly
elliptical, but rather a distorted elliptical pore.
The 31p MAS N M R spectrum of SIZ-1 shows a number of resonances in the region for
tetrahedral phosphorus, as expected. The single crystal structure has five independent
phosphorus sites, and the NMR shows at least four resolvable sites in a broad envelope
252
between -14.6 and -29.3 ppm. SIZ-3 exhibits three resonances in a broad envelope from-20
ppm to -35 ppm, again consistent with the expected three resonances from the XRD structure.
SIZ-4 has three major resonances that are consistent with the structure, the very low field
resonance at -7.25 ppm is due to the phosphorus nearest the fluoride bridges in the structure.
SIZ-5 shows a spectrum similar to those that have been previously reported for A1PO-41.
The 27A1 MAS NMR spectra of all the materials show evidence of tetrahedral aluminium,
with intense resonances at around 40 ppm. However, they each show evidence of higher
coordinated aluminium, consistent with the five coordinated in the case of SIZ-1, SIZ-3 and
SIZ-5 or octahedral (A104F2) coordination for SIZ-4. All are consistent with the XRD
structures that we have reported above or in the literature. The fluorine-19 MAS NMR spectra
for SIZ-3 and SIZ-4 both confirm the presence of fluoride in the materials (resonances centred
at - 118 and - 125 for SIZ-3 and SIZ-4 respectively).
Ionic liquids are relatively polar solvents, making them suitable for the dissolution of the
inorganic components required for the synthesis (good enough to form a clear solution in the
synthesis at 150~ [2].
Many ionic liquid cations are chemically very similar to species that are already known as
good templates (alkylimidazolium-, pyridinium-based ionic liquids etc).
Most ionic liquids show suitable thermal stability. Many are reported to be stable to almost
400~ [13] with even the worst of the common ionic liquids stable to well above the
temperatures we have shown are required for successful ionothermal synthesis.
Zeolites are notoriously difficult to prepare as large crystals. Our experience in preparing
zeotypes in this work indicates that these solvent systems are suitable for the preparation of
highly ordered solids with crystals large enough for single crystal X-ray diffraction studies.
This makes the characterisation of the structural architecture very much easier than is often
the case.
Ionic liquids have negligible vapour pressure and synthetic procedures can be carried out in
open vessels avoiding the high autogenous pressures (up to 15 atm at 200~ and associated
safety concerns that accompany hydrothermal synthesis in sealed autoclaves. Again, this
distinguishes ionothermal synthesis from hydrothermal methods.
Potentially the most important feature of this ionothermal synthesis mechanism is the removal
of the competition between template-framework and solvent-framework interaction that is
present in hydrothermal preparations. The structure directing properties of templates are often
not as specific as we would like, but in a system where an ionic liquid is both solvent and
253
template the negatively charged atoms at the surface of a growing framework will always be
interacting primarily with the templating cation rather than with a mixture of template and
solvent. Fig. 3 illustrates this schematically. Recent modelling studies indicate that the
structure of ionic liquids is characterised by long range correlations and distributions that
reflect the asymmetric structures of the cations [14]. Long range asymmetric effects of this
kind potentially increase the likelihood of transferring chemical information from the
template cation to the framework; a situation that is desirable if full control over the
templating process is to be achieved. This distinguishes this method further from
hydrothermal synthesis where solvation/clathration by water reduces the range of any effects
of the template structure markedly.
Ionothermal synthesis allows a great deal of control over the mineralisers that are present in
the solvent. Mineralisers, such as fluoride (and to a lesser extent other halides) help solubilise
the inorganic components and catalyse T-O-T bond formation (where T are the tetrahedral
framework atoms). The ionic liquids are good solvents on their own, but adding fluoride
increases their solvating power and can change the types of framework that are formed.
The ionic liquids solubilise the starting materials almost completely at the reaction
temperatures, indicating that the synthesis mechanism is a crystallisation from solution rather
than a solid-to-solid-transformation. The dependence of the products on water also gives
some clues as to the mechanism of the reaction. With little water and no HF to act as
mineralises interrupted framework structures can be targeted. On addition of HF or water,
condensed structures are formed with no hanging bonds.
H
O\
/ H H~
O H
..
H ' H
I L-- . , ~ . , Br-
H O O-H
O ( / Br -
Br O\~.,.O I~ ~ "~-'h
H I N H
H--O O O' N O '
10 \
H\ H H \ H I
H O" H--O~ O "1
H--O H O--H
Br-
I O I ~'N \
I H Br- O O
~T\ \ T I I
0 \ 0 -" 0 ~ ' { ' 0 , ~0 ~T\ ,, T
0 .TIO\ " 0 T 0 \ 0 _ " n ~'~'0 0
/ /~ "odT\O'T/O'~ T / ,0 :T~--O...r,'-" 0 T"
/ ? od"O,T d'
Fig. 3. A schematic comparison between a hydrothermal synthesis of a material and an ionic liquid
synthesis. Note that where the ionic liquid is both solvent and template the surface of a growing
crystallite is always in contact with the template, whereas in a hydrothermal synthesis the framework-
water interaction competes with the framework-template interaction.
254
ACKNOWLEDGEMENTS
We thank the CCLRC, S. Teat and J. Warren for access to the Synchrotron Radiation Source
(Daresbury) and the EPSRC and D. Apperley for access to solid-state NMR. REM thanks the
Royal Society for the provision if a University Research Fellowship.
REFERENCES
[1] P. Wasserscheid and W. Keim, Angew. Chem., Int. Ed. Engl., 39 (2000) 3773.
[2] E.R. Cooper, C.D. Andrews, P.S. Wheatley, P.B. Webb, P. Wormald & R. E. Morris, Nature,
430 (2004) 1012.
[3] R.M. Milton, US Patent No. 2 882 243 (1959).
[4] S.Y. Yang, A.G. Vlessidis and N.P. Evmiridis, Microporous Mater., 9 (1997) 273.
[5] C. Baudequin, J. Baudoux, J. Levillain, D. Cahard, A.C. Gaumont and J.C. Plaquevent,
Tetrahedron: Asymmetry, 14 (2003) 3081.
[6] L. Cammarata, S.G. Kazarian, P.A. Slater and T. Welton, Phys. Chem. Chem. Phys., 3 (2001)
5192.
[7] J.G. Huddleston, A.E. Visser, W.M. Reichert, H.D. Willauer, G.A. Broker and R.D. Rogers,
Green Chem., 3 (2001) 156.
[8] C.G. Hanks and R.M. Lynden-Bell, J. Phys. Chem. B, 107 (2003) 10873.
[9] P.Wasserscheid and T. Welton, 'Ionic Liquids In Synthesis', Wiley-VCH, (2003) Ch. 2.
[10] J.W. Richardson, J.J. Pluth and J.V. Smith, Acta Crystallogr., Sect. B: Struct. Sci., 44 (1988)
367.
[ 11] M.M. Harding and B.M. Kariuki, Acta Crystallogr., Sect. C: Cryst. Struct. Commun., 50 (1994)
852.
[12] H.W. Clark, W.J. Rievert and M.M. Olken, Microporous Mater., 6 (1996) 115.
[13] H.L. Ngo, K. LeCompte, L. Hargens, A.B. McEwen, Thermochim. Acta, 357 (2000) 97.
[ 14] M.G. Del Popolo and G.A. Voth, J. Phys. Chem. B, 108 (2004) 1744.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 255
Center for Sustainable and Green Chemistry, Department of Chemistry, Building 206,
Technical University of Denmark, DK-2800 Lyngby, Denmark
Zeolites are crystalline materials, which are widely used as solid acid catalysts and supports in
many industrial processes. Recently, mesoporous MFI-type zeolite single crystals were
synthesized by use of carbon particles as a mesopore template and sodium aluminate as the
aluminum source. With this technique, only zeolites with relatively low A1 contents were
reported (Si/A1 ratio about 100). In this work, the preparation of aluminum-rich mesoporous
MFI-type zeolite single crystals (Si/AI-~I 6-50) using aluminum isopropoxide as the aluminum
source is reported for the first time. All samples are characterized by X-ray powder diffraction
(XRD), scanning electron microscopy (SEM), ammonia temperature programmed desorption
(NH3-TPD), and N2 adsorption measurements. The obtained zeolites combine the high
crystallinity and the characteristic micropores of zeolites with an intracrystalline mesopore
system having mesopore volumes varying between 0.37-0.40 cm3/g.
1. INTRODUCTION
Zeolites are crystalline microporous materials that are widely applied in industry for
adsorption, separation and catalysis. A multitude of chemical reactions of industrial interest is
catalyzed by zeolites, which often exhibit unique properties with respect to both activity and
selectivity [ 1]. Much of the success of zeolites can be attributed to the presence of micropores
with well-defined structures. The exact shape and size of the micropores may result in various
kinds of shape selectivity [2]. Nevertheless, the sole presence of micropores may in many
cases also limit the catalytic performance of zeolite catalysts, particularly when diffusion in
the micropores becomes significantly slower than the catalytic reaction [3]. Several different
preparative strategies have been attempted to circumvent this problem. One of these is to
introduce mesopores into each individual zeolite crystal, usually done by suitable post-
treatments, such as dealumination [4] or desilication [5] that lead to partial disintegration of
the zeolite crystals. However, during the last few years, it has proven possible to introduce a
very high degree of mesoporosity in zeolite crystals in a more controlled manner. These
mesoporous zeolite single crystals are prepared by nucleating zeolite crystals inside a carbon
material so the zeolite crystals grow and partially encapsulate the carbon. After complete
crystallization, the carbon is removed by combustion. The resulting zeolite material has been
shown conclusively to consist of quite large mesoporous single crystals. Compared to other
known approaches for introducing mesopores into the structure of zeolites, the offered
method does not lead to the partial destruction of the zeolite crystal. So far, mesoporous
zeolite single crystals have been reported for zeolite structures MFI [6], FAU [7], and MEL
[81.
256
Recently, the benefits of introducing mesopores into zeolites have been demonstrated
in several cases as a means of improving the zeolite catalysts performance in the catalytic gas-
phase alkylation of benzene [9], and in liquid-phase cracking and isomerization of long-chain
hydrocarbons [8, 10, 11].
The aluminum content of a zeolite is important because it influences the number and
strength of the Bronsted acid sites as well as the ion exchange capacity. In all previously
reported works, mesoporous zeolite single crystals were synthesized by use of carbon
particles as a mesopore template and sodium aluminate as the aluminum source [ 12-14]. In
these works, only zeolites with relatively low aluminum contents were reported (Si/A1 ratio
about 100). Here, we report the preparation of aluminum-rich mesoporous MFI-type zeolite
single crystals (Si/AI-16-50) using aluminum isopropoxide as the aluminum source.
Particular, we show that the use of aluminum isopropoxide facilitates the incorporation of
aluminum into zeolite T-sites.
2. EXPERIMENTAL
several times. Then the product was dried at 110~ for 10h. Finally carbon black was
removed by controlled combustion in air in a muffle furnace at 550~ for 18 h. In this way a
white material is obtained, which by the results of chemical analysis is shown to contain less
than 0.5 wt % of carbon.
Fig. 1. XRPD patterns of (a) Na-ZSM-5, synthesized with aluminum isopropoxide (theoretical Si/Al
=15), (b) Na-ZSM-5, synthesized with aluminum isopropoxide (theoretical Si/AI = 35), (c) Na-ZSM-
5, synthesized with sodium aluminate (theoretical Si/A1- 15), (d) Na-ZSM-5, synthesized with
sodium aluminate (theoretical Si/Al = 35).
In Fig. 2, the nitrogen adsorption and desorption isotherms of the zeolites after
combustion of the carbon are given.
The isotherms of all samples contain a hysteresis loop at relative pressures higher than
p/po- 0.4, which is indicative for mesoporosity. For all samples hysteresis loop have an
upward curvature at relative pressures above 0.8. This upward curvature indicates the
presence of cylindrical mesopores which are connected to the external surface area [ 15].
In Table 1, the BET surface areas and micro- and mesopore volumes of all samples are
summarized.
Table 1
Nitrogen adsorption data of the mesoporous zeolites samples after combustion of carbon
matrix.
Aluminum Alcohol Si/A1 Vmicro Vmeso BET area
Zeolite
source source ratio a (cm3/g) b (cm3/g) c (m2/g) d
aluminum iso-
ZSM-5 14.4 0.11 0.39 383
isopropoxide propanol
aluminum iso-
ZSM-5 31.2 0.11 0.40 390
isopropoxide propanol
sodium
ZSM-5 ethanol 16.8 0.11 0.37 338
aluminate
sodium
ZSM-5 ethanol 34.3 0.11 0.38 361
aluminate
,.., 350
E 300-
0
i.._J
250 -
E
:3
"~ 200-
>
"0
m 150 -
,,Q
L.
b
0 100 -
50
0 0.2 0.4 0.6 0.8 1
i,,,-,i 350
03
E 300 -
i__J
0
250 -
E
=Bin
:3
0 200 -
>
"0
150 -
.rL . d
0 100 -
"0
50
0 0.2 0.4 0.6 0.8 1
Fig. 2. Nitrogen adsorption and desorption isotherms of (a) Na-ZSM-5, synthesized with
aluminum isopropoxide (theoretical Si/AI =15), (b) Na-ZSM-5, synthesized with aluminum
isopropoxide (theoretical Si/AI = 35), (c) Na-ZSM-5, synthesized with sodium aluminate (theoretical
Si/A1 - 15), (d) Na-ZSM-5, synthesized with sodium aluminate (theoretical Si/AI = 35).
260
It is clearly seen from Table 1 that all obtained mesoporous MFI-type zeolite single
crystals have high mesopore volumes varying between 0.37-0.40 cm3/g, and are characterized
by large specific surface areas. Besides that, mesoporous zeolites, obtained with the use of
aluminum isopropoxide as the aluminum source, show slightly better adsorptions
characteristics in comparison with samples, which were synthesized with sodium aluminate as
the aluminum source. Usually mesoporosity is obtained by close packing of nanosized
crystals. Therefore, the mesoporosity of the present zeolites is unique since it is built into each
individual single crystal.
The scanning electron microscopy (SEM) images of the mesoporous zeolite single
crystals after combustion of the carbon are shown in Fig. 3.
According to the SEM analysis, all obtained samples appear to be highly crystalline.
Moreover, the average crystal size as determined from SEM is almost the same for the
different samples.
The presence of non-crystallographic intracrystalline mesopores is clearly visible in
the images just as the high mesoporosity and the relatively large average pore size of the
mesoporous zeolite single crystals. These mesoporous zeolites crystals are unique in the sense
that they contain interconnected micropores and mesopores inside each individual single
crystal.
In Table 2, a comparison between mesoporous ZSM-5-type zeolites, which were
obtained using different aluminum sources, are given.
Fig. 3. SEM images of (a) Na-ZSM-5, synthesized with aluminum isopropoxide (Si/A1-14.4), (b) Na-
ZSM-5, synthesized with aluminum isopropoxide (Si/A1 = 31.2), (c) Na-ZSM-5, synthesized with
sodium aluminate (theoretical Si/A1- 16.8), (d) Na-ZSM-5, synthesized with sodium aluminate (Si/A1
= 34.3).
261
Table 2
Comparison of mesoporous ZSM-5 zeolites obtained using different aluminum sources
Aluminum Si/A1 A1 content, Amount of NH3 Si/A1
Sample source ratio a ~mol/g desorbed,/tmol/g ratio b
ZSM-5 aluminum 14.4 870 740 16.6
isopropoxide
ZSM-5 aluminum 31.2 529 464 32.3
isopropoxide
ZSM-5 sodium 16.8 825 350 43.9
aluminate
ZSM-5 sodium 34.3 451 336 46.9
aluminate
The comparison of the results of NH3-TPD with the aluminum content determined by
elemental analysis, provides information about the part of aluminum, which is responsible for
the zeolite acidity. The difference between these two values roughly corresponds to the non-
framework aluminum, which, consequently, does not generate much acidity. Thus, the acidity
of the samples is created mainly by the framework aluminum content, and each framework
aluminum is responsible for the formation of one acid site. Non-framework atoms are
supposedly present as smaller or larger domains of aluminum oxide. Therefore, the
experimentally observed amounts of desorbed NH3 molecules are lower than the aluminum
content, especially in the case of the samples prepared with sodium aluminate as aluminum
source. In our previous paper [9] we had also observed good correlation between the acidity
of the mesoporous zeolite samples determined by NH3-TPD and IR-pyridine adsorption and
the amount of framework A1 determined by 27A1NMR.
It is seen that increasing the quantity of sodium aluminate in the reaction mixture leads
almost exclusively to an increase of the non- framework aluminum content. The opposite
situation is observed in the case of samples which were synthesized using the aluminum
isopropoxide as the source of aluminum. Here the increase of the aluminum isopropoxide
content in the crystallization mixture leads to a noticeable increase of the framework A1
concentration. Finally, all these results show that the use of aluminum isopropoxide instead of
sodium aluminate as the source of aluminum allows production of mesoporous zeolites of
MFI- type with higher framework aluminum content than previously possible.
4. CONCLUSIONS
ACKNOWLEDGEMENTS
We thank Elena Knyazeva from Laboratory of Kinetics and Catalysis (Moscow State
University, Chemistry Department) for help in nitrogen physisorption measurements.
REFERENCES
1. INTRODUCTION
The current interest in the designed preparation of microporous and microstructured solid-
state compounds based on silicates and metal phosphates relies not only on their complex
structural diversity, but also on possible applications, e.g. as catalysts or molecular sieves
[1,2] or in advanced materials [3,4].
Among these compounds zincophosphates have attracted much attention also because of
the comparatively mild synthesis conditions. Typically, such phosphates are synthesized in
hydrothermal syntheses using the template effect of so-called "structure-directing agents"
(SDA). These mostly are organic amines which are able to form hydrogen bonds involving
the hydrogen atoms of the amino groups and the oxygen atoms of the phosphate framework.
The strength of the total host-guest interaction between the phosphate framework and the
SDA molecules is roughly proportional to the number of hydrogen bonds formed. In addition,
the sterical disposition of the amino groups on an SDA will influence the structure of the
framework, possibly involving the transfer of some symmetry information. In this context, the
use of chiral compounds as SDAs is especially promising, as it could lead to the generation of
chiral frameworks of low symmetry, exhibiting special physical effects, for example non-
linear optical properties [5].
Metal complexes with amine ligands possess several amino groups, and are thus
potentially interesting as SDAs for phosphate frameworks. This is especially true for
cobalt(III) amine complexes, which are known to be very stable. For example, the complex
[Co(en)3] 3§ (en = ethylenediamine) has demonstrated its potential as a suitable SDA for
zincophosphates: Several different composites have been produced, in part also based on the
large number of conformational isomers of the [Co(en)3] 3+ complex (which can be described,
264
for example, by A~x and Aaa~ configurational and conformational terms) [6,7]. In addition,
[Co(en)3] ~+ possesses a large number of amino groups available for hydrogen bonding as well
as a high charge. Based on the success of [Co(en)3] 3+ as an SDA, we now employ complexes
with an increased rigidity, thus minimizing possible isomerization processes (also at harsher
synthesis conditions) and optimizing control in structure-directed synthesis. For this purpose,
we use the related complexes [CodiAMsar] 3+ (diAMsar = diaminosarcophagine: 1,8-diamino-
3,6,10,13,16,19-hexaazabicyclo[6.6.6]icosane) and [Cosep] 3+ (sep = sepulchrate:
1,3,6,8,10,13,16,19-octaazabicyclo[6.6.6]icosane) (Fig. 1). They are derived from [Co(en)3] 3+
and are best described as chelate complexes with encapsulating amino ligands. Their cage-
like structures allow these complexes to withstand ligand exchange reactions, also at higher
temperatures, and provide a more rigid conformation as compared to [Co(en)3] 3+. For
example, these complexes do not undergo equilibrium reactions between different
enantiomers. The additional amino groups at the top and the bottom of the caps are
predominated to form additional hydrogen bonds to the framework and to support the host-
guest interactions. Here we report the synthesis of the novel zincophosphates UH-6 and
UH-8, which have been synthesized hydrothermally using [CodiAMsar] 3+ and [Cosep] 3+ as
SDAs, respectively.
3+ 3+
NH 2
Fig. 1. The structure of the cobalt complexes [CodiAMsar] 3+(left) and [Cosep]3§ (right)
2. EXPERIMENTAL
2.1. Synthesis
Chloride salts of the cobalt complexes were synthesized according to literature
procedures [8-11] and were then transformed to the hydroxide forms. The zincophosphate
UH-6 was synthesized as follows: The hydrochloride of [CodiAMsar] 3+ was passed over
DOWEX 1X8 anion exchanger to form a basic solution of the complex. This solution was
reduced in volume to achieve a concentration of about 0.6 M. Afterwards, it was mixed with
H3PO4 (85 %) and finally ZnCI2 was added, followed by dilution with water to a
concentration of 0.4 M with regard to the complex. The synthesis gel was stored in an oven at
363 K for three days. The molar ratio of the synthesis gel was about 1 ZnCI2 : 1 SDA : 2
H3PO4 : 130 H20. The resulting precipitate was filtered, washed with hot water and dried in
air. The zincophosphate was isolated as orange-coloured prisms, often intergrown. A crystal
suitable for structure determination via single-crystal X-ray diffraction was selected.
UH-8 was synthesized in a slightly different way: To the basic 0.6 M solution of
[Cosep] 3+, ZnCI2 was added before phosphoric acid and no further dilution took place. The
265
molar ratio of the synthesis gel was about 1 ZnC12 : 1 SDA : 2 H3PO4 : 90 H20. The product,
which could only be isolated with the dense zinc phosphate hopeite as a by-product, was
filtered, washed with water and air-dried. UH-8 crystallizes as thin orange needles which
were of sufficient, but limited quality for structure determination via single X-ray diffraction.
Table 1
Crystal data for UH-6 and UH-8
UH-6 UH-8
Empirical formula CI4H41ColN8OI7P4Zn2 C24H74Co2NI6041P1oZn6
Formula weight 907.10 2062.77
Crystal system monoclinic triclinic
Space group; Z P21; 2 P-1; 2
a (A) 9.950(3) 8.9133(8)
b (A) 11.715(3) 19.0614(19)
c (A) 12.836(5) 19.092(2)
or(~ 87.194(12)
/3 (~ 107.27(4) 85.546(12)
3,(o) 78.011(11)
V (A a) 1428.8(8) 3161.6(5)
T(K) 293 293
~MoKot(mm -1) 2.557 3.115
Ocalc (g cm -3) 2.108 2.167
Crystal size (mm 3) 0.07 x 0.74 x 0.03 0.3 x 0.03 x 0.06
20max (~ 51.84 56.12
Reflections measured 19919 55000
Unique reflections (total); Rint 5274; 0.0934 14088; 0.1177
Unique reflections (I > 2ai) 3766 6614
Refined parameters/restraints 427/1 468/0
R1 (I > 2at) 0.0460 0.0985
wR2 (all data) 0.1074 0.2412
Largest difference peal~ole (e A -3) 1.706/-1.368 5.447/-5.063
266
3. R E S U L T S
100
rel. int.
2o
the corners (Fig. 3). The Z n - O bond lengths are in the range from 1.920(5) A to 1.978(4) A,
the P - O bond lengths are in the range from 1.491(4) A to 1.598(5) A. UH-6 has a framework
density of 8.4 T per 1000 A 3. The doubly protonated [CodiAMHsar] 5+ cation is placed
between four parallel chains and linked with the framework via several N-H...O-hydrogen
bonds. Both primary amino groups form hydrogen bonds to three chains while the two
nitrogen atoms of each en (ethylenediamine) unit are linked to one oxygen atom of the
framework (Fig. 4).
Oco
a'q 00
Oc
Fig. 3. Crystal structure of UH-6. The SDA complexes can be seen alternately in the front and in the
back of the chains
267
~Zn
OCo
GP
Oo
ON
6tt
Oc
Fig. 4. Crystal structure of UH-6. Left: The environment of the SDA in UH-6. View along [100]. Only
hydrogen atoms of primary amino groups are shown for clarity. Top right: The building unit of the
framework in UH-6. Hydrogen atoms are omitted for clarity. Bottom right: The propeller-like motif,
viewed along the Zn-Zn direction. Hydrogen atoms are omitted for clarity
OCo
b *~ II~P
~.~ ~ r" Oo
c / ~ ' Qc
Fig. 5. Crystal structure of UH-8. The [PO4] tetrahedra which protrude into the channels are marked
The structure solution via single-crystal X-ray diffraction revealed that one Zn atom as
well as a neighbouring P atom are disordered, as was detected by the rather large thermal
displacement ellipsoids obtained after anisotropic refinement. Since the crystal isolated for the
measurement was only a rather thin needle, the data were not sufficient to fully refine this
disorder and to determine the number of protonated P-O groups on the phosphorus atoms.
The data are, however, reliable with regard to the determination of the topology of the
composite. In the building unit, two non-equivalent [Cosep] 3+ cations can be identified. We
were not able to determine whether the tertiary amino groups of the complexes are
protonated. The corresponding nitrogen atom is strongly s p 2 hybridized as a result of the
strain within the ligand imposed by the coordination to the metal atom [ 11 ]. This corresponds
to a less basic character of the amino group. The rather large C-N-C bonding angles of about
269
>: P3
Fig. 6. Topology of the inorganic framework in UH-8: Zigzag ladder of four-membered rings
113 ~ (compared to an ideal tetrahedral angle of about 109 ~ supports this character; also, there
are no oxygen atoms in the vicinity of these nitrogen atoms which would be available for
host-guest interactions. In contrast, each nitrogen atom of the central e n units of the complex
participates in at least one hydrogen bond to oxygen atoms of the framework.
The framework topology contains an interesting motif which has been observed in a
large number of zincophosphates and is assumed to be a secondary building unit in the
formation process of zincophosphates [14,15]. This motif is a four-membered ring
constructed of alternating Zn and P atoms (Fig. 6). In UH-8 these rings are condensed to form
a zigzag ladder which runs along the a axis as can be seen in Fig. 7 in the center of the unit
cell.
Fig. 7. Framework structure of UH-8. Only connections between the Zn and P atoms are drawn. View
close to [ 100]
270
4. DISCUSSION
The use of the cage-like complex cations [CodiAMsar] 3+ and [Cosep] 3+ as SDAs in the
hydrothermal synthesis of microstructured zincophosphates has led to crystal structures with
topologies which had not been observed before. While the use of racemic solutions of the
complexes has generated the chiral composite UH-6, UH-8 incorporates both enantiomers.
Obviously, the connectivity of the zincophosphate decreases with increasing ability of the
metal complex to form hydrogen bonds to oxygen atoms of the zincophosphate part.
These results open up a wide range of synthetic possibilities. Firstly, using only one
enantiomer of [CodiAMsar] 3+ should result in enantiopure UH-6. Secondly, it would be
interesting to investigate, whether enantiopure [Cosep] 3+ would generate a novel chiral
zincophosphate. Finally, complexes of these cage-type ligands with other metal cations (Ni 2+,
Zn 2+ or Fe 3+) should also be tested in the synthesis of zincophosphates.
ACKNOWLEDGEMENTS
We acknowledge the support of our work by the Fonds der Chemischen Industrie. Also, we
thank Dr. R. Wartchow for performing the X-ray measurements on the IPDS diffractometer.
REFERENCES
The aim of this paper is to systematically investigate the formation mechanisms of ITQ-2
from different MCM-22(P) materials.
Particular attention has been devoted to understand how each delamination step influences
the final materials ITQ-2. In order to characterize the intermediate products of the
delamination steps, various types of analysis have been carried out: XRD, 27A1-NMR,
29 Si-NMR, DSC-TG and chemical analysis. It has been confirmed that the delamination
process is favoured by the decreasing aluminium concentration of parent materials. The
investigation carried out has shown that aluminium always remains in tetrahedral
coordination, because it is protected, during the delamination process, by the organic
cation used in the MCM-22(P) synthesis.
1. INTRODUCTION
In recent years there has been growing interest in synthesizing molecular sieves with large
and ultra-large pores having higher accessible surface [1]. Accordingly, efforts have been
focused on increasing pore sizes [2], in order to allow a wider application of these
materials in the fine chemical, pharmaceutical and petrochemical industrial processes.
A new approach to obtain materials allowing access to large molecules to active sites,
involves the preparation of lamellar zeolites whose structures are successively
delaminated. In this way it is possible to obtain single crystalline sheets of zeolitic nature
where all the potential active sites are accessible through the external surface.
Some layered materials are known to present three-dimensional structures which
exhibit their strongest chemical bonding in only two dimensions. In such materials, the
stronger chemical bonds are formed in two-dimensional planes and the interactions
between the planes are weaker than the chemical bonds holding an individual plane
together.
It has been discovered that certain synthetic crystalline zeolites undergo a
transformation during the synthesis from an intermediate swellable layered state to a non
swellable state having order in three dimensions, obtained through final calcination [3].
As an example, zeolite MWW is obtained after calcination of synthesized MCM-22(P),
where P (Precursor) indicates the product of hydrothermal synthesis. In the swellable
intermediate state it is possible to interpose the swelling agent between the layers of the
zeolitic material, and successively the swollen product is submitted to ultrasound
272
2. EXPERIMENTAL
The MCM-22(P) samples were obtained starting from syntheses mixtures having Si/A1 ratio
20 and 50. The gel compositions were the following:
30 SiO2: x AI(OH)3 : 5.4 NaOH :15 HMI :1350 H20 with x- 0.6 or 1.5.
The synthesis gels were obtained by mixing in the following order: the NaOH solution in
distilled water (30% wt), AI(OH)3 (98% Pfaltz and Bauer), hexamethylenimine (HMI,
Aldrich) and finally silica fume (Aldrich). The mixtures obtained were magnetically stirred
for half an hour at room temperature, then placed into Teflon lined autoclaves and heated in
dynamic conditions at 150~ for 9 days. The crystallization products were washed with
distilled water and dried overnight before characterization.
Following the procedure described by Corma and co-workers [6], the swelling was carried out
by mixing MCM-22(P) with hexadecyltrimethylammonium choloride (CTMABr, Aldrich,
25%), tetrapropylammonium hydroxide (TPAOH, Fluka, 40%) at the following weight ratio:
1MCM-22(P): 5.6CTMABr: 2.4TPAOH: 21.5H20.
Successively, the recovered swollen material was treated in an ultrasonic generator (130W,
25kHz) for different times. Then, the resulting solids were recovered through centrifugation
and after drying calcined at 813 K yielding the final ITQ-2.
2.1. Characterization
The characterization of delaminated material ITQ-2 requires different techniques. In fact,
the diffractometric analysis usually carried out in the identification of the zeolitic phases
allows one to establish the amorphization of the material, but is not sufficient to evaluate the
degree of delamination.
The X-ray powder diffraction patterns were collected using CuKol radiation (Philips PW
1730/10 generator equipped with a PW1050/70 vertical goniometer). The micrographs were
collected utilizing a scanning electron microscope JEOL JSTM 330 A. The chemical
composition of samples was determined by elemental analysis with atomic absorption
spectroscopy (GBC, AVANTA); the experimental error of atomic absorbance measurement
was around 5% or less.
Surface area measurements were carried out on Micromeritics ASAP 2010 system at
liquid nitrogen temperature. Before the adsorption measurement the calcined samples were
degassed at 350~ and 10STorr for 10 hours.
The amount of organic species contained in the products was measured by
thermogravimetric analysis ( TG and DSC-Netzsch STA 409).
273
The NMR spectra were recorded on a Bruker MSL 400 spectrometer. For 298i (79.4
MHz) a 6 gs (0=~/6) pulse was used with a repetition time of 6.0 s, while for 27A1 (104.3
MHz) a 1 ps (9=rd12) pulse was used with a repetition time of 0.1 s and for 13C (100.6 MHz)
a 15 gs (0=~/2) pulse with proton high power decoupling and 6.0 s repetition time.
3. R E S U L T S AND DISCUSSION
Fig. 1 shows the X-ray diffraction patterns of the products recovered after various
delamination steps:
9 swelling of a MCM-22(P) sample with Si/A1 = 50 at 80 ~ (swollen product)
9 ultrasound treatment for 1 hour at 50~ product)
9 calcinations at 540~ (ITQ-2)
The XRD patterns of the ITQ-2 samples markedly differ from those of MCM-22(P),
showing a general decrease in intensity and a broadening of the same peaks consistent
with the reduction of particle size to nanometric scale. A relevant loss of crystallinity
of the material is obtained after the swelling treatment, while the ultrasounds treatment
does not promote relevant amorphization, as also confirmed by electron scanning
microscope analysis. The complete delamination is obtained after the final
I2ounteractio.
a;
~
ra~ SONICATEO
swoLLEN,
Fig. 1. X-ray diffraction patterns of intermediate products obtained during the preparation of
ITQ-2 from MCM-22(P) prepared from gels with Si/AI=50.
274
The crystallinity loss corresponds to a significant decrease of the Si/A1 ratio in the
intermediate products (Table 1), for both those obtained from MCM-22(P) synthesized
from mixtures with Si/A1 ratio 20 and 50.
Table 1
Elemental analysis of intermediate solid products in the preparation of ITQ-2
Si/A1 ratio in Si/A1 ratio Si/A1 ratio Si/A1 ratio after Si/A1 ratio
the synthesis in MCM- after swelling ultrasound after
gel 22(P) treatment calcination
20 19.8 8.5 9.5 9.2
50 30.0 11.9 10.8 11.0
SONICATED PRODUCT
MCM-22(P)
It can be hypothesized, in agreement with Schenkel et al. [12,13], that the swelling of MCM-
22(P) in the presence of surfactant and hydroxide ions is responsible for both the partial
destruction of the zeolitic framework and the solubilization of the giO44- species.
The 29Si-NMR data confirm the progressive desilication in all delamination steps.
275
The spectra of the product recovered during the delamination process show an increase of the
intensities of the lines a t - 1 0 4 a n d - 9 3 ppm assigned to (Si)3Si-OH (Q3 groups) and
(Si)2Si(OH)2(Q2 or geminal groups), respectively and the corresponding decrease of the line
at ca -114 ppm assigned to (SiO)4Si. In addition, each delamination step results in a
broadening of lines in the 29Si NMR spectra.
The 27Al_NMR spectra also performed on the sample recovered after each delamination
step, show the presence of tetrahedral aluminium and alter ~ounteractio the presence of isolated
extra framework aluminium in an octahedral coordination. Decrease of chemical shift could be
related to an increasing distortion of the aluminium atom, caused by a partial destruction of the
structure during the swelling process.
i4() i015 dO "20 b -~20' go '-100'_i40 i46 i06 do' 20 b -~20' -60 '-i00' lz~O
8 (ppm) 8 (ppm)
SWOLLEN PRODUCT
[4d [0d 60'2b b-~20' '60 "100'_140 i40 i0d 60'2()b'~20' '60"i()0'iz~0
15(ppm) 8 (ppm)
Fig. 3.27AI_NMR spectra of intermediate products obtained during the preparation of ITQ-2
from MCM-22(P) prepared from gels with Si/AI=50.
The 27 A1-NMR spectra of the liquid phase recovered for the system with Si/AI=50 after
the swelling procedure show the presence of one line at ca 70 ppm due to Al(Osi)2(OH)2-. The
presence the aluminium in the liquid phase recovered for the system with Si/AI=20 is lower
implying a minor destruction of the zeolite framework (Fig.4). The negligible dealumination is
also confirmed by chemical analysis of the liquid phase; in fact, the measured aluminium
concentration is insignificant compared to the silicon concentration.
It was already reported that several thermal effects are visible in the DSC curves. These
effects can be classified into "low", "middle" and "high" temperature ranges [ 14]. In particular,
the presence of peaks at high temperature ranges is directly related to the non-extractable HMI
molecules. In Fig.5 DSC curves of the products obtained during the delamination process are
reported. The temperature of exotherms at high temperature attributed to the loss of fragment of
276
| ' | ' l ' | ' l ' | l ' | ' l ' | ' | ' l l ' l ' ,
140 100 60 20 0 -20 -60 -100-140 140 100 60 20 0 -20 -60 -100-140
8 (ppm) 8 (ppm)
Fig. 4. 27AI_NMR spectra of liquid phase recovered after swelling of system with Si/A1-50 (A) and
Si/AI=20 (B).
100
j,
" \ \\\ A
uct
40
100
\ 90
\
\ 80
\
70
\ ~ swollen product
60
50
f ~ .,.
40
100
\ % 95
- 90
f ~ 85
~" 80
,,.,
Fig. 5. TG ( .... ) and DSC ( - - ) curves for intermediate products obtained during the preparation
of ITQ-2 from MCM-22(P) prepared from gels with Si/AI=50
277
organic molecules [15] adsorbed on the acid sites decreases during the delamination processes
from 550~ to 420~ This indicates that the amorphization of materials decreases the acid
strength of aluminium sites. The weight loss in the high temperature range is nearly constant in all
the samples indicating that the HMI species are always present as []ounteraction. Probably, the
aluminium is protected by organic cations used in the synthesis of MCM-22(P). This is also
confirmed by thermal analysis. In fact, after all delamination steps, obviously except the
[] ounteractio, the organic cation always remains as Dounteraction in the zeolitic structure.
In Table 2 are summarized the results obtained for the surface measurement by
the N2 adsorption isotherm, at different time of ultrasounds treatments. They show a
progressive increase of total surface and mesoporous area increasing with sonication
time.
Starting from MCM-22(P) precursor with Si/AI=20, a longer ultrasound time
promotes the undesired formation of MCM-41 mesoporous materials.
Table 2
Surface properties of the different samples derived from nitrogen adsorption
isotherms.
Si/A1 ratio in the synthesis Time of Total Surface, Mesoporous Surface,
gel ultrasound mZg-I m2g -I
treatment,
minutes
20 15 832 785
90 840 796
120 985 960
*600 1010 1010
50 15 452 91
90 560 98
120 580 121
600 630 280
1200 760 600
* Mesoporous sample material type MCM-41.
Fig. 6. Scanning electron micrographs of the MCM-22 precursor obtained from gels with
Si/AI= 50 (A); sample of the final ITQ-2 (B).
278
All MCM-22(P) crystals synthesized exhibit typical platelet morphology, while ITQ-2
particles exhibit irregular shape and size, as confirmed by scanning electron
microscope analysis (Fig. 6).
4. C O N C L U S I O N S
The above reported experimental results confirm, in good agreement with those previously
reported in the literature, that the preparation of material ITQ-2 is favoured by decreasing
aluminium content of the precursor zeolite MCM-22 (P). Moreover, the ultrasound treatment
time is a very important parameter to obtain a final product with good adsorption properties.
The sonication time necessary to obtain a significant exfoliation of MCM-22 precursor
obtained from gels with Si/AI=20 is longer compared with that necessary to delaminate
MCM-22 precursor obtained from gels with Si/A1-50.
The 29Si-NMR data confirm the progressive desilication in all delamination steps. The
27A1-NMR performed also on the sample recovered in every delamination step shows the
presence of tetrahedral aluminium and, only after calcination, the presence of isolated extra
framework aluminium in an octahedral coordination.
Probably, the aluminium is protected by organic cations used in the synthesis of
MCM-22(P). This is also confirmed by thermal analysis. In fact, after all delamination steps,
obviously except calcination, the organic cation always remains as countercation in the
zeolitic structure.
REFERENCES
Hydrothermal syntheses leading to clathrasils were carried out under high pressures of
different gases (CO2, SF6), employing organic or organometallic molecules as structure-
directing agents (SDAs). The type of clathrasil formed depends on the chosen SDA and on the
specific gas used. The gas molecules as well as the SDA molecules are occluded in the cages
of the clathrasils, as proven by different methods. The gases act either as "co-SDAs" or can
themselves take the role of a primary SDA, when no other SDA molecules are present. Here,
we report about the influence of high pressures of CO2 and of the inert gas SF6.
1. I N T R O D U C T I O N
Clathrasils are microporous solids consisting of a pure silica framework and belong, together
with the zeosils, to the class of the porosils [1,2]. While the pore geometry of the zeosils is
channel-like, clathrasils possess only cage-like voids which contain the molecules of the
structure-directing agent (SDA). Typically, these SDAs are organic or organometallic
molecules, e.g. 1-aminoadamantane (AAD) or the cobaltocenium cation [3, 4]. The size and
the geometry of the SDA molecules, which become occluded in the voids during
hydrothermal synthesis, have an influence on the resulting pore geometry of the clathrasil
and, therefore, on the type of structure which is formed.
The structure of clathrasils can also be influenced by different gases, when these are
present at high pressures during synthesis [5,6]. The gas species then occupy small pores of
the framework. Usually, the presence of a high pressure of an inert gas will favour the
formation of crystals which are larger and of a higher quality. The gas molecules then act as a
"help gas" [6]. In certain cases, however, the gas molecules may favour the formation of
another structure type. Then, they may be designated as "co-structure directing agents" ("co-
SDAs") [5,6]. This case is observed when a certain structure type does not offer small cages
to host the gas species. For example, under certain reaction conditions, the cobaltocenium
cation as SDA induces the formation of the octadecasil (AST) structure [4]. Using the same
conditions, but applying a high pressure of a noble gas (Ar [5], Kr, Xe [7]), the dodecasil-lH
(DOH) structure is formed. Whereas octadecasil possesses a large cage for the incorporation
of the primary SDA molecules, its smaller [46] cages are too small to host a noble gas atom.
On the contrary, the DOH structure (Fig. 1) contains a large [5126s] cage with a volume of
about 430 fit 3 to host the primary SDA species as well as s m a l l e r [512] and [435663] cages
(volume ca. 80 A 3 each) which are large enough to host the gas atoms. When no organic
compounds are present in the synthesis mixture, inert gases can even themselves act as
primary SDAs. For example, Kr and Xe direct the formation of dodecasil 3C or
melanophlogite [8].
280
Fig. 1. Framework structure of DOH with its different cage types (oxygen atoms omitted for clarity)
Extending our work on the influence of high pressures of gases on the synthesis of
clathrasils, we here present results obtained with the molecular gases CO2 and SF6. The inert
gas SF6 was shown before to act as a primary SDA, inducing the formation of dodecasil 3C
(MTN) [9]. In addition to the small [512] cages, the MTN structure (Fig. 2) exhibits larger,
nearly spherical [51264] cages, which are able to host the SF6 molecule.
(3
2. EXPERIMENTAL
The syntheses of clathrasils were either carried out using standard procedures for the
preparation of fluoride-containing syntheses gels [4] or by dissolving AAD in 2M aqueous
ethylendiamine solution with subsequent addition of tetramethoxysilane (TMOS) [6]. After
preparation, the gel or solution was transferred to a Teflon-lined high-pressure steel autoclave
and a high pressure of the corresponding gas (50-800 bar) was added at room temperature.
Reaction mixtures were heated to 180 ~ for four weeks (or eight weeks when only a gas, but
no other SDA was present). Finally, the product was filtered off, washed with deionized water
and dried in air.
Fig. 3. Structural modelling studies of the C02 molecule in the [435663] cage (left) and in the [512] cage
(right)
282
Fig. 4. Results of the structural modelling study of the SF6 molecule in the [5 ~2] cage (left) and in the
[435663] cage
The synthesis with AAD as primary SDA in the presence of SF6 yields DOH, as could be
expected. However, the hexagonal crystals formed have a strange morphology, possessing a
hole in the center of each hexagonal platelet (Fig. 5). The synthesis of these "doughnut-like
crystals" is reproducible.
A possible explanation for the formation of the holes might run along the following
ideas: As primary SDAs, the AAD molecules exert a strong structure-directing effect and
cause the initial formation of DOH crystals. When the concentration of AAD in the reaction
mixture decreases, SF6 is able to compete with AAD for the occupation of the large cage of
the DOH structure. Due to slight changes in the cage geometry, this will possibly lead to
strain in the crystal, resulting finally in fracture, thus producing the holes.
283
i ......~ ......~iii!!iiiii~i
i~!!!!ii~iiiii~i
2::• ............... !
!. . . . . . . . . . . :,,~i!~~<~.....:.::.~. ~~20 Hm
Fig. 5. Scanning electron microscopy pictures of AAD-(SF6)x-DOH crystals. Below: Hexagonal hole
in the center of a crystal. Above: A part of the crystal that might be broken out of the middle of the
platelet.
284
90
TI%
80
70
6O
50
40
30
20
10
I
In the synthesis using only SF6 as an SDA, the X-ray diffraction pattern shows that MTN
was formed, as reported before [8]. Infrared spectroscopy demonstrates that SF6 is occluded
in the framework (Fig. 6).
A structural modelling study shows that SF6 fits quite well in the [51264] cages of the
MTN structure (Fig. 7). Thermogravimetry of the product shows that about five of the eight
[51264] cages per unit cell of the dodecasil 3C structure are occupied by an SF6 molecule. This
value may, however, be influenced by a small amount of an unidentified impurity. Thus, SF6
can assume the role of a primary SDA when no other SDAs are present.
Differential scanning calorimetry of the MTN crystals shows signals in two temperature
ranges (Fig. 8), indicating that a thermally induced phase transition occurs between 0 ~ and
-5 ~ during heating and between -40~ and -45 ~ during cooling.
2 ."
Fig. 7. Result of a structural modelling study of SF6-MTN: orientation of an SF6 molecule in the
[51264] cage.
285
dQ/ dt endo
- l heating
_ e
-50
T~ ~
Fig. 8. Differential scanning calorimetry of SF6-MTN
Klemperer et al. [11] also found phase transitions in a variety of MTN compounds.
Temperature-dependent measurements of X-ray diffraction patterns of the product give
evidence that there is a distortion of the cubic structure to an orthorhombic structure at about
0 ~ Fig. 9 shows the changes of the 400 peak (cubic system) in the powder diffraction
patterns as a function of temperature. Moving from 140 ~ to lower temperature an increasing
broadening is detected. At T = 25 ~ a shoulder can be observed which at lower temperatures
clearly splits into two different peaks (400 and 040/004) of an orthorhombic structure.
Intensity / a. u.
1200
1000
800
600
' "~~/'~of-~-~,~,~ 140 ~
400 25 ~
5~
0~
200
-10 ~
-50 ~
I I I I I
17.5 18.0 18.5 19.0 19.5
~
Fig. 9. Details of the powder X-ray diffraction pattem of the clathrasil SF6-MTN at different
temperatures.
286
ACKNOWLEDGEMENTS
We thank the Fonds der Chemischen Industrie for support of our work.
REFERENCES
O.V. Shvets a, N.V. Kasian a, S.V. Kolotilov a, S.A. Sergienko a, E.E. Knyaseva b, V.G. Ilyin a
The search of templates from the number of nitrogen-containing organic compounds, mono-
and polynuclear complexes of 3d metal was performed; several bulky templates with
quaternary nitrogen atoms, promising for preparation of new types of extra-large porous
zeolites, were synthesized. The conditions of their preparation were studied and factors,
which have the influence on formation of zeolites with large micropore volume, were
established. Two new zeolite phases were synthesized, XRD spectra of which contained low
angle reflexes in range 4.2-5.7 A New silica/germanium oxide and alumina/germanium
dioxide analogues of ASU-16 zeolite were prepared.
1. INTRODUCTION
2. E X P E R I M E N T
The synthesis was performed using the number of amines, diamines, nitrogen-containing
heterocycles, quaternary ammonium bases, mono- and polynuclear complexes of 3d metals,
as well as the number of specially prepared organic compounds (tetra-4-pyridinylpyrazine and
its derivatives, as well as substituted diamides) as templates in Li, Na and K hydroxide media.
For the synthesis of tetra-4-pyridinylpyrazine and its derivatives bis(4-pyridyl)benzoine was
converted to tetra-(4-pyridyl)pyrazine similarly to reported methods [8]. The latter was treated
288
with alkylating agents (C2HsBr, C2HsI, C6HsCHzBr) to give desired N-alkylated compounds.
Oxamide templates were prepared by reaction of oxalic acid diethyl ester with 3-(N,N-
diethylamino)propylamine or 3-(N-imidazolyl)propylamine followed by alkylation of the
product with ethyliodide [8]. Ni(Dtdien)(C104)2 (Dtdien -- 1,14-diamino-4,1-dimethyl-6,9-
dithia-3,12-diazatetradeca-3,11-diene), Ni4(OH)z(Inaf)4(Dien)2(C104)2 (InafH = 1-oxime of 2-
phenyl-l,2-etanedione, Dien = 1,5-diamino-3-azapentane) were prepared as reported
previously [9, 10], Cuz(Bdado)2(CIO4)2 (BdadoH2 = 2:1 Shift base of 2-oxime of 2,3-
butanedione and benzidine) was synthesized by modification of described procedure [11].
This work presents some results, the most successive in our opinion. XRD spectra for selected
zeolite samples, synthesized in the presence of the number of nitrogen-containing templates,
are shown on Fig. 1-2. These samples have reflexes in the region of rather low angles within
20 range 4.28-5.57 ~ (d = 20.6-15.9 A). It can be expected that formation of cells with
comparatively large parameters at inclusion of template molecules in zeolite structure results
in formation of zeolite framework with extra-large channels or cavities. Such supposition may
be partially confirmed by the results of adsorption measurements, and especially by
adsorption of molecules-sondes. However, final proof of extra-large porous zeolite formation
may be obtained after determination of its crystal structure.
Two new zeolite phases, called IPC-1 and IPC-2, were prepared in this study. For
preparation of IPC-1 zeolite phase in the presence of octylamine, carrying of the synthesis in
KOH medium, followed by hydrothermal treatment (HTT) at 155-160~ during 4 days
represents optimal conditions (Fig. 1). Lesser crystallization times reduced the degree of
crystallinity of zeolite phase, while higher caused generation of up to 40 % of o~-quartz phase
along with target IPC-1. Carrying of the synthesis in NaOH medium in all cases resulted in
formation of more dense zeolite phases, for example, ZSM-11 or ZSM-5 (structures MEL and
MFI). Zeolite phase IPC-1 also crystallized from more alkaline reaction medium at long gel
aging at 70 ~ with mixing, whereas short-time gel aging at room temperature favored to
formation of IPC-2 phase.
Similarly, IPC-1 zeolite phase crystallized in the presence of melamine, optimal
condition is HTT in KOH medium during 10 days at 175 ~ (Fig. 1). Lesser crystallization
times and reduction of pH resulted in formation of impurity phase of tridymite, whereas
longer preparation times caused formation of quartz. Replacement of KOH by NaOH
produced ZSM-11 phase in addition to IPC-1.
The use of tetra-4-pyridinylpyrazine as template resulted in formation of the mixture of
zeolite phases IPC-1, IPC-2 and some quantity of tridymite, whereas the latter didn't form in
the presence of tetra(N-ethylpyridinium) pyrazine iodide (Fig. 1). Similarly to the previous
case, introduction of Ni(Dtdien)(C104)2 in reaction mixture favored to formation of the
mixture of IPC-1 and IPC-2 zeolite phases (Fig. 1). IPC-1 zeolite phase with some admixture
of mordenite (MOR) also formed in the presence of Co(phen)3C12 (phen = phenantroline) in
result of HTT at 155 ~ during 7 days in NaOH medium, whereas increase of pH and
template content led to formation of MEL impurity phase instead of MOR. We also observed
clearly pronounced emission for the samples with high content of cobalt complex, which may
be additional proof of Co ions inclusion in the structure (Fig. 1).
1.5-fold increase of octylamine content in reaction medium resulted in formation of IPC-
2 phase with impurity of tridymite (Fig. 2), whereas 2-fold increase of amine content caused
generation of ZSM- 11 phase and traces of tridymite. Increase of gel aging time produced IPC-
289
1 phase and resulted in partial amorphization of this phase. Relative pH reduction also
allowed to move to the region of IPC-1 phase crystallization.
3 7
2000:
1500 2 6
1000
500 1 5
10 20 30 40 50 10 20 30 40 50
20, degree
Fig. 1. XRD spectra of zeolite samples with IPC-1 structure prepared in presence of octylamine (1 and
2), tetra-4-(N-ethylpyridinyl)pyrazine (3), tetra(N-ethylpyridinium) pyrazine iodide (4),
Ni(Dtdien)(C104)2 (5 and 6), melamine (7), Co(phen)3Cl2 (8) as template: 1 and 2 - HTT 2 and 4
d at 160 ~ correspondently, 6 and 8 - template content differs in 1.5 from 5 and 7 correspondently.
IPC-2 zeolite with the traces of ZSM-11 also formed when tetra-4-pyridinylpyrazine was
used as template in NaOH medium (Fig. 2). Introduction of seeds of IPC-2 phase, led to
crystallization of the mixture of IPC-1 and IPC-2 with admixture of MOR. The use of tetra(N-
ethylpyridinium) pyrazine bromide as template allowed to prepare high crystallinity IPC-2
phase with some impurities of ZSM-11 and tridymite just after 6 days of HTT at 155 ~ in
NaOH medium (Fig. 2), at this case increase of template content was not desirable because it
resulted in transformation of IPC-2 phase into tridymite with MOR impurities. Replacement
of template by tetra(N-ethylpyridinium) pyrazine iodide usually resulted in formation of the
mixture of IPC-2 with IPC-1 under similar conditions (Fig. 2). IPC-2 phase in the most pure
form (with the traces of amorphous phase) could be prepared for 7 days with the use of
tetra(N-phenylpyridynium) pyrazine bromide, in this case increase of template content caused
generation of some quantity of MOR, whereas increase of pH favored transformation of
zeolite phases into tridymite. Increase of synthesis duration to 9 days caused crystallization of
the mixture of IPC-1 and IPC-2 with tridymite impurities and, after 15 days, pure tridymite.
The use of templates of other nature, in particular, oxalic acid bis-(N,N,N-
triethylammonium)propylamide diiodide and oxalic acid bis-(N-
ethylimidazolium)propylamide diiodide in NaOH medium, also resulted in formation of IPC-
2 phase with impurities of ZSM-11 and ZSM-22, correspondingly (Fig. 2). In both cases
increase of synthesis duration caused formation of quartz, and increase of pH of reaction
medium led to tridymite admixtures.
290
. . . . . . . . . . . . . . . . . , . . . . . . . . . , . . . . . . . . . , . . . . . . . . . . . , . . . . . . . . . , . . . . . . . . . . . . . . . . . . . ,.
10 20 30 40 50 l0 20 30 40 50
20, degree
Fig. 2. XRD spectra of zeolite samples with IPC-2 structure prepared in presence of octylamine (1),
tetra(N-ethylpyridinium) pyrazine bromide (2), tetra(N-ethylpyridinium) pyrazine iodide (3), tetra(N-
phenylpyridynium) pyrazine bromide(4), oxalic acid bis-(N,N,N-triethylammonium) propylamide
diiodide (5), oxalic acid bis-(N-ethylimidazolium)propylamide diiodide (6), Ni(Dtdien)(C104)2 (7)
and Cuz(Bdado)2(C104)2 (8).
In similar conditions crystallization of IPC-2 zeolite phase with tridymite impurities could
be caused by introduction of Ni(Dtdien)(C104)2 in reaction medium, here successive increase
of template content gave rise to formation of MOR impurities, and then - the mixture of IPC-
1 and IPC-2 phases (Fig. 2). Similar IPC-2 phases with some portion of amorphous phase
were also prepared in the presence of two other complexes - Ni4(OH)z(Inaf)4(Dien)2(C104)2
and Cu2(Bdado)2(C104)2 (Fig. 2), in this case the attempts to improve crystallinity by increase
of synthesis duration failed because of transformation of IPC-2 phase into MOR and
tridymite.
Two new zeolite phases IPC-1 and IPC-2, prepared by us, can apparently be synthesized
both from sodium and potassium media in the presence of a number of templates with close
sizes, which can be the evidence that the influence of alkali metal cations on formation of
these structures' SBU was not significant. Carrying of the synthesis in high silica reaction
media didn't cause the necessity of inorganic cations participation in compensation of zeolite
framework charge, caused by isomorphous introduction of aluminium ions in the structure,
that is why the cations only had the function of alkaline medium creation and redissolving of
amorphous silica in high silica zeolite. Both phases show clearly expressed thermodynamic
and kinetic instability, because already insignificant variations of synthetic parameters
provoke formation of more dense and stable zeolite phases, such as MOR, ZSM-11, 12, 5, 22,
analcime, and in limit case - quartz and tridymite.
Obviously, both zeolite phases mainly consist of non-planar five-membered rings
(especially IPC-1) with dominating interplane separations (apparently, O-O) of 3.43 and 3.46
291
A for IPC-1 and IPC-2 correspondingly, which may be testified by the presence of single (or
main) reflexes in XRD spectra at 20 about 25.8 ~ However, SBU of IPC-2 is apparently more
manifold, which can be concluded from the presence of a couple of lower interplanar spacing.
Table 1
New zeolite phases synthesized in the presence of various temp!ates.
template T, ~ zeolite
duration, d
1. Octylamine 155.4* IPC- 1(4.43 ~)
2. Octylamine s 155.4 * IPC-2 + Trid~cmite
3. Melamine 175.10" IPC-1 (4.90 A)
4. Melamine 180.6 * IPC-1 + Tridymite
5. Melamine 180. 10 IPC- 1 + quartz
6. Melamine 175.7 # IPC- 1 + MEL
7. Tetra-4-pyridinylpyrazine 155.5 # IPC- I+IPC-2+MOR
8. Tetra(N-ethylpyridinium) pyrazine bromide 155.6 # IPC-2+MEL+TRI
9. Tetra(N-ethylpyridinium) pyrazine iodide 155.6 # IPC-2 + IPC- 1
10. Tetra(N-phenylpyridynium) pyrazine bromide 155.7 # IPC-2(5.60 ~) + am.
11. Oxalic acid bis-(N,N,N- 155.6 # IPC-2 + MEL
triethylammonium)propylamide diiodide
12. Oxalic acid bis-(N-ethylimidazolium)propylamide 155.3 # IPC-2 + TON
diiodide
13. Ni(Dtdien)(C104)2 ~ 155.7 # IPC-2 + MOR
14. Ni(Dtdien)(C104)2 155.7 # IPC-1 + MOR
15. Ni(Dtdien)(C104)2 ~ 155.7 # IPC- 1 + IPC-2
16. Co(phen)3C12 s 155.7 # IPC- 1 + MEL
17. Co(phen)3Cl2 160.7 # IPC- 1 + MOR
18. Ni4(OH)z(Inaf)4(Dien)2(C104)2 155.3 IPC-2+amorphous
19. Cuz(Bdado)2(C104)2 ...... 155.3 IPC-2+amorphous
f~ - lower template content on '/4, s - 1.5-fold higher template content, * and # - K+ and Na+ medium
The studies of zeolite phases with IPC-1 and IPC-2 structures by SEM showed formation
of lamellar crystals (Fig. 3), aggregated in more complex globular structures, and larger
crystals of other zeolites also were found in the case of two-phase systems.
Micropore volume for the majority of samples, synthesized by us, was between 0.10-0.18
cm3/g (for example, see Fig. 4a), which is rather high, supposing that extra-large porous
zeolite with one-dimensional channels formed. Micropore volume of the same samples,
measured by nitrogen adsorption was in range from 0.09-0.15 cm3/g. Somewhere reduced
micropore volume, calculated from nitrogen adsorption compared to adsorption of water or
methanol, may be caused by the presence of some quantity of very narrow pores, which may
be filled by H20 or CH3OH, but not by N2 molecules. Total adsorption volumes, measured
from CC14 adsorption for the most crystalline samples, containing IPC-1 and IPC-2 phases,
were 0.26 and 0.32 cm3/g, correspondingly.
Zeolites, synthesized by us, were characterized by very strong (in zeolite chemistry)
acidic centers. For example, according to the results of preliminary studies, H-form of zeolite
IPC-1 (Fig. 4b) had high-temperature maximum of cumene cracking (method of temperature
programmed reaction) at 400 ~ and low-temperature maximum near 155 ~ evidencing the
formation of high number of weak acidic centers along with significantly lower number of
very strong acidic centers. Hammett acidity of the latter was equal to H0 = -12 (for
comparison, this values for zeolites H-ZSM-5, H-ZSM-11, H-MCM-22, H-B and MOR are -
8.2,-5.6, -8.2, -8.0 and -12.1 respectively) [12]. IR-controlled adsorption ofpyridine vapor on
IPC-1 and IPC-2 samples showed formation of small quantity of Bronsted acidic centers,
stable till 600 ~ Thermal programmed desorption of ammonia revealed the presence of two
types of acidic centers, with corresponding desorption maxima at 175 and 495 ~ for IPC-1,
which well correlated with the results of cumene cracking and confirmed the formation of
strong acidic centers. No clear peak in high-temperature region was detected for IPC-2 by
thermal programmed desorption of ammonia, which may be explained by high diversity of
acidic centers in this structure.
0,15] a / ~ 1,0-
~ 0,8
>" o, lo
S "~ 0,6
0,4
0,05.
0,2
0,00 ................................................... 0,0 9 , . , . , . ,
0,0 0,2 0,4 0,6 0,8 1,0 100 200 300 400
P/Ps Temperature, ~
Fig. 4. Isotherm of methanol adsorption (a) and catalytic cracking of cumene (b) (with mass-
spectroscopy control of 42 (1) and 43 rn/e (2)) on H-form of zeolite IPC-1.
Thermal stability of IPC-1 and IPC-2 zeolite structures was measured by thermal
gravimetric analysis. It was found that partial disordering of these structures began at 560 ~
and 520 ~ with full collapse at 890 ~ and 875 ~ respectively.
One more promising approach to creation of new types of zeolites, including extra-large
porous zeolites, may be presented by isomorphous (or non-isomorphous) introduction of other
elements, which significantly differ by their radius and coordination number, in silica or
germanium oxide structures. Creation of the structure from complex oxide systems allows to
use as primary building units not only tetrahedrons, but other polyhedrons, such as trigonal
bipyramides, octahedrons, three-membered rings, etc., which in turn opens wide ways for
creation of new types of secondary building units (SBU), and, hence, crystallization of new
293
10000.
5000~ ~ ...................
~j' ~ ~ i
. . . . . . . . . . . . . . . . , . . . . . . . . . , . . . . . . . . . . . . . . . . . . . , .~ . . . . , . . . . . . . . . , . . . . . . . . . , . . . . . . . . . , . . . . . . . . . ,
10 20 30 40 50 10 20 30 40 50
20, degree
Fig. 5. XRD spectra of zeolite samples with structure analogues ASU-16 prepared in presence
diaminohexane as template: 1 - pure GeO2; 2, 3 and - Ge/Si = 4, 1.5 and 0.66 respectively; 5 and 7 -
Ge/AI = 20 and 10; 6 and 8 - Ge/Ti = 10 and 5;
As our studies showed, formation of zeolite phases, isostructural to ASU-16, took place at
substitution of up to 40 % of Ge by Si, up to 5 % by A1 and up to 10 % by Ti in reaction
mixture (Fig. 5). Introduction of higher quantities of Si (more than 60 %) resulted in
formation of high crystallinity phases of Ge-containing zeolites with MEL structure. Increase
of A1 content to 10% and Ti to 20 and 40 % of total quantity of polyhedron-forming atoms
resulted in formation of new unknown significantly more dense phases. Said tendencies may
be apparently explained taking in account preferable coordination surrounding of each
element in oxygen-containing units. It is logical to suppose that Si atoms may possess only
tetrahedral positions in described above SBUs, that is why limiting Si content in reaction
mixture, in which the synthesis of zeolites with ASU-16 structure is possible, is equal to 4/7.
Octahedron may be expected to be preferable oxygen surrounding both for titanium and
294
aluminium at the absence of cations in reaction mixture, that is why their limiting content
should be 1/7. Exceeding of these limits should manifestly result in impossibility of
mentioned structure generation because of formation of other SBUs. It is interesting to note,
that introduction of some quantity of NaOH in reaction mixture, containing 10 % of A1,
resulted in formation of two-phase system, containing ASU-16 as one of the phases. This may
be the confirmation of introduction of A1 in tetrahedral positions, where appeared charge of
zeolite framework may be compensated by Na + cations. These results considerably extend the
concept of isomorphism in element-oxygen systems, which, to certain time, first of all
represented introduction of other elements in tetrahedral positions, characteristic for Si in
silica systems. From this viewpoint it can be expected that it is possible to build the structure,
similar to ASU-16, from, for example, 4 tetrahedral SiO4, 1 octahedral TiOsF and 2 trigonal
bipyramidal VO4F SBUs.
4. C O N C L U S I O N
Thus, the influence of chemical nature and the structure of template, reaction mixture pH,
reagents' concentration, the type of alkaline metal, hydrothermal treatment conditions, etc. on
zeolite formation process was determined. Two new zeolite phases were synthesized,
characterized by cell parameters in range 15.78-20.26 A and micropore volume up to 0.18
cm3/g. It was shown that polynuclear complexes could act as templates in synthesis of new
zeolites. In line with determination of structure and optimization of synthetic conditions we
considered the possibility of isomorphous introduction of heteroelements in their structure.
Analogues of ASU-16, containing heteroelements, were synthesized; limiting contents of such
elements, which still allowed to build ASU-16 structure, were determined.
REFERENCES
[1] R.F. Lobo, M. Tsapatsis, C.C. Freyhardt, S. Khodabandeh, P. Wagner, C.Y. Chen, K.J. Balkus,
S.I. Zones, M.E. Davis, J. Am. Chem. Soc., 119 (1997) 8474.
[2] P. Wagner, M. Yoshikawa, M. Lovallo, K. Tsuji, M. Taspatsis, M.E. Davis, Chem. Commun.,
1997,2179.
[3] A. Burton, S. Elomari, C.Y. Chen, R.C. Medrud, I.Y. Chan, L.M. Bull, C. Kibby, T.V. Harris,
S.I. Zones and E.S. Vittoratos, Chem. Eur. J., 9 (2003) 5737
[4] K.G. Strohmaier and D.E.W. Vaughan, J. Am. Chem. Soc., 125 (2003) 16035
[5] T. Cheetham, H. Fjellvag, T. E. Gier, K. O. Kongshaug, K.P. Lillerud and G.D. Stucky, Stud.
Surf. Sci. Catal, 135 (2001) 788.
[6] J.-L. Paillaud, B. Harbuzaru, J. Patarin, N. Bats, Science 304 (2004) 990
[7] A. Corma, M.J. Diaz-Cabanas, F. Rey, S. Nicolopoulus and Kh. Boulahya., Chem. Commun.,
(2004) 1356.
[8] Modem methods of experiment in organic chemistry. GNTI KhimLit, Moscow, 1960, 204.
[9] V.V. Pavlishchuk, S.V. Kolotilov, A.W. Addison, R.J. Butcher, E. Sinn J. Chem. Sot., Dalton
Trans. (2000) 335
[10] V.V. Pavlishchuk, S.V. Kolotilov, A.W. Addison, M.J. Prushan, D. Sehollmeyer, L.K.
Thompson, T. Weyhermiiller, E.A. Goreshnik, J. Chem. Soc., Dalton Trans. (2003) 1588
[11 ] A.S. E1-Tabl Trans. Met. Chem., 27 (2002) 166
[12] V.V. Brei, Theor. Exp. Chem., 39 (2003) 70
[ 13] J. Plevert, T.M. Gentz, A Laine, H. Li, V.G. Young, O.M. Yaghi, and M. O'Keeffe, J. Am.
Chem. Sot., 123 (2001) 12706
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 295
Density Functional Theory calculations are used to probe the energetics of the formation of
key pre-nucleation species implicated in the formation of zeolites. We demonstrate that the
role of solvent and pH is to control specific condensation reactions, especially those involving
cyclization. Our results are compared with available experimental data relating to silicate
species in synthesis gels.
1. INTRODUCTION
The widespread application of zeolites has been the main driving force behind attempts at
synthesising novel materials, leading to the considerable growth in the number of unique
structures discovered [1]: siliceous zeolites are an important sub-group that has seen many
new materials formed. However, the shear complexity of zeolite synthesis (although at first
hand, we simply have a condensation polymerisation reaction) provides many challenges to
the zeolite chemist who wishes to characterise and understand the fundamental processes that
occur during the self-assembly of a microporous crystalline lattice from molecular precursor
species. Nevertheless, it is clear that such a detailed picture offers new opportunities to
improve known zeolite syntheses and to form tailor-made materials.
Many of the silicate clusters present in solution at the pre-nucleation phase of synthesis have
been identified using NMR spectroscopy [2,3]. However the precise charge state of these
species is more challenging to determine and furthermore, the reactions involved in forming
these clusters are very difficult to identify experimentally. Some authors have suggested
structures which are believed to be the building blocks in silicalite-1 [4-6]: although this
proposal has raised considerable debate [3,7]. Scattering methods have identified larger
structures (ca. 2.8 nm) considered primary units in the Si-MFI system [8,9], although their
precise structure is unknown. Given these experimental difficulties, computational methods
provide an alternative method to gain key insights into the formation of the pre-nucleation
species - probing areas that are beyond current experimental methods - and hence evidence
for the mechanisms of zeolite nucleation and subsequent crystal growth.
Thus, over the past few years a number of investigations into the structural chemistry of small
silicate species have been performed. Pereira et al [ 10,11 ] studied neutral silicate structures
up to five silicon atoms using LDA methods. However, one of the outcomes of the methods
chosen in this work - essentially a limit of computer technology at the time - was that linear
condensation reactions were favoured over the formation of cyclic species. Lewis et al [12],
using more approximate molecular mechanics methods, identified the key role of templating
molecules in preventing the collapse of large hydrophobic silicate clusters in the presence of
296
solvent. Thus, it is clear that whilst such methods can provide valuable data on the processes
involved, we must also ensure that, the models and methods used accurately reflect
experimental conditions.
In this work, a model to study the formation of small silicate species, fundamental to the
nucleation of pure silica zeolites, is described. The structures of these species and their related
energetic and thermodynamic properties are calculated with Density Functional Theory
methods. We consider the presence of water using the COSMO (conductor-like screening
model) solvation approach and also account for the high pH of synthesis by considering
deprotonated species.
2. M E T H O D O L O G Y
The calculation of the minimum energy conformation and their related thermodynamic
properties for each of the silicates analysed in this study was performed as follows. A first
approximation of the geometry is obtained from a zeolite crystal. A simulated annealing
procedure is then performed on the silicate cluster, to avoid local minima and bias, with a low
cost Density Functional Theory (DFT) approach (PWC functional). The simulated annealing
is carried out at 700K for 3000 steps of 0.46fs. The resulting structure is then optimised with a
higher level of theory: BLYP functional and DNP basis set. Finally the thermodynamic
properties, at 450K (typical of zeolite synthesis), were calculated using standard statistical
mechanical methods. The solvated models were calculated using the COSMO [13, 14]
approach, with full geometry optimisation of the cluster using the same methods as for the gas
phase calculations. The COSMO model is a conductor-like screening model (continuum
solvation model) where the solute molecule forms a cavity within the dielectric continuum of
permittivity, ~, which represents the solvent. All calculations were performed using DMOL 3
version 2.2 [ 15].
As a first approximation, we model the species and reactions in the gas phase. The
condensation reactions for the formation of silicate oligomers (with an emphasis on cyclic
structures) with up to six silicon atoms are shown in the Fig. 1 (together will all the results for
the solvated structures - discussed below). We consider first the condensation reactions of
neutral species, presented in the first row of each table of Fig. 1. From the free energy of these
neutral reactions, in the gas phase, there is clear tendency to form cyclic species rather than
the linear oligomers. Since cyclic species are found experimentally in the pre-nucleation gel,
these initial calculations suggest that it is essential that we consider the free energy of
formation, rather the simpler case of the enthalpy, if a robust description of the reactions
present is to be achieved.
Whilst clearly, zeolite formation is not a gas phase process, we note that zeolites have been
formed in media much different to water; for example in ionic liquids or in eutectic mixtures
[16]. Similarly, a dry gel conversion method has also been fruitful [17]. Thus, such gas phase
results may help to understand what are the important characteristics during such non-aqueous
synthesis. However, we must of course consider the typical aqueous case.
Inclusion of explicit solvent molecules is expensive, since a large number of solvent
molecules must be considered (with many configurational minima possible) to provide a
reasonable picture of solvation spheres, particularly for the larger silicate species - although
297
,~4 tetramer
M = monomer 3ring Cyclization 4ring Cyclization
D = dimer Reactants T/kS /kG Reactants T/kS /kG
Tr = t i m e r Tr 68 -37 T 83 -59
3r = 3 - r i n g 9 Tr- 63 6 0 T- 62 -25
T = tetramer Tr 77 -61 T 55 -48
4r- 4-ring -5 Tr 70 -21 62 -44
03
Tr-2 67 -22 T-2 69 -37
P = pentamer
5r = 5 r i n g ~.
H = hexamer
6r = 6 r i n g
3ring
Pentamerization Hexamerization
Reactants T/kS /kG Reactants TAS AG
T,M -17 -4 P,M -14 -19
-25 -201
CO T,M- -43-183 ~
O P,M-
T-, M -29 -37 P-, M 1 -22
T, M -20 3 P, M -2 -5
o T,M- -19 -74 ~ ~ P,M- -26 -76 , ~ ..
~03 T-2,M -15:3~~~~ ~ M -15 -40 ,;y~-,,~
we have considered some systems. Thus, we have used a more approximate dielectric
screening method embodied in the COSMO [13, 14] method. The condensation reaction
energetics obtained using this particular model (again with neutral species) are given in the
second row of the two sections of the tables in Fig. 1. The inclusion of the solvation via
this model, results in a significant (and general) decrease in the free energy released during
the condensation reactions. Thus, as the species increase in size, we would expect that the
neutral condensation would become unfavourable. Experimentally of course, polymerisation
(and specifically zeolite crystallisation) typically occurs at extremes of pH (alkaline in the
case of zeolites [18]). Thus, again our results are consistent with experimental observation:
neutral conditions on the whole do not lead to extensive polymerisation.
We now consider further improvements to our model of a gel, to necessarily include a
description of the non-neutral pH. At the highly alkaline conditions typical of zeolite
crystallisation (pH=12.6-14 [18]), the principal effect of the OH-present is to deprotonate the
silicate species. Thus, we present the condensation reactions of the dominant anionic silicate
species at such conditions, both in the gas phase and solvent models (Fig. 1). In the gas phase
we find a general increase in the free energy released during condensation when this model of
high pH is considered; with the exception of the formation of the 3-ring species. Additional
inclusion of solvation (our most "realistic" model considered here) results in a reduction in
the free energy change of these anionic condensations compared to the gas phase (as was the
case for neutral reactions). However, it is clear that the condensation reactions of the anionic
species are considerably more favourable than those of the neutral condensations in solution:
again in agreement with experimental observations.
The net result is therefore, that the solvation and the high pH promote the polymerisation of
larger cyclic silicate oligomers. Furthermore, since these cyclization reactions are more
favourable than the linear growth (e.g. formation of 4-ring compared to formation of a linear
pentamer, both from the linear tetramer) larger solution phases cyclic species (certainly from
6-ring upward) are unlikely to form. We should also note that whilst the principal contributor
to the favourable free energy of the polymerisation reactions is the change in enthalpy (mostly
electronic energy) - thus polymerisation is favoured regardless of whether the product is an
open chain or a cyclic species- it is entropic factors that favours the formation of the cyclic
species presented here [19].
The results presented above suggest that the methods and model, developed above (with the
inclusion of solvation and pH), provides a reasonable qualitative description of a gel prior to
nucleation. Indeed, all of the general trends we obtain, in the context of the relative stability
and reactivity of the various species, agree well with the experimental data. We now compare
these results with the quantitative data available of the experimental deprotonation energetics
of simple silicate species, to assess further the reliability of our calculations. The equilibrium
constants for the successive deprotonation of both silicic acid (the silicate monomer) and
disilic acid (the silicate dimer) have been determined [20, 21 ]: that can be compared directly
to the free energy changes determined here.
The first four deprotonation free energies for the monomer and dimer are plotted in the Fig. 2,
from which it is clear that there is excellent linear correlation between the theoretical and
299
experimental results. We should note that previous studies of deprotonation reactions with the
COSMO model, have considered only organic species [22] and it is gratifying to find equally
good results here, for what is arguably a more complex system given the high charge density
and the partially ionic nature of the bonding. However, we should of course note the
discrepancies in the absolute values of the free energy changes of deprotonation obtained.
Whilst, it may be appropriate to consider this a simple offset, we believe that further
investigation is necessary. For example, concentrations are clearly different: we are
considering an effective infinite dilute solution in our calculations. Similarly, the solvation of
the highly ionic hydroxyl group is less well described by COSMO than the silicate species:
here we favour the inclusion of an additional explicit solvation shell - work that is presently
underway. Nevertheless, the overall agreement is very good and provides further evidence
that the model used is robust, provides certainly semi-quantitative agreement with the
available experimental data and will allow predictions to be made about the formation of
larger silicate oligomers.
3O
_~9= Increasing i o n i ~ l
R2 - 0. monomer
-)4-dimer
-40
-5O
AG (kJ/mol) Theoretical
Fig. 2. Experimental [20] and theoretical deprotonation energies at 298.15K. The most negative
energy corresponds to the first deprotonation free energy.
The 3-ring is the smallest closed structure that can be formed in a silicate structure. However,
the presence of such rings in zeolitic structures is rare; only being present in one high-silica
materials, ZSM-18 [23] and a small number of beryllosilicate and zincosilicates [18].
However NMR studies of reaction gels find a considerable concentration of solution phase 3-
rings [3], a result that is supported by the reaction energetics presented in Fig. 1. However, if
the 3-ring species is apparently stable, why do we not find many more 3-ring containing
structures? Perhaps these rings (which we expect are strained) can readily re-open? Recall
that all of the reactions here are in equilibrium. Thus, we are currently considering the
reaction barriers for these internal condensations.
300
The 4-ring is found in a majority of the known zeolite frameworks; 61 zeolites have the 4-ring
as part of their structure [18]. The calculations in Fig. 1 indicate a favourable 4-ring
formation, supporting the NMR evidence for its occurrence in solution [3].
The 5-ring (shown in Fig. 4), is present in 18 zeolitic structures [ 18], although in contrast to
the 4-ring, there is little suggestion from NMR [3] that these species are present in solution.
However, our calculations suggest that a 5-ring species can be readily formed. Indeed, the
geometry of the 5-ring species is remarkably similar to that in crystalline framework (see
below). Such evidence would suggest that the formation of solution phase 5-rings is therefore
likely, in contrast to the experimental evidence. We note, however, that the population of
pentameric species will be low as the most favourable reaction of a tetramer is that to form a
4-ring. Although this may explain why 4-rings would be more prevalent than 5-rings in
solution, it does not explain, how pentasil materials f o r m - although see below for further
comment. Clearly further study is required here.
Although six-membered rings are found in 39 of the known zeolitic frameworks [18], again as
for the 5-ring, there is no NMR [3] evidence to suggest it exists as a free solution species. Our
calculations suggest that ring closure of a hexameric oligomers is favourable. However, as
mentioned in the discussion above for the 5-ring, the population of free hexamers will be very
low - as the formation of the smaller rings will dominate. We can therefore, consider the
formation of 6-ring structures to result from the fusion of smaller rings rather than growth
from 6-ring species. For example, the six-rings in chabazite may form from the fusion of three
4-rings.
a ~ .... b
Fig. 3. Overlay of the calculated (a) 4-ring (grey) and the one from the MFI crystalline structure and
(b) the calculated 6-ring (grey) and the one from the CHA crystalline structure. The silicon and
oxygen atoms are the larger and smaller spheres respectively.
Figures 3 and 4 show overlays between the calculated (solvated) ring structures and those
found in representative crystalline frameworks. Of note for the 4- and 6-ring structures is that
whilst the Si atom overlap is almost exact, there is less good agreement in the positions of
some of the oxygen atoms. The agreement for the 5-ring is more remarkable. Furthermore,
whilst we have identified a number of alternative conformations for the 4-ring, that result in a
geometry that is significantly different from that found in crystalline frameworks, we have not
located any alternative conformation for the 5-ring. Note also that the 4- and 6-ring structures
301
are considerably more planar than the 5-ring. Thus, one might suggest that the 5-ring, o n c e
formed is more amenable to rapid nucleation, resulting in growth of pentasilic materials and
hence may explain why a significant population of solution 5-rings is not observed. However,
it is clear that further work is required to support such a suggestion.
Fig. 4. Overlay of calculated 5-ring (grey structure) and a 5-ring extracted from the experimental MFI
structure. The hydrogens from the optimized cluster have been omitted for simplicity. The silicon and
oxygen atoms are the larger and smaller spheres respectively.
6. CONCLUSIONS
ACKNOWLEDGEMENT
REFERENCES
The AFI type crystals, grown with co-templates, exhibit distinctly shorter, even flat plate
morphology, opposed to the classic elongated rods. Sorption properties of such flat crystals
are superior to those of the classically prepared materials, independently of an apparent
perfect monocrystal appearance or an imperfect aggregated form. This proves a larger pore
volume accessible to the sorbate molecules, which is a consequence of a higher internal
structure order. The crystal morphology itself plays only an indirect or secondary role for
adsorption process after the main factor, crystal structure. This stands even in spite of a close
mutual interdependence existing between the morphology and the internal structure. The latter
is controlled by synthesis conditions and substituted metal(s), affecting in a synergy.
1. INTRODUCTION
From the point of view of possible applications of the crystals of zeolitic materials, the term
morphology includes a number of aspects and/or features:
- crystal structure as corresponding to symmetry;
- crystal form (aggregate, twin, assembly of single crystals, monocrystal);
- crystal shape (regular, distorted, with defects, etc.);
- dimensions of the crystals in individual crystallographic directions;
- aspect ratios of the dimensions.
This broad meaning of morphology explains the parameters dependent on morphology:
- external surface (largeness and shape perfection);
- internal surface (micropore surface area, occurrence of meso- or macropores);
- access to the pore system for adsorbates;
- diffusion in the pores (depth and rate);
- sorption capacity;
- catalytic properties and activity;
- managing the materials.
That is why the morphology is worth of a better understanding.
Application of co-templates in synthesis gels has been developed as a method for tailoring
morphology of the hexagonal crystals characteristic of AFI type structure and, especially,
their aspect ratio, width to length [ 1-4]. Organic aliphatic acids with the molecule chain length
of 2 to 6 carbon atoms are the best-working co-templates found till now [1-4]. The greatest
304
tailoring possibilities are offered by use of a synergetic effect of the co-templates and 3+ or
4+ metal ions [2-4]. This way, the most flat hexagonal plates with d.'l even over 20 have been
grown for CrAPO-5 [1,2,4]. Positive results have also been achieved for Fe and Si [ 1,4]. In
the presence of the 2+ metal ions, as well as for non-substituted A1PO4-5, the crystals have a
less reduced length and d:l between 0.3 and 1 [1,4]. The latter crystals are, in most cases,
relatively extensively overgrown and do not exhibit an apparent single crystal morphology
common at the substitution of the 3+ ions [1,2,4]. These experimental observations yield an
intriguing question whether such strong effects on the morphology of the AFI type crystals
correspond to or depend on (i) the crystal structure (modified/improved internal order due to
another growth mechanism), (ii) metal substitution (modified/improved internal order due to
another distribution of the force field after introduction of heterocentres), or (iii) both these
factors and/or their synergetic effect.
Several series of AFI type materials, grown with co-templates, in the form of pure A1PO4-5
and its derivatives, substituted with various metals, have been investigated with use of one of
the most sensitive techniques, adsorption. Nitrogen, benzene, and water have been chosen as
the probe molecules with different dimensions and shapes, dipole moments, and interaction
affinity, to get a deeper knowledge about the materials and their pore systems. A comparison
to the "classic", elongated, hexagonal rods [5-10], grown without co-templates, has shed some
more light on adsorption mechanisms [7-9,11], diffusion [12], accessibility to the pores and
adsorbate distribution [ 13,14], and also on adsorbate ordering [6].
2. EXPERIMENTAL
The samples of A1PO4-5 and MeAPO-5 were synthesized both with and without use of co-
templates as described elsewhere [1,2,4,5,8-10]. The thoroughly calcined materials were
characterized by XRD, AAS, TGA, SEM, polarized light microscopy, and various
spectroscopic techniques [1,2,8,10-14]. The isotherms of nitrogen adsorption were measured
with a Micromeritics ASAP 2010 volumetric analyzer at 77 K on samples activated in situ at
673 K under vacuum of at least 10-3 Pa for 10 h. The adsorption isotherms for benzene and
water were determined at room temperature under the static equilibrium conditions, using a
vacuum device equipped with a McBain quartz spring balance and MKS Baratron gauges.
The adsorbates were degassed with a freeze-and-thaw technique.
(
100
z
O
8O
HiE] al3El aa [lib EiH ~ [[HI [11 I]il [NI [11 It It [i Ill I]il IN ni []
13_
!'--
03
E 60
O
t-
O
.m
yo 40
Or)
0 0 0 0.20 M acetic acid
i Z~ i 0.45 M acetic acid
O O O 0.70 M acetic acid
20
v V 0.45 M butyric acid
[] [] [] 0.45 M crotonic acid
I> I> I> 0.45 M maleinic acid
Fig. 1. Isotherms of nitrogen adsorption (open) and desorption (full symbols) for A1PO4-5 samples
synthesized with indicated co-templates. For comparison with the materials prepared with the classic
method, see refs [7-9].
160
140
O O O 0.3 M SiO 2
A z~ 0.3 M Fe(lll) acetate basic
120 0 0 0.2 M Mg acetate
"T
v ~ 0.2 M Co acetate
13.. [] [] 0.2 M FeCI3
~- 100
03 I> I> 0.2 M Fe(II)S04
eo
E
.o. 80
o
60 ~(:00~0O00 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ (:~ ~ 0 0 0 ~
o
60}
"0
9 AA ~ ~h~:::~A A A ~ AA ~ AA ,~ E
40
1.6 ....
1.4
V V V
1.2
0 0 0
O
"1.0 0 0
o
,-- 0.8
.0
g 0.6 O 0
Fig. 3. Isotherms of benzene adsorption on A1PO4-5 samples synthesized with indicated co-templates.
For comparison with the materials prepared with the classic method, see refs [7-9].
1.6
1.4
1.2 0 0
0 0
~ 1.0
0 0
0 0
08 o
w ~oOO 2
_~0.6 A a 0 0 0 0.3 M Si02
a A a 0.3 M Fe(lll) acetate basic
0 0 0 0.2 M Mg acetate
0.4
~> [> I> 0.2 M Fe(ll)S04
0.2
~>t> E> C> ~> ~> I>
0.0 I i I , I ' I i
16 -
14 -
12 -
"T,
O"}
" 10 - O
O
E 0 0 0
t- 8
o I> !>
1> I>
o 6
Fig. 5. Isotherms of water adsorption on AlPO4-5 samples synthesized with indicated co-templates.
For comparison with the materials prepared with the classic method, see refs [7-9].
18
16 -
0 0
0
14 -
0
~c~12 -
o s163 s
0 6 I> !> I>
IE 10 -
0 A 0
C b b b b
O
"~ 8 --
O
"o 6 -
4
o 0 0 0.3 M Si02
- 0 A A A 0.3 M Fe(lll) acetate basic
O O O 0.2 M Mg acetate
0
I> I> I> 0.2 M Fe(II)S04
0 I ' I ' I ' I I I i
Fig. 6. Isotherms of water adsorption on MeAPO-5 materials synthesized with 0.45 M acetic acid and
indicated metal ions. For comparison with the materials prepared with the classic method and other
metals, see refs [1-5] and [7-9].
309
nitrogen that too high contents of heterometals can disturb the synthesis and structure of the
growing crystals. Nevertheless, a higher adsorption found for the co-templated A1PO4-5
materials, following type I isotherms (no type IV isotherms), points out an easier diffusion of
benzene inside the pores as well as larger pore volume accessible to the adsorbate, i.e., a
higher internal order of the crystal structure than in the case of the conventional materials.
4. CONCLUSIONS
Morphology of the AFI type crystals depends on their composition and synthesis conditions.
These two parameters allow the modification and/or control of the morphology by varying
substituted metals and co-templates. The crystals synthesized with co-templates exhibit
superior adsorption properties in relation to the "classic" materials, no matter whether they are
apparently perfect single crystals or of an aggregated (A1PO4-5 and MeZ+APO-5) appearance.
This indicates a higher internal structure order and larger pore volume accessible to
adsorbates. The inhomogeneous distribution of water adsorbed under low relative pressures
implies a structural heterogeneity of the crystals, which seems to be related to the crystal
310
growth progress [14]: the components that form at earlier stages of the growth, i.e., a
dumbbell or "pencil-like" core, can adsorb more water than the others. The co-templates, in a
synergy with the metals, positively influence the structure by broadening the dumbbell core of
the crystal. The kind of the co-template affects the distribution and amount of metals, which
in turn influence the growth and internal order of the crystals. The crystal morphology itself
plays only an indirect or secondary role in adsorption process. Similar dependences have
recently been found for MFI type zeolites [16]. Crystal structure, meant as internal order, is
the main parameter determining the sorption properties. This stands even in spite of a close
mutual interdependence existing between the morphology and the internal structure. The latter
is controlled by synthesis conditions and substituted metal(s), affecting in a synergy.
ACKNOWLEDGEMENTS
The author is thankful to Prof. Dr. M. Rozwadowski, Dr. K. Erdmann, and Dr. R.
Golembiewski, all from the Nicolaus Copernicus University of Torun, Poland, for their
valuable assistance and discussions. The work was partially supported by the Polish
governmental Committee for Scientific Research (KBN) with the grant No 4 T09A 147 23.
REFERENCES
[ 1] J. Komatowski, G. Zadrozna, J.A. Lercher, and M. Rozwadowski, Stud. Surf. Sci. Catal., 135
(2001) 04-P-12.
[2] J. Komatowski, G. Zadrozna, M. Rozwadowski, B. Zibrowius, F. Marlow, and J.A. Lercher,
Chem. Mater., 13 (2001) 4447.
[3] J. Kornatowski, G. Zadrozna, and J.A. Lercher, Stud. Surf. Sci. Catal., 142 (2002) 399.
[4] J. Kornatowski and G. Zadrozna, in "Host-Guest Systems Based on Nanoporous Crystals", F.
Laeri, F. Schtith, U. Simon, M. Wark, Eds., Wiley, Weinheim, 2003, Chapt. 4, p. 64.
[5] J. Komatowski and M. Rozwadowski, Proc. 12th Int. Zeolite Conf., Baltimore, 1998, M.M.J.
Treaty, B.K. Marcus, M.E. Bisher, J.B. Higgins, Eds., Materials Research Society, Warrendale,
1999, Vol. III, p. 1577.
[6] J. Caro, F. Marlow, K. Hoffmann, C. Striebel, J. Komatowski, I. Gimus, M. Noack, and P.
K61sch, Stud. Surf. Sci. Catal., 105C (1997) 2171.
[7] M. Rozwadowski, J. Kornatowski, R. Golembiewski, and K. Erdmann, Langmuir, 15 (1999)
5857.
[8] J. Kornatowski, G. Zadrozna, J. Wloch, and M. Rozwadowski, Langmuir, 15 (1999) 5863.
[9] J. Komatowski and M. Rozwadowski, in Proc. of the 12th Int.. Zeolite Conf., Baltimore, 1998,
M.M.J. Treacy, B.K. Marcus, M.E. Bisher, J.B. Higgins, Eds., Materials Research Society,
Warrendale, 1999, Vol. I, p. 285.
[10] G. Mfiller, E. Bodis, J. Komatowski, and J.A. Lercher, Phys. Chem. Chem. Phys., 1 (1999) 571.
[ 11 ] E. Lehmann, S. Vasenkov, J. K/~rger, G. Zadrozna, and J. Kornatowski, J. Chem. Phys., 118
(2003) 6129.
[12] V. Kukla, J. Kornatowski, D. Demuth, I. Girnus, H. Pfeifer, L.V.C. Rees, S. Schunk, K.K.
Unger, and J. K/~rger, Science, 272 (1996) 702.
[ 13] E. Lehmann, C. Chmelik, H. Scheidt, S. Vasenkov, B. Staudte, J. K/~rger, F. Kremer, G.
Zadrozna, and J. Kornatowski, J. Am. Chem. Soc., 124 (2002) 8690.
[14] E. Lehmann, S. Vasenkov, J. K/~rger, G. Zadrozna, J. Komatowski, t3. Weil], and F. Schfith, J.
Phys. Chem. B, 107 (2003) 4685.
[15] J. Komatowski, Compt. Rend. Chim., 8 (2005), in press.
[ 16] J. Wloch and J. Kornatowski, Langmuir, submitted.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 311
The preparation of a new molecular sieve, CAL-1, was recently reported. It presented a
chabazite-type structure and was prepared starting from a lamellar aluminophosphate (ALPO-
ntu), analogous to silica kanemite, and hexamethyleneimine as structure directing agent.
Calcination of CAL-1 causes the transformation to H-SAPO-34, a molecular sieve for acid-
catalyzed methanol-to-olefins reactions. An ex-situ high-resolution synchrotron radiation X-
ray powder diffraction experiment allowed to monitor the relevant steps of the ALPO-ntu to
CAL-1 transformation and its subsequent calcination to H-SAPO-34.
1. INTRODUCTION
(a)
H (Balls)
9 O (Tubes)
AI (Tubes)
Si (Balls)
9 P (Tubes)
Legend
(a)
Fig. 1. Three-dimensional view of one pore of the CHA-type zeolite, with its acid sites (a) and of the
pore disposition (b).
313
Recently, we were able to prepare a new molecular sieve (named CAL-1) with
chabazite-type structure [ 10], starting from a lamellar ALPO-ntu [11 ], structurally similar to
silica kanemite [12]. Hexamethyleneimine (HMI) and SiO2 were added to ALPO-ntu as SDA,
to obtain this new isotype (see the experimental section for more detail on the synthesis).
This synthesis strategy confers to CAL-1 a peculiar layered packing, producing larger
particles with pseudo-rhombohedral morphology (see Fig. 2a) that, however, retain the
memory of the morphology of the parent lamellar material (indicated by arrows in Fig. 2b). It
is worth noting that this chabazite-related material converted to H-SAPO-34 after calcination
retaining the lamellar morphology of the as-synthesized material, which might confer
interesting catalytic properties. Besides, its low density might result in peculiar mass transfer
features.
2. EXPERIMENTAL
CAL-1 - To obtain the CAL-1 material, SiO2 (Aerosil 200, Degussa) and hexamethyleneimine
(HMI, Aldrich, 99%) were slowly added drop wise to an aqueous suspension of ALPO(kan)
[AIPO3(OH)2(C4H9NH2)], following the procedure described in detail in ref. 10. The molar
composition of the starting mixture was 2 ALPO(kan): 0.4 SiO2:1.5 HMI: 50 H20. The
mixture was aged at room temperature for 48 h, distributed in five autoclaves and then
hydrothermally treated at 473K for 3, 12, 24 and 48 h. After the desired reaction time was
passed, each autoclave was taken from the oven, cooled under tap water to stop the
crystallization. The obtained solid was filtered and washed with copious amounts of water
until the pH was close to neutral and dried at ambient temperature. The 48 h material is the
chabazite-type zeolite CAL-1. H-SAPO-34 was then obtained ("calcined SAPO-34" in Fig. 3)
by heating the CAL-1 slowly to 873K under argon and maintaining this temperature for 6 h
under 02 to remove the organic template.
(a)
(b)
Fig. 2. SEM micrographs of the CAL-1 material; panel b shows the lamellar-like features (indicated
by arrows) of the crystals.
The high-resolution synchrotron radiation X-ray powder patterns (XRPD) from the samples
obtained at various synthesis times, are shown in Fig. 3. It can be seen that, after 3 h of
crystallization, the ALPO(kan) material is already slightly changed, as indicated by the
disappearance of the peaks between 11 ~ and 12 ~ in 20. After 12 h, the CAL-1 crystallization
315
is already started and sharp peaks appear at 20 = 4.89 ~ 6.68 ~ 8.29 ~ 10.66 ~ 12.71, 13.39,
15.78 ~ (d = 9.37, 6.86, 5.53, 4.36, 3.61, 2.91 A). After 24 h the CAL-1 pattern is clearly
present and the low angle peak at 20 = 2.53 ~ (d = 2.91 A) from ALPO(kan) is greatly
reduced.
After 48 h the reaction is complete and the collected XRPD pattern of the final CAL-1 sample
does not find close correspondence to any CHA-type zeolite (Table 1), any of the zeolite
pattems reported in the International Zeolite Association collected structures and, in general,
to any other known pattern in the PDF-2 database. An accurate inspection of the 48 h XRPD
pattern suggested the presence of two phases, one characterized by sharp and intense peaks,
and another one, showing very weak and broad peaks. It was possible to index the peaks of
the CAL-1 main phase and also of the minor phase.
i
Calcined,
LI l J, .... l j ...
48h
5 _,., I I ,.._,Ji CAL-I
. . . . L ~
3h
"-~"---~--'---'-~ T r'--'~--l-'--'-'-~
5 10 15 20
28 [degree]
Fig. 3. X-ray diffraction patterns (taken at the BMlb beamline of the ESRF) of materials obtained
after different reaction times with HMI in the starting mixture and of the H-SAPO-34 sample obtained
after calcination of CAL- 1.
The Rietveld refinement of the collected data allowed the determination of the lattice
parameters (Table 2) of CAL-1. It is worth noting that the CAL-1 cell parameters are similar
to those of a trigonal chabazitic structure, but with a slightly different cell volume with
respect to known H-SAPO-34 precursors. This difference is mainly due to the SDA (HMI)
and to residual butylamine [from ALPO(kan)] precursor, present in the CAL-1 cages.
316
The obtained lattice parameters were compared to those of SAPO-34 related materials
containing a different SDA molecule, from literature data (Table 1) and by us, with different
adopted synthesis strategies (Table 2). SAPO-34(kan) and SAPO-34(ge/) show very similar
lattice parameters, thus suggesting that the preparation route does not influence significantly
the crystal structure, although the resulting morphology is very different, since SAPO-34(kan)
has the lamellar-like features like CAL-1 in Fig. 2b. Conversely, the presence of HMI in
CAL-1 causes also structural adjustments and cell volume changes with respect to SAPO-
34(ge/), likely due to the different SDA size. As a final note, the lattice parameters of SAPO-
34(ge/), determined by single crystal [13] and powder diffraction (this work), reported in
Table 2, are in rather good agreement, indicating that the single crystals chosen for the
structural determination [13] are rather representative of the whole SAPO-34(ge/) sample.
Table 1
Lattice parameters (in the rhombohedral setting as reported in the original papers, for a better
comparison with the low symmetry triclinic polymorph) of the most representative CHA-type
materials from literature data.
Material SDA or Channel Lattice type Space Lattice Parameters
content Group (,it and degrees)
Chabazite [2] Water Trigonal R-3m a = b - c = 9.459
ct = 13= Z = 94.07
Methylbutylamine Methylbutylamine Trigonal R -3m a - b = c = 9.383
SAPO-47 [2] c~ = 13= Z = 94.08
Triclinic Ni[ 1,2-diamino- Triclinic P-1 a =9.085
Chabazite [6] ethane202] 2- b = 9.333
c = 8.936
ct = 86.27
13= 101.26
= 92.72
Table 2
Lattice parameters (in the hexagonal setting) of as-made trigonal CAL-1 and SAPO-34(kan)
com P ared to a SAPO-34(ge/); see ex P erimental
...........................
section for synthesis strategies
...........................................
and labeling.
Material SDA a=b c Volume
A A A3
CAL-1 HMI + 13.702(2) 15.132(2) 2460(1)
butylamine
SAPO-34(kan) Morpholine 13.737(2) 14.968(4) 2446(1)
S APO-34(ge/) Morpholine 13.782( 1) 14.873( 1) 2446.4(4)
SAPO-34(ge/) (*) Morpholine + 13.79(2) 14.83(1) 2443(7)
water
(*) From single crystal data already published elsewhere [13].
The few (about 20), broader and very weak peaks, not assigned to CAL-1 were also
indexed and the resulting lattice parameters resulted consistent with the triclinic form of the
chabazite zeolite, reported by Harding et al. [5]. Whereas the CAL-1 trigonal peaks appear
317
after 12 h of crystallization, the triclinic phase peaks can be observed only in the 48 h pattern,
after a longer crystallization time. It is worth noting that this is rather common, because a
small amount of the triclinic phase often appears during the synthesis of CHA-type zeolites,
when a crystallization times longer than 36-48 h is chosen, independently of the synthesis
strategy and/or SDA. Finally, the absence of amorphous silica in the 48h (and, in general, in
all the XRPD collected data) pattern indicates that the SiO2, present in the starting mixture,
has been incorporated in the CAL-1 framework.
The thermal treatment of CAL-1 causes the removal of the organic molecules and the
formation of a pure trigonal SAPO-34 phase ("calcined SAPO-34" in Fig. 3). A preliminary
FTIR study has suggested that the silicon distribution, and consequently the Bronsted acidity
features, is probably different in CAL-1 and in SAPO-34(ge/). It may be then inferred that the
mechanism of silicon insertion within the chabazite framework in these two CHA zeolites
follows different pathways [ 19].
It is worth noting that no amorphous phases were detected in all the steps of this ex
situ study of the ALPO(kan) "-) CAL-1 reaction, suggesting that the rearrangements take
place in the solid state at a local molecular level. An in situ X-ray diffraction study [20] have
been planned to confirm this indication, since possible intermediate phases could be lost
during the sample preparation.
4. CONCLUSIONS
The cell parameters of the trigonal CAL-1, a CHA-type molecular sieve, were obtained by a
high resolution synchrotron radiation X-ray powder diffraction experiment. A triclinic
polymorph of CAL-1, present as minor phase, has been also detected. As compared to SAPO-
34 prepared employing other synthesis strategies, CAL-1 prepared with HMI as SDA, starting
from ALPO(kan), showed slightly different cell parameters, because of the presence of a
different SDA and a peculiar lamellar-like morphology.
Ex-situ high-resolution synchrotron X-ray powder diffraction experiment allowed to monitor
the relevant steps of the ALPO(kan) to CAL-1 reaction and the CAL-1 calcination to H-
SAPO-34. After 3 h reaction, the ALPO(kan) is only slightly changed whereas after 12 h the
CAL-1 phases begins to appear. After 24-48 h reaction the X-rays diffraction pattern contains
signals of both trigonal and triclinic CAL-1, the trigonal phase being largely predominant.
Amorphous phases were not detected during this ex situ XRPD study of the ALPO(kan)
transformation into CAL-1.
ACKNOWLEDGEMENTS
This work was funded by the Italian MURST in the frame of the "Fondo per gli Investimenti
della Ricerca di Base" (FIRB 2001) and the Brazilian FAPESP ("Funda96o de Amparo h
Pesquisa no Estado de S6o Paulo ", 02/1100-4). The authors are deeply indebted to Dr. W.
van Beek and H. Emerich of the Swiss-Norwegian beamline at the ESRF for the precious
support during the XRPD data collection and refinement.
REFERENCES
[1] W.M.Meier, D.H. Olson, C.H. Baerlocher, Atlas of Zeolite Structure Types, Elsevier, 1996,
London.
[2] M. Calligaris, G. Nardin, L. Randaccio, Zeolites, 3 (1983) 205.
318
1. INTRODUCTION
The synthesis of colloidal zeolites has been drawn the attention recently in specific
application fields such as membrane, fiber and film [ 1-3]. The formation of colloidal zeolites
from a gel phase or from a solution consists of a series of self-assembly process involving
numerous simultaneous and interdependent equilibrium and condensation steps [4]. The
preparation for such colloidal zeolites via a hydrothermal process is governed by chemical
composition of reactant mixture and thermodynamic variables [5-6]. A question of
considerable debate in the systematic studies of the processing during the synthesis is the
nature of the synthesis gel and its role in the nucleation and crystal growth mechanisms,
especially because there are no indications for a universal mechanism. Therefore, TPA-
silicalite-1 without the intermediate of a heterogeneous gel phase by performing the
crystallization from a clear solution became apparent as a model crystalline phase for the
320
study of the growth kinetics and nucleation of zeolites in the work of Cundy et al. [7].
Mintova et al. [8] reported the fact that the particle size distribution during the
transformation from amorphous into crystalline silicalite-1 was depended on silica sources
such as TEOS, Cab-O-Sil, and Ludox LS. Kumar et al. [9] showed that the addition of
promoter like NaHzPO4 can be decreased the time and lowed the temperature of zeolite
synthesis. Kirschhock et al. [10] described the interaction between the structure directing
agents, TPA +, and oligomer silicate precursor species forming MFI framework leads to
selective polycondensation process with a curved hydrophobic SiO2 surface under low
temperature (-100 ~ However, the result has been shown to occur in agglomeration in the
formation process of nanosized particles. From the application point of view, the
agglomeration of particle is an important matter to be solved in future nanosized particle
research.
The aim of the present study is to investigate the acidic promoter effect on the
crystallization kinetics and procedure of spherical colloidal TPA-silicalite-1. By doing so,
main factors to determine particle size and regulate morphology could be elucidated.
2. EXPERIMENTAL SECTION
The silica source used for the preparation of the precursor mixtures was Cab-O-Sil
(99.8 %, particle size 200-300 nm, Cabot Co). The precursor mixtures were prepared by
mixing the silica sources with appropriate amounts of distilled water, tetrapropylammonium
hydroxide(TPAOH) (Aldrich, 1M in water) and NaOH (Aldrich). After several minutes of
vigorous stirring at room temperature, acids were used as additional reactant in order to role
of promoter. The synthesis by acid addition was accomplished at 100 ~ for 24 h. On the
other hand, TPA-silicalite-1 synthesized by conventional method was prepared at 150 ~ for
72 h at x=0 and compared with acid-induced TPA-silicalite-1.
MAS probehead using 4 mm ZrO2 rotors) were recored at 79.5 MHz using a 0.8/,ts (rt/8)
pulse with a 4 s repetition time and 2,000 scans. All 29Si spectra were recorded on samples
spun at 4 kHz and referenced to DSS. A ZetaSizer-3000 (Malvern Instrument Co.) with a 10
mW He-Ne laser and a fixed scattering angle of 90 o was employed in this study. The
collected signal was analyzed with the NNLS (non-negative least square) analysis software
provided with equipment.
Fig. 1 shows the crystal growth rate of H2SO4-added colloidal TPA silicalite-1 synthesized at
100 ~ was fast about 6 times than that of TPA-silicalite-1 by conventional method. The
growth degree of HzSO4-induced colloidal TPA silicalite-1 was measured with the area of
selected peaks (2| = 2 2 - 25~ The first crystallinity of solid product by H2SO4 addition was
detected after 6 h and the peaks were getting more intense up to 12 h, where obtaining 100%
crystallinity corresponding to pure MFI structure. Among TPA-silicalite-1 samples
synthesized with different H2SO4 mole ratios, the size of 0.2 mole TPA silicalite-1 was
observed to be much smaller about 10 times than that of sample synthesized by conventional
method. The samples obtained above 0.4 mole ratio were found to be amorphous phase.
100 (a)
80
>., 6 O
..=.....
._=
r
"w" 40
>.,
L
0
20
0 " I I I I I I I
0 8 16 24 32 40 48 56 64 72 80
"7,
,,,!
r
=,-
q)
..E
(a)
= 9
800
.~, 6 0 0
E
(I)
.N
In
i
r
.,..,
= 400
m
L
0
<: 2 0 0 (a}
B
9 I ""~--'---- I I I
Fig. 3. Average particle size for TPA-silicalite-1 Fig. 4. DLS particle size distribution for the
at various promoters: (a) H2SO4, (b) HC1, (c) H2SO4-added TPA-silicalite-1 hydrolyzed at
HNO3 and (d) NH4OH. 100~
population for colloidal TPA-silicalite-1 synthesized from addition of various 0.2 mole
promoters at 100 ~ The colloidal zeolites inducing HC1, HNO3 and NHaOH showed 1.5-3
times larger size than H2SO4. Colloidal particle populations with a constant average size in
each sample were observed before, and for a short period after, the appearance of the
growing crystal population. Based on the measured final crystal size, it is apparent that the
size of TPA-silicalite nucleated is depended on a sort of promoter.
Showed in Fig. 4 is that colloidal particles with an average size 200nm formed by the
hydrolysis of fumed silica in the presence of the structure directing agent, TPAOH, observed
to be included in a part of 25nm particles in silicalite-1 suspensions. The role of 25nm
colloidal in the crystallization of TPA-silicalite-1 may be as growth centers: either the
colloidal particles grow larger sizes or certain fragments of their surface may act as 'nuclei'.
Iler [13] reported that the polymeric species were characterized with respect to particle size
using a modified technique. In acidifying the silicate solution with H2SO4in the study of Iler,
the TPA + are interacted with the polymeric particles by ion exchange. This process is based
on the actual interaction between the precursor sol and TPAOH, organic structure directing
agent. Furthermore, the promoter addition was encouraged that colloidal molecules able to
interact with the particle surface are influencing nucleation like 25nm particle and growth
through the interfacial free energy due to an Ostwald ripening mechanism.
324
40
- 0.1 mole H2SO4
2O
al~~, = 0.2 mole H2SO4
> -..* 0.3 mole H=SO4
,E, =b.
,,,
-- 0
._m
r
o -20
c~
m'~.m%A ~o (c)
= ~mm
N
~ -40
9 (b)
-60
9 (a)
-80
-105.8
"7.
,m,
-86 - 9 2
>,l
(b)
W
r-
=)
,4-,
_c
-86 - 1 2
(a)
I | a I
ppm
shape of TPA-silicalite-1. Also, the use of acidic promoter induces fast synthesis kinetic of
TPA-silicalite-1 as accompanying the different mutual interaction from existing disposition.
The SEM image showed in Fig. 7 confirmed the size uniformity of colloidal TPA-silicalite-1
about 200nm. This agreed with the results of Fig. 3 and Fig. 4. It showed that H2SO2 addition
4. CONCLUSIONS
The TPA-silicalite-1 used H2SO4 for the synthesis of colloidal TPA-silicalite-1 crystals has
been shown to contain spherical colloidal particles. The formation of these particles is
considered that the silicate polyanions prefer to clathrate the organic structure directing agent,
TPA, in silicalite-1. The H2S04 promoter induction greatly polarizes clathrated TPA-silicate
hydrophobic hydration sphere formed around silicate units. The polarization leads to
accelerate TPA-enclathrated silicate polyanions formation leading to crystal growth.
Moreover, H2804 promoter addition seems to increase polarizing ability of silica oligomer in
precursor sol and regulate the shape ofTPA-silicalite-1.
ACKNOWLEDGEMENTS
This study was supported by Eco-technopia 21 project by KIEST and by International joint
project by KICET through grant GNT01032, whom M. H. Kim would like to thank Prof. S. J.
Jeong for kindly making available his dynamic light scattering at Dong-Eui University.
REFERENCES
bDavy Faraday Research Laboratory. The Royal Institution of Great Britain. 21 Albemarle
Street, London W 1S 4BS, United Kingdom.
Clnstituto de Tecnologia Quimica UPV-CSIC. Avda. de los Naranjos s/n. 46022 Valencia.
Spain.
1. INTRODUCTION
The main short-range interactions between the structure directing agents (SDA) and the
zeolite frameworks which they direct occur between the framework oxygens and the terminal
hydrogen atoms of the molecule. These hydrogens are almost invariably bonded to organic
carbon atoms; despite the diversity in size and shape of the molecules used as SDA for the
synthesis of microporous materials, the chemical character of the interacting atoms has only
seldom been modified. One way of achieving this goal consists of substituting hydrogen by
fluorine atoms in the structure directing agents, where several interesting results have already
been obtained for both zeolite and zeotype systems [1]. In the particular case of
aluminophosphates, we have noticed that fluorine can play a substantial role in the SDA
ability of organic molecules [2]. Its presence modifies the chemical character of the
interactions with the framework, making it more ionic, and thus it can improve the template
ability of the molecule. In addition, it can also play a determining role in the intermolecular
interaction and in the packing ability of the molecules inside the microporous framework [3].
In the present work, we present results on a new molecule, dibenzyldimethylammonium
hydroxide (dbdm), and its fluorinated derivatives in ortho (off), meta (mff) and para (pff)
positions of both aromatic rings, as SDA for the synthesis of the AFI structure in both A1PO
and SAPO compositions. We present the synthesis and characterization of the materials as
328
well as the predicted location of the molecules inside the structure. We also rationalise the
experimental results by means of a computational study.
2. E X P E R I M E N T A L
3. COMPUTATIONAL DETAILS
A computational study was carried out in order to rationalize the different ability of the
aromatic molecules to template the synthesis of the AFI structure. The same protocol
329
previously developed for the benzylpyrrolidine SDA [3] has been used to calculate the
location and interaction energies of the four SDA molecules discussed here.
Molecular structures and the interaction energies of the SDAs with the framework are
described with the CVFF forcefield [4], in which the van der Waals and electrostatic
interactions are explicitly included. The geometry of the AFI structure was previously
optimized with the GULP code [5] using the interatomic potential from Gale and Henson [6],
and it remained fixed during the calculations. The atomic charges for the template cations (net
charge of + 1) were calculated by the charge-equilibration method. The net molecular charge
of +1 had to be compensated by the framework; the usual charges for the framework atoms
were-1.2, 1.4 and 3.4 for oxygen, aluminum and phosphorus atoms; we provided charge
balance by decreasing the positive atomic charge of each phosphorus framework ion from 3.4
to 3.3167.
The interaction energies between the SDA and the framework were calculated by
subtracting the energy of the isolated molecules from the energy of the interacting system
(framework + SDA molecules). For each SDA, the computational protocol consists of the
following procedures:
1) Monte Carlo docking of one SDA molecule in the AFI channel,
2) manual docking of a second molecule, followed by molecular dynamics (NVT
ensemble) and energy minimization operations in order to relax the molecules and locate
them at the optimal intermolecular distance,
3) stacking of subsequent molecules in the AFI channel, without imposing periodic
boundary conditions. This step is studied with a series of molecular dynamics and energy
minimization calculations, and yields the preferred density of SDA molecules in the AFI
channel, which we also refer to as 'theoretical' density,
4) study of the SDA-framework interaction energy for a model of the solid based on
periodic boundary conditions. A density of SDA molecules as close as possible to that
determined by (3) above is used (4 molecules in 4 u.c. in the present case), and the molecules
are allowed to relax and locate in the most stable arrangement by a simulated annealing
procedure.
4. EXPERIMENTAL RESULTS
Hereafter, the samples will be named, by reference to the SDA used in the synthesis, as dbdm
(dibenzyldimethylammonium), offdbdm (bis(ortho-fluorobenzyl)dimethyl-ammonium),
mffdbdm (bis(meta-fluorobenzyl)dimethylammonium) and pffdbdm (bis(para-fluorobenzyl)-
dimethylammonium), preceded by A- for A1PO solids or by S- for SAPO solids.
All four SDAs were able to direct the synthesis of the AFI structure for A1PO composition
without notable differences, except for the presence of some small amounts of other dense
phases in the non-fluorinated and the para-fluorinated molecules. The incorporation of silicon
in the synthesis gel carries more pronounced differences in the structure directing ability of
the four derivatives. The XRD patterns of SAPO solids are shown in Fig. 1. SAPO-5 (AFI-
type structure) crystallizes with the four molecules. However, there are clear differences
among them: the non-fluorinated derivative is the least efficient for the synthesis of the AFI
structure since it leads to the crystallization of AFI with the lowest crystallinity and together
with the A1PO-H3 phase (a dense aluminophosphate with APC-type structure). The
appearance of this dense phase is clear evidence for the lower templating ability of this
molecule since no organic can be accommodated within it [2]. In contrast, pure SAPO-5 with
a high crystallinity appears when any of the fluorinated derivatives is used as SDA (Fig. 1).
330
S-db&n S-offdb dm
' i . . . . . . 9 . . . . . . . 9
~o 1~ ~o ~ ~o ~ 40 ~ lO ~ ~o ~;~ ~o ~;~ 40
2e 2e
mffdb dan
~ 1o 1~ 20' ' 25' 30' ' 35' ' 40' 5' ' 1o 1~ ' 20 ' 25
' 30 35' ' 40'
2e 28
Fig. 1. XRD patterns of SAPO solids obtained with the different SDA after three days of
crystallization. Peaks from APC-type structure are highlighted with *
Therefore, it seems clear that the presence of fluorine in the SDA molecules improves the
structure directing ability of the dibenzyldimethylammonium cation. Furthermore, in contrast
to the behaviour of benzylpyrrolidine and its fluorinated derivatives [3], there are no clear
differences between the dbdm
100 derivatives fluorinated in the
different benzyl ring positions.
~ ~ o A-dbdm
The integrity of the molecules
95 ~ A-offdbdm
~, * A-m ffdbdm
inside the inorganic network was
03
03 A, ~ 9 A-p ffdbdm verified by chemical analysis and
c:,
solid state MAS-NMR
I = 90
spectroscopy. The amount of
organic material inside the solid,
85 known from chemical analysis,
was found to be close to 0.9
molecules per unit cell. TG curves
80
9 i
were similar for A1PO and SAPO
, ,
2~o 400 6~o 8~o solids; those for A1PO solids are
Tern perature (oC) shown in Fig. 2. One very intense
weight loss can be observed at
Fig. 2. TGA analysis of A1PO samples after three days of around 300 ~ and a second one
crystallization of lower intensity around 500 ~
331
\
S-db~__~ ~
1
S- offdbdm._.J -raft,
Fig. 3.27A1MAS- NMR (left) and 31p (right) of SAPO samples after three days of crystallization
SAPO solids were also studied by solid state MAS-NMR spectroscopy. Fig. 3 shows the
aluminum (left) and phosphorus (right) MAS-NMR spectra of the SAPO samples synthesized
with the four different derivatives. The 27A1spectra consist of a main peak at around 37 ppm
corresponding to tetrahedral aluminum in the AFI structure, a signal of lower intensity at 8
ppm of pentacoordinated 27A1, and a broad peak between 0 a n d - 2 5 ppm of octahedral
aluminum. The S-dbdm sample shows a second peak at around 40 ppm superimposed on the
main one, which can be explained by the presence of the APC phase. Furthermore, the
relative intensity of the pentacoordinated and octahedral bands is higher in this sample, which
confirms the XRD results that suggested a high amount of the APC phase in the sample
synthesized with the non-fluorinated molecule.
The 31p MAS-NMR spectra (Fig. 3, right), consist of a main signal at around-30 ppm
characteristic of phosphorus atoms in
-- S-dbdm the AFI-type structure. The S-dbdm
; 2 ; ~: o,~o ~m ...,~-. sample displays also two additional
peaks at -26 and -24 ppm
corresponding to phosphorus atoms in
'" ~ii..,.,
the APC phase.
The incorporation of the silicon
atoms in the SAPO solids was studied
by 298i MAS-NMR (Fig. 4). Silicon
atoms can be incorporated in an
aluminophosphate network via two
possible mechanisms: one silicon can
replace one phosphorus atom
i , I . . . .
generating a negative net charge in the
i '
signal at -89 ppm charasteristic of a Si(4A1) environment, and the latter produces a signal at
around-110 ppm typical of silicon attached to four silicon atoms in its second coordination
shell, Si(0A1). Finally, signals appearing between those limits would arise from the other
silicon environments (Si(3A1), Si(2A1) and Si(1A1)) generated at the border of the silicon
patches. The spectra of the S-dbdm, S-offdbdm and S-pffdbdm samples consist of a broad and
asymmetric signal centered at around-105 ppm indicating the formation of silica patches.
The low field asymmetry suggests the presence of 29Si species with aluminum atoms in its
second coordination shell (Si(nA1) n<l :~) present at the border of the silica patches.
Furthermore, a displacement of the signal towards lower shifts occurs with the para- and,
especially, ortho-fluorine derivatives. This displacement seems to indicate that the presence of
fluorine atoms in the SDA molecules tends to direct the incorporation of silicon atoms
towards the formation of larger silicon islands.
Finally, the incorporation of the fluorinated template molecules inside the AFI structure
was confirmed by 19F MAS-NMR.
5. COMPUTATIONAL RESULTS
In previous work [7] we showed that the most stable arrangement of the dbdm molecules
inside the AFI structure consisted of chains in which the benzyl rings of consecutive
molecules face each other. An optimal angle of -90 ~ is formed between the benzyl rings of
the same molecule (Fig. 5), which are perpendicular to each other. The theoretical density
found is of 1.0 SDA molecule per unit cell of A1PO-5, which is very close to the experimental
value (0.9).
We now studied the different
~
~ . , .,,l~ml~. ~.wJll~l~ '
fluorinated derivatives with the
same computational methodology.
Our results showed that the three
fluorinated derivatives were able to
locate in the form of chains in the
same way as the dbdm molecule (1
SDA molecule/u.c.), with benzyl
Fig. 5. Predicted arrangement of dbdm molecules inside rings of consecutive molecules
the AFI channels (from ref. [7]) parallel to each other. In the
benzylpyrrolidine SDA, fluorine in
para position of the benzyl ring prevented the formation of SDA dimers needed to template
the AFI-structure synthesis, due to a repulsion between the two molecules [3] (see Fig. 6,
top). In the dbdm SDA, the presence of fluorine in para position of the benzyl ring does not
cause any repulsion with the next molecule, as shown in Fig. 6 (bottom). Therefore, the
pffdbdm molecules can pack efficiently inside the AFI-type structure, thus providing a good
SDA for this structure in terms of packing ability. Similar results are found for the ortho-
fluorine derivative. In the benzylpyrrolidine SDA, the configuration with highest packing
efficiency was destabilized by a repulsion between fluorine in ortho position and the atoms
belonging to the pyrrolidine ring [3]. The absence of that ring in the offdbdm molecule, where
it is replaced by two methyl groups, which are less bulky, makes this molecule stable in the
conformation needed to arrange in form of chains, as required for directing the synthesis
towards the AFI-type structure. Finally, also the meta-fluorine derivative (mffdbdm) is stable
in the same chain-like arrangement. These results are able to explain the good behaviour of
the para and ortho-fluorine derivatives of dbdm as SDA for the synthesis of the AFI network
333
6. DISCUSSION
All the dbdm derivatives direct the gel chemistry towards the crystallization of the AFI
structure. The appearance of APC phase shows a poorer templating ability of the molecule to
direct the synthesis towards the crystallization of the open framework AFI structure.
The presence of fluorine atoms in the SDA molecules can affect the incorporation of silicon
in the A1PO framework. Our results indicate that fluorine promotes the formation of larger
silicon patches, which are more hydrophobic, in line with the more hydrophobic character of
fluorinated molecules. However, previous results with benzylpyrrolidine, where no
differences between the all-hydrogen and the meta-fluorinated derivatives were found [2],
prevent us from generalizing this observation.
The presence of fluorine atoms in the SDA molecules can alter the structure directing
ability of the molecule in two ways: modifying its packing ability inside the framework, as
occurred with the benzylpyrrolidine molecule [3], and/or its interaction with the framework.
From our experimental results we know that the presence of these fluorine atoms in the dbdm
334
molecule in any of the ortho-, meta- or para- benzyl ring positions improves its structure
directing ability.
The introduction of fluorine atoms in the dbdm molecules does not affect the packing
ability of the molecule; in these cases, no new inter- or intra-repulsions follow from the
presence of the fluorine atoms, in contrast to the case for fluorinated benzylpyrrolidine [3].
The interaction energy is increased by the presence of fluorine in the meta and para positions
because of higher electrostatic interactions with the framework, making these fluorinated
derivatives better SDAs; however, for the ortho-fluorine derivative the experimental results
suggest a higher interaction energy, but its calculated value is lower than that of the all-
hydrogen molecule. This feature shows that additional parameters should be taken into
account when considering the structure directing ability of a molecule. The interaction energy
with the undoped framework is a reasonable and usually a good parameter to estimate the
structure directing ability of organic compounds, but some other factors apart from the
packing ability, such as kinetics and dopant distribution (silicon) in the framework, can also
affect the nature of the crystal phase obtained. Additional work is required to differentiate the
role played by each of the above contributions.
7. CONCLUSIONS
Fluorine containing organic molecules are a new class of structure directing molecules which
can efficiently direct the crystallization of microporous structures. The presence of fluorine
influences not only the interaction energy with the framework but also the packing ability of
the organic molecules inside the microporous network. Our experimental and computational
work show that there is no general correlation between the position of the fluorine atoms in
the benzyl ring and the structure directing ability; instead, that effect depends on the
arrangement of such molecule inside the microporous structure, and so on the detailed
structure of the molecule studied.
We have also shown in this work that the presence of fluorine atoms can definitely affect
the incorporation of silicon atoms in an A1PO network.
ACKNOWLEDGEMENTS
L. G6mez-Hortigtiela acknowledges the Spanish Ministry of Science for a PhD grant. The
financial support of CICYT (project MAT 2003-07769-C02-02) is acknowledged. We thank
Accelrys for provision of their software.
REFERENCES
This study shows that SAPO-44 molecular sieve may be prepared from A1PO-kanemite as
source of phosphorus and aluminum simultaneously, without contamination with other
phases. The results showed that the rate of the reaction was dependent on the duration of the
ageing time and on dilution. X-rays diffractograms show that the minimum time of
hydrothermal treatment necessary to completely transform A1PO-kanemite in SAPO-44 is 36
h. Monitoring of the reaction after different times of hydrothermal treatment shows that no
complete dissolution of A1PO-kanemite occurs in the process of SAPO-44 production, thus a
local, molecular level rearrangement involving the silica source is proposed for this reaction.
1. INTRODUCTION
SAPO-44 is a small pore molecular sieve with CHA structure. It was first synthesized by
researchers at Union Carbide in 1988 [1 ]. The framework contains double 6-membered rings,
4- and 8-membered rings. The three-dimensional structure has large cages where small guest
molecules can enter, passing through eight-membered ring (8R) windows. In the composition
(SixAlyPz)O2 reported in the literature, x typically varies from 0.04 to 0.2 depending on the
synthesis parameters and structure types. Man et. al. [2,3] proposed a topological model for
the incorporation of silicon in the framework and showed the possibility of silicon island
formation in SAPO. This particular type of framework is effective in several catalytic
reactions including the convertion of various oxygenates into olefin products [4].
The SAPO molecular sieves, like zeolites by one hand, are generally prepared through
the formation of a gel of the A! and Si sources, phosphoric acid, water and an amine that
works as structure directing agent (SDA). Zeolites, by the other hand, are known to have been
also prepared from hydrated layered silicates, like magadiite and kanemite [5,6], differently
from them, silicoaluminophosphates have never been prepared directly from ALPO or SAPO
layered materials, probably because up to very recently there was no ALPO or SAPO material
with structure analogous to the ones of the lamellar hydrated silicates. In 1997, Cheng et al.
[7] reported the preparation of a lamellar aluminophosphate analogous to the hydrated silicate
kanemite. From this material, CAL-1, a CHA-type silicoaluminophosphate was prepared [8].
We report here the synthesis of SAPO-44 from ALPO-kanemite with the addition of
silicon atoms, in the presence of cyclohexylamine and examine the influence of dilution, of
the synthesis time, and of the ageing time before the hydrothermal treatment. Some features
of the material thus prepared are compared with the ones of a SAPO-44 prepared by the
traditional synthesis from the reactants.
336
2. E X P E R I M E N T A L
2.3. Characterization.
Powder XRD patterns were obtained with a Shimadzu XRD6000 with CuK~ at 2 ~ 20 min -~,
with slits of 0.5 ~ 0.5 ~ and 0.3 mm for scattering, divergence, and reception, respectively.
Infrared spectroscopy (FTIR) of the samples was obtained using the KBr disk method (0.5
wt%) at 4 cm -1 resolution using a Bomem Michelson MB 102 spectrometer.
Thermogravimetry measurements (TGA) were obtained in a TA 5100, TA Instruments, with
TGA 2050 Analyzer at a heating rate of 10~ min -t, from room temperature to 1000~ under
argon with a flow of 100 ml min -~. Solid state NMR spectra were recorded on a Bruker AC
300/P for 29Si CP-MAS NMR, 27A1 and 31p-MAS NMR under high power decoupling
conditions. The frequencies were 78.20, 121.50 and 59.62 MHz for 27A1, 31p and 29Si,
respectively. Pulses of 90 ~ were used for all measurements with repetition times of 2 s for
27A1, 30 s for 31p and 5 s for 29Si. The acquisition times were 0,025 s for 27A1, 0,016 s for 31p
and 0,066 s for 29Si. The cross-polarization time was 8ms and the rotor frequency was 4.5
kHz. Scanning electron microscopy (SEM) of the samples was performed in a JEOL 6360-
LV, operating in 20 kV.
3. R E S U L T S AND DISCUSSION
3.1. Synthesis
The XRD patterns of the samples after reaction for different times of hydrothermal
treatment (Fig. 1) were recorded to obtain the information on the evolution of possible
intermediate phases as a function of crystallization time. Fig. lc shows that after 12 h of
hydrothermal treatment SAPO-44 is the only product formed and that 36 h, Fig. l e, is the
minimum time necessary for complete transformation of A1PO-kanemite into SAPO-44. The
crystallinity of samples, however, continues to increase up to 48h, Fig. l f, and decreases if
the reaction time is longer (Fig. l g). In the process of SAPO-44 formation no extensive
amortization of A1PO-kanemite was observed by X-rays diffraction, nor even baselines with
the characteristic amorphous halo are present. The comparison of the diffractograms in Fig. 1f
and lh shows that the SAPO-44 sample synthesized from A1PO-kanemite presents a
diffraction pattern similar to that of the SAPO-44 sample prepared by traditional method
except for the extra signals at 7.52, 14.93 and 19.74 ~ 20 that disappear upon dehydration and
are occasionaly present in the difractograms of CHA-type structures. Fig. 2 shows that the
extent of the reaction is also dependent on the duration of the ageing time.
337
10 20 30 40
20/degrees
Samples
Crystallinity/%
1000cps
88
100
t-
O~
Fig. 2. X-ray diffraction
patterns of SAPO-44
81 synthesized at different aging
times e dilutions: (a) 2 h,
H20/A1203 = 63; (b) 24 h,
i , i , | ,
89 HzO/A1203 = 63; (c) 2 h,
lO 20 30 40
HzO/AI203 = 100; (d) 24 h,
20/degrees HzO/A1203 = I00.
When the gel is aged for 24 h, Fig.s 2b and 2d, a small amount of ALPO-kanemite
remains in the final product (indicated by the dashed line) however, the phase purity of the
product is not affected as no extra phases are detected. The crystallinity of the samples are
also presented in Fig. 2. These data show that H20/SiO2 molar ratio of 100 is the one that
affords the most crystalline sample. Therefore, for the preparation of SAPO-44 from ALPO-
kanemite, 2h of ageing time, 48h of hydrothermal treatment and a dilution of 100 are the best
synthesis conditions to obtain highly crystalline materials at the minimum time needed at
SIO2/A1203=0.5.
the large band at aproximately 3400 cm -~ (Fig. 3A curves e to g). This intensity reduction is
probably brought upon the material by the condensation af adjacent layers in the
transformation to a three-dimensional solid. If an extensive bond breaking was operative in
the reaction, that is, if the transformation was effected by dissolution of ALPO-kanemite one
would expect an intensity increase in this band; that was not observed.
5
c
1600 1400
,
, 0o ebo 4oo
Wavenumber/cm
Wavenumber/cm -1 Wavenumber/cm 1
(c)
(A) (B)
Fig. 3. Infrared spectra of (a) ALPO-kanemite and of samples prepared at increasing crystallization
times: (b) 6 h; (c) 12 h; (d) 24 h; (e) 36 h; (f) 48 h and (g) 72 h.
Intensity reduction was also observed in the bands due to C - - H stretchings at 2850-
3000 cm -~ region. This is due to the decreased amount of organics needed in the SAPO-44 as
compared to ALPO-kanemite (see below in the thermogravimetry results). The changes in the
relative intensity and in the number of bands are probably due to the change in the relative
concentration of n-butylammonium and cyclohexylammonium cations in the final material.
Fig. 3B shows the region where NH and CH bends are expected to appear. In fact, the
decrease in organics content is also apparent in that region, as seen by the decrease in the
intensity and broadening of the band at 1619 and 1549 cm -~ due to NH2 bend and at 1469 and
1390 cm -~ assigned to CH2 and CH3 groups, respectively.
Fig. 3C show the A1PO-kanemite anti-symmetric and symmetric T - - O - - T (T = A1
and P) stretching vibrations at 950-1200 and 600-900 cm -l, respectively. The most clear
differences between the spectra in Fig. 3C curves a to d, where the amount of ALPO-
kanemite is still important, and the ones in Fig. 3C, curves e to g, is the disappearance of the
bands at 904 and 876 cm -I in ALPO-kanemite (dashed lines) upon transformation into SAPO-
44. These bands are probably due to P-OH that condense to form the three-dimensional
framework, much the same way as seen in the hydroxyl region. The total disappearance of
these bands coincides with the definition of bands at 574, 526 and 479 cm -~ in the spectra e, in
Fig. 3C due to double ring vibrations.
Definitive changes are seen in the spectra of Fig. 3A, B and C curves e, which
correspond to the sample obtained after 48h of hydrothermal treatment and where the XRD
had already shown that the transformation of ALPO-kanemite into SAPO-44 was complete
and no reactant remained on the solid.
339
100
-0.6 ~
o~ 90 o~ Fig. 4. (A)
rEZ Thermogravimetry and
0 0.4 ~ (B) derivative
80.
o') < thermogravimetry
0.2 r profiles of (a) ALPO-
z; 70.
\ o~
kanemite and (b)
SAPO-44(Fig. 1e).
60 o.o d
200 400 600 800 1000 200 400 600 800 1000
Temperature/~
The TGA curves of the A1PO-kanemite showed 35.6% weight loss between 200 and
400~ due to n-butylammonium decomposition, which is in agreement with Cheng et. al. [7]
whereas as-synthesized SAPO-44 loses only 10.9% of its initial weight between 350 and
550~ due to cyclohexylammonium decomposition. The weight loss observed for SAPO-44
synthesized by the traditional method is in the range of 20 to 25% [9].
The difference in the temperature of organics decomposition in A1PO-kanemite and
SAPO-44 can be explained by the diversity of the sites where the organic material is bound.
In ALPO-kanemite, the n-butylammonium cation balances the charge of a superficial PO site
as (R-NH3) + - O - - P - [ 7 ] while sites of the type ~ I - - ( R - - N H 3 ) + -O--Si -- are the ones to
retain organics in SAPO-44. The surface sites in SAPO-44 are more acidic [11] thus more
difficult to protonate during template decomposition than ALPO-kanemite, therefore the
decomposition of the organic material will occur at higher temperatures compared to ALPO-
kanemite.
The 27A1-MAS NMR of the as prepared material, Fig. 5A, curve a, shows only one
signal at 37.1 ppm due to tetrahedral aluminum ions. After calcination, Fig. 5A, curve b,
another signal at -11.5 ppm appears which is probably due to octahedral aluminum ions. The
calcination of cyclohexylamine liberates the channels and cavities making it possible for
water to access part of the aluminum sites changing their coordination from tetrahedral to
octahedral and generating the observed signal at higher fields.
Fig. 5B shows the 3~p MAS NMR spectra. In the as-prepared sample, Fig. 5B, curve
a, the peak at -30 ppm is assigned to a P(4A1) site, as already described in the literature [9].
Two signals of lower intensity at-14.5 and-20.2 ppm are assigned to surface P-OH groups.
After calcination, the chemical shift of the most intense peak, due to phosphorous with only
aluminum as second neighbours, is displaced to-27.5 ppm and the two signals corresponding
to the presence of P-OH become a shoulder of the more intense one.
The 29Si CP-MAS NMR is most informative about the silicon sites in this structure.
Fig. 6 shows 29Si NMR of SAPO-44 as-synthesized from A1PO-kanemite, Fig. 6A, curve a,
from traditional method, Fig. 6A, curve b, and calcined SAPO-44, Fig. 6B, curve c. The
various types of silicon sites in this structure are indicated in the spectra [9].
The major peak in the spectra of either as-synthesized materials (from ALPO-
kanemite and prepared from the reactants) or calcined one is located at-90.5 ppm and is
assigned to Si(4AI) in acidic sites that were inserted by the substitution of silicon at
phosphorous sites(SM2) [2]. The peaks at-94.6, -99.51-104.5 ppm are assigned in Fig. 6
according to the literature [9]. The peak at-109.2 ppm is typical of silicon islands in this
structure[9].
Both materials, either prepared from A1PO-kanemite or according to the traditional
method, present silicon islands, Fig. 6a and 6b. Taking into account that both materials were
prepared with Si/A1 molar ratio of 1.0, the comparison of curves a and b in Fig. 6A allows to
conclude that SAPO-44 prepared from the lamellar precursor presents less silicon atoms in
silicon islands than the sample prepared using the reactants by the traditional method.
QOSi(4AI)
/,/!~ Qo sic4~,o
ti\. o 1 sKsA0
1~ Q1 Si(3AI)
PtltQ2si(2AV Q3 Si(1AI)
.j "~,./'k
" ~
z tk _ . a
........ ~ ""~" '~**'%"t
....... ,,~2~-~ ....
-70 -80 -90 -100 -110 -120ppm
-70 -80 -90 -100 -110 -120pprn
Fig. 6. 298i CP-MAS NMR of SAPO-44: (a) as-synthesized from A1PO-kanemite; (b) as-
synthesized according to the traditional method; (c) calcined SAPO-44 prepared from AIPO-
kanemite.
Calcination causes a general broadening of the spectrum, Fig. 6c, the maximum of the
curve shifts to-89.1 ppm and the signals assigned to the different silicon sites become much
less distinct, however it is very clear that no increase is seen in the peak corresponding to the
silicon atoms in islands.
341
(A) (B)
Fig. 7. Scanning electron microscopy images of (A) as-synthesized SAPO-44, (B) calcined SAPO-44.
collectively indicate that the transformation of ALPO-kanemite into the 3D SAPO-44 do not
occur through a complete dissolution of the lamellar material nor through the formation of an
amorphous gel. It is probable that the reaction occurs at the interface solid-solution by local,
molecular-level rearrangements of the type proposed by us in Scheme 1.
The structure proposed for the ALPO-kanemite is the same as the silica-kanemite and
was demonstrated by Cheng et al. [7]. To understand the proposition in Scheme 1, it is
necessary to remember that the cyclohexylamine brings the pH of the reaction to values
around 12 and thus fumed silica begins to dissolve and remains deprotonated.
342
+ 4- R+~ R +
R Si
I O~ \ ~ O
X ...(~/
AIwithexpanded
coordination
,,,,,"
oi
\ /o ,, o -
Formationof
R§ protonatedhexamethyleneimineor fourmemberedrings
n-butylamine.
- SiOR
R "l" , 4-
+
Openingof the six-memberedringof
ALPO-kanemite X .,.0/ oj \
.O
,,,,,""
Scheme 1
These dissolved silica species effect a nucleophilic attack on the aluminum atom
causing either an expansion of the A1 coordination sphere or the opening of a six-membered
ring. Another nucleophilic attack on to the silicon atom affords a four-membered ring. These
four-membered rings are completely non existent in ALPO-kanemite but are part of the
structure of SAPO-44.
ACKNOWLEDGEMENTS
REFERENCES
[1] S.T. Wilson, E.M. Flanigen, ACS Symp. Ser., 398 (1988) 329.
[2] P.P. Man, M. Briend, M.J. Peltre, A. Lamy, P. Beaunier, D. Barthomeuf, Zeolites, 11 (1991)
563.
[3] A.M. Prakash, S. Unnikrishnan, K.V. Rao, Appl. Catal. A 110 (1994) 1.
[4] International Patent, Exxon Mobil, WO 02/26677 A2.
[51 G. Onyestyak, G. Pal-Borb6ly, H.K. Beyer, Microporous Mesoporous Mater. 43 (2001) 73.
[6] F. Feng, J.K. Balkus Jr., Microporous Mesoporous Mater. 69 (2004) 85.
[7] S. Cheng, J. Tzeng, B. Hsu, Chem. Mater. 9 (1997) 1788.
IS] A. Albuquerque, S. Coluccia, L. Marchese, H.O. Pastore, Stud. Surf. Sci. Catal. 154A (2004)
g66.
[9] D.B. Akolekar, S.K. Bhargava, J. Gorman, P. Paterson, Colloids and Surfaces A: Physicochem.
Eng. Aspects, 146 (1999) 375.
[10] L. Marchese, J. Chen, P.A. Wright and J.M. Thomas, J. Phys. Chem. 97 (1993) 8109.
[ 11 ] S. Ashtekar, S.V.V. Chilukuri, D.K. Chakrabarty, J. Phys. Chem. 98 (1994) 4878.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 343
bAdvanced Research Institute of Science and Engineering, Waseda University, 3-4-1, Okubo,
Shinjuku-ku, Tokyo 169-8555 Japan
The role of steam on the crystallization of zeolite beta by the steam-assisted crystallization
(SAC) method was studied. It was found that after an aluminosilicate dry gel with
SIO2/A1203 molar ratio = 100 was treated with steam in an autoclave at 130~ and autogenous
pressure, thermal treatment of a resultant product at 130~ without the addition of water in an
autoclave gave zeolite beta with a high crystallinity. Zeolite beta obtained by this method
consisted of nano-particles with a diameter of ca. 30 nm. It was presumed that intermediates
of zeolite beta was formed during steam treatment, and the dehydration between silanols in the
intermediates during the thermal treatment in the absence of steam possibly caused the
transformation from ill-ordered structures to zeolite beta crystals with a periodicity in
long-range order.
1. INTRODUCTION
Zeolite beta possessing a 3-dimensional 12-membered ring micropore system was found to a
promising catalyst for the cracking of paraffins [1], isomerization of n-heptane [2] and
n-hexane [3], and the disproportionation and transalkylation of toluene and C9 aromatics [4].
We have developed a dry gel conversion (DGC) technique, where a silicate or
aluminosilicate hydrogel is dried and the resultant dry gel is converted into microporous
crystals in vapor. We have proposed classifying the DGC method as two separate methods
[1 ]: (1) the vapor-phase transport (VPT) method in which a dry gel is crystallized in the vapor
of steam and volatile structure-directing agents (SDA) like ethylenediamine, and (2) the
steam-assisted crystallization (SAC) method in which a dry gel containing a non-volatile SDA
like tetraethylammonium (TEA) hydroxide is crystallized in steam. Kim et al. [5] and our
previous research [6] confirmed that the VPT method could be applied to synthesis of various
zeolites such as ANA, FER, MFI and MOR. The VPT method has also been applied to the
synthesis of zeolite membranes supported on porous alumina [7] or self-supported [8]. We
recently reported that MCM-22 zeolite obtained by the VPT method showed a higher
selectivity to p-xylene in the alkylation of toluene with methanol than that obtained by the
hydrothermal synthetic method [9].
Zeolite beta can be obtained from a dry gel containing TEA + cation as an SDA by the
SAC method [10]. In comparison with the conventional hydrothermal synthetic method, the
SAC method enables us to rapidly produce highly crystalline zeolite beta from dry gels with a
wide range of SIO2/A1203 ratio from 7 to infinity [ 11-13]. It was remarkable that the contents
of Si and A1 atoms in the solid products were almost the same as those in the dry gels [ 11].
344
We investigated the crystallization mechanism of zeolite beta under the SAC conditions on the
basis of the results of XRD, FE-SEM and 29Si MAS NMR measurements [12-13], and
concluded that the crystallization of zeolite beta from a dry gel in the SAC method occurs via
solid intermediates made of nanoparticles [13].
In this study, we found that well-crystallized zeolite beta with a diameter of ca. 30 nm
was obtained by the thermal treatment of solid intermediate in the absence of steam. The
influence of steam on the crystallization of zeolite beta will be discussed.
2. E X P E R I M E N T A L
a) i--.-7-1 b)
i
- - Water
Fig. 1. Setups of autoclaves a) for the steam-assisted crystallization method and b) for the thermal
treatment used in this study.
2.2. Characterization
The dried and thermally treated products were characterized by using X-ray
diffractometer (XRD, Miniflex, Rigaku) at 30 kV and 15 mA. The morphologies of the
products were observed using field emission scanning electron microscopy (FE-SEM,
S-4500S, Hitachi) at 15 kV of ACC voltage. The contents of water in the as-made products
were measured in flowing air with a heating rate of 10 ~ min-lusing thermal gravimetry (TG,
DTG-50H, Shimadzu).
345
c~. (ei
:~-- I1
C
{c)
{b)
{a)
1000 ............. ~-: ~ ~q - ' i " - - ' " ' - ~ - " ~ ,,,, ,,,, .......
cps ........
5 t0 t5 20 25 30 35
2 t# (CuK(-;) / d e g r e e
Fig. 2. XRD patterns for (a) the dry gel and (b-g) the as-made products treated at 130~ in the
presence of steam. Treatment periods were (b) 1 h, (c) 3 h, (d) 4 h, (e) 8 h, (f) 16 h and (g) 24 h.
346
Figs. 3d-f indicate that ill-crystalline zeolite beta obtained by the steam treatment at
130~ for 8-24 h was converted to well-crystallized zeolite beta by the thermal treatment in the
absence of steam. We presumed that removal of water from an ill-crystalline product having
a structure resembling zeolite beta by the thermal treatment helps the transformation of the
ill-ordered structure to well-crystallized zeolite beta.
Fig. 3a shows that an amorphous phase was obtained by the thermal treatment of the dry
gel at 130~ without the addition of water in an autoclave. This result strongly suggested that
the crystallization of zeolite beta in the SAC method requires a steaming treatment of a dry gel
at 130~ to form the intermediates having ill-ordered structures.
O0
C)_
tD
-u9
C
(a)
1000 I
cps i. I.. I. x. 1. ~"]-~"~ ~ i J ~ J ~ . I . j.. j .. ~ ~ I . J ~ ~ ! ~ 1 l i
5 10 t5 20 25 3(.) 35
2 0 (CuKcJ.) / degree
Fig. 3. XRD patterns for the products obtained by the treatment at 130~ for various periods in
the presence of steam and successive treatment at that temperature for 12 h without the addition of
water in an autoclave. The periods of steam treatment were (a) 1 h, (b) 3 h, (c) 4 h, (d) 8 h, and
(e) 24 h.
347
1.0 ~m
....
~i~ ~ ~ ~ ~.....
100 nm
Fig. 4. Typical FE-SEM images for (a) the dry gel and (b) as-made products crystallized at 130~
for 3 h in the presence of steam and successively at that temperature for 12 h without the addition
of water in an autoclave.
348
,!
~ A (e)
(<t~
(a)
1000
cps
5 t0 15 20 25 30 35
28 (CuKo)/degree
Fig. 5. XRD patterns for the products obtained by the treatment at I30~ for 3 h in the presence
of steam and successive treatment at that temperature for various periods without the addition of
water in an autoclave. The periods of thermal treatment were (b) 2 h, (c) 4 h, (d) 6 h, (e) 8 h, (f)
10 h and (g) 12 h. (a) No thermal treatment was carried out.
349
100 30
80 - -. 25
E II . 2O e'-
60 -~ -
40 . ', ": Er
.> - 'm_ m 1~ i~ IOE
'~ - - ......... m o
-~ / n 1 o
a:: 20 - 4 5
t
0
0 4 8 12
Crystallization period / h
Fig. 6. Relative crystallinity (rn) and contents of water (i) of products obtained by the
treatment at 130~ for 3 h in the presence of steam and successive treatment at that temperature
for various periods without the addition of water in an autoclave.
~ 80
~ 00 30
25
60
o
1522
>~ 40
10~
2:2o 5
0 0
0 4 8 12 16
Crystallization period / h
Fig. 7. Relative crystallinity (o) and contents of water (o) of the dry gel and products
crystallized at 130~ in the presence of steam.
350
4. C O N C L U S I O N S
Well-crystallized zeolite beta with a diameter of ca. 30 nm can be crystallized at 130~ without
the addition of water in an autoclave from an aluminosilicate solid treated at 130~ for 3 h
under the SAC conditions. We supposed that an aluminosilicate solid having an ill-ordered
structure can be transformed into zeolite beta crystals by the dehydration of silanols in the
solid during the thermal treatment. This method would reduce the production cost of
well-crystallized zeolite beta.
REFERENCES
[11 L. Boretto, M.A. Comblor, A. Corma and J. Perez-Pariente, Appl. Catal. A, 82 (1992) 37.
[2] J.K. Lee and H.K. Ree, Catal. Today, 8 (1997) 235.
[3] Z.B. Wang, A. Kamo, T. Youeda, T. Komatsu and T. Yashima, Appl. Catal. A, 159 (1997) 119.
[4] J. Das, Y.S. Bhat and A.B. Hargeri, Catal. Lett., 23 (1994) 161.
[5] M.-H. Kim, H.-X. Li and M.E. Davis, Microporous Mater., 1 (1993) 191.
[6] M. Matsukata, N. Nishiyama and K. Ueyama, Microporous Mater., 1 (1993) 219.
[7] e.g., N. Nishiyama, K. Ueyama and M. Matsukata, Microporous Mater., 7 (1996) 229.
Is] S.G. Thoma, D.E. Trudell, F. Bonhomme and T.M. Nenoff, Microporous Mesoporous Mater., 50
(2001) 33.
[91 S. Inagaki, K. Kamino, H. Shibuya, E. Kikuchi and M. Matsukata, Bull. Chem. Soc. Jpn., 77
(2004) 1249.
[ 10] M. Matsukata, M. Ogura, T. Osaki, P.R.H.P. Rao, M. Nomura and E. Kikuchi, Topics Catal., 9
(1999) 77.
[ 11] P.R.H.P. Rao, C.A. Leon y Leon, K. Ueyama and M. Matsukata, Microporous Mesoporous
Mater., 21 (1998) 305.
[ 12] M. Matsukata, T. Osaki, M. Ogura, E. Kikuchi and A. Mitra, Microporous Mesoporous Mater.,
48 (2001) 23.
[ 13] M. Matsukata T. Osaki, M. Ogura, E. Kikuchi, Microporous Mesoporous Mater., 56 (2002) 1.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 351
b National Centre for High Resolution Electron Microscopy, Delft University of Technology,
Julianalaan 136, 2628 BL Delft, The Netherlands
# Current address: Dept of Inorganic Chemistry and Catalysis, Debye Institute, University of Utrecht,
P.O. Box 80083, 3508 TB Utrecht, The Netherlands.
Carbon oligomers have been introduced as clusters in silicalite-1 by direct synthesis methods.
From a series of analysis techniques it is concluded that the non-SDA carbon source is
predominantly present as oligomers within the zeolite crystals, stabilized within the silicate
framework. Upon calcination, i.e. removal of the carbon oligomers, a complex macroporous-
mesoporous-microporous pore system was obtained. The present work may serve as a new
route to the formation of a secondary pore system with interesting features regarding various
catalytic applications.
1. INTRODUCTION
Zeolites exhibit unique properties with respect to both activity and selectivity. In general,
activity is determined by the zeolite Bronsted acid sites and by the active metal that may be
deposited into the zeolite, while selectivity is provided by the zeolite micropores. Besides the
beneficial shape selectivity effect of micropores, they may in some cases also limit the overall
catalytic performance due to a restricted molecular transport inside the zeolite crystal. This
so-called diffusional limitation may be alleviated by the introduction of a secondary pore
system that effectively shortens the micropores and therefore the intracrystalline diffusion
path lengths [1]. Typically, such a secondary pore system is created by the removal of
aluminium and/or silicon from the zeolite framework using steaming and/or acid-leaching
procedures [2]. Such treatments cause the loss of active sites and generate non-framework
species, which may induce obstructions and a secondary, undesired, catalytic activity [3]. The
incorporation of a carbon source in the framework of the zeolite and its subsequent removal
may be a new route to the formation of a secondary pore system, which avoids the loss of
active sites and prevents the generation of non-framework species.
Carbon containing molecules are numerously applied in zeolite syntheses. For example
quaternary ammonium ions, amines, alcohols and acetals are widely used as pore filling
agents or templates to induce the formation of the crystalline microporous structure. More
recently, various other forms of carbon, like arrays of mono disperse polystyrene spheres [4],
carbon black [5], and carbon nanofibers [6] have been used as a template in zeolite syntheses
352
Syntheses of silicalite-1 were performed via the fluoride pathway, following an overall molar
composition adapted from literature [13], which allowed an investigation of the effect of the
molar ratio of all the reactants. Pentaerythritol (PE) C5H1204 was added in the mixture after
dissolution in distilled water. The resulting gels, after complete homogenisation, were put into
Teflon-lined 60 cm 3 stainless-steel autoclaves. The samples were obtained by hydrothermal
synthesis at 473K. After synthesis, samples were collected by centrifugation, washed
repeatedly with distilled water, and dried overnight. Incidentally, samples may then be
purified by soxhlet extraction in dichloromethane, in order to remove unused PE. Finally the
samples were calcined in a shallow bed at different temperatures, under static air.
The overall outcome of the experimental study is as such: syntheses carried out in fluoride
media resulted in reproducible new materials and the extent of incorporation of carbon
depends on the synthesis procedure followed, as well as the silica source used. As an example
we present here the outcome of a reproducible synthesis performed in fluoride media with a
composition of 100 SiO2:6.3 TPABr: 3.13 NHaF: 1570 H20:25 PE, where the silica source
SiO 2 is Aerosil 200, TPABr stands for the template tetrapropylammonium bromide (Aldrich),
NHaF is ammonium fluoride (J.T. Baker), and PE stands for pentaerythritol (Fluka).
As a comparison, a pure silicalite-1 was synthesized with the same batch mixture. Both the
new material obtained after incorporation of pentaerythritol in the synthesis mixture and the
pure silicalite-1 were characterised with a series of techniques as mentioned above.
Before and after calcination, samples were analyzed by elemental analysis (ICP, AAS,
NAA), IR, XRD, N 2 physisorption, TGA, SEM and TEM. X-ray diffraction measurements
were performed with a Bruker-AXS D5005 diffractometer. For infrared measurements, the
crystals were measured on a Bruker IFS66. N2 adsorption and desorption isotherms were
recorded on a Quantachrome Autosorb-6B gas sorption analyzer. Prior to the analysis, the
samples were evacuated at 623 K for l0 h. Thermogravimetric analyses were performed on a
TGA Cahn 131. Measurements were performed under air flowing at 50 ml per minute, the
purge gas was nitrogen, with a flow of 50 ml per minute. Scanning Electron Microscopy
353
(SEM) of the products was performed after deposition of a gold film on the sample on a JEOL
Superprobe 733. Transmission electron microscopy (TEM) was performed using a Philips
CM30T electron microscope with a LaB6 filament as the source of electrons operated at 300
kV.
3. R E S U L T S AND DISCUSSION
1 18.5 19 19.5 2
29.2 41.3
34.2
5 10 15 20 2 5 2 theta d e ~ 0 35 40 45 50
Fig. 1. XRD patterns of as synthesized new material and conventional silicalite-1, showing extra peaks
in the new material, existing next to the silicalite-1 pattern
XRD powder analyses performed after calcination at 823K (not displayed on the figure) show
that the new material resembles then cristalline silicalite-1.
Different analyses have been performed in order to ensure the incorporation of carbon in the
structure of the zeolite and determine the form in which it is present when incorporated in the
new material.
The new material was carefully washed with distilled water immediately after
synthesis, and further purified by soxhlet extraction in dichloromethane, as pentaerythritol
dissolves in CH2C12. However, after prolonged soxhlet extraction in this solvent, the new
material showed no change in its XRD diffractogram, as well as its infrared signal and also
354
the total carbon content remained identical. This indicates that the new material contains a
form of PE that could resist dichloromethane extraction. Therefore, oligomers of PE have
been synthesised via thermal-induced polymerisation and analysed by XRD. The
diffractograms of the PE source and the oligomers differ in only one peak: the diffractogram
of the PE source presents a small peak at 20 = 18.9 that is absent on the diffractogram of the
oligomers. One can observe in the XRD diffractogram of the new material (inlet Fig. 1) that
this specific peak is not present, indicating that there are no monomers of PE in the crystalline
form and therefore supporting the hypothesis of the incorporation in the framework of
crystalline oligomers of PE.
Elemental analyses have been performed on the two materials, in order to quantify the
amount of carbon incorporated. The results of these analyses are given in Table 1. A total
carbon content of 8% was measured for the conventional silicalite-1, which can be attributed
only to the template remaining in the framework, and a total carbon content, representing both
the template and the pentaerythritol sources, of 22.7% was found for the as-synthesised new
material.
Table 1
Carbon and silicon contents of the conventional silicalite-1 and the carbon-containing new
material, as-synthesized and calcined at 823K
Thermogravimetric analyses were performed on both the new material purified with
CH2C12, and the conventional silicalite-1, in order to provide information on the thermal
stability of the new material in an oxidative atmosphere. On both samples, the first weight
loss of 32.1 wt% (relative) was observed going up to 553K, and is ascribed to the desorption
of water [15]. Between 703K and 753K, a relative 12.3 wt% loss was observed, accompanied
by a brown colouring and cracking of the crystals. These phenomena are also observed in the
TGA of the conventional silicalite-1, and correspond to the Hoffman degradation of the
template TPA occluded in the micropores. For the conventional silicalite-1, this weight loss
represents 11.8 wt%, which is in very good agreement with the literature and corresponds to
four TPA cations per unit cell [ 15,16]. The third event of the thermal decomposition, solely
observed for the new material, represented 10.3 wt% (relative), and occured between 753K
and 833K. Most likely, this event was due to the loss of the PE species bonded to the
framework within the zeolite crystals. Indeed, we expect that the carbon source must vaporize
to be able to desorb from the crystals, and therefore can only occur at a temperature above the
melting point of PE (533 K). If we consider that the carbon source is made of crystalline
oligomers of PE, then the energy needed for vaporization is even higher. Moreover, one can
see from the second and third events of the TGA that the total amount of carbon detected with
this method of analysis equals 22.6%, which is in very good agreement with the total carbon
content of the new material measured with elemental analysis. In conclusion, we attributed
the third weight loss, at the highest temperature, to the vaporization and desorption of the
10.3% carbon species incorporated as crystalline oligomers of PE in the new material.
355
The above results indicate that the carbon source PE is incorporated in the new material at a
large scale. To check whether the carbon source is interacting or truly bonded to the
framework of the zeolite, more investigations were carried out.
The samples, e.g. pure silicalite-1, the CH2Clz-purified forms of the new material, as
well as the calcined form of the new material were further evaluated using IR measurements,
as displayed in Fig. 2.
Silicalite-1
- \
Pentaergthritol
i./
New material
" ,s ,/~=,.,
~ t
4000 3000 2000 ~00 ~000 4s0
Wavenumber (cm -1)
Fig. 2. Infrared measurements of the conventional silicalite-1, the pentaerythritol source, the CH2C12
purified new material, and the new material calcined at 823K (spectra separated for convenient
reading)
As one can seen on the spectrum of the pure silicalite-1, the typical structure bands for
silicalite-1 appear at 450 (O rocking in SiOSi units), 550-600 (stretching in SiOSi units), 780
and 880 (bending SiOH), 1100 (intense, stretching in SiO4 tetrahedron) and 3420 cm -I
(intense, stretching vOH) [17,18]. In addition, the increased intensity of the bands around
1370, 1450 and 2900 cm -~ are assigned to the C-H stretching and bending vibrations of the
template occluded in the pores of the zeolite [ 19,20].
The spectra of the new material and the pentaerythritol source are much alike. However, a
close-up (not shown on fig. 2) on the spectrum of the new material showed that the typical
silicalite-1 bands at 450, 780 and 1130 cm -~ are present but are very weak in regards of the
strong contribution of the pentaerythritol phase. This strong domination of the PE signal over
the signal of silicalite-1 is commonly explained by the much stronger extinction coefficient of
organic materials per comparison with inorganic materials. This phenomenon complicates the
interpretation of the spectra and prevents from concluding to the presence or not of chemical
bonds between the silicalite-1 and the carbon oligomers. However, silicalite-1 is present in the
new material, since after calcination at 823K or higher temperatures, the new material
resembles pure silicalite-1 (Fig. 2).
Elemental analysis confirms the disappearance of carbon in the calcined sample (Table 1).
Accordingly, IR as well as XRD analyses performed at 393K and 573K demonstrate that the
356
characteristics of the new material are preserved at the measured temperatures, but have
disappeared after calcination at 823K.
Although IR measurements cannot prove whether the oligomers are bonded to the
framework of the silicalite-1, another indication of the incorporation/strong interaction
through Si-O-C coupling is depicted in the SEM picture of the crystals of the new material, in
Fig. 3. Face-enrichment of the crystals of the new material, which normally is not displayed
(Fig. 3, scheme a), occurs by the presence of PE (Fig. 3, scheme b). It is known that
enrichment in crystals faces is due to the inhibition of the fastest growing faces, allowing the
apparition of other faces. A crystal grows in the direction of strong bonds, which form an
interrupted chain through the structure. When a periodic bond chain contains bonds of
different types, the weakest bond present in the chain has a determining influence on the
shape of the crystal [21]. Moreover, silicalite-1 crystals are known to contain structural
defects, as non-intact Si-O-Si bonds within the framework, resulting in internal Si-OH and Si-
O ~groups [22]. In the case of the crystals of the hybrid new material, the enrichment in crystal
faces is assumed to be due to strong interactions of the silanol groups with the pentaerythritol
oligomers. The generation of a new bond chain within the framework, creating extra-faces in
the crystals of the new organo-zeolitic material, would result from the creation of Si-O-C
bonds between the oligomers and the zeolite framework.
(a)
(b)
Fig. 3. Scanning Electron Microscopy of a representative crystal of the new material, showing crystal
face enrichment in the c-direction, and schematic representation of a classic crystal of silicalite-1 (a),
and oligomer-containing silicalite- 1(b)
calcined new material is much higher (0.206 cm-3.g-~) than for the sample of conventional
silicalite-1 (0.158 cm3.g-~).
140
,-- 120
r
E
110 ' 9...... ~;/yI!~
O
It)
"a 100
90
80
70 r i i i i i i i i i i
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Fig. 4. N2 adsorption isotherms on the pure silicalite-1 and the new material, showing a significantly
higher volume adsorbed by the new material than for the classic silicalite- 1.
TEM images of the calcined new material are displayed in Fig. 5. Crystals usually
show pores of around 20 nm and some crystals show a large amount of pores, with sizes
ranging from 50 nm to 100 nm. The porosity of the material may be even higher than what
has been suggested by N2 physisorption, due to the fact that such large pores are beyond the
detection limits of the followed analysis method.
2~:~nm
Fig. 5. TFM images of the calcined new material, presenting pores ranging from 20 nm to 150 nm
Overall, it can be speculated that, based on low pH synthesis conditions which allow
Si-O-C bonding, molecules of PE polymerize and are incorporated as oligomers within the
framework of the zeolite by the condensation of terminal Si-OH groups of silicates with
terminal CHz-OH groups of PE oligomers. It is shown that after removal of the carbon source
upon calcination, enhanced porosity is introduced into the crystals of the new material,
358
resulting in highly porous silicalite-1 crystals. Further investigations are currently being done
and a detailed mechanism for the presented results will be described elsewhere.
5. C O N C L U S I O N
ACKNOWLEDGEMENTS
Within TU Delft, Johan Groen is acknowledged for N2 measurements, Niek van der Pers for XRD
measurements, Delia van Rij for elemental analyses, Loes Schouten for TGA measurements. Sander
van Donk (Albemarle Catalysts) and Tom Visser (Utrecht University) are kindly acknowledged for
helpful discussions.
REFERENCES
[1] S. van Donk, A.H. Janssen, J.H. Bitter, K.P. de Jong, Catal. Rev. 45 (2003) 297
[2] A. Corma, Chem. Rev. 97 (1997) 2373
[3] W.E.J. van Kooten, J. Kapteijn, C.M. van den Bleek, H.P.A. Calis, Catal. Lett. 63 (1999) 3
[4] B.T. Holland, L. Abrams, A. Stein, J. Am. Chem. Soc. 121 (1999) 4308
[5] C. Madsen, C.J.H. Jacobsen, Chem. Commun. (1999), 673
[6] A.H. Janssen, I. Schmidt, C.J.H. Jacobsen, A.J. Koster, K.P. de Jong, Microporous Mesoporous
Mater. 65 (2003) 59
[7] K. Yamamoto, Y. Sakata, Y. Nohara, Y. Takahashi, T. Tatsumi, Science 300 (2003) 470
[8] A. Sayari, S. Hamoudi, Chem. Mater. 13 (2001) 3151
[9] S. Inagaki, S. Guan, T. Ohsuna, O. Terasaki, Nature, 416 (2002) 304
[10] Z.M. Wang, K. Hoshinoo, K. Shishibori, H. Kanoh, K. Ooi, Chem. Mater. 15 (2003) 2926
[11] S. Spange, H. Mtiller, C. J/iger, C. Bellman, Macromol. Symp. 177 (2002) 111
[12] D. Kawashima, T. Aihara, Y. Kobayashi, T. Kyotani, A. Tomita, Chem. Mater. 12 (2000) 3397
[13] H. Robson and K.P. Lillerud, Verified syntheses of zeolitic materials, Elsevier, 2001
[14] M.M.J. Treacy and J.B. Higgins (eds), Collection of Simulated XRD Powder Patterns for
Zeolites, 4th ed., Elsevier Amsterdam, 2001
[15] E.R. Geus, H. van Bekkum, Zeolites 15 (1995) 333
[16] R. Ravishankar, C. Kirschock, B.J. Schoeman, P. Vanoppen, P.J. Grobet, S. Storck, W. Maier,
J.A. Martens, F.C. de Schryver, P.A. Jacobs, J. Phys. Chem. B 102 (1998) 1633
[17] D.P. Zarubin, J. Non-Crystal. Solids, 286 (2001) 80
[ 18] A. Miecznikowski, J. Hanuza, Zeolites, 7 (1987) 249
[19] N.B. Colthup, L.H. Daly, S.E. Wiberley, Introduction to IR and Raman Spectroscopies,
Academic Press London, 1964
[20] R.M. Silverstein, G.C. Bassler, T.C. Morril, Spectrometric Identification of Organic
Compounds, John Wiley & Sons, New York, 1981
[21] P. Hartman, W.G. Perdock, Acta Crystallogr. 8 (1955) 49
[22] J.C. Jansen, J. Schoonman, H. van Bekkum, V. Pinet, Zeolites 11 (1991) 306
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 359
Artificial neural network (ANN) methods were used to predict new clear solution
compositions for the preparation of Na-aluminosilicate zeolite coatings. Synthesis
experiments were performed on stainless steel substrates by using the compositions predicted
by the ANN methods. The coatings obtained were characterized by grazing incidence X-ray
diffraction. As a result, coatings of zeolites P and X could be prepared on the substrates after
only a limited number of experiments.
1. INTRODUCTION
The preparation of zeolite coatings on various substrates bears importance for different
processes related to catalysis, separation, adsorption, sensing, etc [1-4]. In order to obtain
coatings of high quality, suitable clear solution compositions should be used. However, for
many zeolite phases of practical significance, no clear solution composition is available in the
related literature, which can be used for the preparation of zeolite coatings via conventional
syntheses carried out directly in one step. Concerning Al-containing zeolites prepared in
sodium form, pure phase coatings obtained from clear solution compositions by direct
synthesis have only been reported for zeolite A [5], HS [6] and analcime [7], to the best of our
information. Other clear solution compositions available in the literature, yielding stable
colloidal products, may be more fit for the preparation of zeolite coatings via multi-step
synthesis techniques, such as the seed-film method [3]. In this method, the surface of the
substrate is first modified to adsorb a layer of zeolite seeds, after which the crystals are grown
into a continuous film upon hydrothermal treatment. The clear solutions used in this latter
case contain organic templates too.
360
Since searching for new clear solution compositions by carrying out conventional
single or multi-step synthesis experiments may be extremely time consuming, theoretical
attempts may be very valuable in this respect. Due to the unpredictable nature of zeolite
synthesis in many different cases, it seems that relating the type of zeolite obtained to the
initial composition of the reaction mixture should rather be accomplished by methods suitable
to explain complex nonlinear phenomena. Artificial neural networks (ANNs) have the ability
to learn from input data and are very useful for the prediction of complex high-dimensional
data. ANN methods have a broad range of applications and are becoming popular tools in
chemical engineering research, too [8-10]. It was recently shown that ANN methods could
learn efficiently from available zeolite synthesis data in the literature [11]. The deviations of
the estimates made by using artificial neural network methods from experimental results were
quite smaller than those obtained by multilinear and nonlinear regression. The best
predictions of the Si contents of the zeolites that would be obtained from a given starting
reaction mixture were made by the Feed Forward Back Propagation (FFBP) and Generalized
Regression Neural Networks (GRNN) methods while the H20 content was predicted better by
the Radial Basis Function-Based Neural Networks (RBF) method. In general, the amount of
deviation was larger for the prediction of the H20 contents, when compared to the prediction
made for the Si contents, of the zeolites. The use of artificial neural network methods
provides the opportunity of making sensible first guesses for discovering new synthesis
compositions, thus reducing the number of experiments that have to be performed to prepare a
certain type of zeolite
In this study, ANN methods were used to discover new clear solution compositions for
the preparation of coatings of different Na-aluminosilicate zeolites, such as P and X. As a first
step, the FFBP method was applied to a number of compositions. For the compositions that
seemed to suit the molar compositions of the mentioned zeolites best, synthesis experiments
were performed to verify the predictions made by the ANN method. The coatings prepared
were characterized by grazing incidence X-ray diffraction.
2. T H E O R Y
Artificial neural networks are black box models that can perform an estimation using limited
input and output data patterns. In this study, the FFBP method was used to relate the molar
composition of the initial synthesis mixture to that of the final zeolite obtained. The
application of ANN to data consisted of two steps. The first step was the training of the neural
networks. At this stage, the training data, which described the input and output to the network
was presented and the inter-connection weights were obtained. The input data were
normalized before training. The number of hidden layers and the number of nodes in the input
and hidden layers were determined after trying various network structures. The network
structure providing the best result was determined according to the success of the predictions
performed for the data set utilized in the testing stage. In this study, the data set employed in
the training of ANN was a recent version of that used in a previous study [ 11 ], obtained by
361
the inclusion of additional data [12]. The components of the input vector were the ratios of the
reagent molar amounts, namely, HzO/Na20, SIO2/A1203 and H20/SiO2 in the synthesis
mixture, while the components of the output vector were the molar amounts of Si and H20
per mole of A1 in the final zeolite product formed. The network structure providing the best
result in the mentioned previous study [11 ] was employed in the testing stage. In other words,
one hidden layer, 150 iterations and 3 hidden layer nodes were the conditions used. In the
testing period, various new initial reaction mixture composition,s, varied systematically, were
used to predict the type of zeolite that may be formed liom these compositions. The
H20/Na20, Si02/A1203 and H20/SiO2 ratios of the clear solution compositions tested varied
in the ranges of 20-40, 10-60 and 10-150, respectively. Since the prediction ability of the
ANN methods does not seem to be developed appreciably yet for the prediction of the H20
contents of zeolites [l l], only the Si contents determined by FFBP method were taken into
consideration.
A parameter (d) was defined and used as a measure of the success of FFBP method in
the prediction of the Si molar content of the zeolite obtained from a certain reaction mixture.
d was determined by taking into consideration the deviation (%) of Si molar amount
calculated (cca;c) by using the ANN method from the corresponding experimental results (C~xp)
determined in this study.
d,,v was defined as the average of the deviations (d) obtained for the different data used in
prediction.
3. EXPERIMENTAL
As a result of an initial screening of the compositions used in the testing stage, during which,
various new initial reaction mixture compositions, varied systematically, were used to predict
the type of zeolite that may be formed from these compositions, six of the compositions were
decided to be worth investigating further for their suitability to produce zeolites X and P. The
success of the predictions made, was tested by performing synthesis experiments. Coatings
were prepared on stainless steel substrates (AISI 316 grade) in an oven kept at 75-80~ for
various periods of time. Sodium silicate solution (Merck), granular sodium aluminate (Riedel-
de Haen), anhydrous NaOH pellets (Carlo Erba) and deionized water were used in the
synthesis experiments. The stainless steel plates, cleaned in accordance with previously
reported procedures [13] prior to the experiments, were placed vertically in teflon holders and
were immersed in the synthesis mixture sealed in polyethylene containers. Grazing incidence
X-ray diffraction (Philips PW-3020) was used to characterize the coatings synthesized on
stainless steel substrates. XRD was run at an incidence angle of 1~ with Cu Kotl radiation and
was applied for phase identification and determination of the purity of the samples prepared.
362
+
1.6
+ o +
o
1.2 q~ o
0
E
~5 0.8
0.4
I ........... I I .... I ,
0 1 2 3 4 5 6
Data Number
Fig. 1. Zeolite composition predictions by (o) FFBP in comparison to (+) experimental results.
For each reaction mixture composition utilized, synthesis experiments were carried out
for different periods of time till a zeolite phase could be detected by X-ray diffraction. For the
compositions tested, this period generally did not exceed 24 hours at 80~ It was interesting
to note that at lower synthesis temperatures and/or shorter synthesis periods, the precursor
phase recently observed before the formation of many different Na-aluminosilicate zeolites
from clear solutions [6], appeared in the X-ray diffractograms of the samples prepared in this
study, too. The precursor phase was observed to be present in the samples before the
formation of both X and P-type zeolite coatings obtained from different clear solution
compositions. An example depicting the X-ray diffractogram of the precursor phase is shown
in Fig 2. The synthesis was carried out at 75~ for 18 h in this case from the composition
represented by data number 5. The X-ray diffractograms of the zeolite X and P coatings
grown on stainless steel plates by using the compositions represented by data number 3 and 5
are shown in Figs. 3 and 4, respectively. Syntheses were carried out at 80~ for 24 h in these
cases.
364
1000i
800
600
!
400
200
I I . I
5 10 15 20 25 30 35 40 45 50
20 ~
Fig. 2. X-ray diffractogram of the precursor phase obtained after 18 hours of synthesis at 75~
2000
1500
~
-,-,
C
1000
500
5 15 25 35 45
20 ~
Fig. 3. X-ray diffractogram of the zeolite X coating prepared after 24 hours of synthesis at 80~ SS
denotes the two peaks of stainless steel at about 43.5 ~ and 44.5 ~
365
800
600
. ~
" 400
t'--
200
0 [_ ........ I . . . . . I , I . . . . . . . . . I
5 15 25 35 45
20 ~
Fig. 4. X-ray diffractogram of the zeolite P coating prepared after 24 hours of synthesis at 80~ SS
denotes the two peaks of stainless steel at about 43.5~ and 44.5 ~.
As a result of the investigations carried out in this study, guided by ANN methods,
different compositions could be obtained for the preparation of P-type coatings, which may
allow for the selection, by the help of various techniques, such as scanning electron
microscopy, of the most suitable initial reaction mixture composition to obtain high quality
zeolite coatings for the target application.
5. CONCLUSIONS
Coatings of zeolites P and X could be prepared, as guided by the predictions provided by the
FFBP method. It should be noted that no report has been made previously in the literature
about the preparation of pure phase coatings of these zeolites from clear solution
compositions by using conventional synthesis carried out directly in one step. Different
compositions could be obtained for the preparation of P-type coatings, which may allow for
the selection, by the help of various techniques, such as scanning electron microscopy, of the
most suitable initial reaction mixture composition to obtain high quality zeolite coatings for
the target application. The use of the ANN method was observed to decrease significantly the
number of experiments that has to be performed to discover new synthesis compositions. The
inclusion of synthesis time and temperature as additional input parameters in the ANN may
allow a better understanding of the zeolite synthesis and may decrease further the number of
experiments that have to be performed to obtain a zeolite phase.
366
REFERENCES
LTL-type zeolite films were synthesized on glass supports from two initial systems yielding
coin-like and rod-like crystals. The syntheses were realized with and without preliminary
adsorption of nanosized zeolite L crystals. The experiments performed without preliminary
adsorption of seeds resulted in isolated zeolite L crystals formed on the glass surface. The
seeded syntheses provided well-intergrown films, where after weakening of the effect of the
seeds on the zeolite film growth, the orientation of the films was controlled by the
morphology of the growing crystals. Thus, the coin- and rod-like crystals showed a tendency
to crystallize with their c-axis parallel and perpendicular to the support surface, respectively.
The films were characterized by scanning electron microscopy (SEM), grazing incidence X-
ray diffraction (GID) using synchrotron X-ray radiation and nitrogen adsorption
measurements.
1. INTRODUCTION
There is a continuous interest in preparing zeolite films of various structure types due to their
potential application as selective membranes, electrodes, sensors, and optoelectronic devices
[1-3]. In general, the morphology and the size of the zeolite crystals are of great importance
for their performance. The zeolite morphology is of particular significance in the case of
films, where the crystals are fixed and the flows through the layer are predetermined by the
film structure. Other factors such as the uniformity, i.e., the absence of cracks and pin-holes,
the thickness and the orientation of the zeolite films are also of importance for the
performance of the materials.
The control of the orientation of zeolite films has been a topic in a number of
publications with most of the studies devoted to the purely silica MFI-type zeolite. Wang and
Yan have reported the control of the orientation of MFI-type thin films by in situ
crystallization and careful adjustment of the synthesis mixture composition [4]. However, the
application of such a method is limited. More universal way to control the orientation of
zeolite films is by the preliminary deposition of zeolite seed crystals [5-7]. As the oriented
films prepared by pre-seeding are usually very thin, other methods have been developed in
order to keep the orientation as the films grow thicker. Thus, the synthesis of a siliceous
368
ZSM-5 membrane has been reported, in which the secondary growth of the pre-seeded
support was realized in a system containing trimer-TPA template, which enhanced the relative
growth along the b axis thus preserving the b-orientation of the zeolite seed layer even after
film thickening by secondary growth [8].
Zeolite L is the synthetic counterpart of the mineral perlialite, a material with with one-
dimensional large-pore system parallel to the c-axis. The typical morphology of zeolite L is a
combination of two types of hexagonal prisms, (100) and (1-10), and a pinacoidal face (001)
[9,10]. For molecular sieve materials with one-dimensional channel systems, the morphology
and the ordering of the crystals in the zeolite films determine the capacity of the film for a
particular application. Thus, for optimum performance, zeolite L films with the c-axis
perpendicular to the support surface (perpendicular to the surface pores) are needed.
The effect of the synthesis parameters on the formation of zeolite L has been studied by
several research groups. Tsapatsis et al. have used high-resolution transmission and field-
emission scanning electron microscopy to study the formation of zeolite L nanocrystals [11].
Special studies have been devoted to the role of alkali earth cations on the morphological
features of zeolite L crystals, in particular barium and magnesium [12,13]. The detailed
investigation of the effect of the gel composition on the size of zeolite L crystals performed
by Megelski and Calzaferri have resulted in the synthesis of crystals in the size range 30 to
3000 nm [14]. These crystals were further used for the preparation of an optical antenna
system. The quoted investigations provided precious information about the effect of some of
the synthesis variables on the zeolite L crystal size. However, the relation between the
synthesis conditions and the growth of the different zeolite crystal faces and their aspect ratios
has not been explored. For instance, a shortening of the prismatic faces is expected to lower
the diffusion limitations, while an increase of the pinacoidal ones would raise the number of
pore openings per unit external crystal surface. Recently, a detailed study on the influence of
different synthesis variables on the morphology of zeolite L crystals was performed [ 15]. This
investigation revealed that by varying the synthesis conditions, morphologies ranging from
short prismatic (coin-like) to long prismatic (rod-like) crystals could be obtained. The gels
yielding those two type of morphologies were employed in this work. The present
contribution reports on the effect of the crystal morphology on the orientation of zeolite L
films on glass supports.
2. E X P E R I M E N T A L
Zeolite L seeds with a size of 140 nm were synthesized from a gel with the molar composition
0.25K20:0.08AI203:1.0SiO2:15H20:0.005BaO after a hydrothermal treatment at 170~ for 3
days [15]. The seeds were purified by three times centrifugation and redispersion in distilled
water in an ultrasonic bath. The surface charge of pre-cleaned with acetone glass supports was
reversed using a 1 wt. % cationic polymer solution (poly(diallyldimethylammonium chloride),
20 wt.% in water, Aldrich) and then the negatively charged zeolite L seeds were adsorbed.
Pre-seeded or bare supports with a size of 22xl 1 mm were positioned vertically in PTFE-
lined autoclaves and treated for different periods of time at 170~ Two synthesis gels were
employed with the molar composition:
0.25K20:0.08AI203:1.0SiO2:10H20 (gel A)
0.25K20:0.08A1203:1.0SIO2:20H20 (gel B),
the only difference between them being the water content. After the synthesis, the supports
were treated with a 0.1 M ammonia solution in an ultrasonic bath for 10 min to remove the
loosely attached crystals, rinsed repeatedly with distilled water and dried at 80~
369
The films were characterized by scanning electron microscopy (SEM, Philips XL30
LAB6), grazing incidence X-ray diffraction (GID) using synchrotron X-ray radiation (at the
ESRF, Grenoble) and nitrogen adsorption measurements after degassing the samples at 300~
overnight. Specific surface areas were calculated using the BET equation. The size of the
zeolite L nanocrystals employed was determined by dynamic light scattering using a Malvern
HPPS-ET instrument.
Previously, it was found that the water content in the system had the most pronounced effect
on crystal morphology and coin- or rod-like crystals with different prismatic ((100) and
(1-10)) to pinacoidal (001) faces aspect ratios could be obtained varying this content [ 15]. In
the present study, zeolite L films were prepared using gels yielding the two types of crystals,
gel A yielding coin-like crystals and gel B yielding rod-like crystals. The initial experiments,
performed without preliminary adsorption of seed crystals, provided similar results with both
gels. Fig. 1 shows SEM top view images of glass supports treated with gel A. After 14 h of
treatment, an amorphous layer was formed on the support (Fig. 1 A). Upon prolongation of
the hydrothermal treatment to 24 h, isolated well developed zeolite L crystals with a coin-like
morphology can be observed on the support surface (Fig. 1 B). Also, the major part of the
amorphous layer seemed dissolved. Further prolongation of the hydrothermal treatment did
not lead to the formation of continuous films and additional experiments were performed by
seeding the glass supports prior to the hydrothermal treatment with gels A and B.
Top and cross-sectional views of the films prepared with gel A after seeding for different
times of treatment are shown in Fig. 1. The films prepared after 24 h of treatment were
continuous and composed of well intergrown crystals with a coin-like morphology (Fig. 2 A
and B). No distinguishable border between the seeded and the grown layer could be seen in
the SEM micrographs (Fig. 2 B). The thickness of the films was about 1.5 lam. The thickness
of the films slightly changed upon prolongation of the treatment to 34 h to 1.7 gm (Fig. 2 D).
The SEM analysis was used to judge the orientation of the zeolite L films prepared with gel
A. The crystals of the upper part of the film obtained after 24 h of treatment were randomly
oriented (Fig. 2 A and B). On the other hand, a substantial number of the crystals composing
the upper part of the films prepared for 34 h of treatment seemed oriented
Fig. 1. SEM top view images of glass support treated with gel A for 14 h (A) and 24 h (B).
370
~ ~!~i!iiiiiii!il
~ i i~i~i:~:i~,i~'~i ~F~
~iiiii
84184
i~i~i!iii~i~!ii~ii!!ii~i~i
!i!ii~i~i~i!!!|i!iJi~ ~i~i@!ii~ili~i~!!i!i!i~i~i!i~i!i~!!!~i~
ii~iii~!ii!ii~ ~ii~ii!i~i~i~~ iili
: ....
Fig. 2. SEM top view (A, C) and cross-sectional view (B, D) micrographs of LTL-type films prepared
with gel A for 24 h (A, C) and 34 h (B, D) ofhydrothermal treatment.
with the c-axis parallel to the support surface (Fig. 2 C and D). Different behaviour was
observed for the films synthesized with gel B. Top and cross-sectional views of those films
prepared for different times of treatment are shown in Fig. 3. After 24 h of treatment, a layer,
which does not contain the distinctive features of a crystalline material, was formed on the
substrate. Particles with sizes in the range 200 to 500 nm were observed on the surface of this
layer (Fig. 3 A and B). The number of particles and their size increased after 34 h of
hydrothermal treatment and the crystals already have the distinctive morphological features of
zeolite L (Fig. 3 C and D). Those crystals were predominantly oriented with their c-axis
perpendicular to the support surface. Well distinguished border between the two layers can be
seen and the thickness of the zeolite film increased to ca. 800 nm (Fig. 3 D). The secondary
layer crystals were, however, not well intergrown and showed comparatively low density
distribution along the support surface. The formation of the lower layer film can be related to
the effect of the zeolite seeds adsorbed on the surface of the support. A new generation of
nuclei was formed on the top of the seeded layer, which promotes the growth of the upper c-
oriented layer. Further hydrothermal treatment of up to 48 h increased the density of the
secondary crystal layer. At this stage well shaped crystals with a rod-like morphology can be
seen (Fig. 3 E and F). The size of the crystals increased and films with a thickness of ca. 1.8
~tm were obtained (Fig. 3 F). The border between the lower and upper layers was even more
distinguishable in this case and holes can be seen at this border (Fig. 3 F). Again, the majority
of the crystals composing the upper film layer were oriented with their c-axis perpendicular
to the support surface.
371
~:.
, ...: ...
. . ....~,~,:. .
. .... o,....
Fig. 3. SEM top view (A, C, E) and cross-sectional view (B, D, F) micrographs of LTL-type films
prepared with gel B for 24 h (A, C), 34 h (B, D) and 48 h (E, F) ofhydrothermal treatment.
The BET surface area of the non-treated plates was 0.01 m 2 g-i. A gradual increase of the
surface area with the crystallization time was observed for the seeded plates subjected to
hydrothermal treatment with the two synthesis gels (Table 1). Samples synthesized for 34 h
showed higher surface areas than those obtained for 24 h, which reveals that the films are still
growing in the time range 24-34 h. The film growth continues to 48 h when gel B is
employed, while no further growth was detected after 34 h hydrothermal treatment with
gel A. Those results are in accordance with the SEM analysis. Surface area measurements
revealed also that the film synthesized with gel A contained more microporous material,
372
which most probably is due either to the increased density of the films prepared with this gel
compared to the density of the films prepared with gel B.
Table 1
Specific surface areas for_the films prepared with gels A and B for different treatment times.
Time (h) Specific surface area (m 2 g-l) 7
Gel A Gel B
24 5.8 3.8
34 9.2 5.6
48 9.1 7.3
The SEM investigation was coupled with a depth sensitive analysis by measuring the
Bragg reflections with the reciprocal lattice vector parallel to the samples' surface. Depth-
dependent structural information on a certain lattice plane was obtained by keeping the
corresponding 20{hU) constant and varying the incident angle ai for all samples. Reflections
with (hkl) values equal to (002) and (500) were used for these measurements, where the
incident angle ~q was varied from -0.05 ~ to 0.4 ~ The intensity of Bragg peaks as a function of
the incident angle for the film grown with gel A is depicted in Fig. 4 a and b. The shape of the
curves clearly shows that the orientation of zeolite L crystals in the grown film after 24 h of
hydrothermal treatment is random. In contrast, the films grown with gel B showed a preferred
orientation. The well pronounced peak obtained by measuring the (002) reflection clearly
shows that the film is c oriented (Fig. 5 a). These measurements were performed on films
obtained for 24 h ofhydrothermal treatment, when the effect of the adsorbed seeds on the film
orientation is not so pronounced. As was discussed above, at this stage a new
generation of nuclei is already formed. This new generation of nuclei obviously controls the
film growth throughout the second stage of film formation since the c-orientation of zeolite
crystals is constant till the end of the crystallization process. Thus, the results of the gracing
(002) ~/..~(a) ( b)
d d
>.,
~
~z
=o
| |_
-0.1 0.0 0.1 0.2 0.3 0.4 -0.1 0.0 0.1 0.2 0.3 0.4
Incident angle / ~ Incident angle / ~
Fig. 4. Depth dependent distribution of (002) (a) and (500) (b) reflections for the film synthesized with
gel A for 24 h.
373
>h b
. ,..,.q
1 I n | i i I ~ I ~ I a I i
-0.1 0.0 0.1 0.2 0.3 0.4 -0.1 0.0 0.1 0.2 0.3 0.4
Incident angle / ~ Incident angle / ~
Fig. 5. Depth dependent distribution of (002) (a) and (500) (b) reflections for the film synthesized with
gel B for 24 h.
incidence diffraction (GID) study confirmed the c-orientation of the top layer of zeolite films
synthesized for 24 h of treatment from gel B and the random orientation of the films
synthesized with gel A observed by SEM.
4. CONCLUSIONS
LTL type zeolite films were synthesized on glass supports without and with adsorption of
zeolite L seed crystals. Dense and continuous zeolite films could not be obtained without a
preliminary adsorption of zeolite L seeds and only isolated crystals were formed on the
support surface. The films were continuous after preliminary adsorption of 140 nm seed
crystals on the support. Two synthesis gels yielding crystals of coin- or rod-like morphologies
were employed. The films obtained with the coin-like crystals were randomly oriented,
whereas the ones with the rod-like morphology were growing with a preferred orientation
under the effect of the seeds. The majority of the overgrown layer crystals obtained with latter
gel was oriented with the c-axis perpendicular to the support. However, the randomly oriented
films obtained with the gel yielding the coin-like crystals were denser and showed higher
BET surface areas. The study can be used for preparation of zeolite L films with increased
number of pore openings per crystal surface, which is beneficial from application point of
view.
ACKNOWLEDGEMENTS
The financial support from the CNRS-DFG bilateral program is gratefully acknowledged.
REFERENCES
[3] S. Nair and M. Tsapatsis, Handbook of Zeolite Science and Technology, S.M. Auerbach, K.A.
Carrado, P.K. Dutta (eds.), Marcel Dekker, New York, 2003, p. 869.
[4] Z. Wang and Y. Yan, Chem. Mater. 13 (2001) 1101.
[5] J. Hedlund, J. Porous Mater., 7 (2000) 455.
[6] A. Gouzinis and M. Tsapatsis, Chem. Mater. 10 (1998) 2497.
[7] T.H. Metzger, S. Mintova and T. Bein, Microporous Mesoporous Mater. 43 (2001) 191.
[8] Z. Lai, G. Bonila, I. Diaz, J.G. Nery, K. Sujaoti, M.A. Amat, E. Kokkoli, O. Terasaki, R.W
Thompson, M. Tsapatsis and D.G. Vlachos, Science 300 (2003) 456.
[9] T. Ohsuna, Y. Horikawa, K. Hiraga and O. Terasaki, Chem. Mater. 10 (1998) 688.
[10] A. Burton and R.F. Lobo, Microporous Mesoporous Mater. 33 (1999) 97.
[11 ] M. Tsapatsis, M. Lovallo, T. Okubo and M.E. Davis, Chem. Mater. 7 (1995) 1734.
[12] S.J. Tauster, D.E.W. Vaughan and J.J. Steger, US Patent 4552856 (1985).
[13] S. Ferchiche, J. Warzywoda and A. Sacco, Stud. Surf. Sci. Catal. 135 (2001) 188.
[14] S. Megelski and G. Galzaferri, Adv. Funct. Mater. 11 (2001) 277.
[15] O. Larlus and V.P. Valtchev, Chem. Mater., 16 (2004) 3381.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. 2;ilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 375
T. D o u s s i n e a u , *a E. G a v i l a n , a, A. E! M a n s o u r i , a M. Smai'hi, a J.-O. D u r a n d b
and M. G r a n i e r b
a IEM UMR 5635, CNRS, 1919 route de Mende, 34293 Montpellier Cedex 5, France.
b CMOS UMR 5637 cc 007, UM2, Place Eugfne Bataillon, 34095 Montpellier Cedex 5,
France.
1. INTRODUCTION
Inorganic materials with organized and controlled porosity are used in numerous and various
technologies involving nanotechnologies(catalysis, photonics, biotechnologies .... ). This
attractivity is reinforced by the recent availability of these materials as colloidal nanoparticles
which enables the addition of various functionalities at their external or/and internal surfaces.
Generally, these nanoparticles are used as elementary <<units >>for the building of the final
material as a film, a thin layer or a monolithic structure.
The organization of the porosity in these materials enables the confinement of
chromophores in a regularly arranged spatial distribution which also improves their
photophysical properties as well as their photostability [1]. Chromophores entrapped in
molecular sieves are attracting attention with respect to novel applications such as inclusion
pigments [2], components in micro-photonic devices for frequency doubling [3], photo-
induced switching [4] or emission of polarized laser radiation [5]. The spatial organization of
the molecular dipoles in the material enables also to create systems for second harmonic
generation [6] or data recording [7]. In life sciences, nanoparticles bearing optical properties
are developed [8] as ultra-sensitive analysis tools for diagnostics [9], biosensors, biocatalysts
and DNA chips[ 10] for example.
From a fundamental point of view, the availability of colloidal suspensions of dye-
loaded zeolite nanoparticles may provide a unique opportunity to study the behaviour of dye
molecules confined within microporous cavities. In confined spaces, the spectroscopic and
physical properties of dye molecules may be much different from that in solution.
Spectroscopic studies on solid powdered materials such as zeolites commonly resort to diffuse
reflectance techniques which require appropriate theoretical treatment to take into account
376
scattering and other signal modification due to the solid phase. Therefore, if small enough
colloidal zeolite particles are used excessive light scattering problems will be avoided. Thus,
dye-loaded zeolite nanoparticles may provide a good alternative for spectroscopic studies of
confined molecules within microporous cavities.
The preparation of hollow materials often involves a first synthesis step based on
supramolecular templating of structures [11]. Accessibility to the porosity is obtained by a
second treatment which consists on the removal of the structure directing agents (SDA). This
is done either by high temperature calcinations or solvent extraction. In case of nanoparticles,
these last treatments may lead to irreversible particles aggregation [12,13] and modification of
the particles surface properties. Recently, some procedures have been developed to inhibit the
particles aggregation during the template extraction procedure [14]. These procedures enable
the obtention of zeolite nanoparticles with a low polydispersity, particle sizes with diameter
similar to the initial particle size and an accessible microporosity from which the template
used for the synthesis has been eliminated. Therefore, the particles can be used as hosts for
further confinement in their microporosity.
Here, a procedure has been developed for the insertion of fluorescent molecules into the
microporosity of zeolite nanoparticles. Zeolite beta was been chosen as the host ; it is a high-
silica 12-ring pore size zeolite containing a three-dimensional pore system. Two fluorescent
dyes [3-hydroxyflavone (3-OHF) and Rhodamine B (RhB)] have been entrapped in
microporous colloidal zeolite Beta nanoparticles (Fig. 1.). Rhodamine B is a xanthene dye
usually used as an active medium for tuning lasers. Recently it has found a wide application
as water tracing agent, molecular probe and solar collector, etc. 3-hydroxyflavone is a benzo-
7-pyrone derivative which is widespread in the plant kingdom. The presence of hydroxylated
benzene rings in 3-OHF confers antioxidant activities as well as optical wavelength-shifters
properties to the molecule. Both dyes optical properties are considerably dependent on their
environment (pH, polarity) ; therefore, they are excellent probes of their confinement.
2. EXPERIMENTAL
width of the benzoic acid group from the oxygen atom to the opposite hydrogen atom is ca.
0.6 nm [ 16]. 3-Hydroxyflavone is a planar molecule with a theoretical length of 0.9 nm and a
width of 0.5 rim. As zeolite-13 possesses a one-dimensional channel structure with free
opening of about 0.76x0.64 nm 2, the sizes of both dyes are compatible with their insertion in
the zeolite channels.
The dye-loaded zeolite nanoparticles were prepared by liquid-phase adsorption. Typically,
an alcoholic suspension of template-free zeolite nanoparticles (300 mg of nanoparticles in 50
ml ethanol) was added to a chromophore solution (rhodamine B 99%+, or 3-hydroxyflavone
99%, Acros Organics) and stirred at room temperature for 24 hours. The concentration of the
chromophore solution was varied depending on the desired loading (between 10-8 and 10 -4
tool g-i zeolite, Table 1). After reaction, the non-entrapped chromophore molecules were
eliminated by repeated centrifugation cycles (12 000 rpm, 30 minutes, 10 ~ and re-
dispersion by ultrasonication until no absorbance could be detected in the supernatant. After
the last step of the purification procedure, the dye-loaded zeolite nanocrystals were re-
dispersed in a limited amount of water. The zeolite content of this concentrated suspension
was found out by drying at 110 ~ The suspension was further diluted in order to adjust the
concentration to 0.5 wt.% for the spectroscopic studies. Regardless to the chromophore used,
stable suspensions were obtained. For solid-state characterization techniques, the recovered
solid was dried at 80-100 ~ overnight.
After the purification procedure, coloured powders (or suspensions) were obtained. In
order to evaluate the amount and influence of external-surface-adsorbed chromophore
molecules, SDA-filled colloidal particles (as-synthesized colloids) have been submitted to the
same treatments (liquid adsorption and purification procedures) as SDA-free samples.
The thermal and photochemical stability of the loading was evaluated after submitting the
dye-loaded samples to extensive reflux in acidic ethanol at 80~ for 7 days.
Table 1 reports the characteristics of the samples prepared for this study.
a) ,c2,5,2,\T/~/o~N.<.< b) i ..... i
.J---:. . . . . . . . . . . . . . . . . . . . . . . . . . o
Fig. 1. Molecular structures of the chromophores entrapped in the zeolite cages, a) Rhodamine B, b)
3-Hydroxyflavone.
3. R E S U L T S AND D I S C U S S I O N
Table 1
Textural and chemical characteristics of the dye-loaded samples used
..... .....
I (cts)
1 50000
~ I
H 0 , ,
] I I ' I
cationic form
H
400 500 600 700
Tautomer T* wavelength (nm)
~m = 520-575 nm
Fig. 3. (A) Excited state proton transfer process for the 3-hydroxyflavone molecule and cation form of
3-hydroxyflavone. (B) Fluorescence emission spectra of 3-OHF dye in solution (grey line) and
entrapped in the zeolite colloids for various loadings (2.5 10.4 (solid line), 5 10.5 (dotted line), 2.5 10-6
(dashed line)). Excitation wavelength : 345 nm
380
On the other hand, the host system of the fluorophore molecule in solid media is
completely different from the 3-OHF containing liquid mixture. The rigidity of the
surrounding micropore network can not be neglected and it can play an important role in the
deactivation of the normal and tautomeric excited states [24]. In a rigid matrix, torsional
motions of the 3-OHF phenyl and 7-pyrone rings and the matrix reorientation following the
dipole moment change of 3-OHF molecule on excitation are restricted. The energy levels of
the normal and tautomeric forms are then perturbed and different emission features appear as
related to the environing cage rigidity.
/
/
/
/
0,0 570 ' ' '1 ' ' ' '1 ' ' ' '1
I I I
10.7 10~ 10S
500 600 700
Wavelength (nm) Dye concentration (mol/g)
Fig. 4. (A) Fluorescence excitation (grey) and emission (black) spectra of RhB dyes in solution (dotted
lines) and entrapped in the zeolite colloids (loading 4 10-3) (solid lines). Excitation wavelength 9 543
nm (B) Variation of the maximum of the emission peaks as a function of the dye concentration in the
zeolite nanoparticles.
This behaviour has been observed in previous studies for rhodamine B adsorbed on
various solid surfaces [25,26]. The different environment of the dye in an ethanolic solution
or in zeolite cage determines the observed spectroscopic differences [25]. The slightly less
polar nature of the aluminosilicate cage as compared to the solution affects the fluorescent
properties of the RhB [27].
The emission peaks of the RhB loaded particles were continuously red-shifted with
increasing dye concentration from 570 nm to 578 nm (Fig. 4B.). Since there is no chance for
the fbrmation of RhB dimmers in the zeolite beta cages, it is assumed that the continuously
read-shift in fluorescence emission peaks results from a radiationless energy transfer, which is
known as F6rster quenching [28].
381
Table 2
Optical fluorescence emission properties of the dye-loaded nanoparticles. Excitation
wavelength : 345 nm for 3-OHF, 543 nm for RhB.
3-OHF RhB
EtOH 2.5 10"6 5 10-5 2.5 10-4 EtOH 2 10"6 4 10.5 4 10.3
~llax
(nm) 537 451 450 450 568 578 574 571
3.2.4. Photostability
These optical properties are still present after refluxing the suspension 7 days in acidic
ethanol. This demonstrates the stability of the entrapment as well as the good photophysical
stability of the dye-loaded material.
4. CONCLUSION
Microporous zeolite nanoparticles can be used as a host for the confinement of organic dyes
leading to thermally and chemically stable materials. The textural and spectroscopic
characterization of these materials showed the strong dependence of the dye optical properties
with its physico-chemical environment.
382
ACKNOWLEDGMENTS
Funding of this work by the Minist6re Dd16gu6/t la Recherche (ACI Nanosciences 045 270
and 042 130) is gratefully acknowledged.
REFERENCES
[1] G. Schulz-Ekloff, D. W6hrle, B. van Duffel and R.A. Schoonheydt, Microporous Mesoporous
Mater. 51 (2002) 91-138.
[2] I. Braun, G. Schulz-Ekloff, M. Bockstette and D. W6hrle, Zeolites 19 (1997) 128.
[3] J. Caro, G. Finger, J. Kornatowski, J. Richter-Mendau, L. Wemer and B. Zibrovius, Adv.
Mater. 4 (1992) 273.
[4] K. Hoffman, F. Marlow and J. Caro, Adv. Mater. 9 (1997) 567.
[5] I. Braun, G. Ihlein, F. Laeri, J.U. N6kel, G. Schulz-Ekloff, F. Sch0th, U. Vietze, (). Weiss and
D. W6hrle, Appl. Phys. B70 (2000) 335.
[6] H. S. Kim, S. M. Lee, K. Ha, C. Jung, Y.-J. Lee, Y. S. Chun, D. Kim, B. K. Rhee and K. B.
Yoon, J. Am. Chem. Soc. 126 (2004) 673-682.
[7] J. Strickler, W. Webb, Opt. Lett. 1991, ! 780.
[S] I. Gill, Chem. Mater. 13, (2001), 3404-3427.
[9] C.M. Niemeyer, Angew. Chem. Int. Ed. 40 (2001)4128-4158.
[10] J-M Nam, C.S. Thaxton, C. A. Mirkin, Science 301 (2003) 1884-1886.
[ 11 ] B.J. Schoeman, J. Sterte and J.-E. Otterstedt, Zeolites, 1994,14, 110.
[12] L.M. Huang, Z. B. Wang, J. Y. Sun, L. Miao, Q. Z. Li, Y. Yan and D. Y. Zha, J. Am. Chem.
Soc., 2000, 122, 3530.
[13] B.J. Zhao, S. A. Davis, N. H. Mendelson and S. Mann, Chem. Commun., 2000, 781.
[14] M. Sma]'hi, E. Gavilan, J.-O. Durand, V. Valtchev, J. Mater. Chem., 14 (2004) 1347-1351.
[15] R.W. Ramette, E.B. Sandell, J. Am. Chem. Soc. 78 (1956) 4872.
[ 16] D. Wang, Li-Jun Wan, C. Wang and C.-L. Bai, J. Phys. Cem. B 106 (2002) 4223-4226.
[17] N.B. Castagnola and P.K. Dutta, J. Phys. Chem. B, 105, 2001, 1537-1542.
[18] N.B. Castagnola and P.K. Dutta, J. Phys. Chem. B, 102, 1998, 1696-1702.
[19] O.S. Wolfbeis, A. Knierzinger, R. Shipfer, J. Photochem. 21 (1983) 2133.
[20] P.K. Sengupta, M. Kasha, Chem. Phys. Lett. 68 (1979) 382.
[21 ] G.A. Brucker, T.C. Swinney, D.F. Kelley, J. Phys. Chem. 95 (1991) 3190.
[22] S. Carturan, A. Quaranta, G. Maggioni, A. Vomiero, R. Ceccato, G. Della Mea, J. Sol-Gel Sci.
Tech. 26 (2003).
[23] G.J. Woolfe, O.J. Thislethwaite, J. Am. Chem. Soc. 103 (1981) 6916.
[24] D. Avnir, D. Levy and R. Reisfeld, J. Phys. Chem. 88 (1984) 5956.
[25] F. L6pez-Arbeloa, P. Ruiz-Ojeda and I. L6pez-Arbeloa, J. Lumin. 44 (1989) 105.
[26] F. del Monte and D. Levy, J. Phys. Chem. B 102 (1998) 8036-8041.
[27] J.-C. Pouxviel, B. Dunn and J.I. Zink, J. Phys. Chem. 93 (1989) 2134.
[28] M. Ganschow, M. Wark, D. W6rhle and G. Schulz, Angew. Chem. Int. Ed. 39 (2000)161.
[29] U. Narand, R. Wang, P.N. Prasad and F.V. Bright, J. Phys. Chem. 98 (1994) 17.
[30] A.A. Bhuiyan and J.R. Kincaid, Inorg. Chem. 38 (1999) 4759-4764.
[31 ] M. Sykora, J.R. Kincaid, P.K. Dutta and N.B. Castagnola, J. Phys. Chem. 103 (1999) 309.
Studies in Surface Science and Catalysis, volume 158
J. t2ejka, N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 383
1. INTRODUCTION
For industrial uses, the immobilization of biocatalysts offers several advantages that include
reuse, easy product separation, enhancement of enzyme stability and continuous operation.
Adsorption of a biocatalyst in/on a water-insoluble carrier is the easiest and oldest method of
immobilization. The advantages of immobilization by adsorption are the simplicity of the
procedure and the lower enzyme activity loss. The lipase enzyme hydrolyses ester bonds in
tri-, di- and monoacylglycerols. Alkyl esters of fatty acid produced by alcoholysis of
vegetables oils are important intermediates in food and oil processing. Actually, the interest
for alkyl esters as partial substitutes for diesel fuel has increased [1,2]. It is well know that
biodiesel is a non-toxic, biodegradable and renewable energy source. Several researchers have
studied the enzyme immobilization on different carriers by physical adsorption method or by
chemical linkage [3-12]. Zeolites and mesoporous materials are of great interest in the
adsorption processes due to their high surface area (up to 1400 m2/g for mesoporous
materials), hydrophobic or hydrophilic behaviour, electrostatic interactions, mechanical and
chemical resistance. Moreover, the compositional and structural variances of molecular sieves
offer a powerful tool for tuning the carrier properties. The screening results obtained in our
previous studies for various supports show that the Silicalite-1 surface is a good support for
the lipase attachment [13]. Starting from these results, the self-bonded pellets of Silicalite-1
have been prepared. This work presents different procedures for the preparation of Silicalite-1
pellets and compares the effects of the F- or OH- ions, present in the initial reaction mixture,
on the resistance of the so-formed materials. The lipase immobilization is strongly affected by
the pellet preparation procedure. Finally, an explanation of the different enzyme
immobilization behavior of the pellets is proposed.
384
2. E X P E R I M E N T A L
2.1. Materials
A commercial lipase Rhizomucor miehei (RML) - Palatase, from NovoNordisk, Denmark,
was used. The chemicals used for the synthesis of Silicalite-I pellet type materials were:
Silica Gel Precipitated (BDH) or Silica Fumed 200 (Aldrich) as silica sources,
tetrapropylammonium bromide (Aldrich) as a structure directing agent and sodium hydroxide
(Carlo Erba) or sodium fluoride (Adrich) as mineralizing agents.
2.2. Preparation of pellets from solid-state mixture of a precursor gel (direct synthesis)
The batch composition of the synthesis system for obtaining self-bonded Silicalite-1 was:
1 SiO2- 0.12 NaX - 0.32 T P A B r - 0.8 H20
where X stays for O H or F-.
The initial gels were prepared by adding the mineralizing agent to the template. The so-
obtained mixture was homogenized in a mortar. The distilled water and silica were
successively added to the mixture. The mixture was further homogenized. Then the
amorphous gels were prepared in the required pellet form by applying pressures of 70+80
MPa. The so-obtained pellets were put into a special autoclave [14] in which the water phase
was separated from the gel pellet and heated up to 170+2~ under autogeneous pressure for
24 hours.
Dubnoff bath at room temperature. The amount of enzyme immobilized on the support was
evaluated using UV Adsorption Methods [13].
The higher degree of cohesion exhibited by the pellets prepared in the F- media is probably
due to the five-coordinate silicon [SiO4/2F]- units and to the presence of fluorine occluded in
the small [415262] cages of the MF! structure [16]. Consequently, the weight loss after the
leaching test, observed for the samples prepared in the presence of fluoride ions is lower than
the one observed in the case of pellets prepared from OH-media. In fact, the zeolite materials
obtained from alkaline systems usually have a high number of defect groups [17]. The
difference in the cohesion behaviour of samples 3 and 4 is probably due to the specific surface
area of the silica sources (220 m2/g for the sample 3, 550 m2/g for the sample 4).
386
The morphology of the obtained crystals is influenced by the nature of the mineralizing agent
(see Fig. 1). In fact, it is well known that the NaF agent leads to the formation of a platelet
crystal morphology, while using the NaOH agent results in crystals having a rounded
morphology. The length-to-width (aspect) ratio of Silicalite-1 crystals increases with the
decrease of the reaction mixture's alkalinity [18-20]. The obtained pellets have an irregular
morphology shape and a size around 0.7 + 1.4 mm (see Fig. 2).
In the procedure (a) of the indirect pellet preparation way, the hydrothermal treatment to
which the pellets are submitted, fails to increase the crystallinity of the final phase. Moreover,
after calcination the presence of the amorphous phase inhibits the efficient template removal
and the formation of coke is observed. The final morphology is obviously the same as the
Silicalite-1 in powder form and the presence of the non zeolitic phase (the binding agent) is
clearly seen (Fig. 3).
From the results shown in Table 1, it is evident that the better cohesion of the pellets is
obtained if they are prepared by direct synthesis from the F media.
The alkaline treatment of Silicalite-1 causes the degradation of the zeolite and therefore the
decrease of its crystallinity (sample n ~ 7 prepared by indirect synthesis procedure (c)).
Moreover, the morphology and the crystal size of the pellet surface is very irregular due to the
zeolite degradation as well as to the presence of the binding agent (Fig. 4).
The results of the N2 adsorption/desorption analysis show that the pellet synthesis procedure
allows to obtain a very high crystal cohesion because the meso- and macropores in the final
product are negligible (Fig. 5). So, the isotherm type is typical to the one of the microporous
materials (type I). The BET surface is 380 mZ/g. From the Dubinin-Radushkevich's method
the evaluated micropore volume is 0.14 cm3/g.
Fig. 1. SEM image of the pellets surface: (a) sample N~ -(b) sample N~
i
ii!ili!
iiii!il
i!i!!i!ii ....ill..
......
~8!~4i~i!!~ ~i~!~?i~i~i~!ii!~ ~i~
i;i!~;iii~ii;;~!i51i~i!!~U(!i!i~i!!!!ii~+..........., i!!iiiiiii!!i!i!
, i ....
Fig. 2. SEM image of the pellets: (a) sample N~ -(b) sample N~
387
Fig. 3. SEM image of: (a) Silicalite-1 powder sample; (b) pellet surface of the sample N~
150-~
~ i00!
E
"o
v
~, ..............................................
t
m
tD a
E
~3
m
0
> 0.0 0.;2 04 0,6 0~8 t~0
Table 2
Enzyme immobilization test results
Mineralizing Immob.Lipase/Support % Immobilized
Support Agent [mg/g] Lipase
Powder Silicalite- 1* NaOH 228 59
Pellets N~ NaF 275 65
Pellets N~ NaF 274 65
Pellets N ~ 3 NaOH 124 32
Pellets N~ NaOH 145 34
* synthesis condition published elsewhere [13]
The main reasons because OH-materials have higher amount of defect groups than the F-
materials can be resumed as follows: the silica materials prepared in the OH- medium, using
SDA cations, usually have a high amount of Q3 defect sites due to the charge balance needed
for SDA and to the formation of very strong SiO...HO-Si hydrogen bonds that may reduce
the instability of the defective silica network; by contrast in the F-silica materials charge
balance is generally achieved with occluded F- and the low synthesis pH favours a more
complete silica condensation. Additionally, in this case the stabilization of Si-OH groups by
389
hydrogen bonding to Si-O- must be minor (because the concentration of SiO is low).
Moreover, Koller and co-workers [16], using solid state NMR spectroscopy, have shown that
as-made high-silica zeolites, prepared in fluoride medium, contain five-coordinate silicon
[SiO4/2F]- units balanced by the cationic structure-directing agent (SDA). Recent studies [22,
24] have shown that even after calcination fluorine remains occluded inside the small [4 ~5262]
cages found in the MFI structure. The high activation energy prevents the migration of the
fluoride ions, which remains trapped in the D4R cage during the calcinations stage.
Louis and co-workers [25] have demonstrated that by varying the concentration of the
fluoride content in the gel it is also possible to tailor the surface acidity. Particularly, in the
pure silica MFI material the number and the strength of Br6nsted sites are reduced with the
increasing of the fluoride concentration in the synthesis gel.
Another important aspect is that the pure silica phases prepared in fluoride media are strictly
hydrophobic due to their SiO2 composition and the very low concentration of the Si-OH
groups [22, 26]. This has been demonstrated in the case of the all-silica BEA materials by
water adsorption measurements. The superior hydrophobicity of this material compared to
BEA materials containing Si-OH groups was also shown by competitive adsorption of
water/toluene and water/methylcyclohexane [27].
All these aspects concerning the F-materials can explain the different behaviour of the pellets
prepared in the fluoride media with respect to those obtained from the OH-media.
Among the pellet samples n ~ 1 - 3, the higher hydrophobic character of the samples 1 and 2
respect to the one of the sample n~ is the main property that allows the adsorption of a higher
quantity of the enzyme. While, for the sample n ~ 6, the presence of a non-zeolitic phase, such
as the binding agent, reduces the accessibility of the Si-OH groups present on the support
surface to the lipase enzyme, with respect to the Silicalite-1 in powder form.
Since the lipase adsorption process on a hydrophobic support involves the lipophillic center of
the enzyme, the best kind of the support can be evaluated only from the results of the catalytic
tests. In fact, the lipase adsorption promotes a dramatic change of the enzyme structure,
leading to an apparently more ordered enzyme structure [21 ]. This adsorbed lipase is present
in an open form with the active centre accessible for the substrates. By the "interfacial
activation mechanism", the immobilized enzyme usually exhibits a significantly enhanced
enzyme activity [21]. If during the immobilization process the interaction between the silanol
groups of the support surface and the external lid of the enzyme is involved, the most
probable enzyme form is the close one, with the inaccessible active centre.
Present results encourage application of Silicalite-1 pellets as lipase support. The amount of
the retained enzyme is very high and the lipase loading capacity of the Silicalite-1 pellets is
however higher than that observed for other powder supports [13]. Of course, in order to
better characterize the pellet surfaces, ulterior analyses have to be carried out.
4. CONCLUSIONS
Silicalite-1 pellets with high crystal cohesion in aqueous systems have been synthesized. The
most suitable procedure that allows obtaining pellets that have this property is the direct
synthesis involving preparation of pellets from solid-state mixtures of a precursor gel. Among
the pellets synthesized by the first way of the above mentioned procedure, those obtained
using NaF as a mineralizing agent show the best crystal cohesion.
The degree of the enzyme immobilization is higher on pellets prepared in F-media and
without a binding agent. Catalytic tests are in progress in order to evaluate the efficiency of
the enzyme/Silicalite-1 pellet system.
390
REFERENCES
The present study focuses on the effect of the metal-modification method (ion-exchange or
impregnation) and of the type of dealumination procedure (steaming, AHFS) of zeolite-Y on
the quantity, quality and properties of the carbon nanotubes that are formed on the supported
iron or cobalt oxides. Severe dealumination of zeolite-Y by steaming, which induced
secondary meso/macropores, resulted in significant decrease in the carbon nanotube
formation activity of the metal-impregnated zeolite-Y catalysts, while moderate
dealumination by ammonium hexafluorosilicate had a less pronounced effect. The micro-
Raman spectra revealed the presence of high quality as-grown CNTs with low degree of
disordered.
1. INTRODUCTION
Carbon nanotubes (CNTs) possess important properties derived from the combination of
unique dimensional, structural and topological features which makes them outstanding
materials for various potential technological applications [1 ]. Because of their wide range of
applications the fabrication of well-defined and organized arrays of carbon nanotubes, in
sufficient quantities and at a low cost, is of high importance. A general chemical pathway for
their production is the catalytic decomposition of hydrocarbon gases (methane, ethane,
acetylene) over transition metal cations (Fe, Co, Ni) and/or the corresponding metal oxides
embedded in solid supports. Solid supports already employed comprise mainly zeolites [2,3],
mesoporous silica [4-6], silica [3,7], alumina [8-10], graphite [11] and clay minerals [12].
Amongst zeolitic supports, the zeolite type-Y (faujasite) has attracted most attention. It
was shown [2,3,7] that multi-wall carbon nanotubes (MWNTs) with external diarPeters in the
range of 20-30 nm can be formed in relatively large amounts (up to 50 wt.% based on parent
zeolite weight) by catalytic decomposition of acetylene on Fe and Co oxides supported on
zeolite Y. The total carbon (amorphous carbon, carbon nanofibers and nanotubes) formation
activity of Fe/zeolite Y was higher than that of Co/zeolite Y; however, the quality (degree of
graphitization) of the nanotubes produced on the latter catalyst was markedly better. In all
cases, the carbon nanotubes were formed on the outer surface of the zeolitic particles, since
the formation of very small single layer nanotubes that could penetrate the pore openings of
zeolite Y (diameter of 7.4 A) was energetically unfavorable. In addition it was suggested that
392
possible coke formation at the initial steps of the acetylene conversion reaction resulted in
blockage of the pores that restricted further transport of the reactants to the active metal sites
inside the zeolitic micropores [2].
The crystalline structure and microporosity of zeolite Y can be modified through post-
synthesis dealumination procedures, such as steaming at elevated temperatures [13,14],
treatment with SIC14 vapor at moderate temperatures [14,15], treatment with aqueous
solutions of ammonium hexafluorosilicate (AHFS) [14,16,17], and others. Controlled
dealumination of the zeolite framework through parallel destruction and reconstruction steps
can lead to the formation of secondary pores in the range of mesopores (mean diameter 2 - 50
nm) as well as macropores (m.d. > 50 nm) [13,14,18]. Such changes in the textural properties
and the morphology of zeolite particles are expected to have a significant effect on the
formation of carbon nanotubes when the dealuminated zeolites are being used as supports of
the active transition metals. This effect could be related with a more controlled formation of
nanotubes within the secondary mesopores and with a potential improvement in the yield of
carbon nanotubes due to the higher external area of the dealuminated zeolites. This work
presents a systematic study of the effect of the textural characteristics of dealuminated zeolite
Y on the quantity and quality of the carbon nanotubes, which are formed on Fe and Co metal
particles supported on parent and dealuminated samples by ion-exchange or wet impregnation
procedures.
2. EXPERIMENTAL SECTION
hr. The mixtures were then transferred to a rotary evaporator apparatus where they were
further stirred at 70 ~ for 1 hr, followed by evaporation of the solvent at 80~ under vacuum.
The iron and cobalt content of the impregnated zeolite Y samples was approximately 2.5 wt.
%, similar to the exchanged samples. The Fe- or Co-impregnated samples were designated
according to the metal and the zeolite sample (parent or dealuminated) that was used as
support, i.e. FeNaY(imp) or FeUSY(imp). All the Fe- and Co-modified zeolite Y samples
were calcined at 450~ for 4.5 hrs in air prior to the catalytic decomposition of acetylene.
.,..,
FeNaY(ex)
9 u , i ,
t
10 20 30 40'0 20
20/degrees 20/degrees 30 40
Fig. 1. XRD spectra (A) of the parent NaY zeolite and the dealuminated samples, and (B) of the
metal-modified zeolites
275- 350
NaY 36 A 0.6
"-7
250- ~0
3O0
225- / 155 A ~ J
9 ./\ #
9. ,,/ \ /p 250
200 0.2 9
.... ' , . . . . . . . . . . , ~ ,//
...= "T
lo .. ~oo o ~6o9/-/
0 pore dlameter/A , / * /"
> 175
o/ o/
z"
200
150
150
125
lO lOO lOOO
pore diameter/,~ USY
100 , . . . . . . . . . 100
0.0 0:2 014 0:6 0:8 0.0 0.'2 0.4 0:6 0:8 1.0
P/P o
p/p o
Fig. 2. N2 adsorption/desorption isotherms and pore size distribution (insets) of the parent NaY and
steam-dealuminated USY sample
395
Table 1
Structural and porosity characteristics of parent NaY and dealuminated zeolite Y samples
Samples Relative Dealumination Porosity
crystallinity (l) degree (2) _ characteristics(, 3)
SSA Micropore Meso/macropore
(%) (%) (m2.gl) volume (cc.g-1) volume (cc.g-1)
NaY 100 - 908 0.328 0.067
NH4Y 98 - 925 0.336 0.066
USY 63 81 668 0.216 0.266
USY(st) 53 85 565 0.187 0.268
NH4Y- 74 54 684 0.235 0.145
~) By comparing the intensity of the (533) XRD peak of the dealuminated samples with that of the
parent zeolite Y sample. ~2)By comparing the number of framework aluminum atoms (FAI) of the unit
cell of dealuminated zeolites with that of the parent zeolite Y; FAI atoms were determined by the unit
cell size (XRD) and the relationship in ref. 19. (3) From N2 adsorption/desorption experiments; SSA-
Specific Surface Area by multi point BET method.
generated, while treatment with AHFS resulted in the formation of pores with mean diameters
o f - 4 . 5 nm. Except of the different dealumination effectiveness of the two dealumination
methods, steaming generates significant amounts of extra-framework A1 (EFA1) or Si-A1
species which can block the entrance to the remaining micropores, while treatment with
AHFS produces dealuminated samples free of EFA1 and with relatively low amount of
amorphous phases [ 14,17].
Modification (ion-exchange and wet-impregnation) of the parent NaY and dealuminated
samples with iron and cobalt had almost no additional effect on their structural properties, as
it can be seen in the XRD patterns of Fig. lB. Furthermore, careful calcination of the metal-
modified zeolites at 450~ (for 4.5 hrs) and at 700~ (for 15 min) had a small effect on their
structural and porosity characteristics, that is a small reduction in crystallinity and
microporosity (less than 20%), mainly in the samples derived from the parent NaY zeolite.
Table 2
Carbon yield (TGA) and quality of produced carbon nanotubes (Raman spectroscopy)
Sample Raman Ratio Carbon Yield %(1~
. . . . (1D/IG)
FeNaY(imp)-CNT 0.54 73.0
CoNaY(imp)-CNT 0.62 46.2
CoNaY(ex)-CNT 0.58 35.7
FeNaY(ex)-CNT 0.56 30.9
FeUSY(imp)-CNT 0.50 34.4
FeUSY-st(imp)-CNT 0.62 20.8
F eNH4 Y-AHF S( imp)-CNT 0.72 45.8
~ (weight loss/100-weight lossi'100, weight loss measured in the range of 400-700 ~ in TGA curves
Between the FeNaY(imp) and CoNaY(imp) impregnated zeolitic catalysts, the former was
significantly more active, reaching a carbon yield of-70%.
With regard to the dealuminated zeolitic supports and based on their structural and
porosity characteristics shown in Table 1, it can be seen that partial destruction of the
crystalline microporous structure and generation of zeolitic particles with secondary
meso/macropores (in the range of 3.6 nm and 16 nm) results in lower carbon formation
activity (Table 2). The more severe is the steam dealumination, the larger becomes the ratio
between meso/macropore over micropore volume and the higher is the decrease in carbon
yield. On the other hand, dealumination with AHFS which generates less secondary
mesopores in the range of 4.5 nm results in lower carbon yield decrease compared to the
parent Fe-impregnated NaY zeolite. In a previous related study [21], the effect of
dealumination on the diameter of carbon nanotubes had been recognized, the diameter
becoming larger in the presence of large secondary pores generated by steaming.
Effect of the support and the metal-modification method on the quality of CNTs
A very powerful method for the characterisation of the quality of the nanotubes is Raman
spectroscopy, as it was also shown for both SWNTs and MWNTs from previous studies [22-
26]. Fig. 4 shows micro-Raman spectra of as-grown CNTs on the zeolitic supports of this
study. In the first-order Raman scattering region, two intense peaks are observed at 1340-
1350 and 1570-1610 cm l, which are referred to as D- and G-bands, respectively, and are
characteristic of sp2-hybridized carbon material. The G-band corresponds to the tangential
stretching high-frequency EZg mode of highly oriented pyrolytic graphite (HOPG) and
indicates the presence of crystalline graphitic carbon in the CNTs, while the D-band
originates from lattice disorder or defects and by finite size effects of the crystalline graphite
sheets. Furthermore, pyrolytic carbon particles deposited on nanotubes also contribute to the
rise of D-band [24,25]. The relative intensity of D and G bands (ID/IG) can be used as measure
of the degree of graphitization and is characteristic of the quality of the nanotubes. The IIJlG
values of the carbon nanotubes prepared in this study (Table 2) are between 0.50 and 0.72,
which are considerably lower than those reported in the literature (IJIG = 0.7-1.3) for CVD-
grown MWNTs [24-26], revealing the good quality of the produced CNTs using all the Fe-
and Co-modified zeolite-Y samples. No systematic significant variations in the ID/I~ values
could be observed between the different zeolitic supports (parent and dealuminated) or the
different metal (Fe or Co) or the different method of modification (ion-exchange or
impregnation).
397
100 410 ~ 5 0 ~
90 N ',
I,
80 II ;.
9~ 70
60 FeNaY(imp)-CN
FeNaY(ex)-CNT ~
. . . . . . FeUSY(imp)-CNT k~
50
0 2C)0 4()0 6C)0 860 ~ooo
Tern perature/~
Fig. 3. TGA curves of representative zeolite-CNT composites
'
(D
CoNaY(imp)-CNT
u ! r
2200 2000 1800 1600 1400 1200 1000
Raman Shift/cm ~
A lower Raman-shift of the G-band could also be indicative of the good quality of the
nanotubes. The value of 1575 cm -~ for the CNTs formed on Fe-impregnated parent NaY
zeolite (Fig.4) can be attributed to highly ordered graphite sheets. Higher values, such as
those found for some of the samples of this study (i.e. 1595-1600 cm -1) could indicate
increased disorder in the carbon material, although high-quality SWNTs have been reported
where the G-band was composed of several components ranging from 1552 to 1600 cm-" [24].
4. CONCLUSIONS
The carbon nanotube formation activity of Fe- and Co-modified zeolite-Y samples was shown
to be dependent on the structural and porosity characteristics of the zeolite supports, on the
398
REFERENCES
[ 1] P.M. Ajayan, Chem. Rev., 99 (1999) 1787.
[2] K. Hernadi, A. Fonseca, J.B. Nagy, D. Bernaerts, A. Fudala, A.A. Lucas, Zeolites, 17 (1996)
416.
[3] A. Fonseca, K. Hernadi, J.B. Nagy, D. Bernaerts, A. Lucas, J.Mol. Catal. A, 107 (1996) 159.
[4] W.Z. Li, S.S. Xie, L.X. Qian, B.H. Chang, B.S. Zou, W.Y. Zhou, R.A. Zhao, G. Wang, Science,
274 (1996) 1701.
[5] S. Lim, D. Ciuparu, C. Pak, F. Dobek, Y. Chen, D. Harding, L. Pfefferle, G. Hailer, J. Phys.
Chem. B, 107 (2003) 11048.
[6] N. He, Y. Kuang, Q. Dai, Y. Miao, A. Zhang, X. Wang, K. Song, Z. Lu, C. Yuan, Mater. Sci.
Eng., C8-9 (1999) 151.
[7] K. Hernadi A. Fonseca, J.B. Nagy, D. Bernaerts and A.A. Lucas, Carbon, 34 (1996) 1249.
[8] A.K.M.F. Kibria, Y.H. Mo, K.S. Nahm, Catal. Lett., 71 (2001) 229.
[9] N.S. Kim, Y.T. Lee, J.H. Park, H. Ryu, H.J. Lee, S.Y. Choi, J.B. Choo, J. Phys. Chem. B, 106
(2002) 9286.
[10] Z. Konya, I. Vesselenyi, K. Lazar, J. Kiss, I. Kiricsi, IEEE Trans. Nanotechnology, 3 (2004) 73.
[11] V. Ivanov, A. Fonseca, J.B. Nagy, A. Lucas, P. Lambin, D. Bernaerts, X.B. Zhang, Carbon, 33
(1995) 1727.
[12] D. Gournis, M.A. Karakassides, Y. Bakas, N. Boukos, D. Petridis, Carbon, 40 (2002) 2641.
[13] S. Van Donk, A.H. Janssen, J.H. Bitter, K.P. de Jong, Catal. Rev., 45 (2003) 297.
[ 14] C.S. Triantafillidis, A.G. Vlessidis, N.P. Evmiridis, Ind. Eng. Chem. Res, 39 (2000) 307.
[15] L. Kubelkova, V. Seidl, J. Novakova, S. Bednarova, P. Jiru, J.C.S., Faraday Trans. 1, 80 (1984)
1367.
[ 16] M. Neuber, V. Dondur, H.G. Karge, L. Pacheco, S. Ernst, J. Weitkamp, Stud. Surf. Sci. Catal.,
37 (1988) 461.
[17] J.M. Cruz, A. Corma, V. Fomes, Appl. Catal., 50 (1989) 287.
[18] C. Triantafillidis, A. Vlessidis, L. Nalbandian, N. Evmiridis, Microporous Mesoporous Mater.,
47 (2001) 369.
[ 19] L. Kubelkova, V. Seidl, G. Borbely, H.K. Beyer, J. C. S.. Faraday Trans. I, 84 (1988) 1447.
[20] Y. Li, X.B. Zhang, X.Y. Tao, J.M. Xu, W.Z. Huang, J.H. Luo, Z.Q. Luo, T.Li, F. Liu, Y. Bao,
H.J. Geise, Carbon, 43 (2005) 295.
[21] A. Zhang, C. Li, S. Bao, Q. Xu, Microporous Mesoporous Mater., 29 (1999) 383.
[22] M.S. Dresselhaus, P.C. Eklund, Adv. Phys., 49 (2000) 705.
[23] P.C. Eklund, J.M. Holden, R.A. Jishi, Carbon, 33 (1995) 959.
[24] M. Sveningsson, R.-E. Morjan, O.A. Nerushev, Y. Sato, J. B~ckstr6m, E.E.B. Campbell, F.
Rohmund, Appl. Phys. A, 73 (2001) 409.
[25] K. Kwok, W.K.S. Chiu, Carbon, 43 (2005)437.
[26] W. Li, H. Zhang, C. Wang, Y. Zhang, L. Xu, K. Zhu, Appl. Phys. Lett., 70 (1997) 2684.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 399
Laboratoire de chimie des mat6riaux et catalyse. D6partement de chimie. Facult6 des Sciences
de Tunis. Campus universitaire 2092 El Manar. Tunis, Tunisie.
All3 polycation exchanged forms of montmorillonite have been prepared in the presence of
(C12H25O(OCHzCH2)4H). The influence of the (surfactant/Al13polycation) molar ratio on the
properties of the solid obtained has been studied. The clay samples were characterised by
XRD, BET surface area and chemical analysis. The best samples were obtained with an
surfactant/Al~3 molar ratio = 0.045. The study of the acidity of surface groups showed that the
surfactant-Al~3 polycation is transformed into a specific form of A1 oxyhydroxide with
amphoteric properties, having two constants of deprotonation Kal = 3.16.10"6and Ka2-1.58.10-7.
Copper strongly bonded to alumina pillars characterized by ESR is suggested as the active site
in the catalytic wet peroxide oxydation of phenolic molecules at 298K and atmospheric
pressure. A model for the surface reaction was proposed.
1. INTRODUCTION
Clays can be intercalated by metal polycations through ion exchange reactions, yielding
materials with high microporosity and surface area [ 1,2]. The most widely used type of cation
is the Keggin-like All3 oligomer, [A11304 (OH)z4+x (HzO)lz-x](7-x)+ [3,4]. The possibility of
cointercalating organic species together with the pillaring All3 ions has been proposed using
cationic and nonionic surfactants [5,6]. Nonionic alkyl polyether surfactants of the type
RO(CHzCHzO)xH intercalated into All3 montmorillonite give materials with interesting
structure properties [7,8].
The alkyl polyether surfactants contain hydrophobic and hydrophilic chain segments that
can be presented in abbreviated form as CnE• where n is the number of carbon atoms in the
alkyl chain and x is the number of ethylene oxide units. The CnEx-All3 polycation can
dramatically alter the fundamental surface chemistry of the intercalate. The cointercalation of
CI2-14E5 has been shown to greatly improve both the hydrolytic stability of Al13 ions in the
gallery and the layer stacking order [9]. In addition, the alumina pillared derivatives formed by
calcining the intercalates exhibit micropore size distributions that are much more regular than
those observed for pillared derivatives prepared in the absence of surfactant [6].
Potential application of pillared interlayer clays (PILC) in catalytic processes of a redox
nature would require the PILC structure to accommodate transition metal ions known to easily
change their oxidation states. In a series of recent papers pillared clays were used as a support
for the catalytically active copper species [10-12]. Many interesting results have been obtained
for processes of significant practical importance such as the oxidation of organic pollutants. It
seemed of interest to check the ability of copper doped surfactant-modified pillared clays (Cu-
400
S-PILC) to oxidize organic substrates in the liquid phase using hydrogen peroxide.
The present contribution deals with the advantages of a new preparation procedure of Cu-
AI-PILC by cointercalation of Al~3 polycation and C~2E4surfactant in montmorillonite clay. The
influence of the (surfactant/polycation) molar ratio on the structural organisation and surface
area of the solid obtained has been studied. We have characterised the surface hydroxyl groups
in the presence of an aqueous phase, which constitute specific copper ions adsorption sites. We
have chosen degradation of nitrophenol as test reaction. Thus mechanistic model was proposed
according to the nature of eventual active sites.
2. EXPERIMENTAL
2.3. Intercalation
5 g of the clay to be intercalated are dispersed into 1L distilled water under strong
magnetic stirring for 2 h. The desired amount of A1-C~2E4-containing intercalating solution is
then added dropwise to the stirred clay suspension at ambient temperature. Stirring is
maintained for 4 h after completion of A1-CI2E4 addition. The mixture is filtered, washed
several times by redispersion in distilled water, and dried in a shallow bed at 333 K. Five
intercalated samples were prepared, using Al~3 polycation and the four A1-CI2E4 solutions.
2.4. Pillaring
Part of the sample is placed as a shallow bed in a tubular reactor and calcined under
flowing dry air with a linear temperature ramp (2 K rain-t), followed by a 5 h plateau at 773 K.
The resulting materials are called pillared samples.
to CuI-S• (x:1,2,3 and 4). In order to increase the copper content adsorbed, the Cul-Sx-
PILC sample was once more subjected to the exchange / drying / calcination procedure to
obtain sample Cu2-Sx-PILC and when it is subjected to the same operations one more time it
yields sample Cu3-S•
Tablel
Labels of the intercalated and pillared clays
A Initial montmorillonite
AII Intercalated clay with Al~3 polycation
S• Intercalated clay with S• solution (x = 1,2,3,4)
AI-PILC Pillared sample with Al~3 polycation
S• Pillared sample with Sx solution (x = 1,2,3,4)
where CA is the concentration in strong acid added at a given point of the titration, CB is the
concentration in strong base (NaOH) added before titration, [OH-] and [H +] are the
concentrations in free hydroxide and hydronium as deduced from pH readings and a is the
concentration of the solid phase in the suspension (in g/L). Q ma~r be transformed to the
physically more meaningful expression of the surface charge a in C/m by 9 a = Q. F. St,
where F is the Faraday and S is the surface area in m2/g.
In the discussion, square brackets [] wilI denote concentrations in mol/1, while curly brackets
{} will denote concentrations in mole per gram of solid phase.
402
Table 2
Effect of surfactant on the structural organisation of intercalated clays
Table 3
Effect of the molar ratio(nCl2E4/n Al~3) on the properties of PILC
Table 3 shows the d(001) basal spacings, BET surface area, micropore volume and the
content of aluminium in the investigated pillared clays. The use of Cl2E4 surfactant does not
403
affect the character of the XRD pattern of pillared modified clays, but causes a changes in
aluminum content, surface area and pore structure. Comparison of the data obtained for the
SxPILC series with those of AI-PILC shows that a decrease of aluminum content and an
increase of the BET surface area were found for both 0.045 and 0.09 molar ratio value.With
increasing molar ratio above 0.09, the BET surface area and the micropore volume decrease
but the aluminum content increases. On the basis of the above results, the optimal conditions
for cointercalation are obtained for low surfactant loadings. In this study we choose a molar
ratio nC~zE4/n AI~3- 0.045. thus, S~-PILC was selected for the study of surface acidity.
The proton titration curve and the Q -f(pH) isotherm of S1-PILC indicate that the pillared
clay has amphoteric properties since Q is positive at low pH values and changes to negative at
higher pHs. In the "1 site, 2 pK model", the hypothesis is made that all surface sites are
equivalent, and amphoteric in the sense that they can behave as acid sites and as basic sites;
thus, they may exist in three different forms, protonated, neutral and deprotonated, denoted
SOH2 +, SOH and SO-, respectively. The two successive reactions involving these forms may
be written as:
SOH + H + ~ SOH2+ Ka, = ([IF] {SOH}) / {SOH2+}
SOH ~ ~ SO + H + Ka2 = ([H +] {SO-})/{SOH}
with two successive deprotonation constants Kal and Ka2. In this model, the surface charge Q
is Q = {SOH2+}- {SO-} = ([SOH2+] - [SO-])/a
As explained by Stumm [18], the intrinsic values, pKa~ int and pKa2 ~"' may be estimated by
plotting the apparent values as a function of Q, and extrapolating to zero surface charge.
Application of this procedure to S1-PILC provided the following values:
pKal i n ' = 5.5 and pKa2 int= 6.8
The pH of zero net protonic charge, pHpzc, is that for which {SOH2+}= {SO} and therefore Q
rl int~
= 0. It may be checked that pHpzc = 89 (pKat int + pt~a: ), which yields pHpzc = 6.15.
This is of primordial importance for adsorption of ionic species, since pHpzc determines the
border between pH regions where cationic adsorption and anionic adsorption are possible,
respectively. The acido-basic properties of SI-PILC are quite similar to those of an Al-pillared
saponite[ 19].
Table 4
Physicochemical characterisation of the Cu catalysts
Table 4 shows the d(001) basal spacings, BET surface area, micropore volume and the
content of copper in the investigated catalysts. Comparison of the data obtained for the Cu-
doped series with those of S1-PILC shows that an increase in copper content is accompanied
by a gradual decrease of BET surface area and of micropore volume. Since the experimental
procedure used for insertion of copper, that is, cation exchange followed by thorough washing,
prevents retention of copper ions at the outer surfaces of the clay crystallites, the data of
Table 4 may be taken as an indication that copper introduced into the pillared matrix occupies
404
the interlayer area. A part of copper species could remain on the external surface of the
cristallites even after washing. Dehydration/dehydroxylation processes accompanying the
calcination of copper-doped pillared clays leave essentially two possibilities for anchoring the
Cu(II) species. 1) Cu coordinated to the oxygens of the silicate layer, or 2) Cu coordinated to a
pillar. To gain some insight into the location and nature of the copper centers, ESR
investigations, have been undertaken, the ESR spectra of Cu-doped catalysts recorded at 77K
are anisotropic, with clearly defined g• and g//regions (Table 5), as expected for a Cu 2+ in an
axial symmetry. Similar results were reported by Bahranowski et al [ 12] and showed that the
signal can be deconvoluted into two components, one with a resolved hyperfine structure,
associated with isolated Cu 2. ions, the other, broad and unresolved, due to the clustered copper
(II) ions. The calcined Cu-exchanged non-pillared montmorillonite, in which Cu centers are
attached to silicate layers only, has been used as a reference. The ESR signal produced by this
sample (g// = 2.39, g• = 2.08, A / / = 130 G) differs from the resolved spectrum present in the
pillared montmorillonite (Table 5). This leads to a conclusion that the isolated copper ions are
principally anchored at the pillars rather than at the surface of a silicate layer. The lineshape
and parameters of the broad, unresolved signal (g// = 2.31, g• = 2.14) fall into the range
observed for clustered copper ions experiencing exchange interactions [20]. Indeed, the value
of G = g//-(2/g• calculated for this signal is less than 4 which is an indicator of significant
exchange interactions [21]. This result shows that besides the isolated cupric ions giving
resolved hyperfine patterns and associated with the pillars, part of the copper forms clusters
responsible for the broad, structureless component of the ESR spectrum. When the Cu content
in the samples increases, the contribution from clustered species increases as well. Since
preparation conditions of the catalysts involve calcination in air at 673K, the clustered species
are believed to exist as amorphous CuO patches or as CuO in a form of nano-crystallites.
Table 5
g//, g• and A//values for the different samples
All samples are tested as catalysts for degradation of 2-nitrophenol. A preliminary test of
external diffusion control was performed by measuring the reaction rate over different masses
of catalyst with appropriate chosen stirring speed to maintain a constant contact time.
1
0,8
9 Cu3-S1-PILC
0,6
+ Cul-S 1-PILC
a.
Z 0,4 A Cu2-S1-PILC
0,2
0
0 4 8 12
Time/h
Fig. 1. Decomposition of 2-nitrophenol (10-3M) by catalysts (1 g/l), [H202]=2.10-2M, T re 298K, pH=5.
405
The results show that, for a catalyst mass in excess of about 1 g/1 (which also corresponds
to mixing speed of 400 rpm) there are no problems associated with external diffusion
limitations.
Fig. 1 shows the concentration evolution of the 2-nitrophenol solution (103M) in the
presence of the solid catalyst (lg/L) and H202 (2.10-2M). Degradation of 2-nitrophenol is
observed for all copper doped surfactant-modified pillared clays. This result is confirmed by
the titration of nitrate, nitrite and CO2 produced by the reaction and by the disappearance of
phenolic and aromatic protons on the ~H NMR spectra at the end of each reaction. The rapid
initial degradation is seen to be followed by a slower process probably due to the difficult
conversion of N-atom close to the benzene ring into NO3.
When obtained results are compared we conclude that Cuz-S1-PILC is the best catalyst for
the degradation of 2-nitrophenol. Contrary to our expectation, further increase in the Cu
content does not improve the catalytic performance of clay. Instead, a fall in catalytic
activity is observed. On the basis of the physicochemical characterization data, we conclude
that an increase of copper content increases the contribution of clustered cupric ions, which
are catalytically inactive. Clusters of amorphous CuO may also cause a local blocking of the
micropores, thus cutting off some of the isolated active copper species fixed on the pillar.
As described in previous study [22] when Cu-oxides were used as catalysts, the degradation
of nitrophenols is induced by the reduction of adsorbed oxygen giving firstly H202, then OH"
or O2H" The degradation takes place as follow:
C6HsNO3(ads) + HO2" or OH " -~ intermediates ~ C02 + HzO + NO3 + NO2-+ ...
In our case, the first step could be the formation of Cu + in the solid matrix, by interaction with
2-nitrophenol and hydrogen peroxide or only with 2-nitrophenol. Cu+reacts then with H202"
( SI-PILC)-Cu + + H202 ,~ ( SI-PILC)-Cu 2+ + O H ' + OH-
The second step could be the formation of a complex between Cu 2+ fixed on the pillar and 2-
nitrophenol. The mechanism will be as described by J. Kiwi et al. [23] 9
Cu 2+ + 2-nitrophenol .~ [Cu 2§ 2-nitrophenol] ~ e [Cu+--2-nitrophenol] +
[Cu+--2-nitrophenol] + + H202 ~ active species
Active species + 2-nitrophenol ~~ degradation.
To verify the importance of the complexation step in the general process of the
degradation, catalytic tests are realised with nitrobenzene, phenol, 3-nitrophenol and 2-
nitrophenol in the same conditions (Fig. 2). Obtained results show, a slow and limited
degradation reaction for phenol and 3-nitrophenol. For nitrobenzene molecule a simple
adsorption on catalyst occurs. It is obvious that the position o f - N O 2 (ortho) on phenol
406
molecule has the key role of the formation of a stable complex when degrading 2-nitrophenol.
In the case of phenol and 3-nitrophenol an unstable complex [Cu 2+-- OH-C6Hs] and [Cu 2+--
OH-C6H4_ NO2] should be respectively formed.
CONCLUSION
Surfactant-modified pillared clays may be used successfully as Cu support in the catalytic wet
peroxide oxidation of phenolic molecules under mild conditions ( T re= 298 K and atmospheric
pressure). The use of CizHzsO(OCHzCH2)4H at molar ratio nC12E4/nAlI3 = 0.045 or 0.009
improve the BET surface area and the porosity of pillared clay. The study of the acido-basic
properties of Al-surfactant-PILC surface groups for the lowest surfactant loading indicates
well-defined amphoteric properties, with a pHpzc = 6.15. copper (II) ion adsorption should be
possible in a large range of pH on these catalytic supports.
Physicochemical characterisation of copper doped surfactant-modified pillared clays shows
that the catalysts are porous materials with copper species located in the interlayer, present
either as isolated Cu 2+ ions anchored at alumina pillars or as CuO in amorphous or nano-
crystallites form. Catalytic tests show that phenolic molecule seems to be active in the
initiation of the mineralisation process by complexation step.
REFERENCES
A route to synthesize ZSM-5 and TS-1 crystals with a bimodal micro/mesoscopic pore system
has been developed in this study; the successful incorporation of the mesopores within the
ZSM-5 structure was performed using tetrapropylammonium hydroxide(TPAOH)-
impregnated mesoporous materials containing polystyrene-divinylbenzene nanotubes in the
pores, which were encapsulated in the ZSM-5 crystals during a solid rearrangement process
within the framework. Such mesoporous ZSM-5 zeolites can be readily obtained as powders,
thin films, or monoliths.
1. INTRODUCTION
Microporous materials with pore diameters less than 1.5 nm have received much interest due
in large part to their material characteristics such as large internal surface area, unique pore
structure and adsorption capacity and controllability [ 1-5]. However, they have one limitation
as catalysts, regarding their pore sizes that are not efficient to process large molecules in a
nanometer level. Consequently, recent research efforts have been paid to the synthesis of
nanoporous or mesoporous materials with large pore sizes and uniform pore distribution [6-
12]. Obviously, significant advances of the physicochemical properties of these materials can
be expected when the crystalline zeolites are incorporated into the mesoporous frameworks.
Much recent research effort has been paid to the synthesis of the micro- and mesoporous
composites by various procedures [13-19]. The specimens prepared by those researchers
exhibited the micro- and mesoporosity with the improved hydrothermal stability.
Recently, Jacobson et al. have synthesized mesoporous ZSM-5 crystals by restricting the
crystallization of the zeolites to the mesopores resulting from the packing of the nanosized
carbon particles [20]. It was possible for the zeolites to grow around the carbon particles of
the inert matrix within the whole pore system. The mesopores can be created by combustion
of carbon particles. Although this leads to mesoporpus single zeolite crystals, the resulting
pore size distribution was broad. Furthermore, only few useful carbon black materials are
available and careful control of the crystallization conditions is necessary to avoid formation
of nonmesoporous zeolite crystals.
408
Many types of mesoporous carbons have been produced using inorganic templates such as
zeolites, silica sols, SBA-15, MCM-48 and solid core/mesoporous shell (SCMS) silica
spheres. The structure of nanoporous carbons is an inverse replica of the silica templates after
dissolution of them using HF or NaOH solution.
We have synthesized the mesoporous ZSM-5 zeolites by the crystallization of silica
framework of TPAOH-impregnated mesoporous precursors (MCM-41 and SBA-15),
containing nanostructured carbons in the pore channels [21]. Subsequently, small zeolite
crystals grow into the large single crystals, encapsulating the carbon fibers present in the
pores of mesoporous templates. In that case, the alumina sources must be implanted onto the
silica templates for the polymerization of phenol and formaldehyde (or furfuryl alcohol ).
As a result, this method is not helpful for the synthesis of alumina free zeolites such as TS-1.
Herein we report the simple method to synthesize silicalite and TS-1 zeolite with
micro/mesoscopic pore structures using SBA-15 and Ti-SBA-15 as templates. In the
synthesis, the porous walls of siliceous SBA-15 and Ti-containing one were transformed into
the mesoporous ZSM-5 type crystals in the presence of in-situ crosslinked
styrene/divinylbenzene in the mesopores of templates.
2. EXPERIMENTAL
hydroperoxide was performed using a batch-type reactor at room temperature. The conversion
was determined by GC using a capillary column (HP-5 crosslinked 5% PHME Siloxane
30m•215 Film Thickness).
Fig. I. Schematic representation for the transformation ofmesoporous materials into mesoporous
ZSM-5 after formation of crosslinked polymers in the pore channels. Produced mesoporous ZSM-5
has bimodal micro/mesoscopic pores and it is readily obtained in the type of thin film, monolith as
well as powder.
The successful incorporation of the mesopores within ZSM-5 type crystals could be
performed by the synthesis method as shown schematically in Fig. 1.
The polymer chains present in the pores of mesoporous materials could remain firm during
the high temperature crystallization of silicalite and TS-1.
Fig. 2(A-C) displays the powder XRD patterns of calcined mesoporous silicalites and TS-
1. In the wide angle region, a series of peaks up to near 20 = 24 o becomes sharper and
prominent with the crystallization time. The appearance of XRD peaks for microporous
ZSM-5 as well as mesophase suggests that the amorphous silica/titania walls of mesoporous
template are progressively transformed to crystalline phase through the encapsulation of nano
polymers in the crystals.
Transmission electron microscopy (TEM) images reveal the appreciable mesoscopic order
of samples. TEM micrographs of silicalite and TS-1 crystallized at 170 ~ (Fig. 3B, 3D)
display the presence of mesopores penetrating the entire crystals. This mesopore system was
resulted from the packing of nanosized ZSM-5 crystals. The mesoporosity is not observed on
the conventional microporous TS-1 crystal as shown in the same Figure 3C.
Syntheses of related materials with bimodal micro/mesoscopic pores in the form of thin
film have been demonstrated. The Ti containing SBA-15 sol was also prepared using
amphiphilic triblock copolymer and it was coated on the porous silica surfaces.
Consequently, micro/mesoporous TS-1 film could be formed and the mesoporosities
penetrating the whole zeolite crystals are observed on the TEM image of samples.
In addition, mesoporous ZSM-5 monolith can be readily produced using Ti-SBA-15 sol
and polyurethane sponge having connected macropore system as templates.
410
al
Fig. 2. XRD patterns of (A) mesoporous silicalite powder prepared using Ti-containing SBA-15 (B)
from silica SBA-15, and (C) from MCM-41.
A / ,
, . .
Fig. 3. TEM images ofthe calcined materials: (A) SBA-15 used as a starting material, (B)
mesoporous silicalite powders obtained from siliceous SBA-15, (C) microporous TS-1, and (D)
mesoporous TS-1 crystals.
411
Fig. 4. SEM photographs of (A) mesoporous silicalite powders obtained from siliceous SBA-15, (B)
enlargement of photograph (A), (C) microporous ZSM-5 powders, and (D) mesoporous TS-1 crystals.
SEM images are showing the top surface of mesoporous silicalite crystal films deposited on the matrix
(E) and that of pure microporous ZSM-5 film(F).
412
SEM photographs of mesoporous silicalite and TS-1 show that each crystal exhibits the
typical shape of ZSM-5 zeolite, and then it consists of very small aggregates (Fig. 4A, 4B,
4D). TS-1 in the coated layer showed the same morphologies as the mesostructured TS-1
powders as shown in Fig. 4E.
To analyze the pore geometry and structure, Brunauer-Emmett-Teller (BET) analysis was
performed to measure the N2 adsorption-desorption isotherm. Figure 5 displays N2
adsorption-desorption isotherms of calcined mesoporous silicalite and TS-1 powder,
respectively. N2 adsorption-desorption isotherms of calcined mesoporous ZSM-5 powder
gave type IV adsorption isotherm behavior and the steep rises at low relative pressures and at
relative pressures between 0 . 8 - 0.9, indicating the presence of both micropores and
mesopores.
The calcined mesoporous silicalite powder obtained using SBA-l 5 as a template had a N2
BET total surface of 4 5 0 - 530 mZ/g and a pore volume of 0 . 6 - 1.1 cm3/g. The mesopore
surface areas are in the range of 1 9 0 - 220 m2/g. The meso TS-I samples have very high
mesopore surface areas of in the range of 170- 200 m2/g and mesopore volumes between 0.4
- 0.8 cm3/g. The size of mesopores in ZSM-5 powder obtained from various SBA-15
materials was estimated to be approximately 13 nm (Fig. 5B).
The catalytic epoxidation of bulky organic olefin molecule was investigated to characterize
the catalytic performance of prepared mesoporous TS-1 samples.
m '6oo
97, 1.2
450
300 0.8
0 0.0
.~ 0.0 012 0~4 0~6 018 1.0 160 200 300 400
<
Relative Presure (P/P) Pore Diameterq (~,)
~" 600
2.0
--- 450 "7
300
@
/7 N 1.o
> 150,
nm,urn munvm,u n . ~ e ~ ~,,
0 9 9 9 ~ 0.0
0.0 012 014 016 018 1.0 0 200 400 600 800
Relative Presure (P/Po) Pore Diameter ( A )
Fig. 5. Nitrogen adsorption-desorption isotherms for the calcined mesoporous silicalite powder (A, B),
and TS-1 (C, D).
413
For comparison, microporous TS-1 was also prepared and tested as a catalyst. The amount
of titanium in catalysts was almost same. The epoxidation of c~-pinene was performed at room
temperature using a usual batch-type reactor. The mesoporous TS-1 zeolites become much
more active than microporous TS-1 when the bulky olefine is used as a substrate. The
oxidation of c~-pinene to Gt-pineneoxide proceeded 10 times faster over mesoporous TS-1
catalyst than with conventional microporous TS-1. This mesostructured TS-1 catalyst has the
volumes of 0.6cm3/g mesopore and 0.2 cm3/g micropores. Consequently, the difference in the
reaction rates was attributed to the textural mesoporosity in mesoporous TS-1, allowing a
better access of the bulky reagent molecules to the active sites.
4. CONCLUSIONS
In this work, a simple and general procedure has been developed for the syntheses of
mesoporous zeolites with a mean pore size of 5-15 nm. The excellent catalytic properties of a
series of the mesoporous ZSM-5 expand the area for the application of porous materials, and
can be used in a number of commercial processes in the future. Furthermore, such materials
with bimodal micro/mesoscopic pores would be useful for many potential applications, such
as heterogeneous catalysts, supports, sensor and membrane in the type of powder, thin film or
monolith, when the bulky reactant molecules are involved..
REFERENCES
[17] Y.K. Kwon, D.H. Kim, G.-J. Kim, Y.-S. Han and B.-S. Seong, Stud. Surf. Sci. Catal., 146 (2003)
355.
[18] Y.K. Kwon, G.-J. Kim, J.H. Lim, D.H. Kim and B.D. Choi, Stud. Surf. Sci. Catal., 146 (2003)
141.
[19] S.I. Cho, Y.K. Kwon, S.E. Park and G.-J. Kim, Stud. Surf. Sci. Catal., 146 (2003) 137.
[20] C.J.H. Jacobsen, C. Madsen, J. Houzvicka, I. Schmidt, A. Carlsson, J. Am. Chem. Soc., 122
(2000) 7116.
[21] S.I. Cho, S.D. Choi, J-H. Kim and G._J. Kim, Adv. Funct. Mater., 14 (2004) 49.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka, N. Zilkov~ and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 415
Mixed CoNiA1 spinels were prepared by the zeolite precursor route using Co+Ni exchanged
zeolites as starting materials. The spinel phases were obtained by calcination of the exchanged
zeolites at high temperatures. Once prepared the spinel materials show no significant
structural changes upon heating, and no migration of Co 2+ and Ni 2+ from one crystallographic
site to the next one is observed, in strong contrast to spinel materials obtained by conventional
methods. Thus, the spinel crystallites obtained from the zeolite precursors appear to be close
to the thermodynamically favored equilibrium composition and structure immediately after
their formation.
1. I N T R O D U C T I O N
In previous studies, we have shown that transition metal exchanged low silica zeolites are
suitable starting materials for the preparation spinel materials with high surface areas. The
resulting spinels consist of crystallites with sizes in the range of a few nanometers [1]. The
structures of these materials typically contain one transition metal (e.g. Co, Ni or Zn) together
with aluminum as the second metal element.
Spinels usually contain more than one transition metal cation [2,3,4,5,6,7,8]. In
general, oxide spinels have the chemical composition (Al_xBx)tet[AxB2_x]~ with 32 oxygen
atoms per unit cell in an almost perfect cubic close-packed arrangement. The arrangement of
the oxygen atoms allows for the possibility of 64 tetrahedral and 32 octahedral interstices,
which can be occupied by cations of different oxidation states. Thus, 1/3 of the cations
occupy tetrahedrally coordinated sites while 2/3 of the cations occupy octahedrally
coordinated sites. In the normal spinel structure 8 metal atoms (A) occupy tetrahedral sites
and 16 metal atoms (B) occupy octahedral sites. The inversion parameter x can, however,
vary from 0 to 1 depending on the distribution of cations. In the perfectly ordered structure x
= 0 (the normal spinel AB204), and in the inverse spinel B[AB]O4 (x = 1) the octahedral sites
are disordered. For a random distribution of the cations, the inversion parameter has a value
of 0.67. With increasing temperature normal spinels as well as for inverse spinels tend to
disorder.
In the present work, mixed CoNiAl-spinels were obtained by the zeolite precursor
route and their composition and structural properties were investigated. The distribution of the
416
metal cations on different sites within the spinel structure was of particular interest, since this
determines the physicochemical properties of the resulting materials.
2. EXPERIMENTAL
Table 1
Notation for spinel samples from ion exchanged and calcined zeolites
starting material zeolite A Zeolite X
Series I, Series II, Series I, Series lI,
3 x 30 min 5 x 30 min 3 x 30 rain 5 x 30 min
quenched after A-I- 1300-Q A-II- 1300-Q X-I- 1300-Q x-II- 1300-Q
reaching 1300~
annealed at 1300~ A-I- 1300-12 A-II- 1300-12 X-I- 1300-12 X-II- 1300-12
for 12h
2.2. Characterization
The chemical compositions of the exchanged zeolites were determined by energy
dispersive X-ray analysis on an Atomika/Amray 1929 Eco system. The chemical
compositions of the resulting spinels were analyzed in the same way after the amorphous
silica and excess transition metal species were removed by repeated treatment in 1M sodium
hydroxide solution at 90~ (alkaline leaching). X-ray photoelectron spectroscopy (XPS)
measurements were performed with a Kratos HSi spectrometer equipped with a hemispherical
analyzer. The monochromatized Al-K~ X-ray source (E=1486.6 eV) was operated at 15 kV
and 15 mA. For the narrow scans, an analyzer pass energy of 40 eV was applied. The hybrid
mode was used as lens mode. The base pressure in the analysis chamber was 4 x 10-9 Torr. To
account for charging effects, all spectra were referred to C l s at 285.0 eV.
417
The data for the structure refinements were measured on a Stoe STADI P 13-0
diffractometer with Cu-K~ radiation. The intensities were measured between 10 and 120 ~ 2t3
in step scan mode with steps of 0.03 ~ 213 and a fixed divergence slit configuration. For the
structure analysis the program FullProf2000 by Rodriguez-Carvajal [10] was used. The
refinements were performed using space group Fd3m, with a complete occupancy of 8(a) and
16(d) positions by the metal cations. The refined parameters were the lattice parameter, the
scale factor, the profile parameters, the oxygen positional parameter, the site occupancy
factor, and the isotropic temperature parameter. Altogether three isotropic temperature factors
were refined. One for the oxygen atom, one for the cations on the tetrahedral sites, and one for
the cations on the octahedral sites. Due to the strong correlation, the temperature factors and
site occupancy factors were refined alternately. As the X-ray scattering factors of Co 2+ and
Ni 2+ are very similar, it was not possible to determine whether both cations are positioned on
both cation sites. Neutron scattering experiments (not shown here), which allow a definite
differentiation between Co 2+ and Ni 2+, were performed for some of the Co, Ni spinel samples.
The refinement of the neutron data did not give any evidence of a mixed occupancy of the
tetrahedral and octahedral position by Co 2+ or Ni 2+ cations, Co 2+ and A13+ only occupy the
tetrahedral sites, whereas the octahedral sites are always occupied by Ni 2+ and A13+. For that
reason, only the total contribution of two cations on each cation site was determined (i.e.,
Co 2+ and A13+ on tetrahedral sites, Ni 2+ and A13+ on octahedral sites).
3. R E S U L T S AND D I S C U S S I O N
~0.5
m.i.A, m.i.A,
o0. 4 E Co/AI
Co/AI
Ni/AI ~ Ni/AI
0.3 Na/A,
~:: Na/AI
0.2 ---
i:
0.1 l i ,II!Ii Na/AI
0 ...... iiii!iiil--[
CoNiA/ CoNiA/ CoNiX/ CoNiX/
3 x 30' 5 x 30' 3 x 30' 5 x 30'
Fig. 1. Comparison of the C02+/Al3+, Ni2+/A|3+, and Na+/A13+ratios of the zeolites A and X after three
and after five ion exchanges at 70 ~
418
The latter observation was also made for zeolite X, but both cations are exchanged to
almost equal amounts after few exchanges. However, after more than three exchanges a
significantly increased content of Co 2+ was observed, but Co 2+ was not as much preferred by
zeolite X as by zeolite A. The significant over exchanges can most likely be explained either
by salt imbibition, hydroxyl species of the transition metals within the pores of the zeolites or
by a crust of transition metal species (salt or oxidic species) on the external surface of the
zeolite crystals. Information on the chemical composition of the surface of the exchanged
zeolites and on the formation of an external crust was obtained by XPS (Fig. 2).
a) s 2s s ~Si, 2sAI2s/
Ni 3s plAI2p
/
C {/i2 " -Ni3p
/l/0~176
Si 2s AI 2s 3p/~
s 2s s 2p / I Ni
X-I , Co 3p
i i i i i
X-I ~ ~ ~
9 i i i i , i
1000 900 800 700 600 500 400
BindingEnergy[eV]
Fig. 2. a) XPS survey scans of the Co 2+ and Ni 2+ exchanged zeolites A-I and X-I in the binding energy
range between 400 and 0 eV, b) XPS survey scan of the Co 2+ and Ni 2+ exchanged zeolites A-I and X-I
in the binding energy range between 1000 and 400 eV.
419
For both zeolites, sulphur was detected on the external surface of the particles, which was
introduced by the NiSO4 salt solution used for the exchanges (Fig. 2a). The presence of
sulphur (SO4~ indicates a salt crust on the particle surface, which was not completely
removed by the washing after the exchange procedure. The results of the quantitative
chemical analyses obtained from the XP-spectra are compared to the results from the bulk
analyses in Table 2. The quantitative analyses of the data shown in Fig. 2b reveal a higher
concentration of Ni 2+ on the surface of A-I (Co:Ni = 0.6) compared to X-I (Co:Ni = 0.9). The
EDX analysis of X-1 resulted in a bulk Co:Ni ratio which is very similar to the ratio obtained
by XPS analysis of the surface. In contrast, the Co:Ni ratios determined for the bulk (1.5) and
for the surface of zeolite A-I particles (0.6) are different. The spectra show an enrichment of
Ni 2+ on the particle surface (e.g., as sulphate or oxidic phase).
Table 2
Results of the EXD analyses and the results from crystal structure
refinement
Co" Ni" A1 Co+Ni/A1 Co" Ni
Bulk of spinel bulk bulk surface of spinel
composition from from from from from
from EDX Rietveld EDX EDX XPS Rietveld
refinement refinement
A-I 0.37:0.25:1 0.6 1.5 0.6
A-I-1400-12 0.29:0.21:1 1.4
A-I-1400-12 0.42:0.28:1 0.7 1.5
after leaching
A-II 0.68:0.45:1 1.1 1.5
A-II-1400-12 0.34:0.21:1 1.6
A-II-1400-12 0.45:0.26:1 0.7 1.7
after leaching
X-I 0.31:0.31:1 0.6 1.0 1.0
X-I-1400-12 0.25:0.26:1 0.9
X-I- 1400-12 0.40:0.40:1 0.7 1.0 2.7
after leaching
X-II 0.35:0.28:1 0.6 1.3
X-II-1400-12 0.28:0.22:1 1.3
X-I-1400-12 0.53:0.39:1 0.9 1.3
after leaching
than for the spinel X-I- 1400-12 (0.9) as the result of a larger amount of Co 2+ in the tetrahedral
sites. The larger percentage amount of the larger Co 2+ cations in the spinels of series II leads
to an increase of the lattice parameters of spinels obtained from series II.
8.0900 8.0900
OA-1-12 qb r r
o~8.o88o o< 8.o88o r
rq X-1-12
@
-- 8.0860
r
9 9
r r
~ 8.0860 r
E E
~ 8.0840 8.0840
O,,
8.0820 8.0820
o o
9
m 8.0800 8.0800
DX-II-12
8.0780 8.0780 r i i ~ i r
115o 12oo 1 2 5 o 1 3 o o 1 3 5 o 14oo 145o 1 O0 1150 1200 1250 1300 1350 1400 1450
temperature (~ temperature (~
Fig. 3. Refined lattice parameters of the calcined zeolites of series I (left) and series II (right)
determined by Rietveld refinements after cooling the samples from different temperatures to room
temperature. The samples were annealed for 12 h.
From the refinement of the site occupancy factors we know that for the spinels obtained
from exchanged zeolite A, more Co 2+ than Ni 2+ is incorporated into the crystal structure.
Neutron diffraction experiments showed that Co 2+ is located on the tetrahedral positions. The
ionic radius of Co 2+ (0.58 A) in tetrahedral coordination is much larger than that of AI 3+ (0.39
A) [11]. Ni 3+ in the octahedral coordination has a radius (0.69 A) which is only slightly larger
than that of A13+ (0.53 .~) in octahedral coordination. The higher percentage of the larger
cation on the tetrahedral position is the reason for the larger structural expansion of the
spinels obtained from zeolite A. For a pure NiAI204 spinel obtained by sintering oxide
mixtures, it was observed that with increasing temperature, the lattice parameter increases
significantly [4,6]. In contrast, the ternary spinels from the calcined zeolites show only a very
small decrease with increasing calcination temperature. The inversion parameters for the
different spinels from the different series remain rather constant, independent of the
calcination temperatures (Fig. 4). The oxygen parameters of all spinels (not shown here)
increase only slightly with temperature. The change of the oxygen parameter for the spinels
obtained from zeolite A between 1200~ and 1400~ is only 0.0017.
0.600 0.600
0.550 0.550 9
r r
E 0.500 FIX-II-12
r r ~ 0.500
The crystal chemistry of the spinels so obtained is in equilibrium, even for the samples
calcined at lower temperatures. No significant cation re-arrangement takes place during
calcination at elevated temperatures, as is observed for samples prepared by conventional
methods. A probable explanation for this property is the high dispersion of the elements on an
atomic scale in the parent zeolite materials and the rather slow formation of the spinel
crystallites in the solid silica matrix. A comparison of the compositions of the exchanged
zeolites with those of the resulting spinels shows that the final product contains an excess of
Ni and Co that was not converted into spinels and this is probably present as an amorphous
oxidic by-phase.
8.0940 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.0940 .................................................................................................
8.0920 8.0920
,,~" 8.0900 ,,~ 8.0900
8.0880 8.0880
'~ 8.0860 9 8.0860
E 9 E
~: 8.0840 8.0840
8.0820 9 " 8.0820 eA-I
OA-I
.ca 8.0800 9 9 .s 8.0800 OA-II
8.0780 9 '~ 8.0780 9 x-I
8.0760 I-IX-II 8.0760 I-; x-II
8.0740 8.0740
1-1300-Q 1-1300-12 11-1300-Q 11-1300-12 1-1300-Q 1-1300-12 11-1300-Q 11-1300-12
Fig. 5. Comparison plots of the lattice parameters (left) and the inversion parameters (right) of the
spinels obtained by calcination of zeolites A and X at 1300~ for very short (Q = quenched) and long
(12 = 12h) time treatment periods.
4. CONCLUSIONS
the amorphous solid matrix after the collapse of the zeolite structures. The spinels so obtained
have a cationic ordering that is very close to the thermodynamically favored one. Thus,
annealing of the spinels at increasing temperatures does not have a significant influence on
the crystal chemistry of the spinels. This is due the fact that the spinels obtained by
calcination of exchanged zeolites are already in their thermodynamically favored equilibrium
state, which is in strong contrast to the spinels obtained by the mixture and calcinations of
oxides. The spinels obtained from the zeolite precursors are in their most stable configuration
immediately after their formation, and cation re-arrangements, which would result in structure
expansions or contraction, are not observed, even upon heating for prolonged time at high
temperatures. The spinel structures so formed are thus very resistant to heat treatment.
ACKNOWLEDGEMENTS
We thank H. Bongard, MPI ftir Kohlenforschung, for the EDX analyses and the Deutsche
Forschungsgemeinschaft (DFG, Project SCHM 936/3-1) and the Max Planck Society (MPG)
for financial support.
REFERENCES
1. INTRODUCTION
The studies available on membranes of titanosilicate and related materials, reporting on ETS-
4 [1,2], a microporous solid with a pore size of 0.3-0.4 nm, and ETS-10 [3,4], with
micropores of 0.49x0.76 nm, indicate that these novel materials may broaden the field of
application of zeolite membranes and thin films, which are commonly used in separations,
membrane reactors, sensors and now, very recently, in separation and reaction microdevices.
These mixed octahedral-tetrahedral oxides present also enormous possibilities of isomorphous
framework substitution. These framework substitutions would allow fine-tuning the catalytic
and adsorption properties of a given membrane, while preserving its microporous structure.
Up to date, pure titanium silicate membrane (ETS-4 [1-2] and ETS-10 [4]) have been
obtained in the absence of costly organic templates and avoiding calcination treatments,
which often lead to defects in membranes and losing selectivity.
Propylene is one of the most important raw materials in the petrochemical industry
since it finds applications in the synthesis of large variety of products (for example, polymers
and propylene oxide). The main processes to obtain propylene (such as, naphtha cracking,
propane dehydrogenation) give rise to a propylene/propane mixture. Distillation has been
used to separate it but is a high energy consumer process, and then several alternative
technologies have been investigated based mainly on kinetic or steric separations by
adsorption [5]. Membrane separation is another alternative to olefin/paraffin separation
methods on the ground of high selectivity and low energy consumption. Polymeric
membranes have been used for this purpose [6], but they have strong limitations in terms of
mechanical and chemical stability [7]. On the other hand, inorganic membranes have a much
wider range of operation, and therefore constitute interesting candidates for processes in
424
which chemical environment and mechanical restrictions prevent the use of polymeric
membranes. Nowadays, the studies concerning the use of inorganic membranes to separate
the propylene/propane mixture are quite limited [7], and alumina-silica microporous
membranes [8-9], zeolite membranes [10], carbon membranes [11-14] and mesoporous
alumina or microporous silica with Ag [7] have been already tested.
In order to modify separation properties, titanium in the ETS-10 framework has been
replaced by vanadium in this work, producing the vanadosilicate AM-6. ETS-10 and AM-6
membranes have been synthesized, characterized and tested in the separation of the
propylene/propane mixture.
2. E X P E R I M E N T A L
ETS-10 and AM-6 membranes have been prepared on commercial tubular symmetric
supports, made of or-alumina (Inocermic) with 1900 nm pore size, by seeded hydrothermal
synthesis, which was performed into a Teflon-lined autoclaves (volumes approximately: 43
mL or 165 mL). The internal and external diameters of the tubes were approximately 7 and 10
ram, respectively, with 8 cm lengths of which 5 cm were permeable. Before the hydrothermal
synthesis, in both membranes, the supports were seeded by rubbing using ETS-10 seeds. The
seeds had a particle size of approximately 500 nm (measured by photon correlation
spectroscopy (PCS), Malvern Zetasizer 3000 HS), and were synthesised according to the
procedure given by Rocha et al. [15].
ETS-10 membranes have been prepared following the procedure of Lin et al. [4]
through either one-step 48 hours seeded syntheses or two-step 48 hours seeded syntheses, i.e.,
after the seeded support was 48 hours at 230 ~ the membrane was seeded and immersed
again into fresh gel and treated at the same temperature for a further 48-hours period. The
ETS-10 gel has the following molar composition: Na20:4.0, K20:1.2, SIO2:5.5, TiO2:l.0 and
H20:335. AM-6 membranes have been synthesized with one-step synthesis during 48 hours at
230 ~ The AM-6 gel has the following molar composition: Na20:10.0, K20:2.3, SIO2:9.7,
VO2"2.0 and H20:300 [ 16]. The precursor reactants used were TIC13 solution (15 wt. % TiCl3,
10 wt. % HC1, Merck) or Vanadium(IV) oxide sulfate hydrate (VOSO4"xHzO, 99.99 wt. %,
Aldrich), deionized water, hydrogen peroxide for ETS-10 (30 wt. %, Riedel-de Haen),
potassium chloride (99 wt. %, Merck), and a sodium silicate solution (27 wt.% SiO2, 8 wt. %
Na20, Merck).
The obtained membranes were characterized by XRD and SEM. In order to achieve
valuable information on the permeation regimes (the adsorption processes taking place and
the overall quality of the membranes) a temperature programmed permeation (TPP) [17]
system was employed to measure single gas permeance of N2, C3H8 and C3H6 through the
membranes. The membrane was placed in a stainless steel module where it was sealed with
silicone o-rings. Before running TPP, the membranes were heated in inert gas to 230~ in
order to remove any adsorbed species.
Mixed gas permeation tests were carried out in a cylindrical stainless steel membrane
module, which temperature was controlled by a furnace. An equimolar mixture of C3H6/C3H8
was fed to the outside of the membrane and gas that permeated through the membrane was
swept out using a stream of helium. The concentrations of permeants and retentates were
determined using a TCD-GC (Micro GC Varian 4900). The membrane permeance was
defined as moles of gas permeated per unit time and unit membrane area, divided by the
logarithm mean of partial pressure difference between the retentate and permeate sides. The
425
3.1. Characterization
Fig. 1 shows permeances for two different ETS-10 membranes and an AM-6
membrane. The propane and propylene permeances of membrane obtained by one-step seeded
synthesis decreased with increasing temperature. The decreasing of permeance with
temperature is owing to Knudsen permeation through mesopores (inter-crystalline defects)
with perhaps some contribution from larger defects.
In contrast, the propane and propylene permeances of the ETS-10 membrane obtained
by two-step seeded synthesis increased with increasing temperature. The propane permeance
for the AM-6 membrane increased with increasing temperature as well. The increasing of
membrane permeances with temperature can be due to either the adsorption of the organic
compound (the permeance increases because the increase in temperature enhances the
mobility of adsorbed species) or activated diffusion through micropores (Arrhenius-like
dependency of activated transport through micropores).
The AM-6 membrane, shown in Fig. 1, has been synthesized with a larger volume
autoclave than ETS-10 membranes (aprox. 165 vs. 43 mL). The AM-6 membranes
synthesized with the low volume autoclave (43 mL) showed high nitrogen permeance and no
activated flow (not shown), the SEM analysis (not shown) also showed a poor intergrowth
layer of crystals. Therefore, the quantity of gel plays an important role in the AM-6
membrane texture. Probably with the low volume autoclave (43 mL), the reactants are
exhausted at later period of synthesis, which limits the formation of a well-intergrowth layer
of crystals. It should be noted here that an ETS-10 membrane has been also synthesized with
the large volume autoclave with one-step seeded synthesis, and did not show an improvement
in its TPP performance.
4E-7 ' I ' 1 ; I ' I ' I ~ "~1
AM-6 C3Hs
C3Hs
o 1E-7
TPP results suggest that the ETS-10 membranes prepared with two-step seeded
synthesis with the low volume autoclave and AM-6 membranes prepared with the large
volume autoclave can offer the possibility of a mechanism for compounds separation.
Therefore, these membranes have been tested on propylene/propane separation and
characterized by XRD and SEM.
XRD analyses (Fig. 2) reveal that pure ETS-10 or AM-6 phases were prepared as
membranes on alumina supports. The results were obtained by analyzing directly the surface
of the membrane and some peaks coming from the alumina support can be observed.
SEM analyses show (Fig. 3) cross-sectional views for ETS-10 and AM-6 membranes,
with apparently continuous layers, ca. 4-7 lam thick. In all the cases, top view shows a layer of
well-intergrown crystals (1-2 lain) with the corresponding typical morphology.
f AM-6 Membrane
9
i[I
9
1
.....j f[I~k..J ~ I.
-4,,,0
9 ~==,i
-4..0
- 10 Powder
, I I ,
10 20 30 40
2"0 (degree)
Fig. 2. XRD patterns of ETS-10 and AM-6 membranes and ETS-10 powder. (e) Peaks from alumina
support.
a)
b)
Fig. 3. SEM images of cross section and top view of membranes a) ETS-10 and b) AM-6.
428
% ~D
1E-9
% % 1
30 ~
lira A ! r' [ 0
; 2'o ' 4'o ' ' 8O
Time (hr)
Fig. 4. Propylene/propane separation factor (symbol with cross) and propylene (closed symbols) and
propane permeances (open symbols) through an ETS-10 membrane as a function of time. 40 mL/min
of equimolar propylene/propane mixture. Sweep gas helium: 40 mL/min.
3~ [] @ml
4 o
.4-.a
O
o
(D IE-8 [][]1[]
,Ip
~D 3
~D
HrqEH []
[] m
[]
_[]
IE-9 i
'0 .......i ,
0 40 8'0
Time (hr)
Fig. 5. Propylene/propane separation factor and propylene and propane permeances through an AM-6
membrane as a function of time. Symbols and operating conditions same as Fig. 4.
429
50 ' ' ' " I | | | w ||w I ' ' ' ' ' ' " l
+ [] A1203/SiO 2 [8-9]
9 Faujasitezeolite [ 10]
;> 40 A Carbon[ll]
.4-.a V Carbon[12]
eO
(D O Carbon[12]
(D + Carbon[13]
30 • Carbon[13]
o
.4...a
r Carbon[14]
Ag/SiO, [7]
9 Ag/Al203[7]
o
9 ,m..l
20
(D
cr 10
9 ETS-10(30~ this work o 9
9 AM-6(150~ this work
9 I1 []
| i J w wll I | | | |
' ' ' ' ' 1 ' ' ' ' ' ' ' ' 1 ' ' ' ' ' ' ' ' 1 '
Fig. 6. Some examples extracted from the literature of propylene/propane performance for different
inorganic membranes.
Finally, Fig. 6 compares the propylene/propane separation properties obtained for the
membranes synthesized in this work with some of the results reported for inorganic
membranes, extracted from a literature survey. The results from different laboratories are
often difficult to compare because of the different experimental conditions used. In general, it
can be said that the membranes with moderate high selectivity (30-40) have low permeances;
also, as increase the permeance the selectivity decrease. The membranes prepared in this work
have one of the highest permeance reported about 8.10 .8 mol/mZ.s.Pa showing interesting
selectivities. For the ETS-10 membrane the separation factor (6.5) at 30 ~ is one of the
highest reported in the literature for this range of permeance.
4. C O N C L U S I O N S
High-quality and pure titanosilicate ETS-10 and vanadosilicate AM-6 membranes can be
prepared by seeded hydrothermal synthesis on porous alumina tubular supports. XRD patterns
showed that ETS-10 and AM-6 were the only crystalline material present in each membrane.
SEM examination of the membranes revealed layers, ca. 4-7 lam thick with a well intergrowth
of crystals. According to the TPP characterization, these membranes are defect-free, which is
denoted by having activated permeation responsible for the high separation performance
observed.
At ambient temperature, ETS-10 and AM-6 membranes are able to separate at steady
state the propylene/propane mixture with a separation factor ranged between 3.8-6.5 and 2.1-
430
2.4, respectively (when the Knudsen one is of only 1.0), and propylene permeances for ETS-
10 and AM-6 membranes about 8.10 -8 and 8-10 -9 mol/mZ.s-Pa, respectively. This selectivity
may result from the different interactions between the microporous material structures and
propane or propylene. At high temperature (>40 ~ the permeance and separation factor fall
much quicker in ETS-10 membranes than AM-6 membranes due to the pore blocking. A
thermal treatment at 3 50 ~ for 10 hours can recover the permeance and selectivity.
ACKNOWLEDGEMENTS
Authors from University of Zaragoza acknowledge financing from DGA and MCYT, while
those from University of Aveiro thank financial supports from FCT, FEDER, INTERREG
IIIB and POCTI. The acknowledgement is also extended to BSCH and University of
Zaragoza for the research grant awarded to I. Tiscornia.
REFERENCES
[1] C.M. Braunbarth, L.C. Boudreau and M. Tsapatsis, J. Membr. Sci., 174 (2000) 31.
[2] G. Guan, K. Kusakabe and S. Morooka, Microporous Mesoporous Mater., 50 (2001) 109.
[3] H.K. Jeong, J. Krohn, K. Sujaoti and M. Ysapatsis, J. Am. Chem. Soc., 124 (2002) 12966.
[4] Z. Lin, J. Rocha, A. Navajas, C. T611ez, J. Coronas and J. Santamaria, Microporous Mesoporous
Mater., 67 (2004) 79.
[5] J. Padin, S.U. Rege, R.T. Yang and L.S. Cheng Chem. Eng. Sci., 55 (2000) 4525.
[6] R.L. Burns and W.J. Koros, J. Membr. Sci., 211 (2003) 299.
[7] K.A. Stoitsas, A. Gotzias, E.S. Kikkinides, Th.A. Steriotis, N.K. Kanellopoulos, M. Stoukides
and V.T. Zaspalis, Microporous Mesoporous Mater., 78 (2005) 235.
[8] B.N. Nair, K. Keizer, W.J. Elferink, M.J. Gilde, W.J. Elferink, H. Verweij and A.J. Burggraaf,
J. Membr. Sci., 116 (1996) 161.
[9] B.N. Nair, K. Keizer, H. Verweij and A.J. Burggraaf, Sep. Sci. Technol., 31 (1996) 1907.
[ 10] V. Nikolakis, G. Xomeritakis, A. Abibi, M. Dickson, M. Tsapatsis and D.G. Vlachos, J. Membr.
Sci., 184 (2001) 209.
[11] A.B. Fuertes and I. Menendez, Separ. Purif. Technol., 28 (2002) 29.
[12] M. Yamamoto, K. Kusakabe, J. Hayashi and S. Morooka, J. Membr. Sci., 133 (1997) 195.
[13] J. Hayashi, H. Mizuta, M. Yamamoto, K. Kusakabe and S. Morooka, Ind. Eng. Chem. Res., 35
(1996)4176.
[14] M. Yoshino, S. Nakamura, H. Kita, K.I. Okamoto, N. Tanihara and Y. Kusuki, J. Membr. Sci.,
215 (2003) 169.
[ 15] J. Rocha, A. Ferreira, Z. Lin and M.W. Anderson, Microporous Mesoporous Mater., 23 (1998)
253.
[16] J. Rocha, P. Brandao, Z. Lin, M.W. Anderson, V. Alfredson and O. Terasaki, Angew. Chem.
Int. Ed., 36 (1997) 100.
[17] M.P. Bernal, J. Coronas, M. Men6ndez and J.Santamaria, J. Membrane Sci., 195 (2002) 125.
[18] A. Philippou, J. Rocha and M.W. Anderson, Catal. Lett., 57 (1999) 151.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 431
The design of the mesoporous molecular sieves with cage-, and cylindrical-like structures
makes these materials promising candidates in environmental and industrial applications. If
the mesostructures are fabricated into transparent monoliths that are hard, free of cracks, and
macroscopic in scale, then their utilities can be widely expanded. Here we developed a
synthesis strategy, instantly direct templating liquid crystal phase, which is simple in terms of
fabrication time (within minute) and in composition domains, and is efficient to design
mesostructures with cage and cylindrical-like monoliths, large-sized glass, thick-walled, and
uniform mesopores. A family of monolithic structures was feasibly fabricated in large ordered
domains by using triblock copolymer templates. This approach leads to retention of the
ordered structures under extreme hydrothermal conditions for a month or even longer.
1. INTRODUCTION
Ordered mesoporous molecular sieves are one of the most exciting new developments in
materials science and technology in the last decade [1-4]. Features of these mesoscale
architectures included controllable mesopores (2-30 nm) enable these materials to be
promising candidates in applications such as catalysis, separations, sensing, optical, and
electronic systems [5-8]. Since the discovery of the first broad mesoporous family denoted by
M41S, amphiphile-water systems are used as the structure-directing agents to fabricate
mesoscopically ordered materials over a wide range of synthesis conditions [ 1]. Such design
dexterities led to new level of hierarchical design of materials with highly controllable over
the mesophase geometry and pore morphology [2-4]. Among all possible surfactant
templates, the amphiphilic triblock copolymers of Pluronic-types have been used as
prominent systems to develop highly ordered, large pore dimensions (up to 10 nm) without
the use of swelling agents and to develop thicker-wall structures with improved hydrothermal
stability compared with that of the parent MS41 family [9,10].
The synthesis design that achieves a reliable control over the microstructure phase of the
templates and the final mesostructured replicas is expected to open the possibility to design
large mesoscopically ordered domains with a variety of mesophases. Significantly, direct
templating methodology can be used in the design of ordered mesoporous silica monoliths,
and in controlling the surfactant mesophase structure and morphology [11,12]. This direct
templating methodology is advantageous in fabricating hexagonally ordered monoliths with
control over the pore dimensions in the 5-10 nm size range using copolymers F127, P123, or a
mixed combination of triblock copolymer amphiphiles as templates in ethanol under strong
acidic conditions (pH-~l) [13,14]. Important developments in this direct templating
432
2. EXPRIMENTAL SECTION
0.4 g pentadecane and 2 g TMOS were dissolved in a flask by agitation in a water bath (50-60
~ for 1 to 2 minutes, yielding a clear solution (i.e., homogenous). To this mixed solution, 1
g of HzO/HC1 was quickly added. The mass ratio of copolymer: TMOS: H20/HC1 was 1.6: 2:
1. Synthesis of all HOM monolithic samples at various copolymer/TMOS ratios was done
using the same procedure similar to that mentioned in such exemplary composition given with
80 wt%. The amount of the block copolymers was varied at 0.7, and 1.0 in the mixture
composition to fabricate mesophases with copolymer/TMOS mass ratios of 35, and 50 wt%,
respectively, as listed in Table 1. The methanol produced from the TMOS hydrolysis was
removed by using a diaphragm vacuum pump connected to a rotary evaporator at ambient
synthesis temperature in the range of 40-50 ~ Within-10 minutes, the resulting viscous
liquid changed to an optical gel-like material (solid product) and acquired the shape and size
of the reaction vessel [15,16]. The templates were removed by calcination at 450 ~ (1 h
under N2 and then 6 h under O2).
2.3. Analyses
Small-angle powder X-ray diffraction (XRD) patterns were measured by using an MXP 18
diffractometer (Mac Science Co. Ltd.) with monochromated CuKot radiation with scattering
reflections recorded for 20 angles between 0.3 ~ and 6.5 ~ corresponding to d-spacings between
29.4 and 1.35 nm. N2 isotherms were measured using a BELSORP36 analyzer (JP. BEL Co.
Ltd) at liquid N2 temperature (77K) by using the Brunauer-Emmett-Teller (BET) method of
surface area and the Barrett-Joyner-Halenda (BJH) analyses of pore size distributions from
adsorption curve of the isotherms. All samples were pre-treated at 300 ~ for 8 h under
vacuum until the pressure was equilibrated to 10.3 Torr. Transmission electron microscopy
(TEM) images were obtained by using a JEOL TEM (JEM-2000EXII) operated at 200 kV
with a side-mounted CCD Camera (Mega View III from Soft Imaging System Co.). The TEM
samples were prepared by dispersing the powder particles onto holey carbon film on copper
grids. Scanning electron microscopy (SEM) micrographs were obtained by using a Hitachi S-
800 SEM operated at 7 keV. 29Si MAS NMR spectra at room temperature were also measured
using a Bruker AMX-500 operated at 125.78 MHz with a 90 ~ pulse length of 4.7 ~ts. For all
samples, the repetition delay was 180 s with a rotor spinning at 4 kHz. The chemical shift
scale was externally set to zero for 29Si signal by using tetramethylsilane.
Among all synthetic designs, our simple in synthesis time and in composition domains of
[copolymer/C~5-alkane/TMOS/H20] and efficient strategy, namely, the instant direct-
templating method, shows key fact of the fabrication of various types of copolymer-silica
mesophases (Tables 1) in mesoscopically ordered structures, indicating the real control over
the copolymer phases [ 15, 16]. Such successful fabrication is evidence that the formation of
the true transparent liquid crystal phase in both lyotropic and microemulsion systems (one
single phase) was sufficient to retain the anisotropic organization (preferential orientation)
even after interaction between the silica species and the hydrophilic EO blocks to form the
434
rigid condensed framework matrices. However, under acidic synthesis conditions, the
inorganic species is subsequently polymerized by acid catalyst (HC1, pH ~1) to form a rigid
cross-linkage silica network wall around the phase assembly [9, 10]. Results from our
syntheses suggest that sufficiently strong interactions between the silicate species and
hydrophilic EO blocks are necessary to achieve formation of highly ordered copolymer-silica
mesophases, otherwise disordered mesophase structures are formed [1-4]. Here, the silica
polymerization conditions, which were adequate to improve the interactions between silica
and copolymer species, crucially influenced the fabrication of these highly ordered monoliths
(HOM) with rigid network silica structures. However, at l_<pH_<l.3, positively charged
protonated silica species (-Si(OH2) +) and alcohol (methanol) were formed. The alcohol
initially aided the formation of well-homogenized phase [12, 16]. The removal of methanol
by vacuum is necessary to form glassy monoliths during the solidification process [11-16].
The silica species preferentially interacted with the hydrophilic EO blocks to promote the
extended assembly around the crystalline phase and to form viscous gel-like mesophase
structures in rapid gelation (,~10 min) with no effect on the structural periodicity.
29Si NMR data show evidence of the polymerization silica species to form the rigid condensed
framework matrices under our synthesis designs. These data show that two main resonances
at -110 and-101 ppm, and a weak peak at -92 ppm correspond to Si (OSi)4 (Q4), (OH)Si
(OSi)3 (Q3), and (OH)2Si (OSi)2 (Q2) silica species were described the as-made sample with
Q4/Q3 ratios less than 1, suggesting the formation of the silica framework wall with
considerable hydrophilic surfaces. After calcination, 29Si NMR data show a main resonance
at -110 ppm for a Q4 environment of the silica framework and a shoulder at -101 ppm for a Q~
environment from surface hydroxy groups. The higher Q4/Q3ratios of calcined HOM sample
were indicative of fully cross-linked framework walls, and consequently indicative of much
more rigid siloxane linkages [ 18].
In addition, this synthetic approach was used to successfully fabricate 3D cubic
mesostructures into mechanically stable, optically macroscopically transparent, crack-free
monoliths by using our design, despite the materials being prepared in dry air. SEM
micrographs reveal that the cubic monoliths had a large particle sizes (50-150 lam in
diameter). In general, the crystal particles of the monoliths (as-made and calcined forms) were
predominantly plate-like with smooth fine lines on their surface morphology (Fig. 1a, b).
Fig. 1. SEM micrographs of (a) as-made, and (b) calcined cubic Fm3m monoliths (HOM-10)
fabricated in microemulsion system formed by addition of C15-alkane to F68 phase domains at
F68/TMOS of 80 wt%.
For all synthesis designs, the XRD patterns of these HOM monoliths (Fig. 2) show well-
resolved, sharp Bragg peaks that allowed identification of certain HOM structures likely to
have analogous and closely reflected planes (as listed in Table 1). These unique reflection
435
planes were characteristic of highly periodic and 3D cubic phase domains of Ia3d, Im3m, and
Fm3m symmetries with a large lattice constant up to 22 nm for calcined monoliths (Table 1).
Our results also show that following calcination, the structural contractions for all HOM
monolithic lattice dimension constants generally ranged between 10 and 12% depending on
the synthesis compositions of the mesophase domains, yet the structures retained their well-
defined ordering [ 15, 16]. Despite this shrinkage of the HOM frameworks, the intensity and
resolution of the reflection peaks become more intense, indicating a high degree of
mesoscopic order after removal of the copolymer species.
] d111= 11.03 nm
11= 8.8 nm d11o= 9.3 nm
= m
e~
e"
m g,
, I | J , , I , , , i I , , , t I , , | , i i i I I
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5
2e P 20 P 2e/o
Fig. 2. XRD patterns of HOM monoliths of (a) cubic Ia3d (HOM-5) and (b) cubic Im3m (HOM-C1)
structures fabricated by using P123 copolymer at P123/TMOS of 80, 35 wt%, respectively, and (c)
cubic Fm3m (HOM-C 10) structures fabricated by using F108 copolymer at F 108/TMOS of 50 wt%.
Table 1
Structural parameters of mesoporous silica cage (HOM-C) and cylindrical (HOM) monoliths
synthesized in microemulsion mesophases of copolymers; mesopore (Vp)/ micropore (Vm)
volumes, unit lattice dimension (o), BET surface area (Sser), pore size (R), and wall thickness
(w)
Phase synthesis conditions Monolithic Ve Vm a SeEr R W
structure Cm3/g Cm3/g nm m2/g nm nm
Template Copolymer T
/TMOS ~
(Wt %)
TEM images (Fig. 3) are direct, real-space evidence that long-range ordered cage and
cylindrical pores in large-scale domains along the incidences are characteristics the monoliths.
The most prominent feature was the uniform arrangements and continuous ordering along all
directions without distortion, indicating the integrity of the cubically ordered cage
frameworks with Fm3m, Ia3d, and Im3m symmetries (Fig. 3).
Fig. 3. TEM images recorded along the (a) [110], and (b) [211] directions of cubic Fm3m monoliths
(HOM-C10) synthesized by using F108 copolymer, along the (c) [111] and (d) [311] zone axes of
gyroidal cubic Ia3d (HOM-5) monoliths synthesized by using P 123 copolymer, and along the (e) [ 111 ]
and (f) [110] incidences of cubic Im3m (HOM-1) monoliths synthesized by using P123 copolymer.
Fig. 4A shows that wide type-H2 hysteresis loops and well-defined steepness of isotherms
indicate large and uniform cage structures similar to those for prominent FDU-1 silicas [ 19].
The width of the hysteresis loop increased with increasing mesopore size for all monolithic
structures templated by the copolymers. The distinctive similarity in the isothermal shape and
in P/Po (0.45- 0.5 range) of the capillary evaporation of HOM structures fabricated at various
synthesis conditions validated that the sizes of the entrances of the pores that connect the
uniform caged pores were below 4.0 nm. However, isotherms (Fig. 4B) indicate that open
cylindrical channel-like pores featured the monoliths, as evidenced from well-defined type-
H~hysteresis loops [20].
Based on the N2 isotherms, the HOM monolithic structures well-fabricated by using the
instant direct-templating method exhibited appreciable textural parameters of specific surface
area (SsEr), mesopore/micropore volumes, and thick walls (Table 1). The microporosity of
HOM monoliths was assessed (using the ets-plot method) to be 0.06-0.16 cm3/g range [ 19,
20]. The increase of the micropore volumes was strongly affected by the unit numbers of EOm
blocks of the copolymer templates used, rules out the potential effect of the penetration of the
hydrophilic (EO) blocks into the silica wall during fabrication of the silica/copolymer
mesophases (Table 1). In general, the enlargement of the cylindrical and cage pore sizes by
using a higher concentration of copolymers indicates that an increase in micelle aggregation
of triblock copolymers causes a significant increase in pore sizes. For a given copolymer
templates, the pore sizes increased as a function of the core (PO) and corona (EO) block
lengths, whereas the pore size was effectively influenced by the hydrophobic block (PO)
length.
437
500 (A)
800
:Y
"7, 4 0 0 600 '
r
o.
k- 3 0 0
400
E
> 200
i!!iiot
= (a) H O M - C 1 200
100 ~ (b) H O i - C 5 = (a
1 I I I I I I
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
P/Po P/Po
Fig. 4. N 2 adsorption/desorption isotherms of HOM monoliths, respectively. Isotherms A of calcined
cage cubic (HOM-C) structures of Im3m (a), Ia3d (b), and Fm3m (c) synthesized by using P123, F68,
and F108 copolymers, respectively. Isotherms B of cylindrical cubic (HOM-n) structures of Im3m (a),
la3d (b), and Fm3m (c) synthesized by using P123 and F68 copolymers (see Table 1). Isotherms B (
both b, and c) shifted vertically by 100 ml (STP)/g.
Particularly important features of the HOM structures are the thick-walled and high degree of
siloxane unit cross-linked in the frameworks, while no aging times or high temperatures were
used during the instant direct-templating synthesis strategy of monoliths. Those features
would have to stabilize the architecture wall to withstand for a month (or even longer) of
hydrothermal treatment without loss of the mesopore structures, despite the locally disordered
atomic scale of these HOM frameworks. XRD patterns for the cylindrical Ia3d (HOM-5)
sample hydrothermally treated in boiling water up to 20 days (Fig. 5A) show evidence of
retention of high-order structure.
(A) (B)
= 4 days
800 10 d a y s
d211 = 8.1 n m
(d) 20 d a y s
600
~'~ nm
(c) 10 d a y s o0
=
400
200
9; , i , , , , ! . . . . i , , = i t i i i L ! i i i
i i i i
The N2 isotherms (Fig. 5B) precisely show the degree of degradation in the structural
mesoporosity of Ia3d sample after hydrothermal treatment. The H~ hysteresis loops and the
isothermal shapes provide strong evidence that uniformity of the 3D cylindrical geometry was
attained after boiling treatment up to 10 days. For longer treatment times (> 10 days), the
well-developed capillary condensation and the hysteresis loops were narrow, indicating the
degree of structural collapse "distortion", as consistent with XRD profiles (Fig. 5Ad). The
textural parameters such as large mesopore volume, relatively small surface area, negligible
microporosity, thinner-wall framework, and larger pore sizes are evidence of degradation of
the pore wall architectures; however, such degradation did not ultimately lead to loss or
collapse of the mesostructured geometry. This evidence confirms that this synthesis strategy
can be used to effectively fabricate 3D monoliths with thicker pore walls and with a high
degree of cross-linking framework as seen in nature, and thus indeed preserve the key
characteristic of ordered architecture in the monoliths compared with other fabricated
materials that require special synthesis treatment to stabilize their framework [ 18,21 ].
In summary, the ability to synthesize 3D cubic silica monoliths that have controlled pores,
large-sized glass, and highly uniform mesoporosity by using a simple fabrication method
revealed new insights in the design of mesoporous materials. This particular approach,
namely, the direct templating method involving the instantly preformed liquid crystal phase,
shows promise for rapid synthesis to achieve high fidelity of control over the mesophase
topologies of the surfactants. The key result is that this approach enables retention of 3D
structural ordering, despite the severe hydrothermal treatments. Such retention is crucial in
industrial applications.
REFERENCES
[11 C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli and J.C. Beck, Nature 359 (1992) 710.
[2] S. Inagaki, Y. Fukushima and K. Kuroda, J. Chem. Soc. Chem. Commun. (1993) 680.
[3] Q. Huo, et al., Nature 368 (1994) 317.
[4] P.T.Tanev and T.J. Pinnavaia, Science 267 (1995) 865.
[5] M.E. Davis, Nature 417 (2002) 813.
[6] T. Yamada, et al., Adv. Mater. 14 (2002) 812.
[71 A. Corma, Chem. Rev. 97 (1997) 2373.
[8] B.C. Regan, S. Aloni, R.O. Ritchie, U. Dahmen and A. Zettl, Nature 428 (2004) 924.
[91 D. Zhao, et al., Science 279 (1998) 548.
[10] D. Zhao, Q. Huo, F. Jianglin, B.F. Chmelka and G.D. Stucky, J. Am. Chem. Soc. 120 (1998)
6024.
[11 ] G.S. Attard, J.C. Glyde and C.G. G61tner, Nature 378 (1995) 366.
[12] C.G. G61tner, S. Henke, M.C. Weissenberger and M. Antonietti, Angew. Chem. Int. Ed. Engl.
37 (1998) 613.
[13] N.A. Melosh, P. Davidson, P. Feng, D.J. Pin and B.F. Chmelka, J. Am. Chem. Soc. 122 (2000)
823.
[ 14] K.M. Ryan, et al., Langmuir 18 (2002) 4996.
[15] S.A. El-Safty and T. Hanaoka, Adv. Mater. 15 (2003) 1893.
[16] S.A. E1-Safty and Y. Hanaoka, Chem. Mater.16 (2004) 384.
[17] J. Fan, et al., Angew. Chem. Int. Ed. Engl. 42 (2003) 3146.
[18] Y. Liu and Y.J. Pinnavaia, J. Mater. Chem. 12 (2002) 3179.
[19] J.R. Matos, et al., J. Am. Chem. Soc. 125 (2003) 821.
[20] P.I. Ravikovitch and A.V. Neimark, J. Phys. Chem. B 105 (2001) 6817.
[21] A.S. O'Neil, R. Mokaya and M. Poliakoff, J. Am. Chem. Soc. 124 (2002) 10636.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka, N. Zilkov~ and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 439
Alcoholic solutions of cobalt chloride have been used to introduce Co in two different MCM-
41 materials: a pure silica and an aluminosilicate with Si/A1 = 1.25. For both compounds, an
ion exchange occurs between surfactant molecules (and eventually Na § cations) and Co 2+. In
the case of the low-silica MCM-41, Co 2+ occupy cationic sites and are homogeneously
distributed in the mesopores. By contrast, Co species introduced in pure-silica MCM-41 are
located exclusively at the surface of the particles, in the form hydroxide-like thin films. The
latter transform to well-defined Co304 crystals upon calcination at high temperature.
Whatever the composition, the introduction of Co does not alter the long-range mesoscopic
ordering of the parent materials. This method allowed us to obtain [Co]A1MCM-41
compounds with Co contents higher than 16.3 wt.%.
1. I N T R O D U C T I O N
Although cobalt exchanged zeolites have been widely studied, reports on amorphous
aluminosilicate analogs are scarce. Co-exchanged zeolites with MFI, M O R , F E R or BEA
framework types are selective catalysts in NO and N20 decomposition as well as in the
selective reduction of NO by hydrocarbons [ 1]. Moreover, CoZSM-5 and CoBeta catalyze the
selective ammoxidation of ethane to acetonitrile [2]. Co has been introduced in various
mesoporous compounds using co-condensation, impregnation or exchange methods [3-6].
However, the amount of Co, the dispersion of metal particles or the aluminum content was
quite low and compounds were hardly comparable with zeolites. Two major reasons can
account for the difficulty to prepare Co-exchanged low-silica MCM-41 with high Co loading.
First, A1MCM-41 compounds with high A1 content are difficult to synthesize, and most of the
solids are not thermally stable. Second, cation exchanges in aqueous media lead to a partial
dealumination of calcined Al-rich solids along with a loss of long-range ordering [7]. Another
way to introduce Co in MCM-41 is to use the so-called "Template Ion Exchange" method. A
Template Ion Exchange method was previously used to incorporate Fe, Cr or Co in pure-silica
MCM-41 [4,8,9]. In the case of cobalt, surfactant molecules were exchanged for a cobalt
complex [Co(en)2C12] +, leading to a Co:Si ratio of 0.044 in the final solid [4]. In the present
work, we have used a Template Ion Exchange procedure to incorporate Co 2+ cations in two
MCM-41 materials with extreme compositions. The first one is a pure-silica MCM-41, in
which CTMA + cations are associated to S i O defect groups whilst the second one is an
aluminosilicate with Si/A1 - 1.25, in which all surfactant molecules balance the negative
charge of aluminum. The present method allowed us to prepare highly ordered [Co]A1MCM-
41 with metal contents up to 16.3 wt.% Co.
440
2. E X P E R I M E N T A L SECTION
3. RESULTS
Table 1
Structural and textural properties of the different MCM-41 compounds
Solid SBET Pore diameter Pore volume
(m2/g) (A) ,(cm3/g)
A1MCM 550 19 0.28
A1MCM-4C 535 17.5 0.25
SiMCM 1185 32 0.90
SiMCM-2C 1143 31.5 0.89
From chemical and thermal analysis data, it is possible to estimate the chemical formula
of anhydrous materials:
A1MCM: (Nao.32CTMAo.12)Si0.56A10.4402
SiMCM: (CTMAo.18)SiO2
27A1 solid-state NMR indicates that A1 species are mainly tetrahedrally coordinated in as-
made A1MCM. The presence of Si(OA1)4 (Q4) units was confirmed by 29Si MAS NMR: the
spectrum essentially consists in a line at ca.-82 ppm, assigned to Si(OA1)4 units along with a
shoulder a t - 8 7 ppm assigned to SiOSi(OA1)3 species [6]. The absence of signals below-90
ppm indicates the absence of Q2, Q~ and Q0 species, in agreement with the Si/A1 ratio (1.25).
441
A1MCM SiMCM
(d)
(c)
(b)
(a) (a)
I "~'~(~'~"~' I I I I I l l l
1 3 5 7 9 1 3 5 7 9
2 THETA 2 THETA
Fig. 1. X-ray diffraction patterns of as-made (a), calcined (b), A1MCM-4 before and after calcination
(left c and d) and SiMCM-2 before and after calcination (right c and d)
The chemical formula of A1MCM confirms that all surfactant cations balance the negative
charge of A104 tetrahedra. The two as-made MCM-41 materials show well-defined X-ray
powder diffraction patterns, typical of highly ordered hexagonal structures (Fig. 1).
The long-range mesoscopic ordering of the structure was confirmed by TEM. The quality of
XRD patterns remains excellent after calcination at 530~ which indicates that the
compounds are stable at high temperature (Fig. 1). However, the relative diminution of d~10
and d200 reflections with respect to the d~00 reflection for A1MCM shows that Al-rich solids
are less stable than the pure silica product. The major part of aluminum species remains
tetrahedrally coordinated in the calcined product, as evidenced by solid-state NMR However,
a quantitative comparison of 27A1 NMR spectra before and after calcination shows that
approx. 30 % of A1 atoms were extracted from the walls of A1MCM and remained in the solid
in the form o f " N M R invisible" species.
Calcined materials show a type-IV N2 adsorption isotherm characteristic of mesoporous
materials with a narrow pore size distribution. The BET surface area decreases with the A1
content from ca. 1185 mZ/g for pure silica MCM-41 to 550 m2/g for AI~. The same evolution
is observed for the pore volume and the mean pore diameter (Table 1).
2+ 0,2 ]
0,3 O
O 1/2Co2+
0,15
, B
0,2
c
0
, m
0,1
0
0,1
0,05
CTMA +
7q7 " ~ C T M A +"
," , ~ ,v , , 0 I
A
"T
Fig. 2. Evolution of surfactant, Na and Co in AIMCM (left) and SiMCM (right) during exchange
with Co.
Chemical and thermal analyses of the treated solids show that the incorporation of cobalt
is accompanied by a decrease in the concentration of CTMA + and Na + cations in A1MCM-41.
Whatever the value of the exchange level x, the decrease in the [CTMA + + Na § concentration
corresponds to exactly twice the increase in Co 2+ concentration, in agreement with an ion
exchange process (Fig. 2). Moreover, the observed exchange level is comparable to the
theoretical one, which indicates that exchange is complete after 30 minutes. However,
CTMA + and Na + cations are not exchanged simultaneously. The exchange process occurs
exclusively between Co 2+ and CTMA + species until x = 1. Above this value, all organics have
been removed and Na § is exchanged for Co 2+. For x -- 4, all CTMA + and approx. 80% of Na +
cations have been exchanged. Chemical analysis of the corresponding A1MCM-4 sample
gives 16.3 wt.% Co, corresponding to the chemical formula (Na0.05Co0.z0)Si0.56A10.4402.
The nature of Co species introduced in A1MCM has been studied by UV-Vis
spectroscopy. The signal is essentially composed of a broad band at 550 nm, characteristic of
octahedrally coordinated Co 2+ species, without any contribution of a cobalt-oxygen charge
transfer below 350 nm (Fig. 3). The X-ray powder pattern of A1MCM-4, corresponding to a
theoretical exchange level x = 4, shows that the structure is maintained and does not expand
or contract upon exchange, as evidenced by the position of the d~00 reflection (Fig. 1).
Moreover, the fact that d~ ~0 and d200 reflection peaks are clearly visible indicates that the long
range ordering of the solid is not affected by the treatment, which was confirmed by TEM
(Fig. 4). Pictures clearly show the regular array of mesopores and do not reveal the existence
of disordered areas in the large hexagonal particles. Neither Co-oxide particles nor Co-rich
regions could be detected within the aluminosilicate network, in agreement with UV-Vis
spectroscopy.
443
1,2
0,5
1
0,4
,'z,. 0,8
e 0,3
~= 0,6
o
0,2
.~ 0,4
0,1 0,2
O, ,, , , , 0 I I -'5
Fig. 3. UV-Vis spectra of A1MCM-4 (left, bottom), A1MCM-4C (left, top) and SiMCM-2 (right).
EDX analyses performed on a series of particles confirmed the very high dispersion of
Co 2+ species, with a homogeneous Co concentration in excellent agreement with that obtained
by chemical analysis. Exchange of Na + for Co 2+ does not modify the local environment
around aluminum atoms. The 7A1 MAS NMR spectrum of A1MCM-4 is strictly similar to
that of the parent solid, and the absence of signal around 0 ppm indicates that treatments in
alcohol do not dealuminate A1MCM materials.
The color of Co-exchanged A1MCM-41 compounds changes from pink to blue upon
calcination in air at 530~ The corresponding UV-Vis spectra show a triplet at 520, 580 and
630 nm, characteristic of tetrahedrally coordinated Co 2+ cations (Fig. 3). The absence of
signals between 300 and 500 nm is an indication that Co 2+ species are not oxidized to Co 3+ at
high temperature. The absence of cobalt oxide species in Co-exchanged A1MCM-4C could be
confirmed experimentally by re-hydrating the compound at room temperature. Rehydration
restores the original pink color of as-made A1MCM-3 within a few hours. The corresponding
UV-Vis spectrum, which shows a unique signal at 550 nm, is strictly the same as that of the
compound before calcination. In particular, the absence of signal around 250 nm, assigned to
a charge transfer between the metal and surrounding oxygens, suggests that Co 2+ cations
remain isolated in the dehydrated material. TEM pictures of the calcined A1MCM-4C
compound support the previous conclusions. As we could not find any evidence of the
presence of cobalt oxide particles on the surface of A1MCM, it is likely that Co cations are
highly dispersed in the mesopores. Moreover, EDX analyses confirmed the homogeneous
dispersion of cobalt throughout the MCM-41 particles. Pictures also clearly show well-
maintained hexagonal array of mesopores and long-range ordering of the structure. As a
consequence, the X-ray powder diffraction pattern of A1MCM-4C shows not only the main
peak (100) but also three other peaks (110, 200 and 210) (Fig. l-d). The BET surface area of
the solid is 535 mZ/g, a value quite similar to that of the parent calcined A1MCM (Table 1).
444
u,., ,,,,~,
. . . . . ..
Fig. 4. TEM pictures of A1MCM-4 (a), A1MCM-4C (b), SiMCM-2 (c) and SiMCM-2C (d).
The mean pore diameter slightly decreases from 20 A for the parent A1MCM compound
to 17.5 A for A1MCM-4C, again supporting the presence of regularly dispersed metal species
in the mesopores. Calcination of exchanged A1MCM-4 also preserves the coordination of
aluminum atoms inside the amorphous walls. The 27A1 MAS NMR spectrum of the solid
shows a major signal at 55 ppm, characteristic of tetrahedral aluminum, without significant
contribution in the region corresponding to octahedrally coordinated species.
observed in the spectrum of C o 3 0 4 , a mixed oxide with cobalt in both tetrahedral and
octahedral coordination [6]. UV-Vis spectra very similar to that of Fig. 4 have been reported
for g-oxo-Co species present in the channels of Co-exchanged Beta zeolites with high Co
loading [ 11 ]. It is obvious that the exact nature of Co species present in SiMCM-2 cannot be
determined from UV-Vis spectra only. However, some of the bands clearly indicate that Co
species are partially oxidized and most likely exist as cobalt oxide or hydroxide clusters.
TEM pictures of the totally exchanged SiMCM-2 sample show highly ordered SiMCM
with an additional phase located exclusively at the outer surface of the particles (Fig. 4). This
phase, which is not present on pictures of the parent SiMCM compound, is,not homogeneous
in size and shape and contains relatively small particles (< 30 nm diameter) together with
elongated thin films. EDX analysis gives a very high Co concentration in surface species,
contrasting with the bulk of SiMCM particles, in which Co was not detected. Moreover, the
Co:O ratio was found to be close to 1:2 over a series of needles, in agreement with a CoO or
Co(OH)2 stoichiometry. The absence of Co in the mesopores is in disagreement with an ion
exchange between CTMA § and Co 2+ cations. As we could not find any evidence for the
presence of organic or inorganic cations in the pores, it is reasonable to assume that {-SiO-
,CTMA +} sites are transformed to SiOH groups upon treatment. The powder diffraction
pattern of SiMCM-3 shows a series of diffraction peaks between 20 and 60 ~ (20) (Fig. 5).
t,-'
I:
I I I I
20 30 40 50 60
2 Theta (o)
Fig. 5. High angle part of XRD patterns of SiMCM-2 (bottom) and SiMCM-2C (top)
These peaks, which do not exist in the pattern of the parent pure-silica material, can be
unambiguously assigned to the Co-containing phase, and reveal the crystalline nature of the
later. Unfortunately, the reflection peaks did not correspond to any reported Co-containing
compound. However, the blue color of the solid along with the instability of the crystals under
the electron beam are in favor of a hydroxylated compound. Hydroxylation necessitates the
presence of water (the alcohol solution contains approx. 4 wt.% H20) and basic species,
which may be the SiO- defect sites on the surface of the mesopores. Under such conditions,
the removal of CTMA + cations and the formation of Co-containing phases could result from
the following reactions:
4. CONCLUSIONS
Surfactant molecules can be completely removed from the pores of two different MCM-41
materials using alcoholic solutions of Cobalt (II) chloride. For A1MCM, a mesoporous
aluminosilicate with Si/A1 molar ratio = 1.25, Na + ions introduced during synthesis can also
be exchanged, leading to [Co]A1MCM-41 compounds with very high Co contents. The ion
exchange process does not modify the structural and textural properties of the parent solid, in
particular the long range ordering of the mesopores. Co species are highly dispersed in the
mesopores and they remain isolated, even after calcination at high temperature. Co-exchanged
low-silica MCM-41 with such a high Co loading has never been previously reported and can
be considered as the amorphous analog of CoX zeolites.
The behavior of the pure silica SiMCM compound is totally different. Again, CTMA +
cations can be exchanged for Co 2+, but the latter do not remain in the mesopores, as evidenced
by microscopy. Actually, SiO defect groups, present in the pores of SiMCM, are transformed
into silanol groups SiOH in the presence of water and Co species move at the outer surface of
silica particles. They form a blue solid, most likely a Co hydroxide-like compound, whose
structure could not be completely determined. This compound transforms to highly crystalline
Co304 particles upon calcination in air at high temperature.
ACKNOWLEDGMENT
N.L. thanks the French Ministry of Industry for funding within the ZEOMAT project.
REFERENCES
[1] Y. Li, T.L. Slager and J.N. Armor, J. Catal., 150 (1994) 388.
[2] Y. Li and J.M. Armor, J. Chem. Soc., Chem. Commun., (1997) 1013.
[3] A. Jentys, N.H. Pham, H. Vinek, M. English and J.A. Lercher, Microporous Mater., 6 (1996) 13.
[4] B. Echchahed, A. Badiei, F. B61and and L. Bonneviot in Mesoporous Molecular Sieves 1998, L.
Bonneviot, F. B61and, C. Danumah, S. Giasson and S. Kaliaguine (Eds.), Stud. Surf. Sci. Catal.,
117(1998) 559.
[5] V. Parvulescu and B.L. Su, Catal. Today, 69 (2001) 315.
[6] S. Lim, D. Ciuparu, C. Pak, F. Dobek, Y. Chen, D. Harding, L. Pfefferle and G. Haller, J. Phys.
Chem. B, 107 (2003) 11048.
[7] N. Lang and A. Tuel, Chem. Mater., 16 (2004) 2969.
[8] A. Bourlinos, M.A. Karakassides and D. Petridis, J. Phys. Chem. B, 104 (2000) 4375.
[9] Y. Wang, Y. Ohishi, T. Shishido, Q. Zhang, W. Yang, Q. Guo, H. Wan and K. Takehira, J.
Catal., 220 (2003) 347.
[10] N. Lang and A. Tuel, Chem. Mater., 16 (2004) 2969.
[1 I] J. Dedecek, L. Capek, D. Kaucky, Z. Sobalik and B. Wichterlova, J. Catal., 211 (2002) 198.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 447
Faculty of Chemistry, Nicolaus Copernicus University, 7 Gagarin St., 87-100 Torun, Poland
1. INTRODUCTION
Recently, remarkable progress has been made in the synthesis of carbons with periodic porous
structures. This progress has been achieved using templates, such as zeolites [1], ordered
mesoporous silicas [2-4], and silica colloidal crystals [5-7], for the synthesis of ordered
microporous, mesoporous, and macroporous carbons, respectively. The mesoporous carbons
can be utilised as, e.g., catalyst supports, adsorbents, sensors, electrode materials, and
electrochemical double-layer capacitors [2,3,8]. One of the new mesoporous carbon materials,
designated as CMK-1, was synthesised by use of furfuryl alcohol, sucrose, or phenol resin, all
as the carbon source introduced into the cubic mesoporous silica molecular sieve MCM-48
used as the template [2,3].
Due to a developed structure and three-dimensional system of pores, the MCM-48
material appears to be especially interesting precursor for production of carbons with a
periodic arrangement of pores. Usually, the MCM-48 molecular sieves contain relatively
homogeneous pores. Therefore, their applicability as templates for preparation of carbons
considerably differing in the pore diameter, wall thickness, etc., is limited. However,
modification of MCM-48 by introduction of A1 atoms may change its structure and, in
addition, generate acidic centres. Lately, the Al-grafted MCM-48 materials varying in the A1
content and, consequently, in sorption and catalytic properties have been obtained [9]. In the
present work, these materials were used as templates for synthesis of micro-mesoporous
carbons from furfuryl alcohol. Acid sites on the surface of siliceous MCM-48 are necessary
since the polymerisation of furfuryl alcohol is an acid-catalysed process. Introduction of A1 by
grafting was chosen in order to make sure that the A1 atoms were placed on the material
surface and not inside the walls. The latter might easily happen during a direct (hydrothermal)
synthesis [10]. Moreover, as follows from the literature [11 ], when the Si/A1 ratio is below 5,
A1-MCM-41 with the pore system typical of this mesoporous sieve could not be obtained by
means of the direct method whereas the process of grafting was successful.
The aim of this work was to prepare carbons with a developed mesoporous structure and
considerably limited fraction of micropores, using the A1-MCM-48 materials as templates.
Nonpolar (nitrogen, benzene) and polar (ethanol, water) compounds were chosen as probe
molecules for adsorption to learn more about properties of these novel preparations.
448
2. EXPERIMENTAL
The Al-grafted MCM-48 materials were obtained using aluminium isopropoxide dissolved in
n-hexane and aluminium tri-sec-butoxide dissolved in n-butanol (see Table 1). Si/A1 molar
ratios of the reagents were 5 and 32 in the case of isopropoxide (samples I and II) and 5 for
tri-sec-butoxide (sample III). The synthesis details were given elsewhere [12]. The samples of
A1-MCM-48 were exposed to vapours of furfuryl alcohol at room temperature until saturation
was attained. Then, they were heated consecutively at 353 K for 20 h and at 413 K for 22 h.
Next, the samples were subjected to carbonisation at 873 K for 2-3 h and the aluminosilicate
templates were dissolved with use of an aqueous solution of HF at room temperature. Finally,
the carbonaceous products were washed with water and dried. They are designated as C 1, C2,
and C3 and correspond to samples I, II, and III, respectively.
The fractions of Si and A1 in the templates and carbons were determined with the atomic
absorption spectroscopy (AAS) and the C/H ratios in carbons, with the elemental analysis.
X-ray diffraction (XRD) powder patterns were recorded on a Siemens D 5000 diffracto-
meter with use of the Ni-filtered CuK~ radiation. Transmission electron microscopy (TEM)
micrographs were obtained with a JEM 1200 EX (JEOL) instrument operating at 80 kV.
Isotherms of nitrogen ad/desorption were measured at 77 K with a Micromeritics ASAP 2010
volumetric adsorption analyser. Adsorption isotherms of benzene, ethanol, and water were
determined at 298.2 K, using vacuum devices equipped with McBain balances and MKS
Baratron gauges.
Table 1
Conditions for preparation of aluminosilicate templates a
Sample Aluminium compound Solvent Si/A1 of
no. reagents product b
I isopropoxide n-hexane 32 34.5
II isopropoxide n-hexane 5 3.8
III tri-sec-butoxide n-butanol 5 6.9
aSamples were obtained by grafting Al onto siliceous MCM-48 under the specified conditions.
bValues determined with AAS
Grafting of A1 onto MCM-48 was performed using aluminium compounds, solvents, and
Si/A1 ratios of the reagents as specified in Table 1. The objective was to obtain materials that
would differ in acidity and in the ratio of wall thickness to pore diameter. As reported earlier
[9], the concentrations of the Bronsted and Lewis acid sites were, correspondingly, 32 and
150 lamol g-~ for sample I and 130 and 442 ~tmol g-1 for sample III. Sample II demonstrated
acidity similar to that of sample III (110 and 430 pmol g-l, respectively) but its adsorption
abilities were lower (cf. Table 2). Only up to ca. 60% of all grafted A1 atoms formed acid
centres. The remaining A1 atoms were presumably located inside agglomerates the amount of
which increased with the total A1 content of the samples.
The parent and Al-grafted MCM-48 samples exhibited the XRD powder patterns
characteristic of this molecular sieve [9]. The unit cell parameters, calculated from the (211)
interplanar spacings, are listed in Table 2. The table contains also the data obtained from the
isotherms of the nitrogen adsorption and desorption [ 13]. The values of ao, SBET, St, Vp, Vt, and
d~ generally decreased with the increasing A1 content of the samples. The data for sample III
449
somewhat deviated from the observed trend. This sample exhibited relatively large surface
area and volume of pores, in spite of a rather high content of A1 (Si/A1 ratio = 6.9). In this
case, the A1 grafting caused an increase in the surface area of primary mesopores with respect
to the parent sample. The pore wall thickness was calculated with the formula used by
Ravikovitch et al. [14]. The pore wall density was obtained as a weighted average of densities
of amorphous aluminium oxide (2.55 g c m -3) and silicon oxide (2.2 g cm-3). When relatively
small amounts of A1 were grafted onto MCM-48, the pore wall thickness increased rather
insignificantly (see Table 2). Only when Si/A1 was less than 4, then the increase in the wall
thickness became clear.
Table 2
Structural parameters of the MCM-48 materials
Parameter Sample a
0b I II III
ao [nm] 9.10 8.33 8.14 8.38
SsEr [m 2 g-l] 1315 1245 1051 1280
St [m 2 g-l] 1294 1213 1010 1223
Sext [m 2 g-l] 284 259 154 149
Sp [m 2 g-l] 1010 954 856 1074
Vp [cm 3 g-l] 0.718 0.662 0.538 0.686
Vt [cm 3 g-l] 0.958 0.884 0.713 0.841
dav [nm] 2.41 2.34 2.20 2.04
we [nm] 1.14 1.09 1.17 1.06
AWd [rim] 0 0.05 0.15 0.01
"~Sample numbers as in Table 1. bSiliceous MCM-48. ao is the unit cell parameter, SBE~'- the BET
specific surface area, St - total surface area, Sex~- external surface area, Sp - surface area of primary
mesopores, Vp - volume of primary mesopores, Vt - total pore volume, and day - average pore
diameter obtained from the maximum of the pore size distribution, wd = ao [1 -pVp/(1 + pVp)]/3.0919
is the pore wall thickness, p being the pore wall density, whereas Awd = (Wd- Wdref) where Wdref is the
wall thickness if ao would decrease upon incorporation of AI with the same ao/wa ratio as that for the
parent MCM-48. S, Sext, and Vp were obtained with use of the high-resolution as-plot method [ 15] and
LiChrospher Si-1000 macroporous silica gel [16] as a reference adsorbent. Sp is equal to the difference
St--Sex, Vt was derived from a single-point adsorption, taken as the last experimental point of an
isotherm.
It had been assumed that the structure of an aluminosilicate template would relatively
strictly determine the structure of a carbonaceous product. The structural parameters of the
carbons prepared in this work are listed in Table 3. The XRD powder patterns of these
carbons are similar to those of the CMK-1 carbons made by Ryoo et al. from sucrose and also
with MCM-48 as the template [2,3]. According to these authors, dissolution of the template is
accompanied by a structural transformation of carbon: two disconnected, though interwoven,
parts of the carbon framework formed in two channel systems of MCM-48 undergo mutual
displacement after removal of the template, create some contact between them, and fix this
position. This structure transformation has been reflected in a new peak (110) at 20 = 1.6 ~ in
the XRD pattern (Fig. 1), indicating a cubic symmetry. Intensity of the other peak (20 ~ 2.7 ~
depends on the ordering level of a sample. The peak with the highest intensity is accompanied
by another peak (20 ~ 3.15~ which is observed for carbon C2. This clear structure ordering
has been confirmed by the TEM micrograph (Fig. 2). The picture of the ordering of the C2
450
L.:I,
. .
,.__,
I-- C1
%---____
(D
2 3 4 5 6 7 8 9 10
2 0 [deg]
Fig. 1. XRD powder patterns of studied carbons Fig. 2. TEM micrograph of carbon C2.
(cf. Table 3).
carbon appears to be similar to those of CMK- 1 [ 1,2]. Contrary to this sample, the micrograph
of the C 1 carbon shows no ordered structure, which agrees with the relatively low intensity of
the XRD peak at 20 = 2.77 ~ (Fig. 1). The structure transformation mentioned above is
reflected also in shrinking of the carbon materials, evidenced by a decrease in their unit cell
parameter by 4.4-7.5% in relation to that of the aluminosilicate templates (Tables 2 and 3).
Considering the templates obtained from the same source of A1 (samples I and II) on the
one hand and their derivatives (carbons C1 and C2) on the other hand, one can state that
template with a smaller pore wall thickness (wa) and larger pore volume (Vp and G) yield a
carbon with a larger average pore diameter (da~) and, presumably, smaller pore wall thickness
(cf. Tables 2 and 3). However, if the carbon pore walls would be too thin, then the carbon
structure ordering could be lost due to a break of continuity of the walls. The pore walls of
carbon C 1, being thinner than those of C2, are probably thinner than those of C3 as well since
the latter exhibits a larger unit cell, although average pore diameters of C 1 and C3 are similar.
The chemical compositions of carbons C1, C2, and C3 are similar, which is reflected in
comparable values of the C/H ratios and trace amounts of Si and A1 (Table 3).
The data for nitrogen ad/desorption were used to characterise porosity of the studied
templates and carbons (Tables 2 and 3). According to the IUPAC recommendation, pores of a
solid are classified as micropores when smaller than 2 nm, as mesopores when between 2 and
50 nm, and as macropores when larger than 50 nm. The BET specific surface area was
calculated using the standard Brunauer-Emmett-Teller method. Other adsorption parameters
were mostly obtained with both the high-resolution as-plot method [15] and the data for a
relevant reference adsorbent. The results indicate that carbons are weaker adsorbents as
compared to the corresponding templates. For example, the surface areas of the carbons are
below 900 m 2 g-i whereas those of the templates, above 1000 m 2 g-1. On the other hand, it has
been found that micropores (a fraction of ca. 25% with respect to the total pore volume) occur
apart from the mesopores (Table 3). C2 and C3 exhibit well-ordered structures but their
surface areas are somewhat less than those reported in the literature. In order to improve that
parameter, other molecular sieves might be considered as templates for producing carbons.
451
Table 3
Structural parameters of carbons
Parameter Carbon a
C1 C2 C3
ao [nm] 7.59 7.88 7.79
SseT[m 2 g-l] 857 586 542
St [m 2 g-l] 448 364 330
Sext [m 2 g-l] 24 54 45
Sp [m 2 g-l] 424 310 285
Vp [cm 3 g-l] 0.425 0.257 0.247
Vmi [cm 3 g-l] 0.140 0.070 0.084
day [nm] 2.10 1.91 2.07
C/H 2.50 2.65 2.55
Jsi [wt. %] 0.07 0.25 0.24
j~t [wt. %] 0.03 0.01 0.02
aCarbons C 1, C2, and C3 were obtained with use of samples I, II, and III, respectively, ao, SBET,St, Sext,
Sp, Vp, and day denote the same as in Table 2, Vmi is the volume of micropores, C/H - molar ratio of
carbon to hydrogen,3~i and j % - weight fractions of silicon and aluminium, respectively. St, Sex, Vp, and
Vm, were obtained using the high-resolution as-plot method [15] and Cabot BP 280 nongraphitised
carbon black [ 17] as a reference adsorbent.
16
A A
14
$ ~ . o ' o *o
,--. ~
~:~10 ID ~
g 8
V A
~ V
4 V []
E~Z~Z D V D [] [] A []
2 A
A, A A A A A A
o~ I ' I ' I '
14
A
12 a
g
}~ a V V
@ v
m v A
4 V
V []
v V [] [] []
2 [][3 [] [] [] a A[] A
12
a
10 a
AQ
~ 8 oO o oO o ~ o 9 o o o ~oO
-~ o ~o~OO~ ~176~
g6 8
~. zx
~ , v v
a V
~ 4 v
V
V []
i v zx
2'~
D 'E]~j# []
[]
A a
A ID a
[] [] []
0--ii" ~A A A
niu, ~-~A ' I ' l , " t . . . . . . ' t '
Fig. 5. Isotherms of adsorption of nitrogen (O), benzene (C]), ethanol ( V ) , and water ( A ) on carbon
C3. Full circles represent desorption of nitrogen.
453
The shape of adsorption isotherms of nitrogen (Figs. 3-5) is of that between types I and II
of the IUPAC classification. In addition, each of the isotherms shows an ad/desorption
hysteresis loop in the p/ps range above 0.1. This indicates the presence of secondary
mesopores the diameters of which cover a wide range. The secondary mesopores are probably
formed, among other things, by void space between adjoining crystallites of the material. The
wide range of sizes of these pores suggests that the crystallites are relatively small.
The adsorption isotherms of benzene and ethanol (Figs. 3-5) are also of a type between I
and II. It suggests that adsorption of both compounds as well as of nitrogen proceeds via
formation of multilayers at low relative pressures, followed by capillary condensation. The
adsorption processes are similar for these compounds, despite their different physicochemical
properties and adsorbent-adsorbate interactions.
The shape of adsorption isotherms of water (Figs. 3-5) is of that between types III and IV.
It indicates that the surface of the carbon samples is hydrophobic. The polar water molecules
are strongly bonded at the oxygen complexes present on the pore surface. That strong bonding
and subsequent association of water molecules lead to formation of clusters of liquid-like
water around the oxygen centres and to clogging the pores. Thus, the diffusion of water is
hindered and, consequently, water fills the pores only partially, i.e., only the pore sections
near the oxygen centres. Obviously, the clogging effect should increase with a growing
number of the centres and/or with a decreasing distance between them as the water clusters
could grow larger and be anchored on more oxygen centres commonly. Such a picture agrees
well with the observation that the sorption capacities for nitrogen and benzene are higher than
those for water (Table 4).
Table 4
Sorption capacities at p/ps = 0.9
Carbon
Adsorbate C1 C2 C3 C1 C2 C3
cm 3 g-i mol kg -1
nitrogen 0.453 0.386 0.290 15.83 11.13 8.29
benzene 0.409 0.262 0.229 4.58 2.93 2.57
ethanol 0.354 0.302 0.253 5.96 5.08 4.28
water 0.283 0.222 0.193 15.43 11.83 10.25
4. CONCLUSIONS
Active carbons of a developed mesoporous structure with a fraction (ca. 25%) of micropores
can be prepared by polymerisation and subsequent carbonisation of furfuryl alcohol, using
mesoporous A1-MCM-48 molecular sieves as templates that determine or control the carbon
structure.
The obtained carbons are formed from two parts of the framework, corresponding to two
channel systems of the template. These two parts undergo transformation and displacement,
which enables a mutual contact.
The adsorption isotherms of nitrogen, benzene and ethanol on the carbons are of a type
between I and II of the IUPAC classification. However, contrary to the isotherms for the A1-
MCM-48 templates, no step corresponding to adsorbate condensation in pores is observed.
454
The adsorption of water is determined by the presence of oxygen centres, being the
primary adsorption sites. The isotherms are of a type between III and IV, indicating formation
of water clusters at the beginning of the adsorption process.
The carbons exhibit sieve properties in relation to the applied adsorbates.
ACKNOWLEDGEMENT
The work was supported in part by the State Committee for Scientific Research (KBN) with a
grant no. 4 T09A 147 23.
REFERENCES
aInstituto Mexicano del Petr61eo. Eje Central L~zaro C~rdenas 152, Col. San Bartolo
Atepehuacan, C.P. 07730, M6xico, D. F. E-mail:*mguzman@imp.mx
A series of silica mesoporous materials was prepared adding ammonium sulfate into reaction
mixture. The 29Si MAS NMR showed that the Q4signal increased when the sulfate content in
the synthesis mixture increased thus suggesting that the OH/Si of the material decrease hence
confirming that silica had a higher degree of condensation. Accordingly, an increase in the
unit cell parameters and in the wall thickness of the synthesized materials was observed.
1. INTRODUCTION
A self-assembly process has been used for obtaining ordered mesoporous silicate and
alumino-silicate materials, under either acidic or basic conditions [1-4]. The electrical charges
matching of the polymerizing inorganic framework [3, 4] by the surfactant assembly drives
the process.
The main drawback of these materials to be used as adsorbents or supports of catalyst is
the lack of sufficient thermal and hydrothermal stability. It has been shown that the stability
can be improved by increasing the hydophobicity [5] or by decreasing the surface density of
the silanol groups during the formation of the structure [6]. An increase of the wall thickness
results from this stabilizing process [7, 8].
Therefore a simple way to improve the stability of MCM-41 materials is through the
synthesis procedure [9-11] but also by means of post-synthesis modifications [5, 13]. For
instance, it was observed that an electrolyte such as NaC1 added to the gel [12] favored the
condensation rate of silica: the higher the electrolyte concentration, the higher the degree of
condensation as it was previously reported [14]. Igarashi et al. [15] studied the nature of the
silica source on the stability and the effect of hydrolysis. They found that the ratio of the wall
thickness to the amount of uncondensed silanol ratio showed a "strong" correlation with the
stability towards hydrothermal treatments.
This work aims to contribute to the important topic of silica condensation mechanism in
the course of the synthesis of mesoporous materials by showing the effect of introducing
(NH4)2SO4, into the synthesis medium. 29Si MAS NMR spectroscopy was used to study the
MCM-41 samples prepared by hydrothermal synthesis [16].
456
2. E X P E R I M E N T A L P R O C E D U R E
2.1. Synthesis
The gel mixture was prepared using TEOS (Aldrich), NH4OH (Baker), (NH4)2SO4 (Baker)
and CTMABr (Sigma) in the following molar ratios: OH-/SiO2 = 10.99, H20/SiO2 = 67.8,
CTMA/SiO2 - 0.304. The amount of (NH4)2SO4 added was expressed as SO4/Si molar ratio.
It was varied between zero and 1 The gel was heated at 120~ during 24 h. The material was
then filtered, washed to neutrality and dried at 100~ The organic material was bum-off at
550~ for 6h. It was further realized that a fraction of surfactant was displaced by (NH4)2SO4,
and washed out. Therefore, the formal initial surfactant content must be distinguished from
that existing at the end of the preparation, before the final calcination.
2.2. Characterization
XRD patterns were obtained with a D-500 SIEMENS diffractometer equipped with a
graphite secondary beam monochromator. The scanning from 1.5 to 10~ 20 degrees was
carried out with 0.02 ~ and 2s step time.
29Si MAS NMR spectra were recorded on a Bruker 400 MHz spectrometer at 79.49 MHz
with the following conditions: spinning rate, 5 kHz, 90 ~ pulses with ls intervals and 4 mm
zirconia rotors.
The surfactant content was obtained from the weight loss accompanying desorption
endotherm peak of the template.
It was previously reported [16] that adding sulfate within the gel mixture of MCM-41
materials, prior to the hydrothermal treatment, yielded solids with thicker walls. While the
XRD spectra of the as-synthesized samples showed the hexagonal array typical of MCM-41
materials, the intensities of XRD reflections and the resolution of higher order peaks
decreased as the sulfate content increased. The structure withstood calcination at 550~ (Fig.
1a) as long as SO4/Si < 0.3. Moreover, at lower sulfate contents, calcination step increased the
structural order. The dll0 and d200 reflections were easily identified, suggesting a long-range
ordering of the MCM-41 hexagonal framework. The reflection intensity as well as the
increase of the cell parameters with calcination suggested that a restructuration process took
place.
The wall thickness (t) is calculated for all samples, by the methode reported by Galarneau
et al [12] in spite of the poor XRD resolution at SO4/Si > 0.3. Mesostructures like HMS
(hexagonal mesoporous silica) afford an important subset of hexagonal mesoporous molecular
sieves [17]. These materials differ significantly from those obtained by the electrostatic
assembly pathways by thicker walls (2-3 nm compared with ca. 1 nm for MCM-41), while the
small angle X-ray scattering domain is less perfect as evidenced by the absence of defined
dl l0 and d200 reflections. The hexagonal ordering is poor compared to that usually observed in
MCM-41.
In the present study, samples prepared with SO4/Si ratios higher than 0.3 - 0.5 showed a
single dl00 reflection peak (Fig. l a), which became broader and less intense, as in HMS. At
still higher sulfate contents (> 0.5), amorphous products were obtained.
29Si MAS NMR spectroscopy was used to obtain information on the effect of ammonium
sulfate on the silica polymerization process. The results are shown in Fig. l b and Table 1.
457
(a) (b) Q4
Q3
o0
0.01----" ,'f~'/J~'~ X --
Or)
I-"
(D
i'-
. , .
2 4 6 8 ;o -;o ' -~ ' -1;o ' -1'1o ' -~,o ' -1;o
20 (Degree) a(ppm)
Fig. 1. a) X-ray powder diffraction patterns of calcined samples at 550~ and b) 29Si NMR MAS
spectra of calcined samples.
The normalized areas of the Gaussians functions used in the deconvolution 29Si MAS
NMR spectra are expressed as percentages of the band total area.
In agreement with Engelhardt and Michel [ 18], the peak assignment is as follows:
- (Si(Si-O)4 (Q4)at -108 to -109 ppm
- (Si-O)3Si-OH (Q3) at about -100ppm
- and (Si-O)2Si-(OH)2 (Q2) geminal silanols ~ -90 ppm
The relative concentration of silanols can be derived from the following equation [15]:
SiOH(mol%Si) - Q2QZQz + Q 4
+Q~ / x100%
A simple estimate of the degree of condensation of the silicate network can be drawn from
the relative value o f Q4. Table 1 shows that Q4increased (hence the OH/Si groups decreased)
with the concentration of ammonium sulfate in the initial gel. It is also interesting to observe
that the wall thickness follows the same trend. However, while this parameter changes
steadily, the variation o f Q4 was less pronounced when SO4/Si was higher than 0.1. The
decrease in the OH/Si indicates a lower hydrophobicity of the material. In fact, if the surface
area is considered a parallel increase of the OH density measured by the weight loss above
360~ and by 29Si NMR is observed (Table 1).
Previous results [16] showed that the weight loss of these samples between 25-125~
decreased from 5.3 %, for SO4/Si = 0 to 3.3 % for SO4/Si = 0.5 thus suggesting that the
materials have lower water content for higher SO4 concentrations. Fig. 2 shows that when the
ratio sulfate/surfactant increases in the initial mixture, both Q4 and the wall thickness
parameters increase.
458
Table 1
Qn: 298i NMR spectral deconvolution (Gaussians), OH/Si' silanol groups content per Si
calculated from [ 15]. OH,a: OH surface density calculated from 298i NMR, OH,b" OH surface
density calculated from the weight loss above 360~ ..........
Q2 QS Q4 OH/Si OH,a OH,b
SO4/Si* SO4/Surf'* % % % nm 2 nm 2
0 0 3.7 33 64 40.0 5.8 5
*Mixture composition.
If we accept that there is a competition between NH4 + and CTAB + for the negatively
charged silica network, the net effect of the addition of ammonium sulfate would be similar to
that obtained by a reduction of the effective concentration of the surfactant in the reacting
mixture. The consequence is observed in Fig. 3. When the ratio sulfate/surfactant increases in
the initial mixture, the pore volume and the surface area decrease (Table 2). As shown in
Table 1, both the surface density, OH/nm 2 a and b, increase because of the marked decrease
of the surface area.
The competition between NH4 + and CTAB + may find its origin in a Donnan equilibrium
[ 19], between the solutions outside and inside the mesoporous solid. The solid could play the
role of a membrane impermeable to an anionic species, for instance the divalent sulfate anion.
This classic concept could explain the effect of sulfate ion, or of any 'foreign' ion.
Table 2
A" BET surface area, Vmeso" Mesoporous volume and t: wall thickness, measured for the
samples calcined at 550~ and Mol" content in surfactant in the mesoporous material,
(molecule surfactant per nm2)
A Vmeso t Mol
SO4/Si* SO4/Surf'*
mZ/g cc/g nm Surf/nm2
72 O 3.0~,
C/68 2.0~
O O
66 0 Q4
O 1.5~
64 9 t
1 ' I I '
12 ' I '
1.0
.0 0.3 0.6 0'.9 1. 1.5 1.8
SO4/CTAB
Fig. 2. Q4 and wall thickness (t) variations with respect to SO4/CTAB ratio
! 750
0.6
[] Vmeso
eo II [] A
o 0.5- 600 t~
~D
[--1
E m t~
0.4 Ki ~,,L~
0 450
> m
m 0.3-
0
0 -300 >
0 0.2-
to
0.I ' i ' I ' ' I ' I ' 150 ~
0'.0 0.3 0.6 019 1.2 1.5 18
SO4/CTAB
Fig. 3. Variations of the specifique surface area and the mesoporous volume with the SO4/CTAB ratio
460
A simple calculation shows that the amount of physically adsorbed water molecules per
nm 2 is about one per silanol for all samples. Thus the main effect of the sulfate addition is to
increase the silicate condensation as anticipated by the title of this paper. The structural
stability is affected by the swelling of the walls by the increasing silanol content beyond a
limit SO4/CTAB = 0.33, or a wall thickness about 1.5 nm.
4. CONCLUSION
According to the 29Si MAS NMR study, the Q4 signal increases and the surface density in
silanol groups increases with increasing concentration of ammonium sulfate in the synthesis
mixture.
The variation of Q4 is less pronounced when the SO4/Si becomes higher than 0.1. The
hexagonal ordering disappears above this limit as evidenced by the absence of dll0 and d200
XRD reflections. On the opposite, lower sulfate content (0.01) has a beneficial effect since the
structural order increased upon calcinations.
ACKNOWLEDGMENT
We acknowledge the financial support of the Maya Crude Oil Research Program (IMP)
though projects D.00145 and D.00817
REFERENCES
[11 C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.S Beck, Nature 359, (1992) 710.
[2] S.S. Kim, W.Zhang, T.J. Pinnavaia, Science 282, (1998) 1302.
[3] Q. Huo, D.I. Margolese, G.D. Stucky, Chem. Mater. 8, (1996) 1147.
[4] A. Firouzi, D. Kumar, L.M. Bull, T. Besier, P. Sieger, Q. Huo, S.A Walker, J.A. Zasagzinski,
C. Glinka, J. Nicol, D.I. Margolese, G.D., Stucky, B.F., Chmelka, Science. 267, (1995) 1138.
[5] K.A. Koyano, T. Tatsumi, Y. Tanaka, S. Nakata, J. Phys. Chem. B, 101, (1997) 9436.
[6] D. Das, C.-M. Tsai, S. Cheng, J. Chem. Soc., Chem. Commun., (1999) 473.
[71 N. Coustel, F. Di Renzo, F. Fajula, J. Chem. Soc., Chem. Commun., (1994) 967.
[8] R. Mokaya, W. Zhou, W. Jones, J. Chem. Soc., Chem. Commun., (1999) 51.
[9] R. Mokaya, J. Phys. Chem. B, 103, (1999) 10204
[10] P. P6rez-Romo, M.L. Guzmfin-Castillo, H. Armendfiriz-Herrera, J. Navarrete, D.R. Acosta and
J. Ascencidn Montoya, Langmuir, 19, (2003) 3446.
[11] R. Mokaya, J. Phys. Chem. B, 104, (2000) 8279
[12] A. Galarneau, D. Desplantier, R. Dutatre, F. Di Renzo, Microporous Mesoporous Mater., 27,
(1999) 297
[13] Q-H. Xia, K. Hedajat, S. Kawi, Mat. Lett. 42, (2000) 102
[14] R.K. Iler, The Chemistry of Silica, Wiley, N.Y. (1979).
[15] N. Igarashi, K. A. Koyaono, Y. Tanaka, Y. Nakata, K. Hashimoto, T. Tatsumi, Microporous
Mesoporous Mater., 59, (2003) 43.
[16] M.L. Guzmfin-Castillo, H. Armendfiriz-Herrera, A. Tob6n-Cervantes, D.R. Acosta, P. Salas-
Castillo, A. Montoya de la F. and A. Vfizquez-Rodriguez, Stud. Surf. Sci Catal. 142, (2002)
1039.
[17] P.T. Tanev, T.J. Pinavaia, Science, 267, (1995) 865
[18] G. Engelhardt, D. Michel, High-Resolution Solid-State NMR of Silicates and Zeolites. J. Wiley
and Sons, (1987).
[ 19] G. Kortum, J.O.M. Bockris, Textbook of Electrochemistry. Elsevier, (1951 ).
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 461
1. INTRODUCTION
Periodic mesoporous silicas with a cage-like pore structure such as SBA-2, SBA-12, FDU-1,
SBA-16 and SBA-1 are a rapidly growing class of surfactant-templated inorganic materials
[1]. Their unique pore topology consists of a periodic array of mesosized cages (3-10 nm)
which are 3D-interconnected by smaller windows. The size of the cages and windows differs
for each type of mesoporous silica and can be tailored by properly varying the synthesis
conditions. Jaroniec et al. showed that the diameter of the interconnecting channels can be
adjusted in the range of 1 to 6 nm and 1.3 to 2.4 nm for SBA-16 [2] and FDU-1 [lc]
materials, respectively. Wright et al. demonstrated by sorption experiments with differently
sized hydrocarbons that the pore openings of SBA-2 are in the molecular range [3]. Therefore
it can be anticipated that cage-like mesoporous silicas will feature promising candidates for
size and shape selective high-surface materials and, hence, for prominent applications in
catalysis and materials science.
SBA-1 was for the first time synthesized by Stucky and coworkers in 1995 [la]. The
exact pore configuration was determined by electron crystallography [4]. The porous structure
of cubic SBA-1 (space group Pm3n) displays two differently sized cages A and B. A-cages
are slightly larger than B-cages and the ratio of the corresponding cage diameters is
approximately 1.1. The entire structure is of A3B-type. The absolute cage diameters are
estimated to be ca. 3-4 nm depending on the synthesis conditions. The size of the
interconnecting windows is <1.5 nm.
Many applications of periodic mesoporous silicas require a post-synthetic
functionalization of the materials surface. For example, silica grafting via organometallics
(metal alkyls, amides, and alkoxides) can be efficiently exploited for the generation of highly
dispersed catalytically active metal surface sites [5]. Such surface-metallated silicas can be
applied for a wide range of catalytic reactions, e.g., Diels-Alder reactions [6], Meerwein-
462
Ponndorf reductions [7], olefin metathesis [8], epoxidations [9], and olefin polymerization
[ 10]. For most of these examples MCM-41 and SBA-15 has been employed as a mesoporous
support, while SBA-1 has been scarcely investigated in this respect. Hartmann et. al. [11] and
Tatsumi et al. [12] reported on the incorporation of metal cations into the SBA-1 mesoporous
structure by adding metal salts or alkoxides to the synthesis gel. V-/Co- or Mo-SBA-1
materials were synthesized according to this procedure. Ryoo et al. used a salt impregnation
technique to synthesize Zn- and A1-SBA-1 [13]. To our knowledge there is no report on the
grafting of organometallics onto SBA-1 mesoporous silica.
2. E X P E R I M E N T A L SECTION
2.1. Synthesis
Materials. Tetraethyl orthosilicate (TEOS, Fluka) was used as a silica source for the SBA-
1 materials. Hexadecylbromide, octadecylbromide, triethylamine (for surfactant synthesis),
dibutylmagnesium (1 M in heptane), anhydrous YC13, and butyllithium (2.5 M in hexane)
were purchased from Aldrich. 1,1,3,3-tetramethyldisilazane and dimethylchlorosilane were
obtained from ABCR and were used as received. The magnesium and yttrium complexes as
well as the silylated and metallated materials were synthesized and manipulated with rigorous
exclusion of air and water, using high-vacuum and glovebox techniques (MB Braun
MB 150B-G-II; O2 < 1 ppm, H20 < 1 ppm).
Surfactant preparation. Hexadecyltriethylammoniumbromide (Cl6TEABr) and octa-
decyltriethylammoniumbromide (C~8TEABr) were synthesized by the reaction of
hexadecylbromide and octadecylbromide with triethylamine, respectively. For example, 25.0
g of 1-bromooctadecane (75.0 mmol) and 11.4 g (112.5 mmol) of triethylamine were
desolved in 200 mL acetonitrile. The resulting mixture was heated for 24 h under reflux
conditions. While the solution was cooled to ambient temperature the solid product octadecyl-
triethylammoniumbromide precipitated. The mixture was cooled in an ice bath in order to
increase crystallization. The resultant precipitate was filtered, recrystallized from ethyl
acetate, and dried under vacuum.
Synthesis of SBA-1 materials. Two SBA-1 materials with different pore sizes were
prepared according to a literature procedure by Ryoo et al. [ 14], using TEOS as a silica source
and C16TEABr (sample SBA-I-a) or ClsTEABr (sample SBA-I-b) as the surfactants in an
aqueous solution of HC1. The molar composition of the synthesis gel was 1 surfactant : 5
TEOS : 244 HC1 : 3590 H20. The detailed synthesis procedure is as follows: Surfactant
(C18TEABr: 5.00 g, 11.47 mmol or Cl6TEABr: 4.68 g, 11.47 mmol), conc. HC1 (37%, 318 g,
2.79 mol), and distilled water (525 g, 29.17 mol) were combined and the resulting mixture
was vigorously stirred until a homogeneous solution formed. The solution was cooled to 273
K and 12.00 g (57.60 mmol) of TEOS slowly added. Stirring was continued for 4 h, then the
reaction mixture heated from 273 to 373 K, and maintained there for 1 h without stirring. The
solid product was recovered by filtration (without washing) and dried at ambient temperature.
The as-synthesized material was calcined at 540 ~ (air, 5 h) and dehydrated in vacuo (270
~ 10-4 Yorr, 4 h).
Synthesis of metalorganic and silylamine precursors. Bis[bis(dimethylsilyl)amido]-
magnesium {Mg{N(SiHMe2)2]2}2 3: 3.75 mL (3.75 mmol, 2 M in heptane) of
dibutylmagnesium were added slowly to a solution of 1.00 g (7.50 mmol) 1,1,3,3-
tetramethyldisilazane in hexane. After stirring for 4 h at ambient temperature the solvent was
removed in vacuo. The solid residue was crystallized from hexane at -35 ~ yielding 3 as
colorless needles (0.10 g, 92 %). (Found: C, 33.21; H, 9.72; N, 8.75. C8HzaMgN2Si4 requires
463
C, 33.57; H, 9.67; N, 9.68.); 8H (C6D6) 5.03 (sp, 4 H, 3JH,H= 2.7 Hz), 4.94 (sp, 4 H, 3JH,H= 2.8
Hz), 0.37 (d, 24 H, 3JH,H= 2.7 Hz), and 0.36 (d, 24 H, 3JH,H= 2.8 Hz) ppm; ~5c {1H} (C6D6)
6.11 and 5.14 ppm; Vmax/Cm"! 2066s, 1252vs, 1052s, 915s, 775m, 633w and 483w.
Y[N(SiHMe2)2]3(THF)24 was prepared according to ref. [15]. Accordingly, LiN(SiHMe2)2
(2.90 equivalents) was added slowly to a suspension of YC13(THF)3.50 in hexane. After
stirring for 12 h at ambient temperature the reaction mixture was filtered and the white residue
washed with hexane. The hexane phases were combined and the solvent removed in vacuo.
The resulting powder was crystallized from hexane.
Tris(dimethylsilyl)amine N(SiHMe2)3 9 was prepared from LiN(SiHMe2)2 and
dichloromethylsilane according to ref. [ 16].
Silylation experiments. The silylation reactions were carried out by adding 1,1,3,3-
tetramethyldisilazane or hexamethyltrisilazane N(SiHMe2)3 diluted in 2 mL of hexane to
0.200 g of parent SBA-1 or metallated SBA-1 samples which were suspended in 5 mL of
hexane. After stirring the reaction mixture for 18 h at ambient temperature non-reacted
silazane was separated by several hexane washings. The silylated materials were dried under
vacuum and outgassed at 10-2 Torr and 250 ~ (SiHMe2@SBA-I-a/b) or 70 ~
(SiHMe2@Mg(bdsa)x-@SBA-l-a) prior to analysis.
Grafting experiments. A hexane solution of {Mg[N(SiHMe2)2]2}2 3 or
Y[N(SiHMe2)2]3(THF)2 4 (each in 1.5-fold excess relative to surface silanols) was added to a
suspension of dehydrated SBA-1 in hexane within a period of 5 min. The reaction mixture
was stirred for 24 h to 5 d, followed by separation of the solvent and several hexane washings
of thus obtained white powder. The resulting hybrid materials were dried under vacuum and
degassed at 10.2 Torr and 70 ~ for 2.5 h. Not reacted metal complexes were recovered from
the hexane washings by solvent evaporation under vacuum. For microanalytical data of the
organometallic-inorganic hybrid materials, see Table 1.
2.2. Characterization
FTIR spectra were recorded on a Jasco FT/IR-460plus spectrometer using Nujol mulls
between CsI plates (C-H and C-C vibrations are not listed as they overlap with Nujol bands).
Solution NMR spectra were obtained on a JEOL-JMN-GX 400 instrument (400 MHz, ~H;
100.54 MHz, 13C) at ambient temperature in C6D6. IH and 13C MAS NMR spectra were
obtained at ambient temperature on a Bruker AV300 instrument in a magnetic field of 7.04 T
(the resonance frequencies were 300.13 and 75.46 MHz), using 4 mm standard MAS probes.
13C NMR spectra were recorded using cross polarization and proton decoupling. I H and 13 C
NMR spectra were referenced to adamantane (~H: 2.00 ppm; 13C: 29.46 ppm). The following
conditions were used in the measurements: ~H, pulse repetition 2.0 s, spinning speed 8 kHz;
Z3C, pulse repetition 5.0 s, spinning speed 8 kHz. Elemental analyses were performed by the
microanalytical laboratory at Technische Universit~t Mfinchen on an Elementar VarioEl and
on an emission spectrometer plasma 400 (Perkin-Elmer). Nitrogen physisorption isotherms
were performed on an ASAP 2020 volumetric adsorption apparatus (Micromeritics). The
specific surface area as was determined by applying the BET method. The pore size
distribution was obtained on the basis of the BJH method using the Kelvin equation. Powder
X-ray diffraction (PXRD) patterns were recorded on a Philips X'pert PRO instrument in the
step/scan mode (step width: 0.00856, accumulation time: 54.6 s/step, 20. range of 0.50 - 9.99 ~
using monochromatic CuK~ radiation with )~ = 0.15418 nm.
464
Two SBA-1 materials denoted as SBA-I-a and SBA-I-b with pore diameters of 2.2 and 2.0
nm were synthesized with TEOS as a silica source and octadecyltriethylammoniumbromide
(SBA-I-a) or hexadecyltriethylammoniumbromide (SBA-I-b) as structure directing agents
under acidic conditions. The materials were characterized by PXRD (powder X-ray
diffraction) and nitrogen physisorption. The PXRD patterns of the calcined samples were
characteristic of well-ordered materials and representatively shown for material SBA-I-a in
Fig. 1. Sample SBA-I-a derived from the surfactant with the longer alkyl chain exhibits a
larger specific surface area, higher specific pore volume, and as expected a larger pore size
(Table 1). The amount of surface silanol groups of the calcined and dehydrated materials was
determined by surface silylation with tetramethyldisilazane affording SiHMez@SBA-I-a/b 1 /
2 [ 17]. For materials SBA- 1-a and -b silanol group populations of 3.27 and 5.11 mmol per 1g
were calculated, respectively.
According to Fig. 2 the dehydrated SBA-1 materials were grafted with the Lewis
acidic/oxophilic metal centers Mg(II) and Y(III) by exploiting the heterogenously performed
silylamide route. Complexes {Mg[N(SiHMe2)2]2}2 3 and Y[N(SiHMe2)2]3(THF)2 4 were
employed as hexane soluble molecular grafting precursors. The magnesium silylamide can be
easily synthesized from dibutylmagnesium via butane elimination with tetramethyldisilazane
HN(SiHMe2)2 and its dimeric structure was revealed by NMR spectroscopy. The grafting
precursors were reacted with the dehydrated SBA-1 samples at ambient temperature in hexane
for a desired period of time (24 h to 5 d) in order to obtain the hybrid materials
Mg[N(SiHMe2)2]x@SBA-I-a 5 and Y[N(SiHMez)2]y(THF)z@SBA-I-b 6. Fig. 1 shows
possible structures of magnesium and yttrium surface species on SBA-1. In both cases
monopodal and bipodal anchoring of the metal centers is likely to occur while
dimethylsilylation of SiOH groups can be a popular side reaction.
i
.=
, , ,
1 :3 5 7
20 / degrees
Fig. 2. Proposed surface species formed via the immobilization of {Mg[N(SiHMe2)2]E}2 and
Y[N(SiHMe2)z]3(THF)2 on SBA-1 materials
Table 1
Pore parameters and microanalytical data of purely siliceous and grafted SBA-1
Sample asa/ Vtotb/ dp,desC/ wt~ wt%
mZg-1 cm3g-i nm C Mg
SBA-I-a 1560 0.77 2.2 -
SBA-I-b 1190 0.53 2.0 -
SiHMe2@SBA- 1-a 1 1240 0.50 1.6 6.60
SiHMe2@SBA-I-b 2 550 0.23 1.5 9.42
Mg[N(SiHMez)2]x@SBA- 1-a 5 1430 0.61 1.9 3.07 <0.1
Y[N(SiHMez)z]y(THF)z@SBA-1-b 6 760 0.30 1.7 7.78
SiHMe:@Mg[N(SiHMez)z]x@SBA- 1-a 7 930 0.40 1.5 7.33 0.2
SiHMe2@Mg[N(SiHMez)2]x@SBA- 1-a 8 990 0.41 1.5 7.05 n.d.
aSpecific BET surface area. b Single point adsorption total pore volume of pores taken at P/Po- 0.97. c
Pore diameter according to the maximum of the BJH pore size distribution calculated from the
desorption branch. Pure silica samples were calcined at 540 ~ and dehydrated at 270 ~ for 4 h in
vacuo (10 -4 Torr); silylated and metallated samples were outgassed at 250 ~ for 3 h and at 70 ~ for
2.5 h, respectively, in vacuo (10 -2 Torr).
Such grafting reactions can be easily monitored by FTIR spectroscopy (Fig. 3). A
complete silanol conversion is indicated by the disappearance of the sharp band of the OH
stretching mode at 3695 cm -z. The Sill moiety can act as an efficient IR probe to characterize
and identify surface species by means of the peculiar Si-H stretching frequency [17]. We
found that the surface reaction of {Mg[N(SiHMe2)2]2}2 3 with SBA-I-a was by far
incomplete even after 5 days. Approximately 90% of the employed magnesium silylamide
precursor could be recovered from the reaction mixture.
466
---'U-
r.,t3
L_..
i i
L i i i i i
Fig. 3. IR spectra (Nujol): SBA-I-a (a), Mg(bdsa)x@SBA-la (b), SiHMe2@ Mg(bdsa)• (c)
A similar behavior was observed for the reaction of Y[N(SiHMe2)2]3(THF)2 4 with SBA-
1-b. The IR spectrum of Mg[N(SiHMe2)z]x@SBA-I-a 5 still shows a pronounced OH band,
however, no clear signal in the range of 2000 to 2200 cm ] assignable to metal-bonded
silylamide groups or "OSiHMe2" moieties was detected (Fig. 3).
These results are in agreement with the low carbon and magnesium contents of the
grafted samples. For example, less than 0.1 wt% magnesium was found in
Mg[N(SiHMez)z]• 5. The carbon contents of hybrid materials 5 and 6 are much
smaller than the carbon contents of the completely silylated materials SiHMe2@SBA-I-a/b
1/2 (Table 1). Moreover, solid state NMR spectroscopy was applied to examine the surface
species. The 13C NMR spectrum of Mg[N(SiHMe2)2]• 5 did not show any marked
signal intensity even after 11000 scans which is in agreement with the low complex loading.
For comparison, the proton signals of hybrid material 5 at 2.0 and 0.1/4.7 ppm could be
assigned to SiOH and SiHMe2 groups, respectively. Note that outgassing at 70 ~ -2 Torr
was not sufficient to remove all of the heptane as indicated by ~H resonances at 0.8 und 1.4
ppm.
Nitrogen adsorption and desorption isotherms of the metal-grafted samples revealed
considerably decreased pore volumes, surface areas and pore diameters in comparison to the
parent unfunctionalized materials (Fig. 4 and Table 1). The original pore volumes of SBA-I-a
(0.77 cm3/g) and SBA-I-b (0.53 cm3/g) were reduced to 0.66 cm3/g in
Mg[N(SiHMez)2]• 5 and 0.30 cm3/g in Y[N(SiHMe2)z]y(THF)z@SBA-I-b 6,
respectively. However, this "pore filling" is less pronounced compared with the homosilylated
samples 1 and 2 which exhibit pore volumes of 0.50 and 0.23 cm/g, respectively. We assume
that only the external surface of the SBA-1 material including the pore entrances reacted with
the grafting reagents. Apparently, the bulky metal silylamide complexes are not able to pass
467
the small pore openings of the SBA-1 cage-like porous structure and therefore cannot react
with SiOH groups located on the internal material surface.
Furthermore, we found, that the non-reacted SiOH groups of the partly metallated
samples are still accessible to smaller molecules. This was demonstrated by the reaction of
HN(SiHMe2)2 or N(SiHMe2)3 9 with Mg[N(SiHMez)z]x@SBA-I-a 5 affording
SiHMez@Mg[N(SiHMez)z]x@SBA-I-a 7 and 8. All of the IR-detectable silanol groups were
dimethylsilylated within 18 h. The FTIR spectrum shows an additional sharp signal at 2145
cm ~ attributable to the Sill stretching mode of OSiHMe2 moieties.
600
a a
500
'T, 400
to
i
O 300 ( b
r.~
=
t
_=
0
10 20 30 40
>.
200 Pore diameter / ,~
100
1 f i i f
Relative pressure
Fig. 4. Nitrogen adsorption/desorption isotherms at 77.4 K and the corresponding BJH pore size
distributions of parent SBA-I-a (a), Mg[N(SiHMez)z]• (b), and SiHMez@SBA-I-a (c)
(cf. Table 1)
4. CONCLUSIONS
The grafting of bulky magnesium and yttrium silylamide complexes onto the surface of cage-
like mesoporous silica SBA-1 is limited to the external surface area of these materials
468
including the pore entrances. This is probably due to the small windows of the SBA-1 pore
structure. SiOH groups located on the internal pore surface can be reacted with smaller-sized
silylating agents such as tetramethyldisilazane or hexamethyltrisilazane.
ACKNOWLEDGMENT
We thank the Deutsche Forschungsgemeinschafi and the Fonds der Chemischen Industrie for
financial support. We are also grateful to Dr. G. Raudaschl-Sieber for recording the MAS
NMR spectra.
REFERENCES
[1] For example:(a) Q. Huo, R. Leon, P. M. Petroff, G. D. Stucky, Science, 268 (1995), 1324. (b) D.
Zhao, Q. Huo, J. Feng, B. F. Chmelka, G. D. Stucky, J. Am. Chem. Soc., 120 (1998) 6024. (c) J.
R. Matos, M. Kruk, L. P. Mercuri, M. Jaroniec, L. Zhao, T. Kamiyama, O. Terasaki, J. T.
Pinnavaia, Y. Liu, J. Am. Chem. Soc., 125 (2003) 821. (d) Q. Huo, D. I. Margolese, G. D.
Stucky, Chem. Mater., 8 (1996) 1147.
[2] T. Kim, R. Ryoo, M. Kruk, K. P. Gierszal, M. Jaroniec, S. Kamiya, O. Terasaki, J. Phys. Chem.
B, 108 (2004) 11480.
[3] A.E. Garcia-Bennett, S. Williamson, P. A. Wright, I. J. Shannon, J. Mater. Chem., 12 (2002)
3533.
[4] Y. Sakamoto, M. Kaneda, O. Terasaki, D. Y. Zhao, J. M. Kim, G. Stucky, H. J. Shin, R. Ryoo,
Nature, 408 (2000) 449.
[5] (a) R. Anwander, Chem. Mater., 13 (2001) 4419. (b) P. Iengo, G. Aprile, M. Di Serio, D.
Gazzoli, E. Santacesaria, Appl. Catal. A, 178 (1999) 97.
[6] G. Gerstberger, R. Anwander, Microporous Mesoporous Mater., 44 (2001) 303.
[7] R. Anwander, C. Palm, G. Gerstberger, O. Groeger, G. Engelhardt, Chem. Commun., 17 (1998)
1811.
[8] T. Ookoshi, M. Onaka, Chem. Commun., 21 (1998) 2399.
[9] J. Jarupatrakom, T.D. Tilley, J. Am. Chem. Soc,.124 (2002) 8380.
[10] J.C. Miller, D. O'Hare, Chem. Commun., 15 (2004) 1710.
[11] A. Vinu, J. D6de~ek, V. Murugesan, M. Hartmann, Chem. Mater., 14 (2002) 2433.
[12] (a) L.-X. Dai, K. Tabata, E. Suzuki, Y. Yatsumi, J. Mater. Sci. Lett., 19 (2000) 2071. (b) L.-X.
Dai, Y.-H. Teng, K. Tabata, E. Suzuki, T. Tatsumi, Chem. Lett., 7 (2000) 794.
[13] R. Ryoo, M.J. Kim, J.M. Kim, S. Jun, Chem. Commun., 22 (1997) 2225.
[ 14] M.J. Kim, R. Ryoo, Chem. Mater., 11 (1999) 487.
[15] R. Anwander, O. Runte, J. Eppinger, G. Gerstberger, E. Herdtweck, M. Spiegler, J. Chem. Soc.,
Dalton Trans. (1998) 847.
[16] G.H. Wiseman, D.R. Wheeler, D. Seyferth, Organometallics, 5 (1986) 146.
[ 17] R. Anwander, I. Nagl, M. Widenmeyer, G. Engelhardt, O. Groeger, C. Palm, T. R6ser, J. Phys.
Chem. B, 104 (2000) 3532.
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. 2;ilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 469
Universit~it Leipzig, Institut fiJr Technische Chemie, Linn6str. 3, D-04103 Leipzig, Germany
The templated synthesis of porous carbons is a suitable tool to obtain materials with tailored
textural and structural properties. The porosity of the resulting carbons strongly depends on
the pore filling degree of the template. The present studies have been shown that the pore
filling degree is determined by the template pore infiltration and the carbonization step as
well. The release of carbon precursor from inside the template can cause both, an increased or
a decreased porosity of the resulting carbon material. Therefore synthesis conditions have to
be optimized to obtain materials with well defined characteristics. In the present comparative
studies common problems of the templated synthesis of several microporous and mesoporous
carbons have been discussed.
1. INTRODUCTION
Porous carbon materials have been used in a wide field of applications like adsorption,
catalysis or energy storage. Since they have been originally prepared by carbonisation of
natural precursors like wood or coal, its chemical and textural properties are not well defined.
Therefore, many efforts have been taken to develop carbon materials with improved
properties. In the last seven years novel synthesis methods by using solid templates have been
described [1,2]. The synthesis include filling of an inorganic porous material (i.e. the
template) with carbon precursor followed by carbonization and calcination steps. After
removal of the inorganic material a carbon replica can be obtained.
The textural and structural properties of the resulting carbon are mainly determined by the
template. Structural regularity can be transferred from the template to the carbon. Commonly
used templates are zeolite Y or mesoporous materials like MCM-48. As carbon precursors
among others sucrose, furfuryl alcohol and propene have been described in literature [1-4].
As,shown in Fig. 1, the pores of the resulting carbon can be generated from the removal of the
template walls. Furthermore, the porosity of the carbon can be extended by unfilled pores of
the template [5]. Therefore, the pore filling degree of the template is an important factor
determining the properties of the carbon material. It is affected by the infiltration of the
carbon precursor into the template pores as well as the loss of precursor from the pores during
the preparation procedure. The aim of our contribution was therefore to elucidate single steps
of the templated synthesis of several microporous and mesoporous carbons in more detail.
The problems have been discussed with respect to the template pore filling degree and
consequently to the porosity of the carbon product.
470
template carbon
former template wall + unfilled pore
HF
i~: ~
Fig. 1. Templated synthesis of carbons: Formation of the pore system of the carbon material [ 1-5]
2. E X P E R I M E N T A L
Synthesis. Microporous carbon materials were synthesized by using modified zeolite Y (Cu-
Na-Y, H-Na-Y) as template. The zeolites were loaded twice with the carbon precursor
(pyrrole, styrene) using a flow apparatus by injection into the carrier gas stream at 65~
Additionally, ethene was used as a carbon source. The zeolites were loaded twice with ethene
in a batch reactor at 2 Mpa pressure and 35~ followed by heating up to 250~ Afterwards,
the polymer/template composites were calcined at 800~ for 3 hours in nitrogene flow. The
template was removed by dissolving in HF. The zeolite made carbons are denoted as Py-ZC;
St-ZC and Et-ZC, respectively. In case of ethene, the synthesis procedure is described detailed
elsewhere [7]. Mesoporous materials were synthesized by using an amorphous silica
(KOstrosorb 1015) as template. The silica was loaded with pyrrole according its pore volume
by simple mixing. Afterwards the polymerization was initiated by treatment with HCI. The
treatment procedure was varied as described in the section below. The polymer/silica
composites were calcined at 800~ for 3 hours. The template was removed by dissolving in
HF. The mesoporous carbons are denoted as Py-MC. Further experiments have been done by
using of naphthalene as carbon source. In this case, the impregnated silica was kept at the
melting temperature of the carbon precursor for 3 h followed by calcination at 800~ in
nitrogene flow.
Nitrogen adsorption. Nitrogen adsorption-desorption isotherms were measured a t - 196~
using a ASAP 2010 (Micromeritics) analyzer. Prior to the experiments, the samples were
evacuated at 150~ in vacuum for 12 h. Specific surface area and pore size were determined
following the BET procedure. The pore size distribution was estimated applying the density
functional theory using the adsorption branch.
X-ray Diffraction. The XRD powder pattern were recorded using a Philips PW 1877
diffractometer with CuKa radiation, 0.05 ~ step size, and 2.5 s step time of each point.
Thermogravimetry. The thermogravimetry experiments were carried out by using an
equipment STA 409 (Netzsch). A mass spectrometer was used for evolved gas detection. The
following conditions were established: heating rate 10~ min -~, carrier gas N2, sample mass
20 mg.
The textural properties of the used templates and the synthesized carbons are given in Tables
1 and2.
471
Table 1
Textural properties of microporous carbons and the used template
Zeolite HY Py-ZC St-ZC Et-ZC
Bet surface 520 1400 1700 2400
mZg-I
Pore volume 0.35 0.94 1.1 1.5
cm3~;-!
Table 2
Textural properties of mesoporous carbons and the used template
Silica Py-MZ-1 Py-MZ-2 Py-MZ-3
Bet surface 312 1200 1400 1300
mZg-I
Pore volume 1.0 0.88 1.6 2.2
cm3g -1
As already mentioned above, the aim of our contribution was to present a comparative study
of common problems for the synthesis of several microporous and mesoporous carbons. Two
different templates have been chosen: Y type zeolites and an amorphous silica.
10001 " o~
i;- /o
o,~~_..~.o..o..o.o_o-o-.o-o-o-o-~
03
eo
E
o 500.
>
o
2 4 6 8 10 0,0 0,2 0,4 0,6 0,8 1,0
2 O/~ (piP0)
Fig. 2. XRD patterns and low temperature nitrogen adsorption isotherms of carbons synthesized from
(a) ethene and (b) pyrrole
472
The estimated pore volumes were 1.5 cm3g -I (Et-ZC) and 0.94 cm3g-1 for the Py-ZC,
respectively. The quality differences of the resulting carbons can be explained by the better
infiltration of the smaller molecule ethene into the partial filled carbon/zeolite composite
pores. The critical point should be the narrow interconnection between the zeolite cages [3].
The channel size of 0.74 nm is only slightly larger than that of a pyrrole molecule. After the
first loading step the channel size will be further limited due to carbon deposition inside the
zeolite. Therefore a complete filling in case of pyrrole seems unlikely during the second
loading step and the resulting carbon will be less ordered.
Silica. Non modified silicas do not have strong adsorption sites. Therefore, there is no special
driving force to enter the template pores by the carbon precursor. The precursor molecules
interact with the template in competition with used solvents. As a consequence, the precursor
can be washed out from the template pores and deposited on the external surface leading to
non porous carbon. Thus, preparation conditions which avoid an excess of solvent have to be
chosen. As an example in Fig. 3 the adsorption/desorption isothermes of carbons prepared
from pyrrole by different methods are shown. After the template loading with pyrrole the
polymerization was initiated by different ways of HC1 treatment. A gas phase treatment using
gaseous HC1 in a bed reactor resulted in a mesoporous carbon presenting high porosity with a
pore volume of 2.2 cm 3 g-1 (Py-MZ-3). Two capillary condensation steps at relative pressures
of about 0.7 and 0.9 (Fig.3, curve (a) indicate different pore systems. However, if the
pyrrole/silica composite was treated with an a.~ueos solution of HC1, the pore volume of the
obtained carbon decreased up to 1.6 cm 3 g (Py-MZ-2). The material posses only one
condensation step at relative pressure of about 0.7 (Fig.3, curve (b). The decreased porosity
of the material Py-MZ-2 can be explained by a partial removal of carbon precursor from the
template pores by the aqueous solution. Similar results have been found by using sucrose as
carbon precusor [6]. In this case, the resulting carbon possess the highest porosity if the
volume of solvent does not exceed the pore volume of the template (i.e. "incipient wetness").
An excess of solvent lead to partial removal of carbon precursor and consequently to lower
porosity of the final product [6].
1500
F:
o
1000
> 5oo.
Even tough the pore size of the silica template is large (about 12 nm) the infiltration can be
affected by the molecule size of the carbon precursor. In Fig. 4 the adsorption/desorption
isothermes of mesoporous carbons synthesized from pyrrole by different methods are shown.
If the polymerization of the pyrrole takes place inside the template pores a material with high
porosity will be obtained (curve (a), Py-MZ-3). However, a partial polymerization prior to the
template infiltration resulted in a material with lower porosity (curve (b), Py-MZ-1). Due to
their size the pyrrole polymers hardly diffuse into the template pores and deposit partially on
the external surface. Therefore, the resulting carbon posses a lower porosity. This result has
been confirmed by investigation of the templated carbonization of sucrose. Thermal treatment
of the synthesise solution (i.e. pre-polymerization) prior to the template impregnation resulted
in a material with less defined pore system[6].
1500
2.2 cm g - ~ / s
1 ooo
E
0
500
(b) /
0 0.88 cm3g-~
010 012 0140;6 018 1'0
(P/Po)
Fig. 4. Low temperature nitrogen adsorption isotherms of mesoporous carbons synthesized from
pyrrole (a) polymerization inside the template pores (b) polymerization prior to the template filling
3.2. The formation of stable oligomers or polymers inside the template pore system
Zeolites. After infiltration into the template pores the carbon precursor has to form polymers.
However, under the conditions of carbonization, i.e. at higher temperatures, the monomers
can be desorbed. The use of alkali metal ion modified X type zeolites as template lead to an
almost complete removal of aromatic compounds at higher temperatures [8]. From that one
can conclude, that only thermal treatment cannot initiate the polymerisation. Reactive or
catalytic sites are necessary to start the polymerisation. In case of ethene and styrene the acid
zeolite HY is a suitable template to obtain the polymer, there are two reaction pathways to
generate a polymer from pyrrole. This polymerisation can proceed acid catalysed or oxidative
initialized by metal ions like Cu 2+ [9]. By using HY or Cu ion exchanged zeolite Y one can
obtain oligomers or polymers of pyrrole, respectively [10].
In Fig. 5. nitrogen adsorption isotherms of carbons synthesized from pyrrole, ethene
[7] and styrene are shown. For all materials, type 1 isotherms indicating microporosity have
been obtained. The highest microporosity has been found for the Et-ZC. As shown above, it
can be explained by the role of the molecule size. However, in difference to the Py-ZC, this
material and the styrene made carbon posses a pronounced increase in adsorption at higher
relative pressure indicating the existence of mesopores (curves a and b). At higher
temperatures, a competition between carbonization and desorption of carbon precursor should
474
be expected. One explanation for the formation of mesopores is therefore the emptying of
adjacent template pores by an enhanced release of carbon precursor during the calcination.
Such pores can cause an increasing carbon porosity by the formation of mesopores as shown
in Fig. 1. To investigate the release of carbon precursor we have analysed the thermal
behaviour of the different encapsulated polymers. In Fig. 6 (left hand) the curves of
thermogravimetry of polymer/zeolite composites are shown. The experimental conditions
were nearly the same as in case of the carbonization/calcination step during the synthesis
procedure. The mass loss of polystyrene (curve b) is significantly higher than that of
polypyrrole (curve c) and the ethene [7] formed polymer (curve a). From that, one can
conclude, a considerable higher release of carbon precursor should be responsible for the
formation of mesopores of the styrene made carbons.
1000 ~ ~
o"}
" _.,,...-.-.""'""" .... (c) ..
500 ~ ~ ~
> 0 .,
0:0"0:2'0',4"0:6 018 1:0
(piP0)
Fig. 5. Low temperature nitrogen adsorption isotherms of carbons from (a) ethene, (b) styrene and (c)
pyrrole
For the ethene made carbon, the explanation is more complex. On the one hand, this material
exhibits mesopores (similar to the St-C) which should be formed by enhanced desorption of
carbon precursor. On the other hand, the mass loss of the encapsulated ethene formed polymer
is very similar to that of adsorbed polypyrrole (curve a and c). However, pyrrole made carbon
does almost not exhibit mesopores. To more elucidate these facts, differential
thermogravimetry curves (DTG) for polymer/zeolite composites from ethene and pyrrole have
been calculated (Fig. 6. right hand). Evolved gas analysis by mass spectrometry has been
shown, that desorption up to 200~ is mainly attributed to water. Above this temperature, the
desorption is determined by the release of organic compounds. Regarding the release of
organic compounds, both materials possess one desorption maximum. For the ethene formed
polymer, the amount of released organic compounds is similar to the polypyrrole
encapsulated sample. However, since the maximum appears at about 300~ for the
polypyrrole consisting sample, there is a significant shift in desorption maximum of about
150~ to higher temperatures for the ethene formed polymer.
We explain the difference in mesoporosity for both carbon materials by the significant
difference in desorption maximum temperatures. As already mentioned, the carbonization
/calcination is always accompanied by release of carbon precursor from inside the template
pores. However, an additional effect has to be considered. Especially at the beginning of the
calcination, i.e. at lower temperatures, the carbon precursor containing mixture is still
475
o,oo-~~
o~ -10 (a)
E
< -20
%# i organ.compounds
-30 -0,07
6 250 560 750 250 560
Temp./~ Temp./~
Fig. 6. Thermogravimetry and DTG profiles of polymer/zeolite composites from (a) ethene [7], (b)
styrene and (c) pyrrole
Silica. As has been shown in the last section, the polymerisation step has to be initiated. Non
acidic silica does not have active sites. The carbon synthesis using pyrrole or styrene as
precursor was only successful using an external initiator like acids or radicals. Experiments
without initiator lead to a complete desorption of the precursor during the calcination.
On the other hand, the large pore size of the silica allows the infiltration of polycyclic
aromatic hydrocarbons. Due to their chemical nature, polycyclic aromatics should be suitable
precursors for direct carbonization. However, the use of naphthalene as carbon precursor lead
to a partial destruction of the template (not shown here). One reason for that can be the
melting of the precursor inside the template pores. The melting can cause mechanical stress
and consequently lead to the destruction of the pore system.
The importance of the melting of the precursor has been shown for sucrose by
variation of the concentration of sulphuric acid in the synthesis solution [6]. If the synthesis is
carried out without sulphuric acid the polymerisation compete with the melting. Part of the
carbon precursor will leave the template pores and deposit on the external surface. These
depositions lead to non porous carbon.
4. CONCLUSIONS
Generally, the porosity of the resulting carbon materials depend strongly on the template pore
filling degree. A minimum value of pore filling degree is necessary to form a stable carbon
framework. The release of carbon source from inside the template pores can cause two
different effects. Deposition of the released precursor (e.g. sucrose) on the external template
476
surface will lead to non porous carbon and consequently reduce the porosity of the carbon
material. On the other hand, volatile carbon precursor like styrene and ethene will desorb
leaving unfilled template pores. The mesoporosity of the carbons will be increased.
ACKNOWLEDGMENT
Thanks to B. Rulle and G. Kommichau for measurement, Prof. W.-D. Einicke for discussions.
We are grateful to Chemiewerk Bad K6stritz GmbH for supplying with silica materials.
This work was supported by the DFG. Contract No: PA194/13-1 and KL 1202/3-1.
REFERENCES
[11 R. Ryoo, S.H. Joo, S.J. Jun, Phys. Chem. B., 103 (1999) 7743.
[21 Z. Ma, T. Kyotani, A.Tomita, Chem. Comm., (2000) 2365.
[3] Z. Ma, T. Kyotani, A.Tomita, Carbon, 40 (2002) 2367.
[4] T.Kyotani, Z.Ma, A.Tomita, Carbon, 41 (2003) 1451.
[5] A.B. Fuertes, D.M. Nevskaia, J. Mater. Chem. 62 (2003) 1843.
[6] K. B6hme, W.D. Einicke, O. Klepel, Carbon, (2004), submitted.
[7] A. Garsuch, O. Klepel, Carbon, (2004), submitted.
[8] H. F6rster, H. Fuess, E. Geidel, B. Hunger, H. Jobic, Ch. Kirschhock, O. Klepel, K. Krause,
Phys. Chem. Chem. Phys., 1(4) (1999) 593.
[9] A. Garsuch, R.R. Sattler, P.G. Pickup, Chem. Comm., 3 (2004) 344.
[10] T. Bein, P. Enzel, Angew. Chem., 101 (1989) 1737.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 477
A novel method has been developed for the synthesis of Ti-MCM-41 based on the preparation
in a first step of an amorphous SiO2-TiO2 xerogel, followed by addition of a cationic
surfactant and hydrothermal treatment. Ammonium hydroxide (NH3) and
tetrapropylammonium hydroxide (TPAOH) solutions were used as basic agents during both
the preparation of the xerogel and its subsequent hydrothermal transformation. The properties
of the samples thus obtained strongly depend on the method used to prepare the raw xerogel.
When employing NH3 the materials produced present a lack of mesostructure, showing that
this basic agent is not suitable for promoting the xerogel reorganization around the surfactant
micelles. In contrast, the use of TPAOH leads to Ti-MCM-41 samples with a high degree of
mesoscopic ordering and presenting all the Ti atoms incorporated into the walls. This novel
method allows Ti atoms to be first stabilized in isolated positions of a silica matrix, which
avoids the formation of non-desired TiO2 phases, while the Ti content of the final Ti-MCM-
41 is similar to that of the raw xerogel.
1. INTRODUCTION
The discovery of TS-1 zeolite [1] was an important progress to develop catalytic oxidation
reactions by redox titanosilicate molecular sieves. TS-1 zeolite exhibited excellent catalytic
properties for liquid phase selective oxidation of small organic molecules in the presence of
H202. The high activity of TS-1 zeolite is due to the presence of isolated tetrahedral Ti sites
incorporated into the framework. However, this catalyst presents the disadvantage of the
small pore size, which causes the inaccessibility of bulky molecules, limiting its use in fine
chemical processes. Synthesis of Ti-Beta was reported to be effective for oxidations of bulkier
substrates [2], but its applications were limited to molecules smaller than 7A [3].
Consequently, developing new catalysts with larger pores became indispensable.
A new family of mesoporous molecular sieves named MS41 was discovered by
researchers at Mobil Oil in the presence of cationic surfactants through a micelle-templated
pathway [4, 5]. MS41 family is classified into several members with MCM-41 (hexagonal),
MCM-48 (cubic) and MCM-50 (lamellar) as main components. These materials have uniform
pores and their pore size can be tailored in the range 1.6-10 nm through the right choice of the
surfactant used as template, auxiliary chemicals and reactions conditions [6]. Mesoporous
materials overcome the limitation of microporous zeolites and allow the diffusion and
conversion of bulky molecules [7, 8]. Nevertheless, the incorporation of metallic species into
the silica walls is necessary for their use in catalysis. Several successful examples of
transition metal-containing mesoporous silicates have been reported with metals such A1, Ti,
Ga, Mn and Fe [9]. Thus, Ti-MCM-41 has shown to be an effective catalyst for epoxidation
478
of bulky olefins [10]. This material can be prepared mainly by two ways: i) addition of
titanium alkoxides to the MCM-41 synthesis gel [7], ii) grafting of the titanium species
(metallocene complexes) to the mesopore silica surfaces via a post-synthetic treatment
[11,12]. In the conventional synthesis of titanium-containing mesoporous molecular sieves,
organic alkoxides are used as silicon and titanium species. The hydrolysis of the titanium
precursors must be controlled carefully in order to avoid the formation of extraframework
TiO2 [13]. Moreover, since isolated Ti atoms with tetrahedral coordination are the main
responsible for the catalytic properties of these materials, several efforts have been made to
increase the titanium loading in the mesoporous material avoiding the formation of TiO2.
The present paper describes a novel method to synthesize Ti-MCM-41 from amorphous
SiOe-TiO2 xerogels impregnated with a basic solution using cetyltrimethylammonium
bromide (CTAB) as surfactant and followed by hydrothermal treatment. It is inspired in
earlier works reporting the synthesis of different Ti-containing zeolites (TS-1, TS-2, AI-TS-1
and A1-Ti-Beta) by reorganization and transformation of amorphous SiO2-TiO2 xerogels [ 14-
16]. The samples of Ti-MCM-41 so obtained present a well-defined array of uniform
hexagonal mesopores. According to this method, all the titanium added during the synthesis is
incorporated into the siliceous matrix of the xerogel, stabilizing the titanium atoms and
avoiding the precipitation of extraframework TiO2.
2. E X P E R I M E N T A L
Finally, the solids produced were filtered, washed with water, dried at 110~ and
calcined at 550~ for 5 h to remove the organics occluded.
2.3. Characterization
X-ray diffraction (XRD) pattems of the samples were collected on a Philips X'Pert
PRO diffractometer with Cu-Kcz radiation (~. = 1.542 A). Fourier transform IR (FT-IR)
spectra were taken on an ATI MATTSON Infinity Series FTIR spectrometer, using KBr
pellets containing 1 wt. % of the samples. DR UV-Vis spectra were recorded on a Varian
CARY 500 spectrophotometer. The titanium content in the materials was determined by
simultaneous ICP-AES using a Varian VISTA AX instrument. Nitrogen adsorption-
desorption isotherms were measured at 77 K with a Micromeritics TRISTAR 3000 system.
Previously, the samples were outgassed at 300 ~ under nitrogen flow. The surface area was
determined using the BET equation. The pore size distributions were calculated using the
Barrett-Joyner-Halenda (BJH) model applied to the adsorption branch. The t-plot method was
used to determine micropore volume of the samples. Transmission electron microscopy
(TEM) images were recorded on a PHILIPS TECHNAI 20 transmission electron microscope
operated at 200kV. CHN elemental analyses were performed on an ELEMTAR Vario EL III
analyzer.
Two series of samples were synthesized using NH3 and TPAOH as basic agents, respectively,
during both the preparation of the SiO2-TiO2 xerogels and the subsequent hydrothermal
treatment. In addition, an aqueous solution containing the cationic surfactant (CTAB) was
added to the xerogel in order to promote its reorganization into a material with mesoscopic
ordering. For both basic agents, several synthesis times were employed during the
hydrothermal treatment. Samples prepared with NH3 or TPAOH were designated by N-x or
T-x, respectively, where x denotes the time of the hydrothermal treatment in hours.
Low-angle XRD spectra of N-x and T-x samples, as well as of the starting xerogels,
are shown in Fig. 1. The spectra of the xerogels prepared with both basic agents confirm they
are completely amorphous materials, without any type of mesopore ordering. However, T-x
samples exhibit a main diffraction peak, placed at 20 = 2.1 ~176 and minor peaks in the range
20 = 3-4 ~ that can be indexed as (110) and (200) reflections corresponding to an hexagonal
unit cell. These peaks have been typically observed in MCM-41 samples, assessing the
presence of a hexagonal mesopore structure. By contrast, these reflections are absent in the
XRD pattern of N-x materials, indicating the lack of long-range order of the mesopores in the
samples obtained from NH3 impregnated xerogels. Only a slight shoulder is present at 20 = 2~
in the XRD spectra of these samples. Since TPA-containing xerogels may lead to the
formation of TS-1 crystals [14], wide-angle XRD spectra of the T-x samples were also
recorded. A broad bottom reflection appears for all samples, which indicates the absence of
crystalline zeolitic units.
Table 1 summarizes the textural properties of the different samples determined from
the N2 adsorption isotherms at 77 K. Likewise, Fig. 2 illustrates the N2 adsorption-desorption
isotherms at 77 K and the BJH mesopore size distribution corresponding to the raw xerogels
and N-48 and T-60 samples. The SiO2-TiO2 xerogels exhibit a high surface area, although
they present a wide pore size distribution in the mesopore range, which is consistent with the
absence of ordering in these materials concluded from the XRD spectra. For the N-x samples,
a strong reduction in the surface area is observed in regards to the raw xerogel, while the pore
480
size distribution is even wider than in the latter. These results show that ammonia is not a
convenient basic agent for the transformation of the amorphous xerogels into ordered
mesoporous materials.
(a)
N-120
T-120
v
N-72
T-60
4,J
.P,I
N-48
_ - . . . . . . . . . . . . .
T-48
N-xerogel T-xerogel
,,
9 I ' I ' I ' I ' I ' " 1 ' " ' 1 ' 1 ' 1 ' 1 ' 1 " ' ' I ' 1 '
I 2 3 4 5 6 7 8 9 I0 1 2 3 4 5 6 7 8 9 I0
2 0 [Ol 20 (*)
Fig. 1. Low-angle XRD spectra of samples prepared with NH3 (N-x) and TPAOH (T-x).
A completely different behavior is observed for the T-x samples, as they retain the
high surface area of the starting xerogel whereas the N2 adsorption isotherm presents a sharp
inflection at P/Po=0.34, which can be assigned to capillary condensation within mesopores
with a narrow pore size distribution (type IV isotherm). This conclusion is confirmed by the
BJH pore size distribution of sample T-60, shown in Fig. 2.b, with the maximum
corresponding to a pore diameter around 28 A. The textural properties and isotherm type of
the T-x samples agrees well with those typical of MCM-41 samples. Therefore, the presence
of TPAOH is an essential factor to promote the mesoscopic reorganization of the amorphous
xerogel around CTAB micelles. The interplanar distance (dl00), the hexagonal unit cell
parameter (a0) and the wall thickness of T-x samples have been estimated from both XRD and
N2 adsorption measurements, being reported in Table 2. T-x samples exhibit d~00 spacing and
wall thickness about 40 and 18 A, respectively, which are in agreement with the values
reported in literature for MCM-41 materials [ 13, 17]. A trend to increase the wall thickness of
the T-x samples with the time of the hydrothermal treatment is observed.
Since the synthesis of T-x samples takes place in a basic medium with high TPAOH
concentrations, an important subject to be addressed is whether the products obtained are
simply Ti-MCM-41 materials or, on the contrary, they may be really considered hybrid
samples with both zeolitic and mesostructured features, having TS-1 zeolitic units within the
mesopore walls. In this way, t-plot analyses of the N2 adsorption isotherms indicate the
presence of a small amount of micropore volume, as shown in Table 2, for the three T-x
samples. Likewise, in addition to a high content of surfactant, CHN analyses confirm the
existence of small amounts of TPA in some of the as-synthesized T-x samples (Table 2).
However, as the hydrothermal treatment is prolonged, the TPA content is progressively
481
reduced, being negligible for the T-120 sample. This result suggests that the TPA molecules
present in the as-synthesized samples are not occluded in a zeolitic environment, since in this
case the TPA species should exhibit a high stability, which would avoid their extraction
during the hydrothermal treatment. This conclusion is also supported by the absence in the
FTIR spectra of T-x samples (Fig. 3.a) of a band at 550 cm -~, typical of MFI type zeolites,
confirming that zeolite units are not present within the pore walls.
Table 1
Textural properties of T-x and N-x calcined samples
Sample Basic agent
ta (h) SBET (m2/g) Dpb (/~k) Vp c (cm3/g)
8
I
.,_..
700
~ aoo-
o 2001
looi
=::1
'~ o ". . . . . ,i ....... i
0.0 0.2 0.4p/pO.6o 0.8 1.0 10 100 . 1000
700 ..... 8 Pore Diameter (A)
A
600-
(c) I _.,,.~.... N~48 l (d) I N-48
N-xerogel
500-
"~ 6-
o~ 4001
a00~
o 2ooi
~2~
I00 ~
~ o 0!:
::--" ' ~ , .~
> 0.0 " 0'.2 0:4pipo0'.6 0'.8 1.0 10 100 . 1000
Pore Diameter (A)
Fig. 2. N2 adsorption at 77 K: (a, c) isotherms and (b, d) BJH pore size distributions.
482
ICP-AES analyses of the T-x samples show that most of the titanium present in the
raw xerogel remains incorporated into the ordered mesoporous materials, with a slight
decrease in the Ti content as the duration of the hydrothermal treatment is increased (Table 2).
The state and coordination of the Ti species have been probed by DR UV-visible
spectroscopy. Fig. 3.b illustrates the UV-VIS spectra of the T-x samples after calcination. An
intense band with a maximum at 215 nm is observed in the spectra, which is usually
associated with isolated Ti(IV) atoms having tetrahedral coordination. However, the T-x
samples also show significant absorption at higher wavelengths with a slight shoulder at 260-
270 nm, attributed to the presence of Ti atoms in penta- or octahedral coordination [ 18, 19].
This feature, typically observed in Ti-MCM-41 materials, can be interpreted as a consequence
of the amorphous character of the walls of ordered mesoporous materials. Moreover, for all
the T-x samples, the absence of an absorption band at 330 nm indicates that they are free of
TiO2, which is an important fact regarding the possible use of these materials as catalysts in
partial oxidation reactions.
Table 2
Chemical composition and structural properties of the T-x samples
Organic content Wall Vmic
Sample ( wt %) Si/Yi dl00 a0 thickness (cm3/g)
CTAB TPA (,,~) (A) (A)
T-48 48.5 1.8 40.8 41.7 48.2 17.1 0.031
T-60 53.4 0.6 41.3 39.8 45.9 18.1 0.031
T-120 55.5 0.0 45.6 41.4 47.8 18.9 0.039
Fig. 4 illustrates TEM images of N-48 and T-60 samples. In the first case, the TEM
micrograph show a disordered wormhole pore structure, which is in agreement with the above
discussed XRD and N2 adsorption results corresponding to N-x samples. On the contrary, the
T-60 material possesses an overwhelming honey-comb pore structure. Moreover, values of
d~00 - 4.14 nm and wall thickness of 1.83 nm were determined for this sample using these
TEM micrographs, being in good agreement with the results obtained from combined XRD
and N2 adsorption data.
4. CONCLUSIONS
Ti-MCM-41 materials were prepared from SiO2-TiO2 xerogels. The basic solutions
used in both the gelation and the wetness impregnation steps proved to be decisive for the
effective transformation of the xerogel into mesostructured materials. Xerogel treated with
aqueous ammonia led to disordered samples, while the use of TPAOH solutions allowed
highly ordered Ti-MCM-41 to be obtained. Materials prepared from TPA-containing xerogels
exhibited high specific surface area, high pore volume, a narrow pore size distribution and an
outstanding ordered honeycomb-like array. Although these samples present a small amount of
483
(a) (b}
T-120
T-60
T-48
T-xerogel
' I , i ' I ' I ' i '
Fig. 3. FT-IR spectra (a) and UV-Vis spectra (b) of calcined T-x samples.
i: ;iiil;i~ ii!i!ii!i~i!!iii:~i:!iiiiiiiii!!
! i!il:~ ~!
i~.~ I ~ili~iiiiii~i!i~!ii~i~,~ I
microporosity, probably originated by the removal of TPA species, there is no evidence of the
existence of zeolitic units. As the time of the hydrothermal treatment is increased, TPA is
removed from the structure, corroborating it is not confined in a zeolitic environment. The
method here developed permits to control the Ti amount incorporated in Ti-MCM-41, as its
Ti content is similar to that of the starting xerogel. Moreover, the Ti-MCM-41 so obtained is
free of extraframework TiO2 phases. Therefore, a new and efficient method has been
developed for the synthesis of Ti-MCM-41 via a xerogel route.
484
ACKNOWLEDGEMENTS
REFERENCES
Olefins are interesting compounds to anchor onto the surface of silica supports since their
reactivity allow to form a large variety of organic functionalities. In this work, the synthesis
of mesostructured silicas (MCM-41 and SBA-15 type materials) functionalized with butenyl
groups is studied through the use of post-synthetic grafting or direct-synthesis procedures.
Results confirm the higher functionalization degree achieved in the final mesostructured
materials when using direct synthesis procedures. Unlike MCM-41 materials, which become
microporous when high organic content are attached on, SBA-15 type butenyl functionalized
silicas keep the features of a mesoporous solid. Post-synthetic and in-situ oxidations of
butenyl groups during the preparation of the mesostructured silicas have been carried out
using H202 as oxidant, leading to materials functionalized with glycol moieties attached to
inorganic mesoporous materials, which are attractive to be used as supports for anchoring
different catalytic species.
1. INTRODUCTION
One way to modify the physical and chemical properties of mesoporous materials, like
MCM-41 [ 1] and SBA-15 [2], for improving the attachment of different compounds lies on
the incorporation of organic functionalities onto their surface. These organically
functionalized materials have great interest in the development of new catalysts and
adsorbents with a wide range of applications, since surface properties such as hydrophobicity
[3] or catalytic behaviour can be easily controlled [4]. For this purpose, many different
reagents have been used following different synthetic procedures [5] like post-synthetic
grafting of organosilanes [6] or direct-synthesis methods like co-condensation of
organotrialkoxysilanes with tetraalkoxysilanes [7].
The most used procedure for modifying mesostructured supports with organic
functionalities is basec' on the reactivity of the surface hydroxyl groups, used as anchoring
points for the organic functionalities [8]. The most important advantage of this grafting
method is the formation of the mesoscopic structure before functionalization, although the
control over the degree of incorporation of the organic group and the homogeneity of the final
materials is difficult to reach [9, 10].
On the other hand, direct synthesis procedures based on the incorporation of the
organic moieties during the synthesis of the mesostructured solids offer a more uniform
surface coverage with functional groups, better stability of the attached organic chains and a
higher control over the surface properties of the resultant materials [ 10, 11 ].
486
Olefins are useful organic groups, not only because the double bond is quite stable
under the synthesis conditions but it can be turned into a large variety of compounds through
many kind of reactions [12]. As it has been previously reported by Kruk et al. [13] vinyl
groups attached onto the surface of MCM-41 type materials, can be easily transformed into
alkylboranes, alcohols, epoxides and diol groups, leading to materials useful as catalyst
supports. However the M41s materials thus obtained show a strong decrease of the
mesoporosity, especially for large functionalization degrees, which affects their use in many
applications. In this work we describe the synthesis of mesostructured silicas, functionalized
with butenyl chains through grafting and direct-synthesis methods as well as their oxidation to
glycol species, which are useful functionalities to be used as anchoring sites for many
different applications.
2. EXPERIMENTAL
2.1. Synthesis
Grafting procedures to obtain SBA-15 (S-1,S-2) or MCM-41 (S-6,S-7) type materials are
based on dispersing 2 g of the mesoporous support, previously calcined at 550 ~ in 300 ml
of toluene under reflux with the desired amount of organosilane (butenyltriethoxysilane,
BTES, Gelest). The resultant solution is then kept under reflux for 24 h. Finally, the solid is
recovered by filtration and intensively washed with toluene.
Direct-synthesis Butenyl-SBA-15 type materials (S-3, S-4, S-5) were synthesized using
a method derived from that reported by Zhao et al. [2]. First 3.20 g of Pluronic P-123
(Aldrich) were dissolved, under stirring, in 100 ml of a 1.9M HC1 aqueous solution at room
temperature. The homogeneous solution obtained was then heated up to 40 ~ before adding
7.11 g of tetraethylorthosilicate (TEOS, Aldrich). After one hour of TEOS prehydrolisis, the
butenyl functionality (BTES) was added into the aqueous suspension, using a molar ratio
BTES/TEOS = 1:1, 1:2 or 1:5. The resultant mixture was then stirred at 40~ for 20h and
hydrothermally aged at 110~ for 24 hours. Finally, the solid was filtered off and the template
was removed by an extraction treatment with ethanol under reflux and subsequently air-dried.
Direct-synthesis Butenyl-MCM-41 type materials (S-8,S-9,S-10) were synthesized
following a procedure similar to the method reported by Lin et al [ 10]. Thereby, 2.12 g of
cetyltrimethylammoniumbromide (CTABr, Aldrich) were mixed with 80 g of distilled water
and 6.57 g of dimethylamine (DMA, Scharlab), used as basic agent, at room temperature.
When mixture was completely homogeneised, 8.26 g of TEOS were added dropwise to the
suspension. Additionally, after one hour TEOS prehydrolisis, certain amount of the butenyl
precursor (BTES) was added, using molar ratios BTES/TEOS = 1:1, 1:2 or 1:5, and then the
mixture was stirred for 4 hours at room temperature. All the samples were aged in an
autoclave at 110~ for 48 h. The products obtained were filtered off, washed with distilled
water before removing the template by solvent extraction with an acidic solution of
hydrochloric acid in ethanol and finally air-dried.
Double bond oxidations have been carried out through two methods: in-situ (S-12) and
post-synthesis (S-11). The in-situ oxidation procedure consists of adding the hydrogen
peroxide into the synthesis media, following a similar procedure to that reported by
Margolese et al. for oxydizing thiol groups [7] in the synthesis of sulphonic-containing SBA-
15. The oxidant molar ratio was HzO2:BTES = 5:1. On the other hand, oxidation reactions
have been carried out through a post-synthetic procedure by dropping H202 on a suspension
of the functionalized material in acetone under stirring. Typically, 0.5 g of butenyl
functionalized-material was suspended in 300 ml of acetone. Then an aqueous solution of 30
487
wt % H202 was added in a molar ratio H202:BTES = 5:1. The suspension was stirred at 55 ~
for 24 h before filtering off the solid. Finally, the oxidized material was washed with acetone
and air-dried.
Table 1
Synthesis conditions, textural properties and organic content for the butenyl-functionalized
materials
Material Sample Synthesis BTES/TEOS SBET Dp Vt T/(Q+T)
Procedure molar ratio (m 2 g-l) (A) (era 3 g-l) 29SiNMR
Pure silica
SBA-15 --- 652 89.2 0.960 0.0%
S-1 1:2 617 88.8 0.966 0.0%
S-2 Grafting 1"1 570 89.6 0.855 2.3%
SBA-15 S-3 Direct 1"5 572 69.7 0.663 5.4%
S-4 synthesis 1:2 629 66.8 0.606 15.0%
S-5 1"1 560 69.7 0.484 23.8%
Pure silica
MCM-41 1010 27.2 0.750 0.0%
The extent of BTES incorporation into the mesostructured materials was monitored by
means of solid state 29Si NMR. Different resonance lines can be observed for the siloxane [Qn
-- Si-(OSi)n-(OH)4_n, n = 2-4; Q4 at -110 ppm, Q3 at -100 ppm and Q2 at -90 ppm] and
organosiloxane [T n - RSi(OSi)n-(OH)3_n, n = 1-3; T 3 at-65 ppm and T 2 at -57 ppm] species.
Regarding the organic content anchored onto the final materials [T/(T+Q)], it seems clear that
the direct synthesis procedures lead to materials with higher contents on organic
functionalities compared to that achieved through grafting methods. For both synthetic routes,
the final organic content attached onto the silicas is higher as the BTES/TEOS ratio increases,
although higher incorporation efficiencies are achieved for MCM-41 type materials. This
different behaviour is found for both synthetic paths, grafting and direct synthesis, but its
explanation is different depending on the synthesis route. MCM materials are able to
incorporate higher organic contents through grafting procedures since the surface silanol
density is higher than that present on SBA-15 materials [14]. On the other hand, the
differences found for the materials obtained through the co-condensation pathway may be
explained by a different behaviour of the butenyl-silane in the formation of the solid
depending on the pH of the synthesis media, basic for MCM-41 or acidic for SBA-15.
The conclusions drawn from the 29Si NMR spectra about butenyl incorporation in the
mesostructured materials are supported by the results of the thermogravimetric analysis
(TGA) in air. Analysis of the samples show a peak centered at 220 ~ corresponding to the
loss of BTES and the remaining surfactant species after extraction with ethanol. Assuming a
similar surfactant content for all the samples (measured for the pure silica material), BTES
content in direct synthesis SBA-15 have been determined as 3, 17 and 26 molar %, for
samples S-3, S-4 and S-5 respectively
Fig. 1 displays the XRD spectra recorded for the samples obtained through direct synthesis
procedures for both structures, MCM-41 and SBA-15. All of samples show the typical
diffraction pattern of a mesostructured solid, being the main diffraction signal (d~00) located at
lower angles for SBA materials, since the distance between pore rows is greater for these
materials. Regarding to other XRD signals characteristics for a bidimensional hexagonal
ordered structure, like those related to dl~0 or d200 spacings, there is no evidence of their
presence and therefore, these materials show a lower ordering degree than unfunctionalized
samples (not shown).
a) b)
S-3 S'8
$4 S-9
. . . . ,, . . . . . . . . . : ,
110 1,5 2,0 215 310 3,5 4,0 45 5,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 80 9,0 10,0
20 (o) 20 (o)
Fig. l. XRD spectra of the direct synthesis materials, a) SBA-15 materials and b) MCM-41 materials.
489
These results suggest that the co-condensation of butenyl species determines the ordering
degree of functionalized materials. An increase of BTES loading leads to a lower order, as it
is shown for the MCM-41 type materials, where the dl00 signal intensity is largely decreased
as well as its width is increased. On the other hand, all of the SBA-15 materials show a
similar diffraction pattern suggesting a different interaction during synthesis for between the
surfactants used for the synthesis of SBA-15 and MCM-41 materials and the butenyl groups.
Thus, the ordering of MCM-41 seems to be more dependant on the butenyl loading than
SBA- 15 samples are.
Fig. 2 shows TEM micrographs from samples S-4 and S-9. Although these samples have
been synthesized with the same content of butenyl groups, the amount is enough to cause
great differences between both materials. While SBA-15 material (sample S-4) show a highly
ordered honeycomb structure, typical of an hexagonal array with highly regular parallel
layers, MCM-41 material (sample S-9) shows a wormlike porous structure, very different
from the one corresponding to non-functionalized MCM-41 material (not shown).
S-9
~0 r a n
This different behaviour is also observed in the analysis of the textural properties
determined by the adsorption-desorption of nitrogen at 77K. Fig. 3 displays the adsorption-
desorption isotherms (a and b) obtained for SBA-15 and MCM-41 materials as well as their
pore size distribution (c and d) calculated by BJH method. SBA-15 samples present type IV
isotherms, typical of mesoporous solids. The organic content promotes a drastic change on the
textural properties, since the pore volume and surface area decrease when increasing the
organic phase. On the other hand, the step in the adsorption branch of the isotherms remains
almost in the same position, independently of the functionalization degree and therefore, the
maximum of the pore size distribution is similar for all these samples (fig. 3, c). MCM-41
type materials show a different trend when increasing the organic loading, mainly for high
butenyl incorporation. Thus, the isotherms of MCM-41 samples depicted in fig. 3, b) shows
an striking inflection, becoming from a mesoporous solid with a type IV isotherm (S-8) for
the lower butenyl loading to a microporous material showing a type I isotherm (S-10) for the
highest organic content.
490
a.
I-
500,
s,s:'/s, !
/I]- 5S]1oI ~ _
400.
E
_=
300, goo~
o
9"o 200,
r~
0
100.
,,r
o
0,0 0,2 0,4 0,6 0,8 1,0 0,0 0:2 0:4 016 0:8 1,0
(PIPo) (PIPo)
f.-,.,od)
-- S-8
--=-- S-3
l--e--S-4 e-- S-9 ~,
}--,=-- S-5 f1
c) o
QJ
. . . . = . . . . . . . i
Fig. 3. N2 adsorption-desorption analysis for direct synthesis materials, a) SBA-15 isotherms b) MCM-
41 isotherms, c) SBA-15 pore size distribution, d) MCM-41 pore size distribution
Table 2
Synthesis conditions and organic content for the synthesized oxidized materials
Sample Oxidation BTES/TEOS SBET Dp Vt T/(Q+T)
Procedure molar ratio (m2g -l) (A) (cm 3 g-i) 29SiNMR
As it is shown from the table, BET-surface, pore diameter and total pore volume of the
oxidized materials (samples S-11 and S-12, respectively) are quite similar to that obtained for
the non-oxidized sample (S-4). On the other hand, regarding final organic content, the in situ
oxidation path leads to materials with lower functionalization degree.
Fig. 4 shows the solid state 29Si (a) and 13C (b) NMR spectra of the S-4 and S-12 samples.
From ~3C NMR, it is possible to determine the different carbon assignments. Thus, in sample
S-4 (non-oxidized) the peaks labelled with asterisks correspond to the carbon species of non-
removed surfactant P123, while the peak at 120 ppm is attributed to the double bond carbon
from butenyl groups. Sample S-11 hardly presents modifications with regards to S-4,
indicating a low degree of double bond oxidation.
Q4 ,k, 1
Q3 4 3
,1 2~/'1
/
T3 Q2
" S..,4
~'~Si 2
! i i !
i i i
-40 -60 -80 -1 O0 -120 -140 140 120 100 80 60 40 20 0 -20 -40
,5 (ppm) G (ppm)
Fig. 4. NMR spectra of oxidized samples, a) 29Si MAS NMR, b) ~3C CP MAS NMR. Asterisks denote
surfactant compound.
On the contrary, in-situ oxidation, sample S-12 shows the overwhelming extinction of
signal at 120 ppm, which accounts for the total conversion to a glycol group, corroborated by
the 29Si NMR spectrum, where the presence ofT groups evidences the Si-C bonds existence.
4. CONCLUSIONS
Direct synthesis procedures leads to SBA-15 and MCM-41 materials with higher contents on
organic functionalities compared to that achieved through grafting methods. While SBA-15
492
material shows a highly ordered honeycomb structure, typical of a hexagonal array, MCM-41
type material presents a wormhole porous structure, even if the lowest BTES loading is used.
While the butenyl groups do not seem to affect the micelle organization with triblock
copolymer template used in SBA-15 synthesis, the transformation of MCM-41 materials into
microporous samples suggests the hard interaction between butenyl functionalities and the
CTABr template. The conversion of butenyl species into glycol groups can be achieved by in-
situ oxidation method with greater oxidation efficiency, being these modified materials
suitable to be used as anchoring sites for different active phases in the preparation of
heterogeneous catalysts.
ACKNOWLEDGEMENTS
This research has been funded by the "Ministerio de Ciencia y Tecnologia" through the
CICYT project PPQ 2002-02382.
REFERENCES
[1] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartulli, J.S. Beck, Nature, 359 (1992) 710.
[2] D.Y. Zhao, J. Feng, Q. Huo, N. Melosh, G.H. Fredrickson, B.F. Chmelka, G.D. Stucky,
Science, 279 (1998) 548.
[3] A.S. Maria Chong, X.S. Zhao, Angeline T. Kustedjo, S.Z. Qiao, Microporous Mesoporous
Mater., 72 (2004) 33-42.
[4] B.-G. Park, J. Park, W. Guo, W.-J. Cho, C.-S. Ha, Stud. Surf. Sci. Catal., 146 (2003) 489.
[5] D.Y. Zhao, J. Feng, Q. Huo, N. Melosh, G.H. Fredrickson, B.F. Chmelka, G.D. Stucky,
Science, 279 (1998) 548.
[6] T. Tatsumi, K.A. Koyano, Y. Tanaka, S. Nakata, Stud. Surf. Sci. Catal., 117 (1998) 143.
[7] D. Margolese, J.A. Melero, S.C. Christiansen, B.F. Chmelka, G.D. Stucky, Chem. Mater., 12
(2000) 2448.
[8] Ningya Yu, Yanjun Gong, Dong Wu, Yuhan Sun, Qing Luo, Wuyang Liu, Feng Deng,
Microporous Mesoporous Mater., 72 (2004) 25-32.
[9] D.J. Macquarrie, D.B. Jackson, J.E.G. Mdoe, J.H. Clark, New. J. Chem., 23 (1999) 539.
[~0] M.H. Lim, A.Stein, Chem. Mater., 11 (1999) 3285.
[11] A. Stein, B.J. Melde, R.C. Schroden, Adv. Mater., 12 (2000) 1403.
[12] T. Asefa, M. Kruk, M.J. MacLachlan, N. Coombs, H. Grondey, M. Jaroniec, G.A. Ozin, Adv.
Funct. Mater., 11 (2001) 447.
[131 M. Kruk, T. Asefa, M. Jaroniec, G. A. Ozin, Stud. Surf. Sci. Catal., 141 (2002) 197-204.
[14] J. Jarupatrakorn,T.D. Tilley, J. Am. Chem. Soc. 124 (2002) 8380-8388
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilko% and P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 493
A vapor phase transport method is applied to the synthesis of composite zeolitic mesoporous
aluminosilicate. Analytical data suggest that aluminosilicate in mesopore wall is partially
transformed to a zeolitic unit, and the composite shows the zeolitic nature.
1. INTRODUCTION
Since the report by Kresge et al. [1], synthesis of mesoporous materials has widely been
conducted over the world, and recently, application of the mesoporous materials as a catalyst
support and a catalyst itself can be seen in many journals. Though Al-containing
mesoporous silica have been synthesized, the acidity is known as weak as that on amorphous
silica-alumina [2]. Various attempts have been dedicated to create stronger acid sites on
mesoporous materials; for instance, incorporation of sulfonic acid group inside walls of
mesoporous silica has been achieved to produce strong acidic sites in the mesostructures [3].
It must be a good opportunity to plant AI atoms, whose environment is zeolitic, to utilize the
materials for the reactions requiring strong acidity such as catalytic cracking and so on.
Although a composite material of zeolite and mesoporous silica such as MCM-41 has been
endeavored to synthesize [4-8], the possibility that the materials are physically mixed ones of
both component of zeolites and mesoporous materials, cannot be excluded because of their
hydrothermal synthetic process. Furthermore, the synthesis is inevitably limited to materials
which are prepared in accordance with thermodynamic aspects [9] From several points of
view, mesopore system such as hexagonal or cubic, should be tunable, and zeolitic phases
must be variable. Very recently much more attention has been attracted for usage of
mesoporous carbon as a replica toward crystallization of nanosize zeolites having a
mesospacing [ 10,11 ].
Matsukata and coworkers have widely investigated a new synthesis method of zeolites
called dry gel conversion [12]. By this method, dried aluminosilicate gel is transformed
directly into crystalline zeolites by the help of water steam and/or vapor of organic
structure-directing agents (SDA). The merit of this method for zeolitic mesoporous material
synthesis is that nucleation of zeolites takes place on the surface of mesoporous materials
using mesoporous (alumino)silicate as the sources for zeolites. Using the vapor phase
transport method, a zeolitic mesoporous composite material can be easily obtained without
consideration of physical mixture of each component. Also by use of this technique, the
mesopore system and zeolitic phase can be selected as we like. In this study, we will report
494
for the first time, the synthesis of a composite material of SBA-15 and ZSM-5 by the vapor
phase transport (VPT) method, one of the dry gel conversion methods to supply SDAs from
vapor phase.
2. E X P E R I M E N T A L
All silica SBA-15 was synthesized according to the literature [13]. The product was
confirmed by XRD to be well organized in a hexagonal ordering, which was calcined in
flowing air at 540 ~ to remove the surfactant molecules inside the mesopores. Then, it was
impregnated into aqueous solution of sodium aluminate or aluminum chloride to form
A1-SBA- 15 (Fig. 1a).
VPT was carried out in a Teflon-lined autoclave, at the bottom of which
ethylenediamine and triethylamine with H20 by the molar ratio of 1: 7.7:10 according to the
literature [14] were placed apart from A1-SBA-15, at 175 ~ for several days.
In order to support the mesostructure when VPT was carried out, furfuryl alcohol or
sucrose was used, followed by carbonization in the mesopores at 800 or 900 ~ for 6 h
according to the procedure to prepare CMK-3 [15], and then A1-SBA-15/carbon composite
was applied to VPT zeolitization. Carbonaceous materials were then burned out of the
composite at 600 ~ to regenerate mesopores.
The product was characterized by use of XRD, SEM, TEM, FT-IR, and nitrogen
adsorption. The catalytic performance was evaluated by cumene cracking in a pulse-type
reactor at several temperatures.
After several days of VPT, the diffraction assigned to I00 plane of SBA-15 was gradually
decreased, while that to ZSM-5 was increased with the VPT period. Fig. 1 shows the typical
XRD patterns of the products after VPT for 5 and 12 days (Fig. l b and 1c, respectively). 5
days of VPT resulted in the formation of ZSM-5 with a little disordered mesostructure in nm
range of d spacing. Prolonged VPT, 12 days in this figure, preferably formed ZSM-5
-4
o~
c
b
._=
C
I
I I I I I I I _.]
0 1 2 3 4 5 10 20 30 40 400 500 600 700 800 900
2theta (CuKcz) / deg wavenumber / cm -1
Fig. 1. XRD patterns and infrared spectra for A1-SBA-15 (a), after 5 days (b) and 12 days (c) of
VPT synthesis of the A1-SBA-15.
495
Fig. 2. SEM images for AI-SBA-15 (a), after 5 days (b) and 12 days (c) of VPT synthesis of the
A1-SBA-15. The bar indicated in the images corresponds to 1 pm.
crystals and the mesostructure was almost disappeared. The catalytic performance revealed
that the ZSM-5 obtained in this way showed a comparable activity for acid-catalyzed cumene
cracking as ZSM-5 commercially available, indicating that aluminosilicate zeolites can be
crystallized by this method, and that Al-containing mesoporous silica can be used as a source
for zeolites. Fourier transformed infrared spectra are also demonstrated in the same figure.
The absorption at 550 cm 1 was assigned to the framework vibration of ZSM-5, and the
samples obtained by VPT showed an intense peak.
SEM images, as summarized in Fig. 2, showed that the typical morphology of SBA-15
was observed even after ZSM-5 crystallites was observed on the surface of SBA-15. The
crystallites seem to be aggregated, leading to the destruction of the morphology.
In order to maintain the mesostructure of SBA-15, carbonization in the mesopores was
adopted as a filling method of the mesopores, in accordance with the procedure for the
preparation of mesoporous carbon, CMK-3, where sucrose was used as a carbon source and
carbonization was carried out at 900 ~ The XRD patterns of the products obtained after
the VPT of carbon/A1-SBA-15 composite for several days are illustrated in Fig. 3. Fig. 4
summarizes the IR spectra of the samples after the VPT. It is noted that the peak contributed
to the framework vibration of ZSM-5 could be detected even before the appearance of the
X-ray diffraction, meaning that a small portion of ZSM-5, that is a kind of building unit of
ZSM-5, is formed before configuration in a long range order. Adsorptive analyses of
nitrogen (Fig. 5) also support this phenomenon, and the mesostructure was hardly observed
after 12 days of VPT, while the microporous structure was gradually developed.
When using furfuryl alcohol as the carbon source, crystallization into ZSM-5 appeared
to be delayed, and the X-ray diffraction at the low angle near 1 deg clearly remained after 12
days of VPT. It has not been essentially clarified yet whether the delayed crystallization is
due to the carbon source, but when using sucrose as the source, 12 days of VPT resulted in the
major crystallization of ZSM-5 as shown in Fig. 3. It might depend on the micro- or
mesostructure of the carbon/SBA-15 composite. If in the case that micro/meso- porosity
exists in the composite, the SDA for ZSM-5 could easily diffuse through the pore system.
When sodium aluminate was used as an A1 source to insert acid sites, crystallization into
ZSM-5 was observed faster than that using A1C13. It strongly depends on the state of
aluminum, and the environment around aluminum in sodium aluminate was confirmed by
27A1 MAS N M R to be tetra coordination to silica, maybe resulting in the fast occlusion into
496
the framework of zeolite. At this stage, it can be deduced that to control the rates of
nucleation and crystallization on the mesopore walls is the key to fabricate micro/meso
composite materials.
Q.
~@~+~ ' .
C"
I I I I I I
1 2 3 4 5 5 10 15 20 25 30 35 40
2theta(CuKo~) / deg
Fig. 3. XRD patterns for carbon/A1-SBA-15 after 12 h (a), 1 day (b), 3days (c), 5 days (d) and 12
days (e) of VPT synthesis.
[
a
5 I
E
t-
Fig. 4. Infrared spectra for carbon/A1-SBA-15 (a) after 12h (b), 1 day (c), 3 days (d), 5 days (e), and
12 days (f) of VPT synthesis.
497
5 c
0.00.20.40.60.81.0 5 10 15 20
P/Po Pore size D /nm
Fig. 5. Nitrogen adsorption isotherms and the pore size distribution calculated by BJH equation for
carbon/Al-SBA-15 after 1 day (a), 3 days (b), 5 days (c), and 12 days (d) of VPT synthesis. Data
were collected after removal of carbon.
Fig. 6 shows the XRD patterns of A1-SBA-15 as the source of the composite, the
mesoporous carbon prepared at 800 ~ by furfuryl alcohol, and the zeolitic mesocomposite
material using the mesocarbon as a filler after 5 days of VPT. The dl00 of hexagonal
mesostructure was hardly changed throughout the synthetic process, indicating that the
mesoporous carbon is highly ordered and prepared in a well-defined mesostructure of the
AI-SBA-15. The zeolitic mesocomposite maintained its mesostructure even after the partial
transformation of mesopore walls into ZSM-5 by VPT. The typical TEM image of the
mesocomposite is shown in Fig. 7. Crystallite-like materials were hardly observed and the
well ordered mesostructure was found in the sample.
It is noted that the as-made A1-SBA-15 hardly showed the catalytic activity for cumene
cracking at a low temperature of 300 ~ while the materials after the VPT but before
developing a long-range order of zeolite crystal detectable by XRD showed the catalytic
performance. This indicates that some kinds of a tiny building unit of ZSM-5 would be
already involved on the mesopore surface of SBA-15.
498
>,
09
c"
(1)
c-
i I I I I
1 2 3 4 5
2theta(CuKoO / deg
Fig. 6. XRD patterns of A1-SBA-15 (a), CMK-3 (b), and the product after 5 days of VPT synthesis
(c).
Fig. 7. A typical TEM image of the product after 5 days of VPT synthesis of AI-SBA-15.
499
on
~ (~ @/~ vPTcrystallizati~
mesoporous alumint c
REFERENCES
[ 1] C.T.Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, and J.S. Beck, Nature, 359 (1992) 710.
[2] A. Corma, Chem. Rev., 97 (1997) 2373.
[3] D. Margolese, J.A. Melero, S.C. Christiansen, B.F. Chmelka, and G.D. Stucky, Chem. Mater., 12
(2000) 2448.
500
[4] K.R. Kloetstra, H.W. Zandbergen, J.C. Jansen, and H. van Bekkum, Microporous Mater., 6
(1996) 287.
[5] A. Karlsson, M. StOcker, and R. Schmidt, Microporous Mesoporous Mater., 27 (1999) 181.
[6] L. Huang, W. Guo, P. Deng, Z. Xue, and Q. Li, J. Phys. Chem., B 104 (2000) 2817.
[7] Y. Liu, W. Zhang, and T.J. Pinnavaia, Angew. Chem. Int. Ed., 40 (2001) 1255.
[8] M. Ogura, E. Kikuchi, and M. Matsukata, Stud. Surf. Sci. Catal., 135 (2001) 255.
[9] M. Ogura, H. Miyoshi, S.P. Naik, and T. Okubo, J. Am. Chem. Soc., 126 (2004) 10937.
[10] C.S. Carr, S. Kaskel, and D.F. Shantz, Chem. Mater., 16 (2004) 3139.
[11 ] A. Sakthivel, S.-J. Huang, W.-H. Chen, Z.-H. Lan, K.-H. Chen, T.-W. Kim, R. Ryoo, A.S.T.
Chiang, and S.-B. Liu, Chem. Mater., 16 (2004) 3168.
[ 12] M. Matsukata, M. Ogura, T. Osaki, P.R.H. Prasad Rao, M. Nomura, and E. Kikuchi, Top. Catal.,
9 (1999) 77.
[13] D. Zhao, J. Feng, Q. Huo, N. Melosh, G.H. Fredrickson, B.F. Chmelka, and G.D. Stucky,
Science, 279 (1998) 548.
[14] W. Xu, J. Dong, J. Li, W. Li, and F. Wu, J. Chem. Soc., Chem. Commun., 1990, 755.
[ 15] S. Jun, S.H. Joo, R. Ryoo, M. Kruk, M. Jaroniec, Z. Liu, T. Ohsuna, and O. Terasaki, J. Am.
Chem. Soc., 122 (2000) 10712.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 501
Al-rich mesoporous FSM-16 materials (Si/A1 = 5-16) have been synthesized using CTMABr
as a surfactant. Additional mechanistic insight into the formation of A1-FSM-16 materials is
gained from careful characterization of the samples collected at various stages of the
preparation procedure by means of XRD and MAS NMR (29Si and 27A1). These results
suggest that the single sheet structure of kanemite breaks up during the intercalation process
and regains its structure upon the addition of Na-aluminate and then condensation occurs
between the adjacent silicate layers, leading to the formation of mesoporous AI-FSM-16
materials. Fe-exchanged A1-FSM-16 samples were also prepared and their performance
tested in the decomposition of N20 in a fixed bed reactor.
1. INTRODUCTION
Mesoporous materials have stimulated extensive research efforts because of their potential
applications in catalytic, adsorption, separation and nanoscience processes [1]. In the past
decade, a wide variety of mesoporous materials have been synthesized using different types of
surfactants [2]. FSM-16 materials [3], in particular, have received considerable attention
because of their higher thermal and hydrothermal stabilities than the most extensively studied
MCM-41 (hexagonal, p6mm) materials. The exceptional thermal and hydrothermal stabilities
were attributed to their thicker pore walls. Therefore, these FSM-16 materials are of
industrial significance for catalytic applications. The synthesis and characterization of
FSM-16 and A1-FSM-16 materials are well documented in the literature [4-8]. In addition,
the mechanism of the formation of FSM-16 materials has been extensively studied using
various characterization techniques (XRD, EDXRD, MAS NMR and TEM), which have
provided clearcut evidence for the topochemical folded sheet mechanism involving the
condensation of the reactive silanol groups present on the adjacent silicate layers in CTMA-
kanemite complex [4, 9-10]. This mechanism leads to the formation of a hexagonal array of
channels with thicker pore walls. However, the intralayer condensation of the silicate sheets
502
2. EXPERIMENTAL
2.3. Characterization
X-ray powder diffraction (XRD) patterns, Transmission Electron Micrographs (TEM)
and N2 adsorption isotherms were obtained as described previously [14]. The 29Si MAS
NMR spectra were acquired on Bruker spectrometers (MSL 300 and MSL 500) with different
measurement times (>20 h) in order to achieve acceptable signal-to-noise ratios. The 27A1
MAS NMR spectra were acquired on a Bruker Avance 400 spectrometer under identical
conditions. Chemical analyses of the samples were accomplished using Inductively Coupled
Plasma Emission analysis (ICP, Perkin-Elmer 400).
mesh size) of the catalyst. Helium was used as the carrier gas. In the present study, a
modified residence time of 36 g min mo1-1 was used. The catalyst was activated prior to the
reaction in Helium atmosphere at 550 ~ for 2 h. Both the inlet and outlet concentrations of
N 2 0 were monitored using a nondispersive infrared analyzer (Hartmann & Braun, Uras 10 E).
The percentage conversion of N20 was calculated as follows:
First, we studied the influence of AI source on the synthesis of A1-FSM-16 materials. The
physico-chemical properties of the Al-rich FSM-16 materials prepared using different A1
sources are given in Table 1. Among the A1 sources used, Na-aluminate is found to be the
best source for the preparation of Al-rich FSM-16 material with a Si/AI ratio of 5. The XRD
patterns of the A1-FSM-16 materials showed a single peak at around 1.81-2.06 ~ (20) (not
shown). All the XRD patterns were similar although there were differences in the relative
intensity of the main reflection. As shown in Table 1, the A1-FSM-16 material (sample 1)
prepared using Na-aluminate as A1 source exhibits lower surface area and pore volume than
the other two AI-FSM-16 materials. It seems that the extent of the decrease in surface area
and pore volume is due to the extent of A1 incorporation into the A1-FSM-16 material.
Additionally, it could be due to the thicker pore walls (25.8 A, sample 1). Therefore, Na-
aluminate was chosen as the A1 source for further studies.
Table 1
Physico-chemical properties of Al-rich FSM-16 samples prepared from kanemite a
aSynthesis gel composition: 0.30 Na20 : SiO2:0.05 A1203 : 0.10 CTMABr : 100 H20, bNa-aluminate,
CAluminum sulphate, dAluminum nitrate, eWall thickness = (d-spacing)-(BJH pore diameter).
kanemite, a new peak at around 1.98~ (20) with very low intensity can be seen (curve 'c' in
Fig. 1A), which indicates the formation of mesoporous A1-FSM-16 material even at room
temperature. This result suggests that the single sheet structure of kanemite initially breaks
up during the intercalation process and regains its structure upon the addition of Na-
aluminate, and then condensation occurs between the adjacent silicate layers, leading to the
formation of mesoporous A1-FSM-16 materials. This result is consistent with the
observations made by Chen et al. [9]. Fig. 1A also includes the XRD patterns of the as-
synthesized and calcined A1-FSM-16 (curves 'd' and 'e') samples. Both the XRD patterns of
the samples are identical but the peak appears somewhat more intense and sharper for the
calcined A1-FSM-16 sample than for the as-synthesized A1-FSM-16 sample. In general, three
or four well-defined peaks are observed for FSM-16 samples with very low A1 content (high
Si/Al ratio) [3-4]. The absence of higher order diffraction peaks indicates a slightly
disordered hexagonal packing of channels in the A1-FSM-16 samples.
The samples of Fig. 1A were further characterized by solid state 29Si- and 27A1 MAS
NMR measurements. The results are shown in Fig. 1B and 1C, respectively. The spectrum of
kanemite clearly shows a sharp peak at around-97.6 ppm (curve 'a' in Fig. 1B), which is
assigned to Q3 silicon environments and is in good agreement with previous data of Kimura
et al. [12] and Kato et al. [16]. The 29Si MAS NMR spectrum of CTMA-kanemite reveals a
change in chemical shift from -97.6 ppm to -94.9 ppm (curve 'b' in Fig. 1B). Such a change
in the chemical shift reflects a structural change in the silicate framework of kanemite upon
intercalation. Moreover, the original spectrum reappears (curve 'c' in Fig. 1B) after the
addition of Na-aluminate into CTMA-kanemite and further corroborates our XRD results. The
spectra for both as-synthesized and calcined AI-FSM-16 (curves 'd' and 'e' in Fig. 1B)
materials exhibit a broad and overlapped resonance (between -85 to -120 ppm) from which
the quantification of Q3 and Q4 species is difficult. Broad Q3 and Q4 resonances indicate the
disordered nature of the as-synthesized and calcined A1-FSM-16 samples.
The 27A1 MAS NMR spectra of both kanemite and CTMA-kanemite (curves 'a' and
'b' in Fig. 1C), did not exhibit any characteristic signal between 100 t o - 5 0 ppm, indicating
that the aluminium is absent within the detectable limit of NMR instrument. Interestingly, the
sample collected after the addition of Na-aluminate into CTMA-kanemite shows a single peak
at around 56 ppm assigned to the tetrahedral A1 species (curve 'c' in Fig. 1C). This result
suggests that the AI species present in solution phase are rapidly incorporated into the
framework ofkanemite. The spectrum of the as-synthesized A1-FSM-16 (curve 'd' in Fig. 1C)
sample shows a relatively sharper peak, indicating an increase in structural ordering.
However, the calcined A1-FSM-16 (curve 'e' in Fig. IC) sample shows a much broader signal
than that of the as-synthesized A1-FSM-16. Calcination of the A1-FSM-16 sample at 550 ~
did not result in any change in the chemical shift, indicating that A1 remains unchanged.
Fig. 2 shows the TEM images of the mesoporous A1-FSM-16 materials (Si/A1 = 5 and
16) obtained from kanemite. It can be seen that the A1-FSM-16 materials display a hexagonal
array of channels. It appears that the hexagonal channels are slightly disordered in agreement
with the XRD measurements. The structures of these A1-FSM-16 materials are different from
those of mesoporous materials obtained from kanemite by Inagaki et al. (hexagonal; pH =
8.5) [4] and Kimura et al. (lamellar; pH = 10.9 and square channels; pH = 4.0-6.0) [11-12].
This could be due to the different self-assembly pathways of CTMA ions that occur at
different pH conditions.
505
(e)
~. (d)
_=
9 i' 9 i 9 i 9 " i ,
10 20 30 40 50
20 [o]
B C
(e)
(e)
~ ~ (d)
~__../~. (c)
.......... _~_____+ (c)
(b)
A - 9 4 . 9
(a)
-97"6 A
.............. ~ ~ (a)
, u , f , i , u , i , i , J , i , i , i , i , i , i , i ,
Fig. 1. (A) X-ray powder diffraction patterns of: (a) kanemite and the samples withdrawn at different
stages of transformation ofkanemite into mesoporous AI-FSM-16 materials, (b) after the intercalation
of CTMA ions into kanemite (CTMA-kanemite) at 25 ~ for an intercalation period of 1 h, (c)
addition of Na-aluminate into CTMA-kanemite at 25 ~ for an intercalation period of another 1 h, (d)
as-synthesized A1-FSM-16 at 100 ~ (0.30 Na20 : SiO2:0.05 A1203 : 0.10 CTMABr : 100 H20) and
(e) calcined A1-FSM-16 at 550 ~ for 12 h; (B) 29Si MAS NMR spectra of the samples of (A); and
(C) 27A1MAS NMR spectra of the samples of (A).
506
(a) (b)
Fig. 2. TEM images of AI-FSM-16 materials with different Si/AI ratios: (a) 5 and (b) 16.
Fe-containing FSM-16 materials are also known to be effective catalysts for reactions
such as isomerization of 1-butene [ 17] and hydroxylation of phenol using aqueous solution of
H202 as the oxidant [ 18]. In our group, the direct hydroxylation of toluene and benzene with
N20 using Fe-containing ZSM-5 type catalysts has been studied intensively [19-20]. In
continuation of our previous studies, we have investigated the decomposition of N20 over Fe-
exchanged A1-FSM-16 samples. For comparison, the decomposition of N20 was carried out
over a H-A1-FSM-16 sample under similar reaction conditions. The chemical compositions of
the samples used for this study are given in Table 2. It appears that more A1 was lost (after the
ion-exchange treatment) as a consequence of dealumination. The contents of iron in the
Fe/A1-FSM-16 (0.025 M) and Fe/A1-FSM- 16 (0.1 M) samples are 0.15 and 0.28 wt.%,
respectively.
Table 2
Chemical composition of the catalysts used for N20 decomposition
H-A1-FSM- 16 (0.1 M) b 12 13
Fe/A1-FSM-16 (0.025 M) c 12 35 0.15
Fe/A1-FSM- 16 (0.1 M) c 12 38 0.28
25
20
~ 15
j/
C J
o 5
0
0
450 475 500 525
Reaction temperature (~
Fig. 3. N20 decomposition over H-AI-FSM-16 (=), (Fe(0.15 wt.%)/AI-FSM-16) (A) and (Fe(0.28
wt.%)/A1-FSM-16) (*) catalysts. Reaction conditions: catalyst - 0.2 g, pressure = 1 atm and
modified residence time = 36 g min mol ~
4. C O N C L U S I O N S
Al-rich mesoporous FSM-16 materials (Si/A1 = 5-16) have been synthesized by intercalating
kanemite with CTMA ions and sodium aluminate at room temperature followed by
hydrothermal treatment at 100 ~ XRD and 29Si MAS NMR results provide additional
mechanistic insight into the formation of A1-FSM-16 materials. In addition, XRD and TEM
reveals that the A1-FSM-16 materials obtained from kanemite possess a slightly disordered
hexagonal structure. The catalytic tests reveal that Fe-exchanged A1-FSM-16 materials are
active for N20 decomposition. Further studies on the characterization of Fe/A1-FSM-16
materials and their catalytic performance in the hydroxylation of toluene using N20 as
oxidant are ongoing.
REFERENCES
[1] C.T. Kresge, M. E. Leonowicz, W. J. Roth, J. C. Vartuli and J. S. Beck, Nature, 359 (1992)
710.
[2] U. Ciesla and F. Schtith, Microporous Mesoporous Mater., 27 (1999) 131; M. L. Occelli and
S. Biz, J. Mol. Catal., A: 151 (2000) 225; I. K. Mbaraka, D. R. Radu, V. S.-Y. Lin and B. H.
Shanks, J. Catal., 219 (2003) 329; A. Taguchi and F. Schtith, Microporous Mesoporous
Mater., 77 (2004) 1.
[3] T. Yanagisawa, T. Shimizu, K. Kuroda and C. Kato, Bull. Chem. Soc. Jpn., 63 (1990) 988;
S. Inagaki, A. Koiwai, N. Suzuki, Y. Fukushima and K. Kuroda, Bull. Chem. Soc. Jpn., 69
(1996) 1449.
[4] S. Inagaki, Y. Fukushima and K. Kuroda, J. Chem. Soc., Chem. Commun., (1993) 680.
[5] K. Kuroda, J. Porous Mater., 3 (1996) 107.
[6] Y. Sakamoto, S. Inagaki, T. Ohsuna, N. Ohnishi, Y. Fukushima, Y. Nozue and O. Terasaki,
Microporous Mesoporous Mater., 21 (1998) 589.
[7] S. Inagaki, Y. Fukushima, A. Okada, T. Kurauchi, K. Kuroda and C. Kato in: R. von
Ballmoos, J. B. Higgins, M. M. J. Treacy (Eds.,), Proceedings from the 9 th International
Zeolite Conference, Vol. I, Butterworth-Heinemann, 1992, p. 305.
[8] S. Inagaki, Y. Yamada and Y. Fukushima in: H. Chon, S.-K. Ibm, Y. S. Uh (Eds.), Progress
in Zeolite and Microporous Materials, Studies in Surface Science and Catalysis, Vol. 105 A,
Elsevier, Amsterdam, 1997, p. 109.
508
[91 C-Y Chen, S-Q. Xiao and M. E. Davis, Microporous Mater., 4 (1995) 1.
[lo] S. O'Brien, R. J. Francis, A. Fogg, D. O'Hare, N. Okazaki and K. Kuroda, Chem. Mater., 11
(1999) 1822.
[11] T. Kimura, D. Itoh, N. Okazaki, M. Kaneda, Y. Sakamoto, O. Terasaki, Y. Sugahara and
K. Kuroda, Langmuir, 16 (2000) 7624.
[12] T. Kimura, T. Kamata, M. Fuziwara, Y. Takano, M. Kaneda, Y. Sakamoto, O. Terasaki,
Y. Sugahara and K. Kuroda, Angew. Chem. Int. Ed., 39 (2000) 3855.
[13] T. Shigeno, K. Inoue, T. Kimura, N. Katada, M. Niwa and K. Kuroda, J. Mater. Chem., 13
(2003) 883.
[14] T. Selvam, M. K6stner, G. T. P. Mabande, W. Schwieger, N. Pffinder and R. Schl6gl,
J. Porous Mater., (2004) accepted.
[15] K. Beneke and G. Lagaly, Am. Mineral., 62 (1977) 763.
[16] M. Kato, T. Shigeno, T. Kimura and K. Kuroda, Chem. Mater., 16 (2004) 3224.
[17] J. K. A. Dapaah, Y. Uemichi, A. Ayame, H. Matsuhasi and M. Sugioka, Appl. Catal., A: 187
(1999) 107.
[18] M. M. Mohamed and N. A. Eissa, Mater. Research Bull., 38 (2003) 1993.
[19] B. Vogel, C. Schneider and E. Klemm, Catal. Lett., 79 (2002) 107.
[20] U. Hiemer, E. Klemm, F. Scheffler, T. Selvam, W. Schwieger and G. Emig, Chem. Eng. J.,
101 (2004)17.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 509
Two ordered mesoporous carbons have been obtained by a casting procedure, starting from a
SBA-15 silica. XRD spectra showed that the ordered structure of the parent SBA-15 silica is
maintained and that the samples are ordered carbon replicas. Micro-Raman spectra showed
that rather homogeneous powders are obtained, exhibiting a consistent graphitized carbon
phase along with an amorphous one, notwithstanding the relatively low temperature adopted
during the carbonization process. BET surface area and pore diameter analysis indicate that
these materials could have potential applications as media for gas storage, for example CH4
and H2.
1. INTRODUCTION
Carbons are widely used as industrial adsorbents because of the hydrophobic nature of their
surfaces, high surface area and good thermal stability. The surface properties may also be
modified by oxidation or doping with metals. Commercial carbons, obtained by thermal
treatment of cheap materials, like coal or bitumen, suffer from several drawbacks, i.e. the
wide and poorly controlled distribution of pore sizes and the presence of impurities (sulphur,
metals...) in the precursor. Carbons prepared by synthesis and characterized by a tailored
porous structure are thus attractive for the development of advanced separation systems,
catalysts and in the field of gas storage media [1-7].
Hydrogen storage, for example, is a challenge for material science, as to this purpose
either systems exhibiting a proper reaction with H2 or systems with only slight interaction
with the H2 molecules are possible [8]. As far as physisorption is concerned, materials with a
large specific surface area like activated or nanostructure,1 carbon and carbon nanotubes
(CNTs) [9-20] are eligible candidates.
510
With respect to other nanoporous materials under current investigation, e.g. A, X and
Y zeolites characterized by different pore architecture and composition [21], porous carbons
present the advantage of a lower specific weight.
The synthesis of ordered meso- and microporous carbons involves the use of
mesoporous silica or zeolites as a template [1, 22-29]. The procedure consists in the
infiltration of the inorganic template with an appropriate carbon source, the carbonisation of
the precursor and subsequent removal of the inorganic template using HF solution. In some
cases, the carbon source is the organic template used to obtain the mesoporous silica itself.
Depending on the type of mesostructured silica used, carbons with different structures
and pore sizes are prepared: this work report on the synthesis and the characterization of
ordered mesoporous carbons obtained using a SBA-15 silica as inorganic template. In
particular, samples, synthesised according to a literature recipe [30], have been characterized
in detail as far as their textural and morphological properties are concerned, seeking for
materials with high surface area and low specific weight to be applied in challenging fields,
like hydrogen storage for example.
2. E X P E R I M E N T A L
The parent SBA-15 silica has been synthesised according to ref. [30] by using a non-ionic
oligomeric alkyl-ethylene oxide surfactant as structure directing agent, Pluronic P123. The
adopted carbon source was sucrose, as it was found to give the best structurally ordered and
the most reproducible results. After template removal by means of calcinations at 773 K in
flowing air, mesoporous silica has been impregnated with a sucrose aqueous solution, dried
and then divided into two fractions, which underwent carbonization respectively by means of
(i) treatment under a vacuum at 1173 K; and (ii) treatment under a N2 flow at 1173 K. The
obtained samples are referred to as C-V and C-N2, respectively. Complete silica removal was
performed by leaching with a 5% weight HF aqueous solution, as checked by X-ray
fluorescence spectroscopy.
XRD spectra were collected on a X'Pert Phillips diffractomer using Cu k~ radiation, in
the following conditions: range - (0.7 - 4 ) 20; step width 20 = 0.02; time per step = 10 secs.
BET surface area and pore size analysis were carried out on powders previously
outgassed at 363 K by means of N2 adsorption at 77 K on a Micromeritics ASAP 2100
instrument.
Field Emission Scanning Electron Microscope pictures were collected on a High
Resolution FE-SEM instrument (LEO 1525) with a Gemini Field Emission Column.
Raman spectra were recorded on a Micro-Raman (Renishaw) system, equipped with
an Ar laser (514.5 rim).
corresponding to the negative replica of the parent SBA-15. The expected loss of long range
order, as observed by the disappearance of reflections at higher 20 values, is however limited.
With respect to the parent SBA-15, a shift of d~00 peak towards higher 20 values is observed,
in agreement with literature data [31,32]. The corresponding lattice parameter decreased
down to 96 A (Table 1), supporting the idea of the pores filling by the carbon precursor and
the consequent formation of smaller structures with respect to the silica template.
12000
=
"" 8000- ti iA 2
o !
0
m 4000
C
c
0
1.0 1.5 2.0 2.5 3.0 3.5 4.0
2O
Fig. 1. XRD patterns of the parent SBA-15 silica (dotted curve) and of C-N2 (curve 1) and C-V (curve 2)
samples.
Table 1
Textural properties of parent SBA-15 silica and templated carbons as determined by
combined XRD and N2 adsorption at 77 K.
Isotherms obtained with carbon replicas still show the presence of a steep increase in
adsorption in the low pressure range, so confirming the occurrence of some disordered
microporosity, whereas in the higher pressure range the isotherm shape changes dramatically.
The micropore volume, as evaluated by the t-plot, is however very small (Table 1).
The hysteresis loop, clearly seen with SBA-15, is much less pronounced with C-N2
sample and nearly absent with C-V sample. The absence of hysteresis loop is ascribed to the
smaller size of the channels, as it is not expected for pore sizes as small as 35 A, even in well
ordered MCM-41 mesoporous materials [31].
a."-"8oo ]6oo
I~~176 /o~
/
O0
/ o.
, --A-- C-V
=4oo
o
"----' / 0.0 0.2 0.4 016 0 . 8
=A_-&-&--A-~
"o 200
-8
< 01~ . . . . O' '
oo 0:2 0:4 0'6 8 10
P/Po
Fig. 2. N2 adsorption/desorption at 77 K isotherms on samples C-N2 (squares) and C-V (triangles).
Inset: isotherm on parent SBA-15. Full symbols refer to adsorption; white symbols to desorption.
500 -
400
o
>,, 3OO
o0
r
200 2
u
C
100
0 i i i
900 1200 1500 1800
Wavenumbers (cm
Fig. 5. Micro-Raman spectra recorded on C-N2 (curve 1) and C-V (curve 2) samples.
In the present case, the coupling with a microscope allowed the variation of the
incident beam spot from 50 to 2 It. This did not lead to changes in the spectra, thus showing
that, at this scale, homogeneous powders have been obtained.
On both samples (Fig. 5), two main bands are seen, with maxima around 1353 and
1596 cm -~, due to the D- and G-vibrational modes of partially graphitized materials,
respectively, broadened because of disorder.
Notwithstanding the rather low temperature during the carbonization process (1173 K
in both cases), partially graphitized materials are obtained, as shown by the higher relative
intensity of G-band compared to D-band.
A more graphitized material is likely be obtained by increasing the temperature.
515
4. C O N C L U S I O N S
By impregnating a SBA-15 silica with a sucrose aqueous solution, two porous carbons have
been synthesised. These carbons are characterized by high BET surface area, ordered
mesoporous structure and rope-like morphology resembling that of the SBA-15 silica used as
template.
Notwithstanding the rather low temperature of the carbonization process (1173 K
under vacuum or in a N2 flow), a partial graphitization of the walls of mesoporous channels
has been detected by means of Raman spectroscopy.
On account of the high porosity and low specific weight, such materials could be
promising media not only as supports and adsorbents, but also for gas storage. As already
done with order disordered carbon materials, they might be used as such, or after convenient
functionalization or metal doping.
REFERENCES
[11 J.-S. Lee, S. H. Joo, R. Ryoo, J. Am. Chem. Soc. 170 (2002) 1156.
[2] H. Suda, K. Haraya, J. Phys. Chem. B 101 (1997) 3988.
[3] H. Suda, K. Haraya, Chem. Comm. (1997) 93.
[4] H. Foley, Microporous Mater. 4 (1995) 407.
[5] F. Rodriguez-Reinoso, Carbon 37 (1998)407.
[6] S. Flandrois, B. Simon, Carbon 37 8 (1999) 165.
[7] R. Ryoo, S. Joo, S. Jun, J. Phys. Chem. B 103 (1999) 7743.
IS] A. Zattel, Mater. Today (2003) 24.
[9] A. Chambers, C. Park, R.T.K. Baker, N.M. Rodriguez, J. Chem. Phys. B 102 (1998) 4253.
[10] A.C. Dillon, K.M. Jones, T.A. Bekkedahl, C.H. Kiang, D.S. Bethune, M.J. Heben, Nature 386
(1997) 377.
[11] P. Chen, X. Wu, J. Lin, K.L. Tan, Science 285 (1999) 91.
[12] C. Carpetis, W. Peschka, Int. J. Hydrogen Energy 5 (1980) 539.
[13] J.S. Noh, R.K. Agarwal, J.A. Schwarz, Int. J. Hydrogen Energy 12 (1987) 693.
[14] K.A.G. Amankwah, J.S. Noh, J.A. Schwarz, Int. J. Hydrogen Energy 14 (1989) 437.
[15] R. Chaine, T.K. Bose, Int. J. Hydrogen Energy 19 (1994) 161.
[16] M. Conte, P.P. Prosini, S. Passerini, Mat. Sci. and Eng. B 108 (2004) 2.
[17] M. Hirscher, M. Becher, M. Haluska, F. von Zeppelin, X. Chen, U. Dettlaff-Weglikowska, S.
Roth, J. of Alloys and Compounds 356-357 (2003) 433.
[ 18] H.-M. Cheng, Q.-H. Yang, C. Liu, Carbon 39 (2001) 1447.
[19] F. Lamari Darkrim, P. Malbrunot, G.P. Tartaglia, Int. J. Hydrogen Energy 27 (2002) 193.
[20] A. Zt~ttel, P. Sudan, Ph. Mauron, T. Kiyobayashi, Ch. Emmenegger, L. Schlapbach, Int. J.
Hydrogen Energy 27 (2002) 203.
[21] J. Weitkamp, M. Fritz, S. Ernst, Int. J. Hydrogen Energy 20(12) (1995) 967.
[22] S. Jun, S. H. Joo, R. Ryoo, M. Kruk, M. Jaroniec, Z. Liu, T. Ohsuna, O. Terasaki, J. Am. Chem.
Soc., 122, 10712, 2000.
[23] T. Kiotani, T. Nagai, S. Inoue, A. Tomita, Chem. Mater., 9, 609, 1997.
[24] P.M. Barata-Rodrigues, T.J. Mays, G.D. Moggridge, Carbon 41 (2003) 2231.
516
[25] H. Darmstadt, C. Roy, S. Kaliaguine, S.J. Choi, R. Ryoo, Carbon 40 (2002) 2673.
[26] S.H. Joo, S. Jun, R. Ryoo, Microporous Mesoporous Mater. 44-45 (2001) 153.
[27] J. Kim, J. lee, T. Hyeon, Carbon 42 (2004) 2711.
[28] A.B. Fuertes, S. Alvarez, Carbon 42 (2004) 3049.
[29] A.B. Fuertes, D.M. Nevskaia, Microporous Mesoporous Mater. 62 (2003) 177.
[30] D. Zhao, J. Feng, Q. Huo, N. Melosh, G. H. Fredrickson, B. F. Chmelka, G. D. Stucky, Science,
279, 548, 1998.
[31 ] R.M. Barrer, N. McKenzie, J.S.S. Reay, J. Coll. Sci. 11 (1956) 479.
[32] D. Zhao, J. Sun, Q. Li, G. Stucky, Chem. Mater. 12 (2000) 275.
[33] P.C. Eklund, J.M. Holden, R.A. Jishi, Carbon, 33, (1995) 959.
[34] S.K. Doom, L. Zheng, M.J. O'Connell, Y. Zhu, S. Huang, J. Liu, J. Phys. Chem. B, 109 (2005)
3751.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 517
a Laboratoire de Rfactivit6 de Surface (UMR 7609 CNRS), Universit6 P. & M. Curie 75252
Paris, France
Two routes for the preparation of high surface area silica-alumina materials with strong
Bronsted acidity and open porosity are examined: one uses the organisation of 13-zeolite seeds
with surfactant micelles, and the other the delamination of Al-Magadiite, a naturally occurring
sodium phyllosilicate. The textures of these materials are characterised using Nz-sorption and
the localisation of most of the AI in tetrahedral substitution in the silica framework is shown
using 27A1MAS NMR. These two materials show good activity for the cracking of cumene.
Furthermore, the characterization of their acidity by FTIR of adsorbed CO reveals the
presence of strong and medium Bronsted acid sites.
1. INTRODUCTION
Materials presenting strong acidity in an open porosity have potential applications in the
hydrocracking of bulky molecules. Two main routes are currently explored: the first one is
inspired by the MCM-41 and SBA-15 syntheses but, to overcome the low Bronsted acidity
generally observed for these materials, the walls are either made of, or lined with, zeolite
seeds [1-4]. The second one also uses zeolite-like materials as precursor but proceeds via the
delamination (also called exfoliation) of the transitory lamellar phases sometimes observed
during zeolite synthesis, leading to e.g. ITQ-2 (delamination of MCM-22(P)) or ITQ-6
(delamination of PRE-FER) materials [5,6]. The preparation of these materials is often
delicate because it requires to master the two steps of the synthesis, i.e. the preparation of the
zeolite-like precursor (zeolite seeds or lamellar zeolite phase) and either its swelling and
delamination, or its organisation in a mesoporous material via a template.
In a previous work [7], we have successfully replaced the lamellar zeolite phase used for
the synthesis of ITQ-2 and ITQ-6 materials by a naturally occurring sodium phyllosilicate,
namely Magadiite (Na2Si14Oz9.nH20) [8], which can be easily synthesised with an AI
substitution [9,10]. We have shown that Del-A1-Magadiite possesses remarkably high surface
area, close 500 m2.g~ and Bronsted acidity, and that the local connectivity of SiO4 tetrahedra
in the starting A1-Magadiite is p artlall
" y p reserved in
" Del -AI-Magadiite (29Si MAS NMR:
conservation of the Q~/Q4 ratio (1/3); FTIR and Raman spectroscopy : conservation of the
spectral signature of strained siloxane rings) [7]. Regarding this support, we will in this paper
2. E X P E R I M E N T A L
~MCM-41: The zeolite seeds solution was prepared by hydrothermal treatment at 140~ for
a given time noted tseed and included between 0 and 35h of a mixture containing A1OOH,
SiO2, tetraethylammonium hydroxide (TEAOH), NaOH and H20.
Hexadecyltrimethylammonium bromide (CTAB) was subsequently added to the zeolite seeds
solutions and HCI was added afterward. The final mixtures, whose composition are
AlOOH/SiOz/TEAOH/NaOH/CTAB/HC1/H20=0.39/10/5.9/0.27/2/0.38/200 (molar ratios),
were hydrothermally treated at 140~ during 24h. The solid phases were recovered by
filtration, washed, dried and calcined at 550~
The structure and texture of the samples were characterised by N2 sorption and XRD,
whereas adsorption of CO at low temperature and cumene cracking were used to characterize
their acidity (number and nature of the Bronsted acid sites). The N2 adsorption desorption
isotherms were collected on a ASAP 2010 analyser. Prior to analysis, the samples were
degassed (p<l Pa) at 150~ during 5h. The contributions of microporosity, mesoporosity and
external surface to the overall surface area were estimated from a t-plot (Harkin-Jura) analysis
of the adsorption curve (0.3 nm < t < 0.5 nm mesoporous and external surface; 0.7 nm < t < 1
nm external surface, t being the statistical thickness). The 27A1NMR spectra were recorded on
a Bruker Advance 400 spectrometer operating at 104.3 MHz, with a rt/8 pulse, a pulse delay
of 0.3 s, a spinning rate of 10 kHz and about 3600 scans. The transmission FTIR spectra of
adsorbed CO were obtained on a Bruker Vector 22 spectrometer using a MCT detector
(resolution 2 cm -I, 32 scans per spectrum). The self-supported wafer of about 15 mg/cm 2 was
first calcined in situ under N2 (100 mL.min -1) at 723 K and then evacuated (10-2-10 .3 mbar) at
723 K for l h. The cell was cooled down to ca. 100 K and pulses of CO were admitted.
519
Cumene cracking was carried out in a fixed-bed microreactor under atmospheric pressure at
573 K. Prior to the activity measurement, each catalyst was pretreated in situ in N2 at 773 K
for 2h. The partial pressure of cumene (Aldrich, 99%) was Pcumene=7.3 kPa, the flow rate of
N2 50 mL.min l and the amount of catalyst 50 mg. The only observed products were propene
and benzene. The conversion was measured after the steady state was reached (ca. 4h). All
sample present similar deactivation curves.
3. R E S U L T S
3.1. Del-AI-Magadiite
The surface area of the AI-Magadiite and the Del-Al-Magadiite are reported in Table 1.
The starting material shows a small, mostly external surface area, typical for a clay mineral.
After delamination and calcination, the surface area increases considerably (by a factor of 20).
The adsorption-desorption isotherm of A1-Magadiite is characteristic of materials with a
broad pore size distribution. The decomposition of the surface area in meso- and micro-
porous and external surface area using t-plot (see Table 1) clearly shows that most of the
surface is external or mesoporous. The 27A1NMR spectra of the starting A1-Magadiite and of
the Del-A1-Magadiite are shown in Fig. 1. They both display only one peak located at ca. 52
ppm, characteristic of A1 inserted in tetrahedral substitution in the silica framework (no peak
located at ca. 0 ppm and characteristic of extraframework A1 is detected). A broadening of the
peak toward high field can however be clearly observed after dealumination.
~/~ ite
J i i i i i i
The Bransted acidity of the delaminated A1-Magadiite was studied by means of FTIR
of adsorbed CO and of a model cracking reaction. The cumene cracking activity (Table 1) of
Del-AI-Magadiite is more than half (62 %) that of a H-BEA zeolite (Si/AI=30) and about 4
times higher than that of a commercial silica-alumina (Si/AI=5). This result indicates the
presence of a large amount of medium and/or strong Bronsted sites. FTIR was used in order
to determine the nature of the Bronsted sites. In the OH stretching region, Del-A1-Magadiite
exhibits, after dehydratation, a strong, sharp band at 3748 cm -I attributable to silanol groups,
and a broader one at 3615 cm l (result not shown). The latter can be tentatively assigned to
520
Bronsted acidic Si-OH-A1 bridging groups, since they resonate at a frequency close to that
observed for such groups in H-ZSM5 (3616 cm -l) [11,12]. After addition of the first pulses of
CO, a partial loss of the v(OH) band located at 3611 cm -I can be readily observed (Fig. 2). It
is accompanied by the appearance of a broad band located at ca. 3300 cm ~ in the v(OH)
range, and of a peak located at 2176 cm i in the v(CO) range. The Av(OH) value (Av(OH)= -
320 cm -~) together with the Av(CO) value (Av(CO) = 34 cm -1, reference: liquid-like CO at
2142 cm -~) confirm that this site is indeed a strong Bronsted acid site, whose structure is most
probably the same as in zeolite (i.e. Si-OH-A1 bridges). At higher CO coverage a second band
is observed in the v(OH) range at ca. 3455 cm -~. Its appearance is accompanied by a
broadening of the v(CO) band toward lower wavelength. This v(OH) frequency corresponds
to Bronsted sites of moderate strength usually observed in amorphous silica-alumina [13].
The maximum intensity of this band is about two times that of the strong Bronsted sites (3300
cm-~). At high CO coverage a second and third band can be observed in the v(CO) range at
2158 cm -~ (2139 cm -~ resp.) together with a broad band at 3640 cm -1 in the v(OH) range.
These bands are usually observed in amorphous silica and assigned to CO H-bonded to silanol
{2158 cml; 3640 cm -l} and physisorbed CO (2141 cm l ) [14]. At the maximum CO
coverage, the intensity of the band associated with strong Bronsted sites (3300 cm -~) is about
one third of the intensity of the same band on a BEA zeolite reference sample (Si/AI=30)
(result not shown).
0.2
~o
0.06"
0.04
8 O
,..0
8
0.02
-0.1
0.00"
I I I I I i i Ii Ti I| 1'! T! 1!
2250 2200 2150 2100 2050 2000 3800 3600 3400 3200 3 00
Wavenumber (cm -I) Wavenumber (cm j )
Fig. 2. Difference spectra of Del-AI-Magadiite after progressive CO introduction; left v(OH) range;
right v(CO) range.
3.2. ]3-MCM-41
The influence of the duration of the thermal treatment of the zeolite seeds solution on the
texture, structure and acidity of the final material where examined for samples prepared with
different pretreatment time (tseed). The XRD peaks of the BEA-zeolite structure are only
detected for tseed> 16h, whereas the cumene cracking activity increases constantly with/seed and
is equal to 80% of the activity of the crystalline 13-zeolite after 16h pre-treatment (Fig. 3). N2-
sorption shows the absence of microporosity for tseeds< 16h, whereas a large contribution of
the microporosity to the overall surface area is detected for tseeds> 16 h.
521
Table 1
Physico-chemical characteristics and cumene conversion of the Del-A1-Magadiite and [3-
MCM-41 samples
Si/A1 Surface ( m2.g -l) cumene
ratio micro meso external conversion (%)
AI-Magadiite 30 24 6
Del-A1-Magadiite 30 84 231 198 51
I3-MCM-41 t~eea=l0h 27 590 204 50
tseed-16h 30 0 687 271 68
t~eect=24h 98 105 26 74
H-BEA 30 356 0 289 82
silica alumina 5 575 12
80
BI ,~
60 J / :/
9
/,
40 9 /
i. / .......
9 o
/ 9
20
MCM-41 A A
__ __- ,z
500-
~'~ 400-
~D
2 300-
o 200 . . . . . . . '
9 1
100]
0 i i i i
53.6 ppm
~-MCM-41
i i i r i i i
o,0
0.15
f
0.10 00000
I.
O
6
< 1' 8
0.05 .0.05-
-0.01 -
0.00
I I I I
' I I I I I I ' ' I
2250 2200 2150 2100 2050 2000 3800 3600 3400 3200 3000
Wavenumber (cm-l) Wavenumber (cm-I)
Fig. 6. Difference spectra of 13-MCM-41 after progressive CO introduction; left v(OH) range; right
v(CO) range
The sample with tseed=l Oh was further characterised: small angle X-ray diffraction reveals
the presence of a mesostructure (d(100) = 4.4 nm), whose order is however lower than in
MCM-41 (result not shown). The surface area and pore size are close to the values reported
for MCM-41 materials, but pore size distribution is broader than in MCM-41 and the capillary
condensation occurs in a wider range of P/P~ value than in MCM-41 (Fig. 4).
The 27A1 MAS NMR spectra of this sample together with the spectra of the H-BEA zeolite
are shown on Fig. 5. As for the Del-A1-Magadiite sample, only the peak assigned to A1 in
tetrahedral position is observed (that is AI in tetrahedral substitution in the silica framework),
523
indicating the presence of Bronsted acid sites. This peak is broader and its position is slightly
shifted upfield compared to that of the reference BEA zeolite. The nature of the Bronsted
acid sites was further investigated with FTIR of adsorbed CO. The spectra of the dehydrated
sample (result not shown) shows a strong band at 3750 cm -1 (v(OH) for Si-OH groups)
together with a weak and broad one at 3620 cm -I (v(OH) of Si-OH-AI groups). The difference
FTIR spectra after CO adsorption are shown on Fig. 6A (v(OH) range) Fig. 6B (v(CO) range).
These spectra are very similar to those recorded upon adsorption of CO on Del-A1-Magadiite
(see Fig. 2). They indicate the presence on this support of both strong (v(OH) = 3310 cm -1)
and medium (v(OH)= 3450 cm -l) Bronsted acid sites, and the intensity of the v(OH) band
associated with strong Bronsted sites is close to that of the Del-A1-Magadiite sample.
However this sample displays a smaller proportion of mild Bronsted acid sites.
4. DISCUSSION
Both Del-AI-Magadiite and J3-MCM-41 can be regarded as made of the assembly of building
units: in the case of Del-A1-Magadiite, the building unit is a sheet of Al-Magadiite, generated
during the swelling and delamination process: the swelling step increases the distances
between the layers of the starting material, whereas the delamination leads to a loss of the
coherence between the layers (that is to the formation of the A1-Magadiite sheets); in the case
of J3-MCM-41, the building unit is a BEA zeolite nanoparticle. The assembly process leading
to the final material is also different: in the case of Del-A1-Magadiite the formation of the
final material proceeds by the random condensation of the A1-Magadiite sheets during the
drying and/or calcinations steps, leading to a house of card structure. On the other side, [3-
MCM-41 formation proceeds via the auto-organisation of the zeolite nanoparticles with the
cationic surfactant. These two assembly processes lead to different textures: the house of card
structure of Del-Al-Magadiite is responsible for the broad pore size distribution of this
support and for the large contribution of the external surface (close to 40 %) to the overall
surface area. On the other side, the organisation of the zeolite seeds around the surfactant
micelles during the 13-MCM-41 synthesis leads to regular pore size and to a large contribution
of the mesoporous surface to the overall surface ( >70 %).
The two building units (BEA nanoparticles and A1-Magadiite sheet) have in common to
present strong acidic properties. This is of course the case for the BEA nanoparticles, but it is
indeed also true for AI-Magadiite. A1-Magadiite shows indeed a remarkably high cumene
conversion with regards to its low surface area (6% conversion for a surface area of 24 m2.g
]), indicating a high density of accessible Bronsted sites and/or a strong acidity of these sites.
This result is in agreement with previous FTIR studies of A1-Magadiite [10], which showed
the presence in the v(OH) range of a band located at 3610 cm i , usually observed on zeolite
and assigned to OH groups bridging a silicon and an aluminum atom (Si-OH-A1).
Upon delamination the surface area of the A1-Magadiite is multiplied by 20 whereas the
acidity is multiplied by 12. This indicates that the strong Bronsted acid sites initially present
on the AI-Magadiite layers are either not all accessible or that the Bronsted acid sites undergo
a modification during the delamination-calcination procedure. The CO FTIR study shows that
the delaminated Magadiite still contains strong Bronsted sites but also an important
proportion of medium Bronsted sites (usually found in amorphous silica alumina). This is a
clear indication that the nature of some of the Bronsted sites was partially modified upon
delamination, and it could explain the relative evolutions of the surface area and acidity.
The CO FTIR study of the [3-MCM-41 also shows the presence of strong and medium
524
Bronsted sites. This could be due to a partial collapse of the zeolite seeds upon calcination, as
it is well established that zeolite seeds do not have a high thermal stability [ 15]. Nevertheless,
the fraction of strong Bronsted sites is not negligible and much higher than in an amorphous
silica-alumina. For these two materials the 27A1 solid state NMR shows the presence of a
single peak located at 50-53 ppm and characteristic of aluminium atoms located in tetrahedral
substitution in the silica framework. Compared to the reference materials (either A1-Magadiite
for the Del-A1-Magadiite sample or a BEA zeolite for the [3-MCM-41) a broadening of the
27A1 N M R signal can be observed, which indicates a broader distribution of sites.
5. C O N C L U S I O N
Two high surface area materials with open porosity were successfully prepared, either by the
assembly of BEA zeolite seeds or by the delamination of an Al-substituted sodium
phyllosilicate. These two materials possess good cracking activity. The investigation of the
nature of the Bronsted sites by FTIR of adsorbed CO showed that these two materials possess
both strong and medium Bronsted acid sites. The strong Bronsted sites are correlated with the
presence of a v(OH) IR band located at ca. 3610 cm -l, characteristic of the presence of Si-OH-
A1 bond, that are usually observed in zeolite materials. The [3-MCM-41 seems however to
contain a lower proportion of the medium acid sites. The textures of these two materials are
clearly different: the house of card structure of Del-A1-Magadiite leads to a large pore size
distribution, whereas the [3-MCM-41 possesses a narrow pore size distribution, only slightly
broader than that of a MCM-41 material. These two materials are, thanks to their large pore
size and strong Bronsted acidity, promising supports for hydrotreating reactions.
REFERENCES
Ill D.T. On and S. Kaliaguine, Angew. Chem. Int. Edit., 41 (2002) 1036.
[2] Y. Liu, W. Z. Zhang and T.J. Pinnavaia, J. Am. Chem. Soc., 122 (2000) 8791.
[3] F.S. Xiao, Y. Han, Y.Yu, X.J. Meng, M. Yang and S. Wu, J. Am. Chem. Soc., 124 (2002) 888.
[4] Z.T. Zhang, Y. Han, L. Zhu, R.W. Wang, Y. Yu, S.L. Qiu, D.Y. Zhao and F.S. Xiao, Angew.
Chem. Int. Edit., 40 (2001) 1258.
[5] A. Corma, U. Diaz, M.E. Domine and V. Fornes, J. Am. Chem. Soc., 122 (2000) 2804.
[6] A. Corma, V. Fornes, J.M. Guil, S. Pergher, T.L.M. Maesen and J.G. Buglass, Microporous
Mesoporous Mater., 38 (2000) 301.
[7] B. Zebib, J.-F. Lambert, J. Blanchard and M. Breysse, (to be published).
[8] H.P. Eugster, Science, 157 (1967) 1177.
[9] W. Schwieger, K. Pohl, U. Brenn, C.A. Fyfe, H. Grondey, G. Fu and G.T. Kokotailo, Stud.
Surf. Sci. Catal., 94 (1995) 47.
[10] G. Pal-Borbely and A. Auroux, Stud. Surf. Sci. Catal., 94 (1995) 55.
[11] H.V. Brand, A. Redondo and P.J. Hay, J. Mol. Catal. A, 121 (1997) 45.
[12] S. Coluccia, L. Marchese and G. Martra, Microporous Mesoporous Mater., 30 (1999) 43.
[13] W. Daniell, U. Schubert, R. Glockler, A. Meyer, K. Noweck and H. Knozinger, Appl. Catal. A,
196 (2000) 247.
[14] T.P. Beebe, P. Gelin and J.T. Yates, Surf. Sci., 148 (1984) 526.
[15] R. Ravishankar, C. Kirschhock, B.J. Schoeman, D. De Vos, P.J. Grobet, P.A. Jacobs and J.A.
Martens, Proc. International Zeolite Conference, 12th, Baltimore, 3 (1999) 1825.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~ and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 525
X.-Z. W a n g a b *, W.-H. Li b, J.-Y. Lin a, H.-L. Fan a, C.-S. Tian a, B. Z h o n g b and K.-C. Xie a
aKey Laboratory of Coal Science and Technology, Ministry of Education and Shanxi Province,
Institute of Special Chemicals, Taiyuan University of Technology, Taiyuan, 030024 China
E-mail: wangxiaozhong@tyut.edu.cn
bState Key Laboratory of Coal Conversion, Institute of Coal Chemistry, the Chinese Academy
of Sciences, Taiyuan, 030001 China
In the present work, a series of synthesis parameters, including those that may affect the size
of surfactant micelles and that may affect the relative rate of hydrolysis and condensation of
tetraethyl orthosilicate (TEOS), were judiciously adjusted to study their direct influence on
the formation and structure of our previously reported bimodal mesopore silica (designated as
BMS). It is found that both the framework and textural mesopores of BMS silica can be
tailored over a fairly wide size range, but on the whole the textural mesopore size is more
sensitive to the change of the synthesis parameters than that of the framework mesopore size.
The change of three synthesis parameters, such as increasing the amount of ethanol,
decreasing the chain length of surfactant or increasing the alkali/silica molar ratio, would lead
to the mesostructure of the silica obtained to transform from initial BMS into MCM-41.
1. INTRODUCTION
Since the first synthesis of mesoporous MCM-41 materials [ 1], there has been an unparalleled
activity in the design and synthesis of a variety of mesoporous solids with different structural
characteristic. So far, mesoporous M41S materials with hexagonal or cubic structure
characteristic have shown potential importance in various practical applications. However, it
seems that HMS mesoporous materials [2] may be of even larger significance in catalytic
applications due to the presence of a complementary textural mesopores, which would
facilitate mass transport to framework mesopores and has shown the importance in improving
catalytic processes [3]. In earlier investigations [4], we found that through controlling gelation
other than precipitation in a reaction system which used usually to prepare MCM-41
mesoporous materials, a hierarchically structured porous silica gel monolith with well-defined
bimodal mesopore size distribution characteristic ( i.e BMS silica) can be formed at ambient
conditions. To the best of our knowledge, this is the first report of a porous silica material
with bimodal mesopore distribution characteristic, which corresponds to two discernable
inflections at low and high relative pressure, respectively, of the N2 adsorption- desorption
526
isotherm. Further research confirms [5] that the bimodal mesostructure of BMS silica consists
of both the framework mesopores resulting from the surfactant template and the textural
mesopores resulting from the interparticle porosity, and the control of the relative rates of the
hydrolysis and condensation reaction of TEOS and thus of the gelation plays a critical role for
the formation of BMS silica mesostructure. Obviously, smart adjusting of various synthesis
parameters, which would affect the micelle sizes and/or the relative rates, might be an
effective means of controlling structure of the BMS silica. Since numerous applications of
BMS silica are based on their pore structure, which comprises pore size, pore size distribution,
pore volume, surface area, etc, thus, in this paper we examine the possibility of tailoring both
the framework and textural mesopores of BMS silica by adjusting the properties or
concentration of various precursor composition, which would also provide simultaneously us
a better understanding of the structural and textural evolution during the
surfactant-encapsulated silica gel formation.
2. EXPERIMENTAL SECTION
2.1. Synthesis
The synthesis of BMS silica followed the same procedures as previously reported [4.5],
except that various synthesis parameters, typical such as the surfactant alkyl chain length
(Dodecyl(Cl2)-, Myristyl(Cl4)-, Cetyl(Cl6)- and Stearyl(C18)-trimethylammonium Bromide
(TAB)), swelling agent 1,3,5-trimethylbenzene (TMB), catalyst structure and properties
(typical such as some small molecule organic amines), solvent polarity (H20 and ethanol
(EtOH)), TEOS concentration and TEOS/surfactant molar ratio and so on, were adjusted
respectively in a certain range with other composition concentration constant to control the
gelation of reaction system. The gelation was determined visually by the fact that the solution
no longer exhibited bulk fluid behavior. The gelation time was defined as a time between the
start of hydrolysis and the sol-gel transition where the bulk fluidity of the sample is lost. The
final gel composition is: 1.0TEOS : 0.18CI6TAB(CI2,CI4,CIs) : (0-0.84)TMB : (0-5.5)EtOH :
0.46NH3(0.064, amines) : 75H20. After gelation, all of the samples were aged for 5h at room
temperature, and the resulting silica wet gels were washed repeatedly with distilled water in a
centrifuge, dried in air at 343K for overnight, to give the surfactant-contained BMS silica
samples. To remove the template, as-synthesized silica samples were ground into fine
powders and calcined in air at 2K rain ~ to 823K for 6 h.
2.2. Characterization
The powder X-ray diffraction patterns (XRD) were recorded using a D/max-2500 powder
diffractometer with Cu-K a radiation (40kV,100mA), 0.02~ size and 1 s step time over the
range 1~ 0 <8 ~ N2 adsorption isotherms were measured at 77K using a ASAP2000 analyser.
The volume of adsorbed N2 was normalized to standard temperature and pressure. Prior to the
experiments, samples were dehydrated under vacuum (about 10 .3 Torr) at 623K
for 12h. The pore-size distribution was calculated using the desorption branches of the N:
adsorption isotherm and the Barrett-Joyner-Halenda (BJH) formula [6].
527
Initially, we attempt to divide the synthesis parameters into two kinds, one is those that may
affect only the size of surfactant micelle, such as varying the surfactant alkyl chain length or
adding swelling agent TMB, which would thus affect only the framework mesopore size of
the resultant BMS silica, and another is those that may affect mainly the relative rate of
hydrolysis and condensation of TEOS, such as varying the composition concentration, solvent
polarity or catalyst properties and structure etc, which would thus affect mainly the size and
packing of the silica primary particles produced and final affect the textural mesopore size of
the BMS silica obtained. However, the experimental results show that this idea may be
ill-considered, and that any a slight change in synthesis parameters would influence not only
the micelle sizes but the relative rates to a certain degree, which enables both the framework
and textural mesopores of the BMS silica to be adjusted simultaneity over a fairly wide range.
We first investigate the effect of varying the surfactant chain length on the formation and
structure of BMS silica, which has been used previously to tailor the pore size of mesoporous
MCM-41 materials. As we replace the usually used C16 surfactant with C~2, C14, C~8 as
template, respectively, under other composition constant conditions, gel occurs in the
presence of surfactants with alkyl chains having 14 or more carbon atoms. Moreover, the
gelation time decreases rapidly with the decrease of chain length. In contrast, for the shorter
surfactant alkyl chain (n~< 12), precipitation rather than gel is formed rapidly as the precursors
are mixed, even though in the case that the alkalinity of the reaction system is further
decreased to slow the condensation rate of hydrolyzed silica species. This indicates that the
presence of the surfactant in the synthesis also affects the relative rate of hydrolysis and
condensation of TEOS, expected that templates the mesostructure of the resultant silicas.
Moreover, this influence is reinforced with the shortening of surfactant chain length. All
samples prepared under this conditions show qualitatively equivalent XRD diffraction
factures with a strong, relatively broad reflection at 1.5-2.0~ 0, which indicates a lack of
long-range crystallographic order or finite size effect. However, the positions of the intense
reflection are dependant on the change of the alkyl chain length, shifting toward higher 2 0
angle with the decrease of chain length, which is consistence with the general tendency
f
-B b
"~ 1200 "~
800 t
d c '
~"
"6 400 f ~E~
> 0 I i I i
0
0 0.2 0.4 0.6 0.8 1.0 10 100 1000
P/Po Pore Size (A.)
Fig. 1. N2 adsorption isotherms and corresponding BJH pore size distribution curves of BMS silica
samples prepared using C12(a), C~4(b), C~6(c), CIs(d) surfactant as templates, respectively.
528
3200 g 24,
f - 4,Textural Mesopore
e
20 Framework Mesopore
2400
16
o c
"
o 1600 b s 12
< 8!
0
>o 800
00 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
P/Po TMB/TEOS mol ratio
Fig. 2. N2 adsorption isotherms (A) and their framework and textural pore sizes (B) for BMS silicas
prepared at different TMB/TEOS mole ratio: a.0; b.0.065; c.0.194; d.0.332; e.0.451" f.0.58" g.0.84.
observed in the MCM-41 synthesis. This influence of varying the surfactant chain length on
the mesostructure of the resultant silica can be seen more direct from the N2 adsorption results
provided in Fig. 1. It is clear that the decrease of the chain length of surfactant leads to the
adsorption step at p/po-0.2-0.4, which indicates the filling of the framework-confined
mesopores, to shift toward lower p/po, while, at the same time, the adsorption step at
p/po=0.8-1.O, which indicates the filling of the textural mesopores, to shift toward higher p/po.
Corresponding, the framework and textural mesopore sizes of the BMS silica obtained vary
from 2.94 to 2.33nm and from 10.5 to 24.3nm, respectively, as shown in the BJH plots in Fig.
lB. In the case of C12 template, the adsorption step at higher p/po disappears due to the
formation of precipitation, corresponding to a disappearance of textural mesopore.
When the amount of swelling agent TMB is slightly increased from 0 to 0.84 mol with C~6
surfactant as template, the framework mesopore sizes of the resultant BMS silica increase
from 2.71 to 3.65nm due to the swelling effect of TMB to micelles, which is consistence with
the results obtained in the synthesis of MCM-41. However, at the same time, the adsorption
steps denoting the textural mesopore filling shift toward lower p/po and decrease clearly in the
gradient (Fig. 2A). As a result, the average size of the textural mesopore decreases gradually
and the distribution becomes wider upon an increase of the amount of TMB (Fig. 2B).
Depending bn the fact that the interparticle voids of BMS silica are also filled using partial
surfactant [5], the introduction of TMB should also swell the interparticle voids, thus the
textural mesopores. Perhaps, this swelling role would lead to a more relaxed packing of the
primary silica particles, thus a much larger shrink has occurred during calcination to removal
surfactant template.
Increasing the TEOS/C16TAB molar ratio from 5.4 to 16.3 or increasing the TEOS
concentration at constant TEOS/C~6TAB molar ratio from 9.9 to 24.8wt% results in the
textural pore size of the resultant BMS silica to increase markedly from 18.9 to 45.5nm and
from 18.9 to 37.7nm, respectively, while the framework pore size is not affected obviously
(Fig. 3). This indicates that the change of TEOS concentration or TEOS/C~6TAB molar ratio
would mainly affect the relative rate of hydrolyze and condensation of TEOS rather than the
529
A
~ 6 a ,...-.i
4
o o
> >
2
o o
! i i i i i iii i i i r rl rll
micelle sizes in the synthesis of BMS silica, which would thus affect the silica particle size
and its packing, and final affect its textural mesopore size. It should be noted that the degree
of the textural pore size increase resulting from the increase of TEOS concentration at
constant TEOS/C~6TAB ratio is far smaller than that from the increase of TEOS/C~6TAB ratio.
Though the reason is not fully clear yet, it is probably related to the optimal matching
between hydrolyzed silica species and CI6TAB micelles, superfluously hydrolyzed silica
species would stay mainly in bulk solution, which would be favorable for the formation of a
larger silica particle with thicker pore wall, thus a larger textural pore size.
It is well-known that catalyst has a direct influence on the structure and properties of the
silica products obtained not only in the synthesis of conventional silica gel [7.8] but also in
the synthesis of mesoporous MCM-41 [9.10]. When the previously used catalyst ammonia is
replaced by some small molecule organic amines with different structure characteristic,
typical such as ethylamine (EA), n-butylamine (BA), n-hexylamine (HA), 1,2-ethanediamine
(EDA), 1,4-butanediamine (BDA) and 1,6-hexanediamine (HAD), it is found that the
concentration range of these catalysts used suitable to BMS silica synthesis is obvious
different. This is probably related to the alkalinity difference resulting from their structure and
properties difference. As can be seen in Fig. 4, which shows the appearance of the silica
products prepared by various amounts of different base catalysts addition with a constant
other composition concentration, ammonia has the widest concentration range for the
0.064 0.064
Fig. 4. Appearances of sol-gel silica products prepared with different base as catalysts.
530
8
8 -B
6 E
b
6
~D
E d
4 = 4
O 0
> >
2 2
9 9
? 100 1000 ~0
| , | | |,|,
100 1000
Pore Size (A) Pore Size (A)
Fig. 5. BJH pore size distributions for BMS silicas prepared with different bases as catalyst. (A) a.
NH3; b. EA; c. BA; d. HA. (B) a. NH3; b. EDA; c. BDA; d. HDA.
synthesis of BMS silica. According to the physical appearance of the resultant products, three
different typed of behavior can be classified, although these regions are significantly different
for different catalysts. Following the increase of the amount of catalysts used, the resultant
silica products all change gradually from an initial opaque gel monolith via a viscous liquids
between gel and precipitation (i.e. intermediate) to a rapid formed precipitation, which is
similar to the results resulting from the increase of ammonia/silica molar rate reported in
previous work [5]. For the convenience of comparing, the amount of different catalysts used
is selected carefully in such a region, typical as 0.064, to ensure the formation of a silica gel
rather than precipitation. The results show that the gelation time is also influenced by the
catalysts used in the synthesis, gradually decreasing with an increase of the carbon atom
number in organic amine molecules, indicating a increasing of condensation rate. However,
the decreased degree of the gelation time is larger for diamine catalysts than that of
monoamine catalysts. This implies a difference in the relative strength of each base in acting
as a catalyst for the hydrolysis and condensation of TEOS. Generally, the weaker the
alkalinity strength of the catalysts used is, the wider the concentration range suitable to the
formation of a homogeneous gel is.
Fig. 5 provides the BJH plots for the BMS silica samples prepared using different catalysts.
It can be seen from the Fig. 5A that the peak intensity of the framework mesopore distribution
of the samples prepared using EA, BA and HA as catalyst all is weaker, and the
corresponding pore volume is also clearly smaller (<0.4cm3/g) than that of previously
synthesized BMS silicas [4.5]. This is probably related to the similar and lower alkalinity
provided by these catalysts, which results in a slower condensation rate and a weaker
interaction between surfactant micelles and silicate species and thus a more disordered
framework mesopores. In contrast, the corresponding peak intensity from the samples
obtained using diamine molecules as catalyst increases clearly in the order of EDA, BDA,
HAD (Fig. 5B). This implies an in turn increase of the alkalinity intensity, which would
induce a more strong interaction between micelles and silicates, and final a more ordered
framework mesopore can be formed. This influence can be also seen further from the change
of their textural mesopore distribution, the increase of textural pore size from 16.35, 21.03 to
22.3nm for BMS silica samples prepared using EDA, BDA and HAD as catalysts,
531
A 6 B a b c
2
i
?o ....... ;oo .... - ';ooo ?o 100 1000
Pore Size (~) Pore Size (~)
Fig. 6. BJH pore size distributions of BMS silicas prepared with different H20/TEOS molar ratio (A)
a.250; b.150; c.100; d.25 and with different EtOH/TEOS molar ratio (B): a.0; b.1.38" c.2.75" d.4.12;
e.5.49.
respectively, is clearly larger than that samples prepared using EA(16.22nm), BA(18.52nm)
and HA(20.94nm) as catalysts. It is clear that the stronger alkalinity would lead to a faster
condensation rate, thus a mesostructure with a larger primary particle size and a looser
packing geometric, thus a larger textural pore size can be formed.
When the amount of solvent water is increased over a wide range typical from 25 to 250
(H20/TEOS molar ratio), which behaves as a reducing of surfactant concentration, but more
as a lowering of pH values of reaction system, the gelation time increases clearly.
Corresponding, the textural pore size of the resultant BMS silicas decreases markedly from
44.4 to l4.3nm (Fig. 6A), while, at the same time, the framework pore size also decreases
from 3.10 to 2.66nm. The decrease of the framework mesopore size may be related to the
decrease of micelle size resulting from the decrease of CI6TAB concentration, while the
decrease of textural mesopore size indicates even smaller primary silica particles to be formed
due to the decrease of the relative rate of hydrolysis and condense of TEOS. However,
increasing the amount of co-solvent ethanol typical from 0 to 2.75 (EtOH/TEOS molar ratio)
would result in a rapid decrease of the gelation time. The textural pore size thus increases
markedly from 18.6 to 36nm (Fig. 6B), while, at the same time, the framework pore size
decreases from 2.85 to 2.35nm. Further increasing the amount of ethanol over 4.12, the
appearance of the products obtained transforms rapidly from an initial opaque gel to a white
precipitation. Correspondingly, the mesostructure of the resultant silicas varies from the initial
bimodal mesopore distribution (BMS silica) to a hexagonal single mesopore distribution
(MCM-41). The phenomenon may be explained by that, on the one hand, the introduction of
the polar co-solvent ethanol would increase the critical micelle concentration of C~6TAB,
which makes the micelles size, cell size, and thus the framework mesopore size decrease, on
the other hand, this introduction also makes the TEOS dissolve more easy, and thus leads to a
more fast hydrolysis and condensation of TEOS, which is favorable for forming a larger silica
primary particles and taking a looser packing of these particles and thus a larger textural
porosity. In extreme conditions, the precipitation occurs before the gelation due to the further
increase of primary silica particle size. In addition, other some methods such as adjusting the
ammonia/silica molar ratio [5.11], post-synthesis hydrothermal treatment [12] or varying the
532
aging and drying conditions [13] and so on can be also used to tailor both the framework and
textural mesopore sizes and their properties of BMS silica over a certain range.
4. C O N C L U S I O N S
Present study shows that due to the key to synthesis BMS silica is the gelation control of
reaction system, the properties and the relative amount of each precursor in the starting
compositions can influence both the structural and textural properties of the resultant BMS
silicas to a certain extent. Not only is the gelation rate affected, thus the size and packing
geometry of resultant silica primary particles, but also the micelle size also influenced by this
variable, which enables the bimodal mesostructure of BMS silica to be tailored over a wide
size range. These unique mesostructure characteristics of BMS silica make it useful as
catalyst supports, in situations where diffusion limitations could negatively affect yields and
selectivities. The combination of the conventional sol-gel method and the surfactant-
templated mesoporous silica synthesis method would provide simple and versatile synthetic
approaches to orient the design of new multistructured porous materials with a controlled pore
structure on all scales.
ACKNOWLEDGEMENT
This research was supported by the National Natural Science Foundation of China (Grant
No.20073029 and 20371034) and the Youth Science Foundation of Shanxi Province (Grant
No.20032010).
REFERENCES
[1] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli and J.S. Beck, Nature 359 (1992) 710.
[2] P.T. Tanev and T.J. Pinnavaia, Science 267 (1995) 865.
[3] W. Zhang, M. Froba, J. Wang, E Tanev, J. Wong and T.J. Pinnavaia, J. Am. Chem. Soc., 118
(1996) 9164.
[4] X.Z. Wang, T. Dou and Y.Z. Xiao, Chem. Commun., (1998) 1035.
[5] X.Z. Wang, W.H. Li, G.S. Zhu, S.L. Qiu, D.Y. Zhao and B. Zhong, Microporous Mesoporous
Mater., 71 (2004) 87.
[6] E.P. Barrett, L.G. Joyner and P.P. Halenda, J. Am. Chem. Soc., 73 (1951) 373.
[7] S.M. Jones, J. Non-Cryst. Solids., 291 (2001) 206.
[8] E. Framery and P.H. Mutin, J. Sol-Gel. Sci. & Tech., 24 (2002) 191.
[9] X.Z. Wang, T. Dou, Y.Z. Xiao and B. Zhong, J. Natur. Gas Chem.., 8 (1999) 216.
[ 10] W. Lin, Q. Cai, W. Pang, Y. Yue and B. Zou, Microporous Mesoporous Mater., 33 (1999) 187.
[11] X.Z. Wang, T. Dou, Y.Z. Xiao and B. Zhong, Stud. Surf. Sci. Catal., 135 (2001)199.
[12] X.Z. Wang, T. Dou and B. Zhong, in: Proceedings ofthe International Symposium on Solid
State Chemistry in China, J.S. Chen and S.H. Feng (eds), Changchun, China, (2002) 227.
[13] X.Z. Wang, W.H. Li, B. Zhong and D.Y. Zhao, in: Proceedings ofthe 14thInt. Zeol. Conf., E.V.
Steen, L. Callanan and M. Claeys (eds), Cape Town, South Africa, (2004) 555.
Studies in Surface Science and Catalysis, volume 158
J. t~ejka, N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 533
bAdvanced Research Institute of Science and Engineering, Waseda University, 3-4-1, Okubo,
Shinjuku-ku, Tokyo 169-8555 Japan
Rapid synthesis of Si-MCM-41 was accomplished in successive two steps; (1) aging of a mixture of
silica, hexadecylammonium bromide and water with an aqueous ammonia solution for a short
period of 30 min, and (2) successive evaporation-to-dryness of the resultant mixture at 80~ in about
2 h. The yield of Si-MCM-41 reached 99% after calcination at 550~ The product obtained by this
method gave a well-ordered 2d-hexagonal structure with high thermal stability and water-resistance.
It was noteworthy that the d~00 value of calcined Si-MCM-41 was almost the same as that of
as-made Si-MCM-41, suggesting that pore wall of as-made product was dense and rigid.
1. INTRODUCTION
Since Kresge et al. [1] and Yanagisawa et al. [2] firstly reported the synthesis of ordered
mesoporous silica materials by using a cetyltrimethylammonium-type surfactant as a template, the
mesoporous silica materials have attracted much attention because of its uniform mesopores, high
surface area and thermal stability. MCM-41-type mesoporous materials were generally prepared by
hydrothermal reactions of silicates in an aqueous NaOH solution in the temperature range from 100
to 150~ [ 1]. MCM-41 containing transition metals such as Ti, V and Mo, has been found to be
catalysts for production of fine chemical [3], and oxidation of organic molecules to valuable
oxygenates [4-5].
There have been literatures reporting synthesis of MCM-41 in an alkali-flee system. Chen et
al. [6] reported synthesis of aluminosilicate MCM-41 using an aqueous ammonia solution to adjust
the pH of reactant gels, where tetramethylammonium hydroxide (TMAOH) and sodium species
were present. Zhao et al. [7] synthesized highly ordered MCM-41 using an aqueous ammonia
solution without any organic and inorganic salts for 72 h at 100~ Recently, syntheses of
mesoporous silica using an aqueous ammonia solution at room temperature have been investigated
[8-9]. Wang et al. [8] reported that a double-mesoporous silica having two kinds of pore sizes of
2.6-3.0 and ca. 19 nm was prepared using an aqueous ammonia in an alkali-flee system at room
temperature. Tsutsumi and co-workers [9] suggested that the usage of ammonia as a catalyst and
tetraalkoxysilanes as silica sources provided a fast and convenient method for the preparation of
highly ordered MCM-41. In their procedure, when the mixture of tetraethoxysilanes (TEOS),
surfactant and aqueous ammonia solution was stirred for only 1 h at room temperature, the
precipitates of silica possessed a 2d-hexagonal structure.
In this study, we developed a rapidly crystallization method of high-ordered Si-MCM-41
using an aqueous ammonia solution in an Na-free system. The thermal stability and water-resistance
of Si-MCM-41 produced in this method was investigated.
534
2. EXPERIMENTAL
The water-vapor resistance of calcined MCM-41 was tested at 20~ for 24 h under saturated
vapor pressure of water in a glass desiccator.
2.2. Characterization
All as-made and calcined products were checked about their crystallinity and phase purity by
X-ray powder diffraction (XRD) on RINT 2100 (Rigaku) using CuKct radiation at 40 kV and 30
mA. Textural properties of the calcined mesoporous products were determined from N2 sorption
experiments at -196~ on BELSORP-mini (Bel Japan Inc.). Specific surface area was calculated
by using the Brunauer-Emmett-Teller (BET) method [10]. Pore size distributions of the calcined
products were determined from adsorption branch by using the Barrett-Joyner-Halenda (BJH)
method [ 11].
3.1. Product phases of silica prepared using parent mixtures of various compositions
We synthesized silica materials from the mixtures having various SiO2/H20 molar ratios and
surfactant concentrations with the presence of an aqueous ammonia solution by using a rapid
crystallization method which is aging for 30 rain at 20~ and successive evaporation-to-dryness at
80~ till drying up. Fig. 1 summarizes the synthetic results. A 2d-hexagonal structure was
formed in the limited conditions in the ranges of SiO2flrI20 molar ratio from 103 to 104 and the
concentration of C16TMAB from 0.01 to 0.10 mol dm 3. A lamellar structure was obtained
regardless of the concentration of TEOS, when the concentration of C~6TMAB exceeded 0.1 mol
dm -3. The formation of lamellar phase was caused by the formation of plate-type micelles of
silica-surfactant complexes under the conditions containing an excess amount of surfactant in the
parent mixture. No periodic structure was obtained with the concentrations of Cl6TMAB < 0.01
mol dm -3 at SiO2/H20 ratio > 104.
535
106
9~o 1 05
A A lid
o
E 104 7-1
OO4
-1- FI
'a
O
10 3 7-1
Fig. 1. Productphases synthesized by using the parent mixtures having various concentrations of
TEOS and CTAB. e, 2d-hexagonal; n, lamellar and A, amorphous.
536
100
o0
Q.
o
2"
or)
c-
O
c-
100
5000 I
cps
~,.~ ~ a)
1 2 3 4 5 6 7 8 9 10
2 0 (CuKot) / degree
Fig. 2. XRD patterns for the products synthesized at pH = 11 using (a) aqueous NaOH solution and (b)
aqueous ammonia solution.
Table 1
Product phases and yields of silica in the course of aging and drying treatments
Synthetic conditions Remained Product
No. Aging period Drying period pH solvent Phase Yield of silica
/ min / rain / vol% / wt%
1 5 --- 11.2 - I O0 P6mm 21.7
2 30 --- 10.7 96.8 P6mm 43.4
3 30 5 9.2 81.8 P6mm 96.1
4 30 30 8.1 68.7 P6mm 96.6
5 30 60 7.1 45.5 P6mm 95.4
6 30 90 7.0 10.5 P6mm 91.8
7 30 120 . . . . . . P6mm 99.8
Aging and drying treatments were carried out at 20 and 80~ respectively.
The solid products (No.l--6) were recovered by filtration, and the dried product (No.7) was directly
recovered without filtration.
537
100
oo
Q.
o
110 200 c)
(D 210
t-
110200 b)
. 210
looooI
cps
110 200
210
a)
1 2 3 4 5 6 7 8 9 10
2 0 (CuKot) / d e g r e e
Fig. 3. XRD patterns for the as-made products synthesized by the rapid crystallization method, a) No. 1,
b) No. 2, c) No. 4 and d) No.7.
600
, :1) :1
| f ;) i',\ ', I
~,~ 500
2"
F:
- (a)
'1 I
:1 I
i-400
.~ _.. d ) . z 300
,.Q ~,c~2.5
s
..,g E 2.0
c) c-
"6
~ 200 1 ~E 1.5
o 1.0
n
~0.5
~ 100 >
~o.0
0 5 10
Pore diameter / nm
10000
cps I I I I I I i l i I t t t t ,'i-,',-, ;'; : : . ; : z--;-. ;-4. ' - L - - : . . : - : - :
0
0.0 0.5 1.0
1 2 3 4 5 6 7 8 9 10 Relative pressure, P/Po
20 (CuKo0 / degree
Fig. 4. XRD patterns for the (a) as-made and (b) Fig. 5. (a) A nitrogen adsorption isotherm for
calcined products (No. 7) aider aging for 30 rain and calcined Si-MCM-41 (No. 7) obtained by aging for
drying for 120 min. (c) As-made and (d) calcined 30 min and drying for 120 min. e, adsorption and o,
products (No. 3) after aging for 30 rain and drying desorption. (b) A pore size distribution determined
for 5 min. from the adsorption branch using the BJH method.
Table 2
The d spacing values and thickness of pore wall of as-made and calcined products having
p6mm structure
dl00 dl00 Pore diameter* Thickness of BET
No. (as-made) (calcined) Adl00 (BJH) mesopore wall surface area
/nm
' / nm / nm / nm / nm / m2 g4
1 4.18 3.76 0.42 2.45 1.89 599
2 3.94 3.87 0.07 2.55 1.92 932
3 3.99 3.49 0.50 2.80 1.23 1507
4 4.01 3.78 0.23 2.60 1.76 1130
5 4.18 4.03 0.15 2.83 1.82 1236
6 4.09 4.01 0.08 2.45 2.18 923
7 4.11 4.03 0.08 2.45 2.20 1233
*Mesopore diameter of MCM-41 was determined from adsorption branch using the BJH method.
539
c)
or)
e-
b)
b)
looooll
cps ,11 -,"" "",'"'"-
a, i 1| i ill! I i ill I i Ill tT"l"~t~, , , , ,.; r-r. T: ;.~
a)
; : ; ;;-;~
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
2 8 (CuKo0 / degree 2 0 (CuKo0 / degree
Fig. 6. XRD patterns for Si-MCM-41 (No. 7) Fig. 7. XRD patterns for Si-MCM-41 (No. 4)
after aging for 30 min and drying for 120 rain. (a) after aging for 30 rain and drying for 5 min. (a)
as-made, (b) calcined and (c) water-treated as-made, (b) calcined and (c) water-treated
Si-MCM-41. Si-MCM-41.
Figs. 6b and c indicate that a well-ordered 2d-hexagonal structure remained alier exposing
calcined Si-MCM-41 (No. 7) under saturated water vapor at 20~ for 24 h, whereas Figs. 7b and c
show that calcined Si-MCM-41 (No. 4) was broken after the water-resistant test. We deduce that
such high water resistance of Si-MCM-41 obtained by a rapid crystallization method was caused
by a highly dense silicate structure in the mesopore walls.
4. CONCLUSIONS
5. ACKNOWLEGDEMENTS
This study was partially supported by Grant-in-Aid for the 2 I COE Research Program "Practical
Nano-Chemistry" from MEXT, Japan, and Core Research for Evolutional Science and Technology
(CREST).
REFERENCES
ill C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli and J.S. Beck, Nature, 359 (1992) 710.
[2] T. Yanagisawa, T. Shimizu, K. Kuroda and C. Kato, Bull. Chem. Soc. Jpn., 63 (1990) 988.
[3] ET. Tanev and M. Chibwe and T.J. Pinnavaia, Nature, 368 (1994) 321.
[4] K. Wada, K. Yoshida and Y. Watanabe, J. Chem. Soc. Faraday Trans., 91 (1995) 1647.
[5] A. Corma, M.T. Navarro, J.E Pariente, J. Chem. Soc. Chem. Commun. (1994) 147.
[6] C.-Y. Chen, H.-X. Li and M.E. Davis, Microporous Mater., 2 (1993) 17.
[7] X.S. Zhao, GQ. Lu and GJ. Millar, Catal. Lett., 38 (1996) 33.
[8] X. Wang, T. Dou and Y. Xiao, Chem. Commun., (1998) 1035.
[9] a) M. Gran, K.K. Unger, A. Matsumoto and K. Tsutsumi, Microporous Mesoporous Mater., 27 (1999)
207; b) A. Matsumoto, H. Chen, K. Tsutsumi, M. Gran and K.K. Unger, Microporous Mesoporous
Mater., 32 (1999) 55.
[lo] S. Brunauer, P.H. Emmett and E. Teller, J.Am. Chem. Soc., 60 (1938) 309.
[11] E.P. Barrett, L.G Joyner and P.E Halenda, J. Am. Chem. Soc., 73 (1951) 373.
[12] e.g., a) N.K. Raman, M.T. Anderson and C.J. Bfinker, Chem. Mater., 8 (1996) 1682; b) L. Pasqua,
E Testa, R. Aiello, ED. Renzo and E Fajula, Microporous Mesoporous Mater., 44-45 (2001) 111.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 541
I. Diaz, *a V. Gonzfilez-Pefia, a C. M f i r q u e z - A l v a r e z , a J. P ~ r e z - P a r i e n t e , a
and E.S. K i k k i n i d e s b'c
1. INTRODUCTION
Porous alumina is a very interesting material with broad applicability as catalyst, catalyst
support and adsorbent. Despite the textural porosity with wide pore size distribution, and the
rather amorphous nature of transition aluminas they are widely used as supports for various
large-scale technological processes [1,2]. Due to their valuable potential application, the
emergent field of ordered mesoporous materials tried soon to extent the MCM-41 synthesis
route to porous alumina. However, mesoporous alumina represents a much more complex
problem compared to the synthesis of mesoporous silica. Thus, alternative efforts have been
devoted during the last decade to tailor, organise and stabilise their pore structure. This topics
have been properly covered by recent reviews [3,4]. Here, we rather analyse the use of
transmission electron microscopy (TEM) as a tool for investigating the porous structure of
aluminas synthesised in the presence of non-ionic surfactants. Usually, a combination of X-
ray powder diffraction, nitrogen adsorption isotherms and transmission electron microscopy is
used to characterise organised mesoporous alumina. However, for all the mesostructured
aluminas reported in the literature, among those that are stable after calcination, the X-ray
diffractograms show a single reflection in the low angle region, preventing from extracting
any structural information. The combination of that type of X-ray diffraction patterns and N2
adsorption/desorption results have been interpreted in terms of "wormlike" channel motif
based on the analysis of transmission electron microscopy (TEM) images [5,6]. However,
recent TEM studies show that these X-ray patterns and N2 isotherms results could be related
to a wide variety of porous structures [7,8]. Nonetheless, the analysis of the published images
does not allow observing any difference in the pore structure as the synthesis conditions or
calcination temperature change. Since TEM imaging itself does not bring any light to the
complexity of the porous architecture of these disordered materials, additional tools and
542
deeper and more systematic study of the images are needed in order to use TEM as an
efficient technique to characterise disordered solids. In this attempt to carry out a more
systematic study of organised porous aluminas by TEM, we have combined two-dimensional
(2-D) TEM images with three-dimensional (3-D) stochastic reconstruction in order to gain a
deeper insight on the pore structure of these porous materials [9].
The basic principle in the reconstruction methodology is that both the actual and the
model structure should have identical statistical properties, defined over an infinite set of
moments of the pore space. These structural properties can be measured in thin 2-D sections
of the actual material and used as an input for the creation of the simulated 3-D structures,
under the assumption of statistical homogeneity. In practice, for a large number of disordered
porous solids, matching only the first two moments (namely porosity and two-point auto-
correlation function) is adequate for a realistic representation of the pore structure [10]. The
results allow evaluating not only the pore size but also the connectivity of the pore structure,
which is an indication of the pore architecture. Yet, the characterization of disordered porous
solids requires efforts devoted to 3-D reconstruction and simulations of the porous network in
order to predict permeation and diffusion properties. A resulting property from the diffusion-
permeation simulations is the tortousity factor, which is a semi-empirical parameter that
accounts for structural effects such as pore shape, orientation, connectivity, etc., that lead to a
lengthening of the diffusion path and slow down the diffusion process [11,12]. In essence the
tortuosity factor gives an indication on the pore accessibility of the material.
In this work, we analyze the influence of synthesis conditions, such as temperature,
solvent, additives or chemical modification of the precursor, on the porous alumina structure
analysed by transmission electron microscopy (TEM). Stochastic reconstruction models have
been employed to generate three-dimensional structures of so called "wormlike" porous
structure. Furthermore, effective diffusivities are computed using a step-by-step random walk
process and tortousity values are determined for each structure.
2. EXPERIMENTAL SECTION
used, were dissolved in a part of the organic solvent. In order to perform synthesis in reverse
micelle systems, the solution was prepared using cyclohexane as solvent and aluminium
precursor/surfactant molar ratios of 1-2. The hydrolysis was carried out at room temperature
by slowly adding, under stirring, a mixture of water (and an amine or ammonium fluoride
when used) and the remaining solvent. The resulting mixture was stirred for 3 h at room
temperature, and then kept at room temperature or 95 ~ for 24 h. After this, the resulting
gelatinous product was filtered, washed with ethanol and dried at 40 ~ for 2 days. When the
ageing was carried out at room temperature, the dried gel was subsequently heated at 95 ~
for 6 h in an open polypropylene bottle. The surfactant was then removed by Soxhlet
extraction with ethanol for 15 h. The extracted solid was dried at 40 ~ for 2 days and then
calcined in air flow at 550 ~ for 4h. Some samples were further calcined in an oven at
temperatures from 600 to 850 ~ for 3 h.
X-ray powder diffraction patterns (CuKa radiation) were collected using a Seiffert XRD
3000P diffractometer with a curved graphite secondary monochromator, operating at low
angle (1-10 ~ using a primary automatic divergence slit and a 0.2 mm detector slit. Nitrogen
adsorption/desorption isotherms were obtained using an ASAP 2000 Micromeritics apparatus.
The samples were evacuated at 350 ~ for 24 h before analysis. Mercury intrusion
porosimetry data were obtained using a Micromeritics Poresizer 9320 mercury porosimeter.
Porosity was calculated as: %porosity = 100 9 (1-bulk density/skeletal density). For the
transmission electron microscopy (TEM) observations, the samples were crushed in an agate
mortar, dispersed in acetone and dropped on a holey carbon microgrid. Several microscopes
have been used during the progress of this research: JEOL JEM 3010 microscope operating at
300kV (d = 1.6 A), JEOL JEM 2000Fx (d = 3.1 A) operating at 200kV and a Philips 120
Biotwin microscope operating at 120 kV equipped with a Gatan CCD camera. The images
were recorded either by CCD camera or on films and digitised with a normal reflection
scanner or a transmission scanner, delimitating the resolution of the final images.
oe--~
'Note that < - > indicates spatial average. Rz(u) is a measure of the probability that two
points at a distance u are both in the pore space. For an isotropic medium, Rz(u) becomes
one-dimensional as it is only a function of u=lul [ 13,14].
In order to determine the above structural properties, we need to take first each TEM
image and transform it into a binary matrix of zeros and ones for each pixel that belongs to
the solid or void phase, respectively. This procedure was done in the present study using the
software Scion Image Beta 4.02 available for free by Scion Corporation. A TEM image is a 2-
D projection but contains 3-D information at the grayscale so the sample thickness and the
defocus value will affect the final TEM images. Thus an ideal 3-D reconstruction should be
done from a thin section TEM image. In this regard, prior to the 3-D reconstruction, the 2-
point auto-correlation functions were calculated for a number of different TEM images of the
544
same material (3 to 5 images per sample), with various magnitudes of thickness and focus in
order to verify the consistency of our results. Nevertheless, we are aware that systematic
studies accompanied by image simulations for a series of defocus and thickness values should
be done, in order to corroborate the validity of our approximation.
For the micro- and mesostructured aluminas synthesised in the presence of poly(oxyalkylene)
block polymers, the X-ray diffractograms show a single reflection in the low angle region but
a wide variety of N2 adsorption/desorption isotherms. Fig. l a shows a TEM image of a
sample synthesised in butan-2-ol at 25 ~ in the presence of Triton X-114 and calcined at
550 ~ which possesses BET surface area =520 m2.gl, mean pore diameter =5 nm and pore
volume = 0.88 cm3.g ~, as determined from the nitrogen adsorption-desorption isotherm [15].
No discernible ordering of pores can be distinguished in the TEM images, and the material
appears to be built up of very tiny corrugated platelets, which resemble the crystal habit
reported for pseudoboehmite [1,16]. The individual platelets have an estimated thickness of 2
nm, which corresponds approximately to the packing of three individual boehmite-like layers,
each one having a thickness half of the b parameter of the unit cell ofboehmite [15]. Based on
extensive previous studies, no correlation was observed between pore size and surfactant's
head volume when the synthesis was carried out in butan-2-ol. A plausible explanation for
this result is that the surfactant molecules do not form micelles in butan-2-ol under the
conditions used in these experiments. In fact, the results reported by Ray [17] suggest that the
non-ionic surfactants used in these syntheses would not form micelles in alkanol solutions.
However, the use of direct (1,4-dioxane solution) or reverse (cyclohexane solution) micellar
systems has no significant effect on the disordered porous structure of the alumina. The
mesoporous aluminas prepared using either solvent and calcined at 550~ exhibit an intricate
crosslinking of corrugated oxide-hydroxide platelets, similar to that of the aluminas
synthesised in butan-2-ol (Fig. 1). Moreover, the samples prepared from sols obtained in any
of the three solvents exhibit N2 isotherms having similar shape and close BET surface area
values [8].
In order to obtain aluminas with higher thermal stability and increased acidity, we have
tested the effect of alkylamines and fluoride ions as modifiers added to the synthesis gel in
butan-2-ol [15,18]. The structure of the stable modified alumina depends on the calcination
temperature. Thus, in a sample synthesised in butan-2-ol at 25 ~ in the presence of Triton X-
114 and N,N-dipropylamine (amine/aluminum molar ratio 0.25), calcined at 550~ the
already mentioned corrugated platelets are observed, while further calcination at 700 ~ leads
to a stable material formed by small polycrystalline agglomerates, which evidences a partial
condensation of the boehmite-like layers. Due to this condensation, the BET surface area and
pore volume decrease from 520 to 230 mZ.g-I and from 0.72 to 0.60 cm3-g-1, respectively,
while the mean pore size increases from 4.6 to 8.0 nm [19]. The N2 adsorption-desorption
isotherms of aluminas synthesised in the presence of ammonium fluoride and calcined at
550 ~ are characteristic of mesoporous solids. However, the TEM observations demonstrate
that this alumina consists mainly of agglomerated particles formed by very small densely
packed platelets, showing textural porosity similar to that observed in the samples synthesised
in the presence of amines and calcined at higher temperatures [8]. One step ahead was taken
aiming to tailor both aluminium coordination and the pore structure of the alumina [20,21].
This strategy is based on the chemical modification of the aluminium alkoxide precursor with
chelating ligands, in order to retard the hydrolysis and condensation reaction rates [22].
545
(e
~r
i!50nm
"~":;::_. TYU-:L:i.L ii~;:<~Onm
o ,
>5
-'~ .
For this purpose, the hydrolysis of aluminum sec-butoxide was carried out in the presence
of ethyl acetoacetate and triethanolamine in different solvents [20]. No change was observed
in the X-ray diffraction patterns, although significant modifications were shown by the N2
adsorption results compared to those samples prepared without chelating agents. These
samples are microporous. Moreover, it has to be stressed that this change in the porous
structure is accompanied by a decrease in the surface area from 500 to around 300 m2"g-~.
From the TEM images reported in Fig. 1 it is possible to observe that both triethanolamine
and ethyl acetoacetate produce a denser and isotropic structure, in sharp contrast with the
corrugated platelets observed in the materials prepared without chelating agents. The lower
surface area of the aluminas confirms this increased density. Despite the overall "wormlike"
appareance of the TEM images, the shape of the isotherms evidenced small changes in the
porous network. Although there is no appreciable difference in the porous structure shown in
the images, a deeper TEM study including focus and thickness series, and higher
magnification evidences slightly different features and contrast distribution.
Stochastic reconstruction methods have been applied to elucidate the possible differences
due to the tortuosity or connectivity of these porous aluminas [9]. The method generates 3-D
stochastic representations that have the same basic structural properties with the input 2-D
TEM images. Fig. 2, indicates by visual inspection that the method generates 3-D stochastic
representations that are in qualitative agreement with the input 2-D TEM images (shown in
Fig. 1). However, further quantitative measurements of the different structural properties
have to be obtained in order to draw a more definite conclusion about the proposed
reconstruction methodology.
Fig. 2.3-D reconstructed images of porous alumina. Black colour corresponds to void space and grey
colour to solid phase. The synthesis and calcination conditions are those indicated in Fig. 3 caption.
Fig. 3A presents the two-point auto-correlation functions measured on the TEM images of
each material (dots), compared to the auto-correlation functions calculated on the respective
3-D reconstructed images (lines). The agreement is very good in all cases, which was
expected due to the imposed matching of the auto-correlation functions during the
reconstruction process. It is interesting to observe that samples synthesised with the same
chelating agent show similar behaviour in their auto-correlation functions despite the solvent.
However, the reliability of these calculations strongly depends on the resolution of the TEM
images. We would like to stress once again the necessity of high quality TEM results to
further extract information about the porosity structure in disordered matrices. Therefore, in a
future work, we will explore the comparison between these results with the respective pore
size distributions that come out from a BJH analysis of N2 adsorption-desorption isotherms.
From the stochastic reconstruction and the diffusion measurements, it is possible to
distinguish between the pore architecture generated by the chemical modification with
triethanolamine or ethyl acetoacetate in different solvents. Table 1 summarises the diffusion
studies for four selected samples. The calculations show that the material prepared in butanol
547
and using triethanolamine as modifier (sample e) has a tortuosity very close to the ideal value
of 3. This value indicates that the gas molecules diffuse according to a truly random walk in
which the steps are uncorrelated and occur with equal probability in any direction. If the pore
structure were represented as a network of cylindrical pores this tortuosity value would
correspond to a nearly infinite connectivity among the pores [23]. On the other hand,
modification with ethyl acetoacetate in the same solvent (sample g) leads to a material with a
much higher tortuosity (around 5), indicating lower connectivity and thus lower accessibility
to the pore structure. It is interesting to observe from Table 1 that there is an almost
monotonic pattern between porosity and tortuosity, something that is more clearly seen in Fig.
3B. It is evident that the tortuosity decreases as porosity increases in a non-linear fashion in
accord to previous theoretical and experimental studies. A small deviation in the pattern that
is observed in sample b, prepared in cyclohexane with triethanolamine, is most likely due to
the coarse resolution of the original TEM image and its resulting reconstructed representation.
Therefore the results obtained on the specific sample should be considered with caution.
lu A 10- B
9~i o(e)
o 0,8 - i(f)
~= 9 (g)
0,6 .(h)
x = 1.30 ~-i.335 ~k
0,4
o 0,2 '
4 6 1 9 9 w 9 w 9 9 e
-0,2 J o,1 1
Distance, Bm
Porosity
Fig. 3. A: Two-point auto-correlation functions measured on the TEM images of each material. Solid
lines correspond to the measured functions on the respective 3-D reconstructed images. B: Tortuosity
v e r s u s porosity relationship in porous aluminas. The synthesis and calcination conditions are those
indicated in Fig. 3 caption.
Table 1
Diffusion properties computed from random walk simulations on the reconstructed
domains of porous aluminas. The synthesis and calcination conditions are those indicated
in Fig. 3 caption.
Sample Porosity, ~ Tortuosity factor,
e 0.487 3.15
f 0.466 4.10
g 0.373 5.02
h 0.416 3.83
548
4. C O N C L U S I O N S
REFERENCES
[1] C. Misra in Industrial Alumina Chemicals, ACS, Washington (1986).
[2] H. Topsoe, B.S. Clausen, F. E. Massoth, Hydrotreating Catalysis, Springer, Berlin (1996) 310
[3] F. Scht~th, Chem. Mater. 13 (2001) 3184.
[4] J. (2ejka, Appl. Catal. A 254 (2003) 327.
[5] S.A. Bagshaw, T. J. Pinnavaia, Angew. Chem. Int. Ed. Engl. 35 (1996) 1102.
[6] W. Zhang, T. J. Pinnavaia, Chem. Commun. (1998) 1185.
[7] J. Cejka, P. J.Kooyman, L. Vesel~i, J. Rathousk~,, A. Zukal, Phys. Chem. Chem. Phys. 4 (2002)
4823.
[8] I. Diaz, V. Gonz~ilez-Pefia, C. M~irquez-Alvarez, J. P6rez-Pariente, Collect. Czech. Chem.
Commun. 68 (2003) 1937.
[9] I. Diaz, V. Gonzfilez-Pefia, C. Mfirquez-Alvarez, E. S. Kikkinides, Microporous Mesoporous
Mater. 68 (2004) 11.
[10] J. Yao, P. Frykman, F. Kalaydjian, J. F. Thovert, P. M. Adler, J. Colloid Interf. Sci. 156 (1993)
478.
[11] S. Reyes, K. F. Jensen, Chem. Eng. Sci. 40 (1985) 1723.
[12] M.P. Hollewand, L. F.Gladden, Chem. Eng. Sci. 47 (1992) 1761.
[13] P.M. Adler in Porous Media: Geometry and Transports, Butterworth, London, 1992.
[ 14] S. Torquato in Random Heterogeneous Materials: Microstructure and Macroscopic Properties,
Springer-Verlag, New York, 2002.
[15] V. Gonz~ilez-Pefia, I. Diaz, C. M~irquez-Alvarez, E. Sastre, J. P6rez-Pariente, Microporous
Mesoporous Mater. 44-45 (2001) 203.
[16] T. Ono, Y. Ohguchi, O. Togari, Stud. Surf. Sci. Catal. 16 (1983) 631.
[17] A. Ray, Nature 231 (1971) 313.
[18] V. Gonz/dez-Pefia, C. M/lrquez-Alvarez, E. Sastre, J. P6rez-Pariente, Stud. Surf. Sci. Catal. 135
(2001) 1072.
[19] I. Diaz, V. Gonz~ilez-Pefia, C. M/Lrquez-Alvarez, E. Sastre, J. P6rez-Pariente, Zeolites and
Microporous Crystals 2000. Proc. Int. Symp., Sendai, August 6, 2000, p. 200
[20] V. Gonz~ilez-Pefia, C. M/lrquez-Alvarez, E. Sastre, J. P6rez-Pariente, Stud. Surf. Sci. Catal. 142
(2002) 1283.
[21 ] V. Gonz~ilez-Pefia, C. M/Lrquez-Alvarez, I. Diaz, M. Grande, T. Blasco, J. P6rez-Pariente,
Microporous Mesoporous Mater. 80 (2005) 173.
[22] C.J. Brinker, G.W. Scherer in Sol-gel Science. Academic Press, San Diego 1990.
[23] J. K~irger, D. M. Ruthven in Diffusion in Zeolites, John Wiley & Sons, New York (1992).
Studies in Surface Science and Catalysis, volume 158
J. Cejka, N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 549
1. INTRODUCTION
Recently SBA-15-type materials have attracted much attention due to their appealing textural
properties, high surface area, and appreciable thermal and hydrothermal stability. However
these siliceous materials lack necessary active sites, which limits their application in catalysis
and adsorption. Therefore the functionalization of SBA-15 by surface modification of
incorporating organic / inorganic compounds has been became a focus of recent research
works. Many articles reported the partial substitution of framework silicon with heteroatoms
A1, Ti, Zr, V, and Sn in SBA-15 materials [1-7], whereas the organic anions of sulfonic or
carboxyl were employed to modify the surface properties of SBA-15 [8,9]. Recently the
highly dispersed noble metal nanoparticals such as Au, Pt, Zr that possess special catalytic
properties were introduced on mesoporous silica SBA-15 [ 10-11 ] through different synthetic
processes, including "post-synthesis" and "direct-synthesis" [6,12]. Many new methods were
also reported. For example, Newalkar et al. [13] prepared ZrO2/SBA-15 material under
microwave-assisted hydrothermal system, while a large amount of heteroatoms such as AI
and Ti were grafted onto SBA-15 by a simple "pH adjusting" method [ 14]. Specifically in the
application of mesoporous materials, functionalized SBA-15 with organosilanes has been
utilized to sequestrate and release protein [15], immobilize enzymes [16], remove heavy
metals in pollutant solutions [17], and play important role as chiral heterogeneous catalysts
[18] and solid acid [19]. Furthermore, SBA-15 materials have also been used as matrixes to
template the formation of metal nanowires and mesoporous carbon molecular sieves [20, 21].
However, the basic mesoporous silica materials are still ignored. For solid bases the pending
problem is the leaching of basicity in aqueous solution, which limits their application in
industry. Although loading magnesia on A1-SBA-15 was realized to obtain basic mesoporous
material [12], the "post synthesis" method usually needs to synthesize the host SBA-15 at first
and then to introduce the basic guest by impregnation, which costs time and energy, some
550
times causes waste solution. In order to obtain a novel material with strong basicity and
excellent water-resistance, we tried a "direct synthesis" strategy to prepare the mesoporous
basic materials HT/SBA-15 directly from strong acidic solution by in situ coating hydrotalcite
(HT) into SBA-15. The basicity and porosity of the resulted composite were assessed by CO2
-TPD, FTIR, 27A1 MAS NMR, XRD and Hammett indicator method.
2. EXPERIMENTAL
The composite of n%HT/SBA-15, denoted as nHT/SBA where the n represents the mass
percent of HT coated on SBA-15, was prepared as follows: 2 g triblock copolymer P123
(EOz0POv0EO20, Aldrich), a calculated amount of magnesium acetate and aluminium nitrate
were dissolved in 75 g 1.6 M HCI, then 4.25 g TEOS was added whilst stirring at 313 K. The
molar ratio of mixture TEOS / P123 / Mg(CH3COO)2 / AI(NO3)3 / HCI / H20 was 1: 0.02: 2x:
x: 6:192, where x was the molar ratio of TEOS to AI(NO3)3, and the x value was 0.02, 0.05,
0.08, 0.18, 0.25 and 0.31 for the samples coated by HT with mass percentages of 5, 10, 15,
30, 35 and 40, respectively. The mixed solution was stirred for 24 h at 313 K and then heated
at 373 K for another 24 h under static condition. Finally, the slurry was evaporated with
stirring at 353 K, and the obtained solid was dried at 353 K followed by the calcination at 823
K for 6h to remove the template and convert the metal compounds to metal oxides.
The XRD patterns were acquired on an ARL XTRA diffractometer using Cu Ks
radiation in a 20 range. The nitrogen adsorption and desorption isotherms at 77 K were
measured using a Micrometritics ASAP2000 system in which the samples were degassed for
10h at 573 K in degas port of the adsorption analyzer. The Brunauer-Emmett-Teller (BET)
specific surface area was calculated using adsorption data in the relative pressure range from
0.04 to 0.2, and the total pore volume was determined from the amount adsorbed at a relative
pressure of about 0.99. In HRTEM measurements, the specimens were dispersed in ethanol
and placed on holey copper grids.
In CO2-TPD experiments, 50-mg sample was pretreated in a flow of N2 (99.99%) at 873
K and kept in this temperature for 2h and then cooled to 298 K to adsorb CO2 (99.999%).
After the physically adsorbed CO2 was purged by a flow of N2 at 298 K for 2h, the sample
was heated up to 873 K at the rate of 8 K.min -~, and the liberated CO2 was detected by an on
line Varian 3380 chromatograph with a thermal conductivity detector (TCD).
27A1 MAS NMR spectra of samples were recorded with a Bruker MSL400 spectrometer
operating at frequencies of 104.26 MHz. A Bruker MAS probe head was used with a zirconia
rotor of 4 mm in diameter. All measurements were performed at room temperature with a
spinning rate of 10 kHz. 27A1 spectra were traced by a single pulse sequence combined with a
high-power IH decoupling during signal acquisition. The flip angle of the pulse and the
recycle delay were about rt/12 and ls.
The total basicity of the samples was measured by titration. 50 mg sample was dissolved
in 20 ml 0.02M HC1, and the remaining acid was titrated by 0.02M NaOH aqueous solutions.
In the experiment to assess the water stability of solid base, 50mg sample was washed by
distilled water and filtered until the solution did not show basicity in titration; then the
residual basicity of sample was titrated by use of 0.02M HCI.
these isotherms are type IV with a clear H 1-type hysteresis loop at high relative pressure that
is the characteristic of mesostructure. Although no obvious shift was found on the desorption
isotherms of the modified samples, their pore volume progressively decreased as the coating
amount of HT increased that indicates the location of guest species inside the channel of
SBA-I 5. Fig. 2 displays the TEM images of 15HT/SBA sample in which the ordered one-
dimensional uniform channels are clear and no observable separated particles exist in the
channels of the host. As shown in the inset of Fig. 3, the Low-angle XRD patterns of all the
modified samples demonstrate well resolving diffraction peaks that can be indexed (100),
(110), (200) diffraction being characteristic of the typical P6mm space group, identical to that
of SBA-15. Moreover, the intensity of peak (100) in HT/SBA samples is obviously stronger
than that of the parent SBA-15, this clearly elucidates the addition of magnesium and
aluminum salts effectively improves the ordered structure of SBA. There is no new peak
observed in the wide-angle XRD pattern of HT/SBA-15 (Fig. 3A), except for a broad peak
centred at 2 0 of 24 ~ that is the characteristic of SBA-15, which confirms the high dispersion
of hydrotalcite in SBA-15. Even the coating amount of hydrotalcite reached 40 wt.-% there
was still no oxides crystalline phase, mirroring the high dispersion of the coated species in the
channels of SBA- 15 mesoporous materials.
r~
A d BI
=i c
d 5
= c
.~ b
9 , ! 9 , .
10 20 30 40 50 60
2 Theta/degree
Fig. 3. Wide-angle XRD patterns (A) and low-angle XRD patterns (B) of (a) SBA-15, (b) 15HT/SBA,
(c) 30HT/SBA, and (d) 40HT/SBA.
552
Table 1
Textural properties of SBA- 15 and HT/SBA- 15 composites
Sample SBET/m 2 g-I S*BET/m2 g-I Vp/cm3g-I Dp*/nm ao/nm Wall thickness
/nm
SBA- 15 902 902 1.25 6.3 11.34 5.04
5HT/SBA 745 784 1.13 6.7 11.58 4.88
15HT/SBA 482 567 0.81 7.7 11.86 4.16
30HT/SBA 393 561 0.69 7.8 11.32 3.52
40HT/SBA 309 515 0.58 7.6 11.47 3.87
S'BET is the estimated surface area of SBA-15 component in the HT/SBA-15 composites.
Dp*iS adsorption average pore diameter.
Table 1 lists the BET surface areas and textural properties of the HT/SBA samples. In
comparison with that of the host SBA-15 (902 m 2 g-l), the surface areas of modified samples
progressively decrease as the coating amount of guest increased, due to the addition of
magnesium and aluminium salts in the synthesis system. For the surface area of HT/SBA-15
composites, the estimated surface area of SBA-15 component is quite large if the contribution
of the HT guest is ignored, say, 784 m 2 g-l for 5HT/SBA and 567 m g-~ for 15HT/SBA
sample. However, the measured data is obviously smaller than the estimated value, which
means the occupation of guests inside the channel of the host. Judged on the result of XRD,
the introduced guest is highly dispersed over SBA-15 indeed, no matter form hydrotalcite or
oxide. Another reason for the difference between the estimated value and the measured data is
that SBA-15 contains many microspores and the incorporated guests prefer to fill the
microspores of host [22], which consequently makes the surface areas of HT/SBA-15 samples
sharply decrease. The micropore area of HT modified SBAS-15 composites, calculated by the
slope of the t-plots, decreased from 94.3 to 18.2 m2 g-I as the coating amount of HT changed
from 5 to 40 wt.-%. Moreover, the lattice constant and the pore size of modified SBA-15 kept
increasing as HT coating amount rose, but contrarily the wall thickness decreased with the
increase of coating amounts as shown in Table 1. The enlarged pore of HT/SBA-15 originates
from the impact of salt on micellization and micelle structures, because the aggregation
number of spherical micelles can be increased in aqueous salt solution, which in turn
influences the wall thickness of the sample. The addition of the salts causes dehydration of
ethylene oxide units from hydrated PEO shell from the side of PPO core, leading to an
increase in core radius of spherical micelles [23]. The increase of lattice constant may be well
interpreted by the frameworks substitution of "bigger" aluminum for silicon. As shown in Fig.
4, only one peak at c a . 54 ppm was observed on the 27A1 MAS NMR spectra of 5HT/SBA,
10HT/SBA and 15HT/SBA samples, which reflects the tetrahedraUy coordinated aluminium
species incorporated into the framework of SBA-15 and these framework aluminium usually
play a role of acidic sites in adsorption and catalysis. When the coating amount on SBA-15
achieved 30 wt.-%, a new peak appeared near 0 ppm, which represents the octahedrally
coordinated extra-framework aluminium species. Moreover, as the coating amount of HT
further increased, the intensity of both peaks raised, the intensity of the peak near 0 ppm even
exceeded that at 54 ppm on the 40HT/SBA sample. It is thus concluded that aluminium
species are incorporated into the framework of SBA-15 entirely when the coating amount of
HT is relative low, no extra-framework aluminium species formed until the coating amount of
guest exceeded 30 wt.-%. Such expansion of unit cell parameter was also found on some
heteroatom such as zirconium and titanium substituted SBA-15 mesoporous materials [1,13].
m,,-i
ra~
100-
9o- 96.
r./3
,.O
92.
o
i
"~
c~ ~o.,
88 ~
0 ~O~'o-..O~
' O,....O~.o._..__o_O_ O
84
0 5000 10000 15000 20000
Volumn of water passed / ml g-~
Fig. 7. Remaining basicity of 30HT/SBA sample versus the amount of water passed through
about one third of the latter (5.57 mmol(OH) g-l). For the sample of 40HT/SBA, the data
obtained by CO2-TPD experiment (1.37 mmol(OH ~ g-i) is only one sixth of the value
detected by titration (8.18 mmol(OH-) g-l). These differences result from the formation of
multilayers of basic species in the HT/SBA-15 composites. In such structure the basic species
covered by upper layers do not expose to CO2, but react with acid if the top layers were
resolved in aqueous titration process. Thus, the data of CO2-TPD just mean the amount of
basic sites exposed on the top of overlapped structure while titration data represent total
basicity of composite. In the FTIR spectra of HT/SBA-15 composites (Fig. 6), the band at
1448 cm -~ that assigned to magnesia appears when the coating amount reaches 30 wt.-%,
beside the characteristic one at 1080 cm -1 assigned to Si-O-Si groups of SBA-15. It is the
important evidence that magnesium species coated on the mesoporous materials and formed
the basic compounds like MgO. The small solubility of magnesia in water enables this novel
solid base to exhibit high water-resistance. As is evident in Fig. 7, some 85% of the total
basicity remained on the sample of 30HT/SBA- 15 after about 20000 ml g-~ water passed over.
555
(100)
• ) a
b 9~ a
0
9 v...~
Y.
d . .
1.0 115 2.0 215 310 3.5 4.0 1.0 1.5 2.0 2.5 3.0 3.5 4,0
2 ]heta / degree 2 ]1"eta / degrees
Fig. 8. XRD patterns of functionalized materials of (a) 5HT/SBA, (b) 15HT/SBA, (c) 35HT/SBA and
(d) 40HT/SBA (A) before and (B) after hydrothermal treatment for 24h
Such excellent water-resistance property will be useful for expanding the application of solid
bases in those processes involving water.
The hydrothermal stability of the modified samples is also evaluated. After the HT/SBA- 15
composites are treated in boiling water for 24h, their mesoporous structure still remains. Fig.
8 shows the XRD patterns of functionalized materials after hydrothermal treatment for 24h.
5ttT/SBA and 15HT/SBA samples exhibit three diffraction peaks assigned to (100), (110),
(200) characteristic of typical patterns of the two dimensional hexagonal pore ordering of the
p6mm space group, identical to that of SBA-15 hosts. No doubt they possess excellent
hydrothermal stability. When the coating amount was up to 35 wt.-%, however, the sample
exhibit no diffraction peaks in low-angle XRD patterns after to be treated in boiling water,
revealing that the hexagonal mesoporous structure is destroyed by this hydrothermal
treatment. The possible reason leading to this result is the thin wall thickness of
functionalized materials with high coating amounts.
The advantage of this method that we used in this paper is directly introduce both acidic
sites, resulted from the inserting of A1 in the framework of SBA-15 as judged from MAS
NMR results, and basic sites derived from the high dispersion of magnesium species on SBA-
15 within one-pot process. Utilizing the strong acidic condition of the synthetic system to
control the incorporation behaviour of two metal species is crucial to realize the multi-
functionation of mesoporous silica. Generally, the magnesium and aluminium salts cannot
form hydrotalcite in acidic environment, so that they play different roles in the modification
of SBA-15: the aluminium in the salt can insert into the framework of SBA-15 like those
alumination agent used for synthesize Al-containing mesoporous silica. While magnesium
species coated on SBA-15 like that reported by Wei et al [12]. It is interesting that co-
existence of A1 and Mg ions in the synthetic system does not hinder or disturb each other, and
they play their own role in different ways. The other important role of two precursors,
magnesium acetate and aluminium nitrate, is as the counter ions to participate the self-
assembly process by enhancing the specific interaction between metal ions and nonionic
surfactant therefore lead to better ordering either in mesostructure or morphology in nonionic
surfactant templating systems. This functionated mesoporous material possesses high surface
area and uniform pores as well as multi active sites that enable it to be used as a potential
candidate in prospect in adsorption-separation and catalysis field.
556
4. C O N C L U S I O N
The basic mesoporous materials HT/SBA-15 are synthesized through "direct synthesis"
method. The samples possess high basic strength of H_- 22.5, much higher than that of parent
SBA-15. Magnesium species can disperse highly in the channels of SBA-15 and the
aluminium species can incorporate into the framework of the host. The calcined composites
maintain the ordered mesoporous structure identical to that of SBA-15 host. They also exhibit
excellent water-resistance and hydrothermal stability, which will expand their application in
industry.
ACKNOWLEDGEMENT
The authors gratefully acknowledge to NSFC (20273031 and 20373024), and Analysis Center
of Nanjing University financially support on this work. The authors want to thank Professor
Hayashi of Research Institute of Instrumentation Frontier, National Institute of Advanced
Industrial Science and Technology (AIST) for the NMR experiments.
REFERENCES
[1 ] Y. Han, F.S. Xiao, S. Wu, Y.Y. Sun, X.J. Meng, D.S. Li and S. Lin, J. Phys. Chem. B, 105
(2001) 7963.
[2] Y. Li, W.H Zhang, L. Zhang, Q.H. Yang, Z.B. Wei, Z.C. Feng and C. Li, J. Phys. Chem. B, 108
(2004) 9739.
[3] A. Tuel and L.G. Hubert-Pfalzgraf, J. Catal. 217 (2003) 343.
[4] H.M. Ding, H. Sun and Y.K. Shan, J. Photochem. Photobiol. A: Chem. 169 (2005) 101.
[5] S.Y. Chen, L.Y. Jiang and S.F. Cheng, Chem. Mater. 16 (2004) 4174.
[6] Z. Luan, J.Y Bae and C. Kevan, Chem. Mater.12 (2002) 3202.
[7] P. Shah, A.V. Ramaswamy, K. Lazar and V. Ramaswamy, Appl. Catal. A: General, 273 (2004)
239.
[8] J.A. Melero, G.D. Stucky, R. Grieken and G. Morales, J. Mater. Chem. 12 (2002) 1664.
[9] C.M. Yang, B. Zibrowius and F. Schtith, Chem. Comm. (2003) 1772.
[10] C.M. Yang, P.H. Liu, Y.F. Ho, C.Y. Chiu and K.J. Chao, Chem. Mater. 15 (2003) 275.
[ 11] A.H. Janssen, C.M. Yang, Y. Wang, F. Schul th, A. J. Koster and K. P. de Jong,
J. Phys. Chem. B 107 (2003) 10552.
[12] Y.L. Wei, Y. Cao, J.H. Zhu, X.W. Yan, Chin. J. Inorg. Chem. 19(2003) 233.
[ 13] B.L. Newalkar, J. Olanrewaju and S. Komarneni, J. Phys. Chem. B 105 (2001) 8356.
[14] S. Wu, Y. Hart, Y.C. Zou, J.W. Song, L. Zhao, Y. Di, S.Z. Liu and F.S. Xiao,
Chem. Mater. 16 (2004) 486.
[15] Y.J. Han, G.D. Stucky and A. Butler, J. Am. Chem. Soc. 121 (1999) 9897
[16] J. Lei , J. Fan , C.Z. Yu, L.Y. Zhang, S.Y. Jiang, B. Tu and D.Y Zhao, Microporous
Mesoporous Mater. 73 (2004) 121.
[17] A.M. Liu, K. Hidajat, S. Kawi and D.Y. Zhao, Chem. Commun. (2000) 1145.
[18] S.J. Bae, S.W. Kim, Y. Hyeon and B.M. Kim. Chem. Commun. (2000) 31.
[ 19] D. Margolese, J.A. Melero, S.C. Christiansen, B.F. Chmelka and G.D. Stucky, Chem. Mater. 12
(2000) 2448.
[20] Y.J. Han, J.M. Kim and G.D. Stucky, Chem. Mater. 12 (2000) 2068.
[21] S. Jun, S.H. Joo, R. Ryoo, M. Kruk, M. Jaroniec, Z. Liu, T. Ohsuna and O. Terasaki, J. Am.
Chem. Soc. 122 (2000) 10712.
[22] S.H. Joo, R. Ryoo, M. Kruk and M. Jaroniec, J. Phys. Chem. B, 106 (2002) 4640.
[23] N.J. Jian, V.K. Aswal, P.S. Goyal, P. Bahadur, Colloid Surf. A 173 (2000) 85.
Studies in Surface Science and Catalysis, volume 158
J. t~ejka, N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 557
The syntheses and characterization of mesoporous silica and functionalized mesoporous silica
obtained by neutral polyoxiethylene surfactants are reported. Syntheses were carried out
according to a new interfacial procedure based on a slow-rate not-catalyzed
tetraethylortosilicate hydrolysis aimed to tailor the particle size of the final product. Synthesis
systems are constituted by a two-level byphasic emulsion and mesoporous silica particles
grow at the interface of the organic (upper) and the inorganic (lower) regions. The newly-
formed particles migrate because of the gravity force to the bottom of the synthesis vessel, far
from the upper silica-feeding solution. Typical particle size is at the nanometer scale. Solids
with high thermal stability, high pore volume and specific surface SBEVup to 900 m2/g are
obtained at room temperature starting from synthesis systems not including a mineralizing
agent such as H +, O H or F-ions.
1. INTRODUCTION
In the last years, in the field of highly ordered mesoporous silicates much effort has been
devoted to the development of organic-inorganic supramolecular self-assemblies and their
subsequent polycondensation. These syntheses involved organic structuring agents such as
long chain quaternary ammonium cationic surfactants [1,2] and anionic surfactants [3]. Also
non-ionic surfactants, such as long chain alkylamine [4] and polyethylene oxides-type
copolymers [5-7], were shown to produce synergistic assembly of silica. The hybridization of
the inorganic precursor allows the development of innovative multifunctional materials
having complex architectures and combining properties of organic, bio and inorganic
components.
Hybrid inorganic-organic materials are produced when chemically active groups are
covalently linked to the inorganic framework of mesoporous materials either by post-synthetic
grafting or by simultaneous condensation of siloxane and organic siloxane precursors, the last
containing a non-hydrolysable Si-C bond [8-14].
This method was also applied for bulky chromophores [15]. The synthesis of
mesoporous hydrophobic materials by simultaneous removal of surfactant template and
modification of the external surface has been achieved by treating uncalcined MCM-41 with
trialkylchlorosilanes[ 16].
The preparation of mesoporous silicate and alumino-silicate by different synthesis
methods are reported together with their surface modification by reaction with benzoyl
chloride [17]. Mesoporous materials synthesized using sodium silicate and an industrial
558
neutral surfactant activated by calcination or solvent extraction are modified by reaction with
benzoyl chloride to compare surface properties of both samples [18].
Tetraethoxysilane (TEOS)-induced assembling of highly ordered mesoporous silica
structures has been carried out by using hydrophobic polybutylene oxide moiety di-block and
tri-block copolymers as structure-directing agents [19]. Polyoxyethylene tridecylethers with
different polyoxyethylene chain length have been used as templating agents in the synthesis of
mesoporous materials [20]. The effects of surfactant/tetramethoxysilane (TMOS) molar ratio,
heating time, temperature, and the number of oxyethylene units have been studied. Synthesis
of ordered and disordered mesoporous materials obtained by assembling nonionic
polyoxyethylene cetyl ether and TMOS as silica source has been systematically studied with
regard to the effect of heating time and temperature on pore diameter of obtained materials
[21 ]. The influence of both concentration and surfactant/TMOS molar ratios was also studied
on the same systems [22]. Two different synthesis pathways for different surfactant
concentration have been proposed. Well-ordered CMI-1 materials with hexagonal
arrangement of channels or disordered wormhole mesostructured materials can be obtained.
Recently, the internal structure and the morphogenesis of the particles at the micrometer level
have been studied with regard to surfactant/silica molar ratio as well as hydrothermal
treatment [23], while the effect of pH of the synthesis gel on pore size distribution of ordered
and disordered mesoporous silica has been evaluated for concentrated and diluted surfactant
solutions [24]. Mesoporous macroscale structures were obtained by oil-water interface
templating systems at acid pH. TEOS is used as silica source and is hydrolyzed just at the oil-
water interface where mesostructures are formed reflecting on a larger scale size and shape of
the oil spheres because the inorganic materials condense around them. Stirring is one of the
mechanism for controlling emulsion properties through modification of long-range
hydrodynamic forces. Under slow stirring fiber-type morphologies are predominantly
observed while with increasing stirring rate more and more spherelike particles are formed.
The two level byphase control was also used on a larger scale to produce self-supporting
membranes at a static interface between an aqueous and an inorganic phase [25].
Successively, a method has been described in which mesoporous silica particles of
macroscopic dimensions functionalized with organic ligands can be produced from alkaline
medium in a one-step emulsion process [26]. Furthermore, hard mesoporous silica spheres
have been described according to a procedure closely related to that initially reported for the
synthesis of MCM-41 [1-2]; the primary difference is that phase separation and emulsion
chemistry which result from the hydrophobic nature of tetrabutylorthosilicate and butanol are
used in order to create the desired morphology [27]. More recently, the conditions favouring
the formation of mesoporous film over bulk material through the separation of the ingredients
into two stratified phases, both in acid and in basic media, have been defined [28].
In this work mesoporous materials and functionalized mesoporous materials are
obtained from two-level byphase emulsions. The upper one is an organic phase and contains
the silica source; the lower phase is a water or water-ethylene glycol solution of the neutral
surfactant. Micelles are feed by the silicon alcoxide, precursor of the inorganic coverage, in
the interfacial region and, after condensation, the particles move to the bottom part of the
synthesis vessel.
2. EXPERIMENTAL
2.1. Synthesis
The neutral polyoxyethylene( 10)isononylphenylether surfactant, C9H19-C6H4-
(OCHzCH2)10OH named Nonfix 10, was used as received by SASOL. Tetraethylortosilicate
559
2.3. Characterization
X-Ray powder diffraction patterns were measured on a Philips PW 1710 diffractometer
using Cu-K~ radiation (40 Kv, 20 mA) over the range l~ ~ The N2 adsorption-
desorption volumetric isotherms at 77 K were measured on a Micromeritics Asap 2010
apparatus. Samples were pre-treated under vacuum at 300~ to a residual pressure of 2 ktmHg.
560
Surface area of the samples was calculated by BET linearization in the pressure range 0.05 to
0.2 P/Po. Lattice pore volume was obtained from the amount of nitrogen gas adsorbed at the
top of the rising section of the isotherms of type I or IV. The amount of surfactant was
measured by TG analysis. The measurements were carried out with a Netzsch STA 409
instrument between 20 ~ and 850 ~ at a ramp of 10 ~ in air with a flow rate of 10
ml/min. Micrographs were collected by a SEM Jeol JSTM 330A.
The upper organic phase dissolves the silica source while the lower inorganic phase contains
the surfactant so that the composite material is assembled in the interfacial region.
The particle growth process stops when silica particles move by sedimentation to the bottom
of the synthesis vessel far from the upper organic silica-feeding solution.
Different synthesis systems have been investigated and ethyl acetate, heptane or n-octane are
used as solvents for the organic phase while pure water or water and ethylene glycol are used
for the surfactant solution. In the synthesis of 3-aminopropyl-mesoporous silica, no organic
solvent is used for the upper phase constituted by TEOS and APTES. Table 2 shows physico-
chemical properties of as-made and activated samples. Both hydrolysis and condensation of
TEOS are considerably made slower from ethylene glycol. In system 5 the lower phase begins
to condense after 18 days aging. XRD patterns of the obtained samples show the typical dloo
reflection of mesoporous materials. Reticular plane distance from X-ray diffraction (do)
decreases after calcination (de) in the case of H20/ethyl acetate (Fig.la, Table 2), HzO/i-
octane, H20 /heptane and H20-ethylene glycol/i-octane synthesis systems, 1, 2, 3 and 5
respectively.
Table 2
d-spacing values, do, for as-made samples, dc, for calcined samples (from XRD analysis),
specific surface areas SBETand pore volume of obtained mesoporous silica samples.
System do (A) dc (A) SBET Pore Volume (0,6 Organic/SiO2
As made Calcined (m2/g) P/Po, cm3/g) mass ratio
1 59,44 55,34 735 0.62 1,66
2 70,90 55,34 690 0.63 1,23
3 70,90 55,34 719 0.68 1,66
5 85,29 69,23 895 0,98 1,22
6 65,14 69,23* 463 0.48 0,56 (0,12")
*The value is referred to material obtained from system 6 'after solvent extraction of surfactant.
This effect is assigned to a partial structural collapse of the porous solids induced by high
temperatures [29]. Oppositely, dc for solvent extracted material (Fig. lb, Table 2) is slightly
increased if compared to do for as-made material in the case of synthesis system 6 as a results
of the restructuring effect on silica wall. Fig. 2 shows XRD patterns of as-synthesized (a) and
calcined (b) mesoporous silica obtained from synthesis system 2. A large shoulder is present
in the XRD pattern of as-synthesized materials. In this case the organization of micelles with
different diameters into silica architectures can be hypothyzed. The shoulder is also present in
Fig. 2 b although calcination softens the contour of the XRD pattern. The same behaviour is
561
displayed by materials obtained from H20/heptane synthesis (system 3, XRD patterns not
shown).
(a) (b)
d
,.a
(D
10 2 4 6 8 10
Fig. 1. XRD patterns of (a) calcined mesoporous silica obtained from synthesis system 1 (Table 1)
and (b) extracted mesoporous silica obtained from synthesis system 6 (Table 1).
(a) (b)
=i
aL
~D
2 4 6 8 10 0
Fig. 2. XRD patterns of as-synthesized a) and calcined b) mesoporous silica obtained from synthesis
system 2 (Table 1).
Obtained samples show SBET specific surface around 900 m2/g and high pore volume
up to 1,2 cm3/g. Fig. 3 shows nitrogen adsorption-desorption isotherms for samples obtained
from H20/ethyl acetate (a) and water/TEOS-APTES (b) synthesis systems. In both cases a
type IV not reversible isotherm is obtained, typical of capillary condensation in regularly-
sized mesopores. Fig. 4 shows adsorption-desorption isotherms for materials obtained from
H20/i-octane (a) and H20/heptane (b) synthesis systems. Two main mesopore filling steps are
detectable in both isotherms. Similar organic content, reported in Table 2, and surfactant
decomposition pattern for materials obtained from H20/ethyl acetate, H20/i-octane and
H20/heptane synthesis systems are exhibited from DTG and DSC analyses (not shown).
Nitrogen adsorption-desorption isotherms show that calcination of materials obtained
from synthesis systems 1 and 5 produces solids characterized by a regular unimodal porous
system while isotherms obtained from calcined samples 2 and 3 show a bimodal porous
system characterized by two capillary condensation mesopore filling steps. The second step
corresponds to the filling of very large mesopores.
562
400
Fig. 3. Nitrogen adsorption-desorption isotherms of samples obtained from interfacial (a) water/ethyl
acetate and (b) water/TEOS-APTES synthesis systems.
Fig. 4. Nitrogen adsorption-desorption isotherm of samples obtained from interfacial (a) water/i-
octane, (b) H20-heptane synthesis system.
Data obtained from XRD evidence that the properties of the interfacial region affect
surfactant chemistry. Regularly monosized micelles are obtained from synthesis systems 1
and 5, while micelles with different size are formed during the assembly of the material in the
case of sample 2 and 3. This is clearly shown by the observation of the XRD patterns of as-
synthesized and calcined materials obtained from synthesis systems 2 (Fig. 2, (a) and (b)).
The large shoulder at about 2-4 20, representative of two main reflecting elements, confirms
the co-existence of two templating systems with different diameters in accordance with the
nitrogen adsorption on calcined materials showing two distinct capillary condensation steps.
The occurrence of heterogeneous formation of surfactant templating-micelles can be
explained on the basis of solvent inclusion in the hydrophobic core of the micelles although
the formation of reverse micelles represents a possible route.
Bimodal porous systems occur when the solvent in the silica-feeding solution is i-
octane or heptane. These solvents easily interact with hydrophobic tales of surfactant
molecules increasing micelles diameter. Reverse micelles formation could also be favoured
by lipophilic i-octane or heptane organic phases which could also increase micelle diameter.
Furthermore polyoxyethylene region of neutral Nonfix 10 surfactant can present different
"coiling degrees" and micelles with different size can be organized in solids obtained from
the same preparation so that the low "coiling degree" of polyoxyethylene fraction of
surfactant molecules could explain the occurrence of large mesopores in bimodal porous
materials.
563
In the case of system 5 the high stability of micelles formed in H20/ethylene glycol
mixed phase does not favour migration in the upper phase and consequently micelles
inversion, so that the final material is characterized by unimodal pore system.
Although bimodal mesopore systems have been created by means of several post-
synthesis treatments starting from MCM-41 mesoporous materials [30,31], in this case the
bimodal porosity comes from byphasic emulsion and the templating effect, recognized in as-
synthesized materials, is mantained during the calcination process. The obtained materials
show regular morphology with particle size at the nanometer scale, as reported in Fig. 5.
4. CONCLUSIONS
ACKNOWLEDGMENTS
The authors acknowledge the financial Support from the Ministry of Instruction, University
and Research (P.R.I.N. Project "Nanostructured Materials with Controlled Porosity for
Innovative Technological Applications")
564
REFERENCES
[1] C. T. Kresge, M.E, Leonowicz, W.J. Roth, J. C Vartuli, J. S. Beck, Nature ,359 (1992) 710.
[2] J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T-W. Chu,
D.H. Olson, E.W. Sheppard, S.B. McCullen; et al. J. Am. Chem. Soc., 114 (1992) 10834.
[3] Q. Huo, D.I. Margolese, U. Cielsa, P. Feng, T.E. Gier, P. Sieger, R. Leon, P.M. Petroff, F.
Schtith, G.D. Stucky, Nature, 368 (1994) 317.
[4] P.T. Tanev, T.J. Pinnavaia, Science, 267 (1995) 865.
[51 G. Attard, J.C. Glyde, C.G. G61tner, Nature, 378 (1995) 366.
[6] S.A. Bagshaw, E. Prouzet and T.J. Pinnavaia, Science, 269 (1995) 1242.
[7] D. Zhao, J. Feng, Q. Huo, N. Melosh, G.H. Fredrickson, B.F. Chmelka, G.D. Stucky, Science,
279 (1998) 548; D. Zhao, Q. Hou, J. Feng, B.F. Chmelka, G.D. Stucky, J. Am. Chem. Soc.,
120 (1998) 6024
[8] S.L. Burkett, S.D. Sims and S. Mann, Chem. Commun., (1996) 1367.
[9] M.H. Lira and A. Stein, Chem. Mater., 11 (1999) 3285.
[10] D. Brunel, Microporous Mesoporous Mater. 10 (1999) 329.
[11] M. Park, S. Komarneni, Microporous Mesoporous Mater., 25 (1998) 75.
[12] D.J. Macquarrie, M. Rocchia, B. Onida, E. Garrone, P. Lentz, D. Brunel, A.C. Blanc,
F. Fajula, Stud. Surf. Sci. Catal. 135 (2001) 29-P-31.
[131 D.J. Macquarrie, D. Brunel, G. Renard, A.C. Blanc, Stud. Surf. Sci. Catal., 135 (2001) 29-P-
10.
[14] D.Brunel, F.Fajula, J.B. Nagy, B. Deroide, M.J. Verhoef, L. Veum, J.A. Peters and H. Van
Bekkum, Appl. Cata|. A, 213 (2001) 73.
[15] C.E. Fowler, B. Lebeau and S. Mann, Chem. Commun., (1999) 201.
[16] V. Antochshuk and M. Jaroniec, Chem. Commun., (1999) 2373
[171 L. Pasqua, F.Testa, R. Aiello, J. B.Nagy, G. Madeo, Phys. Chem. Chem. Phys, 5 (2003) 640.
[lS] L. Pasqua, f. Testa, R. Aiello, Stud. Surf. Sci. Catal., 146 (2003) 497.
[19] C. Yu, Y. Yu, L.Miao, D. Zhao, Microporous Mesoporous Mater., 44-45 (2001 ) 65.
[20] J.L. Blin, A. Becue, B. Pauwels, G. Van Tendeloo, B.L. Su, Microporous Mesoporous Mater.,
44-45 (2001 ) 41.
[21] J.L. Blin, A. Leonard, B.L. Su, J.Phys. Chem. B, 105 (2001) 6070.
[22] J.L. Blin, A. Leonard, B.L. Su, Chem. Mater., 13 (2001) 3542.
[231 A. Leonard, J.L. Blin, M. Robert, P.A. Jacobs, A.K. Cheetham, B.L. Su, Langmuir, 19 (2003)
5484.
[24] A. Leonard, J.L. Blin, P.A. Jacobs, P, Grange, B.L. Su, Microporous Mesoporous Mater., 63
(2003) 59.
[25] S. Schacht, Q. Huo, I.G. Voigt-Martin, G.D. Stucky, F. Schuth, Science, 273 (1996) 768.
[26] R.I. Nooney, M. Kalyanaraman, G. Kennedy, E.J. Magin, Langmuir, 17 (2001) 528.
[27] Q. Huo, J. Feng, F. Schuth, G.D. Stucky, Chem. Mater., 9 (1997) 14.
[281 L. Faget, A Bergman, O. Regev, Thin Solid Films, 386 (2001) 6.
[29] L.Pasqua, F. Testa, R. Aiello, F. Di Renzo and F. Fajula, Stud. Surf. Sci. Catal. 135 (2001)
06-P-28.
[301 M. Grun, K.K. Unger, A. Matsumoto, K. Tsutsumi, Microporous Mesoporous Mater., 27
(1999) 207.
[31] Y. Zhong-Yong, S. Bao-Lian, Colloid Surface A, 241 (2004) 95.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 565
Mesoporous silica SBA-15 has been incorporated with various amounts of Sn and A1 by post
synthesis grafting technique, which results in excellent structural integrity and enhanced
acidity. Physico-chemical characterization was done by various spectroscopic methods to
understand the chemical nature of incorporated Sn and A1. XRD and BET results show that
the hexagonal structural of the SBA-15 is maintained during the grafting with no loss of long
range structure ordering. 119Sn NMR spectrum shows the presence of Sn in essentially
octahedral environment, while 27A1MAS NMR results confirm that a large proportion of the
A1 is in tetrahedral position. The resulting Sn and Al-grafted materials possess considerable
catalytic activity in the transesterification of diethylmalonate (medium acidity) and
isopropylation of naphthalene (strong acid sites), respectively.
1. INTRODUCTION
Silica-based mesoporous materials have attracted a great deal of interest due to their high
potential as supports for catalytic applications. The microporous materials on the other hand,
have smaller pore size, which often restricts the accessibility of active sites to larger substrate
molecules. The discovery of mesoporous materials, designated as SBA family consisting of
thermally stable, hexagonal array of silica, has overcome the difficulties encountered by their
microporous analogues, as they possess larger pore dimensions (6-30 nm), higher surface
areas (700-1500 mZg-l) and substantial number of silanol groups. The direct hydrothermal
"synthesis of A1- and Sn-SBA-15 mesoporous materials in strongly acidic medium (2 M HC1)
often leads to negligible incorporation of A1 and Sn in SBA-I 5. One of the alternative routes
involves the post-synthesis grafting and therefore, much current research focuses on the post-
synthesis modifications of purely siliceous SBA-15 to generate acid sites on the inner walls of
the mesoporous framework [1-2]. This is possible due to the quasi-amorphous nature of the
walls of SBA-15 and the presence of a high density of silanol groups. It is widely accepted
that the guest species are grafted onto SBA-I 5 via the silanol groups. Therefore, grafting in
aqueous media is desirable, because the host SBA-15 is fully hydrated thereby presenting a
maximum number of silanol groups as potential grafting sites. From such a viewpoint, and for
the synthesis of solid acid catalysts of good structural integrity and well-developed acidity, we
have worked out an efficient method of preparation of Sn-SBA-15 by reacting SBA-15 with
tin chloride and tin acetate on the one hand, and AI-SBA-15 by reacting SBA-15 with
aluminium chlorohydrate and aluminium chloride in aqueous media and in non-aqueous
environment, on the other.
566
We report herein, in detail, the properties of resulting tin and aluminium incorporated
mesoporous silica molecular sieves, their chemical compositions, structural ordering,
porosity, and acidity. The catalytic activity of Sn-SBA-15 samples for the transesterification
of diethylmalonate (DEM) with the various alcohols and that of A1-SBA-15 in the
isopropylation of naphthalene have been investigated.
Esterification of carboxylic acids and transesterification of esters have wide academic
as well as industrial importance. Recently, transesterification is reported to be an important
route for the production of biodiesel. Liquid phase, Friedel-Crafts isopropylation of
naphthalene is widely used in the chemical industry to manufacture 2,6-di-
isopropylnaphthalene (2,6-DIPN), an important chemical intermediate for making monomers
of advanced polymers. The alkylation is traditionally performed with alkyl halides using
Lewis acid catalysts such as A1C13, or with alcohols using Br6nsted acids, typically HF or
HzSO4. However, homogeneous catalysts are not easily separable after the reaction and
generate a large amount of acid waste. Mesoporous A1-SBA-15 molecular sieve, being
sufficiently acidic, is a promising and alternative catalyst for the isopropylation of
naphthalene.
2. EXPERIMENTAL
with resonance frequencies of 111.8 and 78.21 MHz, and reported relatively to tetramethyl tin
and A1 (H20)63+, respectively.
~ ~-~A-15
H-IO
S , i , , , i . , . i
Fig. 1. XRD patterns in the region 0-5~ 20 of (a) Fig. 2. Nz adsorption isotherms of SBA-15, A1C1-
ACH-10 (b) AIC13-10 10, Sn-SBA-15 (5), and ACH-10.
tending towards lower p/po were obtained (Fig. 2). The progressive decrease in specific
surface area and pore volume suggests that these oxides exist as a thin film anchored inside
the mesopores of SBA- 15, which is corroborated by the TEM data. This is accompanied by an
increase in the pore wall thickness. The decrease in surface area is due to the blocking of the
micropores by the incorporated Sn and A1, which is predominant in samples prepared using
Sn acetate and ACH (Table 1). The t-plot analysis shows gradual disappearance of the
micropores on incorporation of metals in SBA-15. Combining this with the above BET
analysis, we presume that the decrease in the total surface area during Sn and A1 addition
points to a reaction between the silanol groups (part of the micropore structure) and the
precursors of Sn and A1.
Table 1
Physicochemical Characteristics of SBA- 15, Sn-SBA- 15 and A1-SBA- 15 samples
Si/A1 dloo a, nm S e E r Micro Pore Micro Pore diam. Wall
(nm) (m2g-1) S.A gol. Vol. DBJH(rim) thick.
(mZg-l) cm 3 g-Z cm 3 g-i (nm)
oe 725 162 1.57 0.07 7.1 4.5
Sn-SBA-40 a 10.17 11.74 670 194 1.16 0.08 6.9 4.8
Sn-SBA-5 a 1 0 . 1 9 11.77 573 173 0.97 0.07 6.8 5.0
Sn-SBA-40 b 10.27 11.86 538 370 0.85 0.07 6.3 5.6
Sn-SBA-5 b 1 0 . 3 5 11.96 421 282 0.66 0.06 6.3 5.7
ACH-40 c 9.83 11.35 480 32 1.03 0.02 8.0 3.3
ACH-10 c 9.87 11.40 428 20 0.93 0.01 8.3 3.1
A1C1-40d 9.82 11.34 696 86 1.42 0.04 7.2 4.1
AICI-10 d 10.0 11.56 557 59 1.05 0.02 6.8 4.5
aSamples prepared using Sn chloride as metal precursor, bSamples prepared using Sn acetate as metal
precursor. In case of Al, ~Samples prepared by using ACH, d Samples prepared by using A1C13.
Transmission electron micrographs of SBA-15, Sn and A1-SBA-15 show the hexagonal array
of uniform channels where tin nanoparticles were observed as dark objects between the walls
of SBA-15 whereas aluminium nanoparticles are not seen (Fig. 3a, b, c and d).
Fig. 3. Transmission electron micrograph: (a) SBA-15, (b) Sn-chloro-SBA-15 (c) Sn-acetate-SBA-15
and (d) ACH- 10
The FTIR studies in the hydroxyl region show that the surface silanol groups are
responsible for metal incorporation [5]. TPD of ammonia shows a broad desorption pattern
indicating a large distribution of different types of acid sites (Fig. 4). Deconvolution of the
569
profile results in three distinct peaks which are described as follows: (1) the desorption at 373
to 523 K is assigned to weak acid sites due to surface hydroxyl groups; (2) the peak between
523-673 K is attributed to moderate structural acid sites and (3) the peak at 673 to 873 K is
assigned to strong structural acid sites. From a comparison of the TPD profiles and the peak
intensities (Fig. 4 a and 4c), it is clear that A1-SBA-15 samples present stronger acid sites than
Sn-SBA-15 samples.
i! ,, .. -.._....
,, "~'-~
323 373 423 473 523 573 623 673 723 ffT~ 473 573 673 773 373 473 573 673 773
Temperature (K) Temperature (K) Te~vgeramre (K3
Fig. 4. TPD of NH3 from (a) Sn-ChloroSBA-15 (5) sample, (b) A1-SBA-15 samples and (c)
deconvoluted profile of TPD of ACH-10
Fig. 5 shows the FT-IR spectra of adsorbed pyridine on A1-SBA-15, that exhibit all
expected bands due to weak Lewis bound pyridine (1585 cm -I) and the small peaks due to
hydrogen-bonded pyridine (1447 cm -1 and 1605 cm]). The band at 1485 cmlis attributed to
pyridine associated with both Lewis and Br6nsted acid sites. Absence of 1545 cm -I band
indicates absence of Br6nsted acidity generated by the pyridinium ion. Hence 1485cm -I band
reflects only Lewis acid sites. Also, the higher intensities of the bands at 1436 cm -I indicate
that the surface acidity of the samples is predominately of the Lewis type. As expected
increasing the evacuation temperature reduces the number of acid sites. A majority of acid
sites generated on A1-SBA-15 are found to be Lewis acid sites. The intensities of these peaks
gradually decreased with the evacuation temperature.
~D
" ~ ~ /I'J ~/
!L.s'., ~ <~
\., 423 K
,, ;.., .j,<s-.,,../
{'--1 . . . .
t
/
~ I t'l
....-,--'1J--''v"
<I k...,
,, .r.,.s,,~",
7"t~ .j , -"-
.J " '"~.,--,...............
~ ft\... 323 K
I , I , l , I , , I ,
XVavenumber (cm9
The isotropic chemical shift at/5 - - 6 0 4 ppm in the l l9Sn MAS NMR spectrum of Sn-
SBA-15 samples reveals the presence of Sn in octahedral coordination, while 27A1 MAS NMR
spectra confirm that a large proportion of the A1 is inserted into the tetrahedral positions
within the framework. The tetrahedral to octahedral aluminium ratio is found to increase with
higher aluminium incorporation as shown in Fig. 6. The shoulder around 30 ppm in ACH-40
sample is assigned to penta-coordinated aluminum atoms.
"2.
- 604
9 \ ~ b
-__/'J~-'Xj'~'-~' ..:t
; ~I "
I,~ ~ ;~ ~ * ~' ~ I~ ~ I~'''I r'~' llll ]II II ''''I ~''' '"- ' I ' ' ' ' I'~''I'~'I i Ij I I 1 I
' . . . . . . . . .,~ 1 (7~ f']~ 1~0 1 ;0 5 'o 'o - ;0 150 1 0 0.......................
50 0 -50
ppm
Fig. 6. 119Snand 27A1MAS NMR spectra of(a) Sn-SBA-15 (5), (b) ACH-40 and (c) A1C13-40
There are probably two types of Sn species in SBA-15, the ratio of which depends on
Sn content. The first one is stabilized atomically (well-dispersed, Si-O-Sn-O-Si- groups) on
2+ 9 9 9 4+ 2+
the walls, which form Sn upon reduction treatment (easily reducible from Sn to Sn ) [5].
The second types are large site entities of extra-framework, octahedral Sn 4+, distributed in the
external pore structure, which are seen as 45 nm particles in TEM. However, the presence of
structural Sn 4+ (tetrahedral) in SBA-15 is not evident from the NMR data [11].
The presence of silanol OH group of the wall of SBA-15 which contribute to the micropores
in the sample are being used up to react with Sn, which ultimately develop into thin oxide
films on the walls of SBA-15. At higher concentrations of Sn, small SnO2 particles are
homogeneously dispersed on the walls of SBA-I 5 channels. Similar interactions between the
silanol groups and the A1 precursors lead to the formation Si-O-A1-O-Si- linkages, which on
calcination result in A1 incorporation in the framework (mainly tetrahedaral), as seen from
27A1 NMR data.
3.2. C a t a l y t i c activity
3.2.1. Catalytic activity of Sn-SBA-15 samples
In the present study, we have examined the effectiveness of Sn-SBA-15, as a mild acid
catalyst for the transesterification of dimethyl malonate with a number of alcohols and found
that the reactions are well catalyzed under mild conditions to give both trans mono- as well as
diketo esters. The results are summarized in Table 2.
Even though the transesterification is catalyzed by SBA-15 itself, the activity
increases with Sn-containing SBA-15 samples. There was no conversion of DEM below 323
K. Above 343 K, the reaction rate increases to reach a very high conversion of DEM (>90 %)
at 383 K. There is, thus, a good catalytic effect of Sn on the reaction, presumably due to the
presence of acid sites of medium strength, as evidenced from the results of TPD of NH3 (Fig.
4a).
571
Table 2
Catalyticactivity of Sn-SBA-15 samples in transesterification of diethylmalonate
Substrate Catalyst Alcohol Conversion Selectivity (%)
(wt.%)
Mono-ester Di-ester
Diethyl SBA-15 n-Butanol 26.7 77.8 22.2
Malonate SnSBA-40 a n-Butanol 75.2 65.0 35.0
(DEM) SnSBA-5 a n-Butanol 92.0 41.3 58.7
SnSBA-5 a n- Hexanol 84.8 41.5 58.7
SnSBA-5 a n-Octanol 80.8 43.9 56.1
SnSBA-5 a Cyclohexanol 82.4 35.8 64.2
SnSBA-5 ~ Benzyl alcohol 42.2 34.8 65.2
SnSBA-5 a Al|yl alcohol 73.3 86.6 13.4
SnSBA-5 b n-Butanol 41.0 85.1 15.0
SnSBA-5 b n- Hexanol 38.3 85.0 15.0
SnSBA-5 b Cyclohexanol 42.3 79.0 21.3
SnSBA-5 b Allyl alcohol 32.2 83.4 16.6
"Sn incorporation using Sn chloride as metal precursor, bSamples using Sn acetate as metal precursor.
Conditions: Batch reactor; temp.- 383K; DEM to alcohol = 1" 3 mole; catalyst = 130 mg; time =24 h.
Table 3
Influence of catalyst type (AI-SBA-15) on isopropylation of naphthalene*
Catalyst Conv. Selectivity to products (mol %)
Type (wt %)
et-MIPN ]3-MIPN 2, 6-DIPN 2, 7- DIPN Others
It is noted that with all the samples, the monoalkylation leads to more of a-MIPN than the
preferred 13-MIPN. This leads subsequently to a moderate distribution of DIPN among 2,6 and
2,7-isomers. Obviously, the mesoporous dimentions of A1-SBA-15 do not bring about the
shape selectively associated with large pore zeolites, such as mordemites in this reaction [ 12].
5. CONCLUSIONS
Tin and aluminium incorporated SBA-15 of good structural ordering can be prepared by a
post-synthesis grafting method. XRD and BET data show that incorporation of Sn and A1
results in a modest decrease in surface area and pore volume, which is accompanied by an
increase in wall thickness. 27A1 MAS NMR indicates that a large proportion of A1 is inserted
into tetrahedral position, which is confirmed by TEM. In the present study, we have examined
the effectiveness of Sn-SBA-15, as an acid catalyst for the transesterification of
diethylmalonate and found that the reactions are well catalyzed under mild conditions. There
is thus a good catalytic effect of Sn on the reaction, presumably due to the presence of acid
sites of medium strength, as evidenced from the results of TPD of NH3. On the other hand, AI
incorporation into the framework of SBA-15 generates materials with Lewis acid sites. This is
confirmed by the considerable catalytic activity exhibited by Al-grafted materials for the
alkylation reaction. The results discussed above demonstrate that incorporation of A1CI3 on
mesoporous SBA-15 silica can create substantial density of acidic sites, which are
catalytically responsible for the liquid phase isopropylation of naphthalene, even though
selectivity to desired products (I3-MIPN and 2,6-DIPN) is not high under the experimental
conditions employed.
ACKNOWLEDGEMENT
The authors are grateful to Dr. P.R. Rajamohanan for NMR spectra, to Ms. Renu Pasricha for
TEM and to the Department of Science and Technology, New Delhi, for financial support.
REFERENCES
[1] Z. Luan, M. Hartmann, D. Zhao, W. Zhou, L. Kevan, Chem. Mater., 11 (1999) 1621.
[2] M. Cheng, Z. Wang, K. Sakurai, F. Komata, T. Saito, T. Komatsu, T. Yashima, Chem. Lett.,
(1999) 131.
[3] D. Zhao, J. Feng, Q. Huo, N. Molish, G. H. Fredrickson, B. F. Chmelka, G. D. Stucky,
Science., 279 (1998) 548.
[4] D. Zhao, Q. Huo, J. Feng, B. F. Chmelka, G. D. Stucky, J. Am. Chem. Soc. 120 (1998) 6024.
[5] P. Shah, A. V. Ramaswamy, K. Lazar, Veda. Ramaswamy, Appl. Catal. A., 273 (2004) 239.
[6] R. Mokaya, W. Jones, J. Mater. Chem., 9 (1999) 555.
[7] V. R. Choudhary, K. Mantri, J. Catal., 205 (2002) 221.
[8] R. Mokaya, W. Jones, J Chem. Soc. Chem. Commun., 22 (1997) 2185.
[9] X. S. Zhao, Max G. Q. Lu, C. Song, J. Mol. Catal. A 191 (2003) 67.
[10] X. S. Zhao, G. Q. Lu, C. Song, Chem Commun., (2001) 2306.
[11] M. Renz, T. Blasco, A. Corma, V. Fornes, R. Jensen, L Nemeth, Chem. Eur. J., 1 8 (2002)
4708.
[12] Y.Sugi, Y. Kubota, T. Hanaoka, T. Matsuzaki, Catal. Surv. Japan, 5 (2001) 43.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 573
A.A. Campos, C.R. Silva, M. Wallau, L.D. Dimitrov and E.A. Urquieta-Gonzfilez*
Universidade Federal de $5o Carlos, Departamento de Engenharia Quimica, Caixa Postal 676,
CEP 13565-905, $5o Carlos - SP, Brasil. E-mail: urquieta@power.ufscar.br
1. INTRODUCTION
The study of mesoporous molecular sieves is being of increasing interest in order to find
alternative catalysts and adsorbents for processing and separating bulky molecules. However,
mesoporous molecular sieves, such as MCM-41 and MCM-48 [1], or SBA-15 [2], possess a
weaker acidity and a lower hydrothermal stability than zeolites [3], therefore the improvement
of these properties is a prerequisite for their application in industrial processes. An enhanced
activity for the cracking of cumene was observed by Kloestra et al. [4] after recrystallisation
of A1-MCM-41 in glycerol, which was exchanged with tetrapropylammonium cations (TPA+).
Campos et al. [5] also found that the activity for the isomerisation of m-xylene increases when
SBA-15 exchanged with TPA + was recrystallised in glycerol. Here we will describe the
recrystallisation in aqueous glycerol of SBA-15 impregnated with tetrapropylammonium
hydroxide (TPAOH) and sodium aluminate (NaA102) and the catalytic properties of the
recrystallised solids using the isomerisation of m-xylene as a test reaction.
2. EXPERIMENTAL
dried at 60 ~ The occluded polymer was removed by heating the sample for 4 h at 120 ~
under nitrogen flow and subsequently under air flow for 20 h at 540 ~
2.3. Characterisation
The obtained solids were characterised by X-ray diffraction on a Siemens D5000
diffractometer using Cu-Ka radiation in the range between 0.5 ~ to 10~ (20) (low angle XRD)
and between 5~ and 40 ~ (20) (high angle XRD).
The N2 sorption isotherms were obtained using a Quantachrome Corporation instrument
(Nova-1200). The specific surface areas were estimated by the BET equation [6]. The pore
size distribution were obtained from the desorption branch of the isotherm using the Barrett-
Joyner-Halenda (BJH) method [7], while the micropore volumes were estimated by the t-plot
analysis in the adsorption branch of the isotherm [8].
The infrared spectra were recorded using the KBr pellet technique on a Perkin Elmer
(Spectrum 100) FTIR Spectrometer from 1050 to 400 cm -~ with 32 scans with 4 cm -~
resolution and an interval of 1 cm ~.
To obtain the Scanning Electron Microscopy (SEM) micrographs a Zeiss DSM960
equipment operated at 30 kV was used.
The SBA-15 and the recrystallised samples were tested as catalysts in the isomerisation of
m-xylene using 100 mg of the solid in a continuous fixed bed reactor, a reaction temperature
of 350 ~ and a weight hourly space velocity (WHSV) of 4 h l. The reaction products were
analysed by a Shimadzu GC 17A Gas Chromatograph equipped with a FID detector and one
packed column (chromosorb W 80/100) with 5 % DOP and 5 % bentonite.
3.1. SBA-15
3.1.1. Nitrogen adsorption
The isotherm of the adsorption of nitrogen on SBA-15 and the BJH pore size distribution
are,shown in Fig. 1 and Fig. 2, respectively. The observed isotherm of type IV, according to
the IUPAC nomenclature [9], proves that the obtained SBA-15 is a mesoporous material with
a narrow pore size distribution. The specific surface area of 888 m2/g is in the range expected
for SBA-15 [2]. However, the observed isotherm apparently does not show a limiting
adsorption at high P/P0 and the hysteresis loop can be classified as an intermediate between
H 1 and H3 type, where the latter is an indication for the presence of slit-shaped pores [9]. In
deed, the BJH plot (Fig. 2) confirms the presence of a second type of mesopores in the SBA-
15. Here two types of mesopores can be observed. One with a mean pore radius of 33 A,
typically for SBA-15 materials [2], and one with a mean radius around 19 A, which might be
attributed to slit-shaped pores between the SBA-15 particles.
575
"~
-n 700
---[]-- dessorption ~_~.~ 19A
.E. 600
(p
--,--adsorption i - [ ] - / , -'-
!
/
E 5OO - - - - [] /
t
i
i
i,, L. _
o> 400 _i
"0
.~ 300
L_
0
200
j,J
"0 i 9 i 9 i 9 i 9 i 9 I |
::3
100
, i
t-
i 110
" i'o
o20
Fig. 3. Low-angle XRD pattern of SBA-15.
TPAOH and NaA102 in the pores of the SBA-15. Furthermore, a wider pore size distribution
is observed for sample T0, which indicates that even at 60 ~ the silica framework is partly
dissolved by the impregnated hydroxyl anions.
When the impregnated SBA-15 samples are recrystallised at 150 ~ a nearly linear
decrease of the specific total pore volume (Vtot) and the specific mesopore volume (Vmeso.)
with the recrystallisation time is observed (Table 1). This decrease of the specific mesopore
volume indicates the collapse of the SBA-15 structure and consequently the collapse of the
mesopore system during the recrystallisation, as it is also indicated by the broadening of the
mesopore size distribution, which can be observed from the BJH plots given in Fig. 5. On the
other hand, the recrystallisation leads to the condensation of silanol groups of the silica
framework, as it can be seen by the decrease in the intensity of the band at 960 cm t in the
FTIR spectra (Fig. 8) discussed in section 3.2.3. This condensation increases the density of
the silica and therefore results in a reduction of the observed specific pore volume. Davidson
[10] reported that condensation during thermal treatment of the SBA-15 framework is
accompanied by a decrease of the SBA-15 microporosity. Such a decrease of the specific
micropore volume can be observed in Table 1 during the initial states of the recrystallisation,
as it was the case of sample T12, where the specific pore volume decreased from 0.043 mL/g
for the parent SBA-15 to 0.009 mL/g. Type H3 hysteresis loops, which are typical for slit-
shaped mesopores [9], were observed for sample Y12 and T24 (not shown in Fig. 4),
indicating the partial collapse of the regular SBA-15 porous structure with increasing
recrystallisation time.
8,\ 19
500
"~ = adsorption TO
-- f_,~-.~,~ ~ .T12
t_
E"1400 ~desorption
o
E 300 ~.~T12
,~ T48
>o200 172
~100
0
The type of the hysteresis loop of sample T48 appears as an intermediate between H3 and
H4 and for sample T60 (not shown in Fig. 4) and T72 hysteresis loop type H4 is observed,
which indicates the formation of predominantly microporous material still containing slit-
shaped interparticle mesopores [9]. It can be observed from Fig. 5 that with the increase in the
recrystallisation time, a third type of regular pores appears in the samples T24, T48, T60 and
T72. The mean pore radii of these pores are around 8 A, which is at the borderline between
micro- and mesopores. From the XRD results discussed in section 3.2.2, it can be seen that
these samples already contain crystalline ZSM-5. Thus, these pores might be attributed to the
577
micropores inherent to the ZSM-5 structure. However, for the ZSM-5 structure pore radii in
the range of 2.6 to 2.8 A [11] are expected. This discrepancy between the crystallographic
micropore radii of the ZSM-5 structure and the pore radii observed in Fig. 5 might be due to
the unsuitability of the BJH model for the determination of micropore radii. Nevertheless, the
formation of an additional amorphous silica phase containing mesopores with pore radii in the
range of 8 A cannot be excluded.
Although the crystallinity of the ZSM-5 increases with the recrystallisation time (see Fig. 6
and Fig. 7), the specific micropore volume (Vmic,o.), reported in Table 1, is apparently
independent of the ZSM-5 crystallinity. A possible explanation for this unexpected behaviour
of the specific micropore volume could be the transformation of the parent SBA-15 not only
into crystalline ZSM-5 but also into an additional amorphous silica phase. As amorphous
silica has a higher density than SBA-15 and ZSM-5, the presence of amorphous silica will
reduce the specific pore volume observed for the bulk. Furthermore, amorphous silica
precipitated on the ZSM-5 crystals could reduce the accessibility into the micropores.
Table 1
Summary of the textural properties of SBA-15 and recrystallised SBA-15 and their
performance as catal~csts for the isomerisation of m-x~'lene
9 Sample SsEr(m2/~) Vtot. (mL/g) Vmeso.(mL/~) Vmi (mL/~) Conv.* [%] o/p-ratio*
....
m-.--m
:::i
,m
t"
TO .~ T72
T12
t"
C
T24
T48
T60
T72
9
-t
547 c m
TO
.~i. I T12
~J
T24
.=_ / T48
E /
T60
C
ca
i,.
~- ~ T72
aluminium into the silica structure is suggested'by the 27A1 MAS NMR spectrum of sample
TO (not shown), which demonstrates aluminium in nearly exclusive tetrahedral co-ordination.
4. CONCLUSIONS
When mesoporous SBA-15 was impregnated with TPAOH and NaA102 and recrystallised at
150 ~ in aqueous glycerol (30 %), plate-like ZSM-5 crystals with high aspect ratios
(length/height) are formed. For the recrystallised samples the vibration of the distorted double
six-rings in the ZSM-5 structure are observed at 547 cm -1, which indicates the incorporation
of aluminium into the ZSM-5 structure and a Si/A1 framework ratio of around 40. The
vibration of the distorted double six-rings were even observed after 12 h of recrystallisation,
which indicates the formation of ZSM-5 units that were not detected by XRD. However, SEM
micrographs showed the mesoporous SBA-15 and the microporous ZSM-5 as segregate
particles evidencing that direct transformation of the SBA-15 pore walls into microporous
ZSM-5 is unlikely. The results of the m-xylene isomerisation suggest that the recrystallised
samples have catalytically active sites, which are consequence of the ZSM-5 formation or
could be generated in the mesopore walls by the aluminium insertion or by the formation of
amorphous aluminosilicate.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the financial support provided by CNPq, Brazil (grant
477.759/2003-3). L. D. D. and M. W. give also acknowledgements to PVE-Program/Capes,
Brazil. Further acknowledgement is given to BASF Corporation for providing the Pluronic |
P123.
REFERENCES
[1] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli and J.S. Beck, Nature, 359 (1992) 710.
[2] D. Zhao, J. Feng, Q. Huo, N. Melosh., G.H. Fredrickson, B.F. Chmelka and G.D. Stucky,
Science, 279 (1998) 548.
[3] D. Trong On, D. Desplantier-Giscard, C. Danumah and S. Kaliaguine, Appl. Catal. A, 253
(2003) 545.
[4] K.R. Kloestra, H. van Bekkum and J.C. Jansen, Chem. Commun., (1997) 2281.
[5] A.A. Campos, L. Martins, L. Lima de Oliveira, C. Ramos da Silva, M. Wallau and E.A.
Urquieta-Gonzalez, Catal. Today, submitted.
[6] S. Brunauer, P.H. Emmet and E. Teller, J. Am. Chem. Soc., 60 (1938) 1553.
[7] E.P. Barret, L.G. Joyner and P.P. Halenda, J. Am. Chem. Soc., 73 (1951) 373.
[8] ' B.C. Lippens, B.G. Linsen and J.H. de Boer, J. Catal., 3 (1964) 32.
[9] K.S.W. Sing, Pure Appl. Chem., 54 (1982) 2201.
[10] A. Davidson, Curr. Opin. Colloid Interface Sci., 7 (2002) 92.
[ 11] W.M. Meier, D.H. Olson, Ch. Baerlocher, Zeolites, 17 (1996) 146.
[12] B.D. Cullity, Elements of X-ray diffraction, Addision-Wesley, Reading, 1967, p.99.
[13] P.A. Jacobs, E.G. Derouane and J. Weitkamp, J. Chem. Soc., Chem. Commun., (1981) 591.
[14] E.M. Flanigen, H. Khatami and A. Seymenski, in E. M. Flanigen and L.B. Sand (eds), Adv.
Chemistry Series 101, American Chemical Society, Washington, D.C. 1971, pp. 201-228.
[15] M.A. Camblor, A. Corma and J. P6rez-Pariente, J. Chem. Soc., Chem. Commun., (1993) 557.
[16] S.M. Champbell, D.M. Bibby, J.M. Coddington, R.F. Howe and R.H. Meinhold, J. Catal., 161
(1996)338.
[17] E. Astorino, J.B. Peri, R.J. Willey and G. Busca, J. Catal., 157 (1995) 482.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 581
M.W. Anderson a, G.J.T. Tiddy b, C.C. Egger a, P. Hughes a, K.A. Brakke c and J.L. Casci d
aCentre for Microporous Materials, School of Chemistry, The University of Manchester, P.O.
Box 88, Sackville Street, Manchester M60 1QD, UK
bSchool of Chemical Engineering and Analytical Science, The University of Manchester, P.O.
Box 88, Sackville Street, Manchester M60 1QD, UK
CMathematics Department, Susquehanna University, Selinsgrove PA 17870, USA
dSynetix, Johnson Matthey, Billingham, UK
Writing on the near-perfect form of the beehive, D'Arcy Wentworth Thompson (1860-1948),
wrote in his life's work, 'On Growth and Form' [1 ],
"it is doubtful whether.., such mechanical forces.., could
produce the nearly smooth surfaces.., and the close approach to
a minimal configuration...
it seems much more likely that.., the walls assume their
configuration when in a semi-fluid state, while the watery pulp
is still liquid or the wax warm under the high temperature of the
crowded hive"
Fig. 1. Beehive wax and the structure ofMCM-41 viewed in the electron microscope
582
Fig. 2. Soap films inside an octahedron constraint - the areas are minimal because creating area costs
energy but the surface on the right is stable and the surface on the left metastable.[3]
also tends to a minimum. This minimum can either be a global minimum or a local minimum
relating to a stable or meta-stable surface respectively. A similar phenomenon is observed
when a wire frame is dipped into a soap solution to produce a soap film constrained only by
583
the boundary of the wire frame, Fig. 2. The complex three-dimensional soap film produced
has a minimal area as the free-energy of the surface is minimised. For one wire frame there
will be one stable surface and a number of meta-stable surfaces. Apart from the desire of the
surface to minimise area the curvature of the final surface, is related to the forces at the
surface. In order to understand this phenomenon, we need a definition of curvature, or
"crookedness", of a surface as referred by Newton, Fig. 3. In two dimensions, the curvature K,
at a point on a curved line is related to the radius, r, of a circl, ,:ascribed at this point such that
In more mathematical terms the curvature is the rate of change of the inclination of the
tangent, ~t, with arc length, s, at any given point, but the result is the same, that
d~, 1
K'-- --
ds r
for a three-dimensional surface the two principle curvatures may be defined at any point on
the surface, K~ and K2, which are the maximum and minimum curvatures. The planes
associated with these curvatures are not necessarily orthogonal. Three examples of surfaces
are shown in the Fig. 4: a surface such as part of a bubble where the two principle curvatures
have the same sign; a surface such as part of a cylinder where one curvature is zero; a saddle-
like surface where the curvatures have opposite sign. From these two curvatures it is
convenient to define two parameters: the mean curvature H
H: 1 K2) "~
5(~,+ =
Fig. 3. Curvature of a line defined by radius of inscribed circle left or as rate of change of
tangent angle ~ with arc length s [4].
the Gaussian curvature K
K - - K " 1 x K" 2
584
The Gaussian curvature determines the form of the surface as the first surface has positive K,
the second surface has zero K and the third surface has negative K. The mean curvature for
each type of surface will also vary but for the saddle type of surface an interesting possibility
exists when the principle curvatures are equal and opposite in sign. In such a case the mean
curvature of the surface is zero.
1 1
r, r2
.'.H = 0
Now consider the minimal surface of a soap film constrained by a wire frame. What can be
known about the curvature of such a minimal surface? If the soap film is open, without
bubbles, then pressure on both sides of the film is the same. Intuitively, it would seem
therefore, that it is impossible to develop curvature as the force on both sides of the film is the
same. Indeed this is the case and is summed up in the Laplace-Young expression which
relates the curvature to the excess pressure, p, across the film
p- cr.f +
where cyf is the surface tension of the film. Consequently, for an open film, without bubbles,
the excess pressure p = 0
1 1
.'.--+--= 0
r, r2
1 1
or
and H =0
In other words the mean curvature of every point on the surface is zero. The surface must
either have the form of a saddle or be flat. Positive mean curvature can only be generated by
Fig. 4. Curvature on a surface is defined by two plane curves with maximum and minimum curvature.
Left, as in a bubble has two positive curvatures; middle, as in a cylinder has one curvature zero; right,
is a saddle with one positive and one negative curvature[4]
585
having an excess pressure across the film which is achieved by closing a volume and
containing this pressure in a bubble. For the beehive D'Arcy Wentworth Thompson was
suggesting not only that the surface of the hive wax is minimal but also that the curvature, the
"nearly smooth surfaces" has been defined not by the bee but by a relaxation of soft wax to a
configuration related to Laplace-Young type principles.
Fig. 5. The catenoid surface constrained between two wire frames refined using "Surface Evolver"[5]
starting with a crude surface and then minimising area.
586
e
_Ix~+y~+z~) =c
N.B. this is the same as x2+y~-+z2=c which is obtained by taking a natural logarithm. The
sphere may then be moved to a new position h, k, 1, in the unit cell by the transformation
e
-t/x h~2 +Iy~ 2 +/z ,~2t =C
Fig. 6. Two exponential functions added together give two separate bodies when they are far
enough apart but a single continuous wrapped surface as the bodies approach.
i
Fig, 7. Left 21 exponentials giving the spheres and ellipsoidal micelles in the Pm-3n cubic surfactant
structure. Right, increasing the size of the micelles to give the wrapped surface of SBA-1 which is the
crude starting point for further refinement[7].
This exponential mathematics is useful because when exponentials are added together
multiple objects are created which when they approach form a continuous wrapped surface,
Fig. 6. This will be exactly the case for mesoporous materials which wrap around and micelle
template.
Fig. 8. Refinement using "Surface Evolver" of the crude SBA-1 surface defined using exponential
functions. The smoothing function was squared mean curvature which was minimised to near zero.
SBA-I[8] is a good case study for such an approach. The material which is synthesised
in acidic media has a structure based upon a cubic arrangement of micelles. There are two
types of micelle, a spherical micelle and an oblate ellipsoidal micelle. One unit cell of
micelles can be described by adding 21 exponential functions together for the 9 spheres and
12 ellipsoids. By increasing the radius of the micelles, a continuously wrapped surface can be
defined which is the crude starting representation for SBA-1, Fig. 7. This surface can then be
refined using "Surface Evolver" using the squared mean curvature as the smoothing function,
Fig. 8.[9] The final structure, Fig. 9, can be verified against experimental data such as x-ray
diffraction, electron diffraction and electron microscopy by filling one side of the surface with
electron density representing the silica walls.
Fig. 9. Atomistic model of SBA-1 from which the diffraction and microscopy data may be
obtained.
588
This general approach could in principle be used for any mesoporous structure as well
as disordered and intergrowth structures as any starting surface may be defined. Like the
beehive, the wall structures of these mesoporous materials appear to have well-defined
curvature and minimal area, therefore, would seem to form while in a soft fluid state before
hardening into the final structure. Also like the beehive the structure is hidden, and electron
micrographs, while containing all the information about this complex hidden structure, can
fool the eye about the pore structure without complete structural refinement[ 10]. The surfaces
will not necessarily conform to a zero mean-curvature surface because of the attractive intra-
micellar, and intra-silica forces which are maintaining the integrity of the individual phases.
However, because the interface between the silica and surfactant phases can, to a first
approximation, be considered uniform we might expect the interface to have constant
curvature. Surface Evolver is particularly good for evolving surfaces of any pre-defined
curvature, indeed it is, to our knowledge, the only way to approach this problem for the
complex surfaces often experienced in mesoporous materials, Fig. 10.
Fig. 10. Series of constant mean curvature surfaces for SBA-1 structure. The curvature is inset
in units of A ~ based on a 100A unit cell.
The key components in the formation of a mesoporous silica material, such as SBA-1, which
is synthesised in an acidic medium are (i) the initial hydrolysis of the silica source; (ii) the
precipitation of a mesophase structure and any reorganisation; (iii) the condensation and
hardening of the structure. Some of these processes are quite rapid, on a timescale of
minutes, whereas others take place over hours or days depending upon the temperature. As a
consequence suitable techniques must be chosen in order to access these different processes.
One further consideration is the dilute nature of typical surfactant preparations. This makes in
situ monitoring of kinetic processes quite challenging.
6000
a 4o00
zI = 3.7 rain
2000
7500
b 5~176176 ,.~
.- "t~ - 1.4 rain
2500 ~ ,-
0
AB 0 1000 2000 3000 4000 5000 6000 7000 8000
. . . . . . . . . . . . . . _,. ,,.w} time / s
ppm
Fig. 11. Time-resolved in-situ, room temperature, ~70 NMR carried out at pH=-0.7 (a without
surfactant, b with surfactant) showing the evolution in time of the water signal.
enriched ~70 water. The reason for the decrease in the intensity of the water signal is
primarily due to substantial binding of the water to silica species as the silica source
(tetraethylorthosilicate) hydrolyses[13] The water binding is so substantial that the water
activity is sufficiently decreased to effect onset of precipitation of a silica/surfactant/water
mesophase despite the very low concentration of surfactant in water (about 2 wt%). The
590
timescale for this initial hydrolysis reaction and concurrent water binding is of the order of 1-
2 minutes under normal synthesis conditions.
~ff..
Fig. 12. Penetration scans (brown images) of cetyltriethylammonium bromide with 5 wt%
bromohexadecane in water showing presence of an isotropic mesophase below 32.8~ at
concentrations between the hexagonal mesophase and isotropic solution. The top right figure shows a
penetration scan of cetyltriethylammonium chloride in water showing the presence of three isotropic
mesophases at concentrations between the hexagonal mesophase and isotropic solution. In all cases
the concentration of surfactant decreases from right to left on the microscope slide.
2.2. Ordering
The onset of order can be followed either by magnetic resonance techniques or using
diffraction. 14N NMR can be very sensitive as long as the quadrupole splitting of the
mesophase of interest is not too large. Very often for kinetic work it is sufficient to monitor
the change in intensity of an isotropic peak in order to establish the rates of kinetic processes.
For SBA-1 the surfactant employed is hexadecyltriethylammonium bromide (HTEABr).
Unusually, this surfactant does not form the micellar cubic I/phase, associated with SBA-1,
in a surfcatant water system. Nevertheless, we have shown[ 13] that this mesophase can easily
be induced either by the addition of small impurity phases such as bromohexadecane or by
exchanging bromide for chloride (SBA-1 synthesis is carried out in concentrated hydrochloric
acid solution) see the penetration scan in Fig. 12.
591
14N NMR of the range of surfactant mesophases (HTEABr) show the usual behaviour
although, somewhat unusually for this particular surfactant, the quadrupole splittings are
coincidentally quite small, of the order of a few hundred Hertz, Fig. 13. To follow the
kinetics of formation it is sufficient to follow the intensity of the isotropic micellar signal
which reduces as the l~ phase is formed which has a larger half-width. This increased half-
width is brought about by the presence of incompletely averaged quadrupole tensors for
surfactant within the ellipsoidal micelles. The time constant for ordering monitored by 14N
NMR is of the order of 13 minutes, Fig. 13, which is similar to that found from energy
dispersive x-ray anaysis, Fig. 14.
.... ,...,,...,...;....~...,,.-......
: ........ - ...........
'. ...... 9 . .
~ [| ", ~ .... : ~ ~,:=., ~j~~--~.... !~:~:~i~i:ili:i~::!:
:~,~:~ :./-~: ........::~ !
9. so , z .:to0,z "= / ~176 " - . * :) .... "-., . g ' ) % , ; ; ) > =~* .~ ' ~ !~, " , . ,
. . . . ,) ; .... ,, .
Fig. 13. Left: 14N NMR spectra for HTEAB/water mixtures showing spectra for the 5 phases,
lamellar, bicontinuous cubic, hexagonal, P,-~,and isotropic. Right: decay of 14N NMR signal as a
function of time for pH=-0.7. Insets: XRD patterns of the final materials.
3O
sB,~-~ t T .~8 A _ , , ,, ,, ,
............... :[
.
..'__
zs ~ i -(
i
lO
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85
Fig. 14. Growth of 38A diffracted peaks with time, at RT for SBA-1 synthesis monitored by energy
dispersive x-ray analysis.
Combining the structural and kinetic pictures of the growth of SBA-1 leads to a final picture
which indicates initial hydrolysis on a timescale of about 2 minutes. This is couple with a
significant reduction in water mobility, and hence water activity, which in turn triggers the
precipitation, salting out, of an ordered mesophase on a timescale of 13 minutes. A soft solid
is created which then undergoes substantial reorganisation to a low curvature surface
incorporating windows between cages. The soft solid is transformed to a hard solid through
subsequent heating to 90~ Fig. 15. The manipulation of a soft solid and the approach to a
592
minimal configuration for SBA-1 is very reminiscent of the description given by Wentworth
Thompson for the beehive.[ 1]
HTEAB , TEOS
as m~ceites l hydrolysis complete in
2 rains ~70
-
SiJicic Qcids
9 ,~, oligomersiation
,s~t~ca.oI!~3om~rs
. . . . . . . . salting out ,~4N NMR
Reorganisation - s t r u c t u r e has
creation of windows z e r o curvature
Har'd .SBA-t
Fig. 15. General kinetic picture representing the different steps involved in forming SBA-I[13].
REFERENCES
[1] D'Arcy Wentworth Thompson, "On Growth and Form: The Complete Revised Edition", Dover
Publications (1992).
[2] (a) C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli & J.S. Beck, Nature 359, (1992) 710:
(b) S. Inagaki, Y. Fukushima & K. Kuroda, Journal of the Chemical Society-Chemical
Communications (1993) 680.
[3] C. Isenberg, "The Science of Soap Films and Soap Bubbles", Dover Publications (1992).
[4] S. Hyde, S. Andersson, K. Larsson, Z. Blum, T. Landh, S. Lidin and B.W. Ninham, "The
Language of Shape" Elsevier Science (1996).
[5] K.A. Brakke, Philosophical Transactions of the Royal Society of London Series a-Mathematical
Physical and Engineering Sciences 354 (1996) 2143.
[6] M. Jacob and S. Andersson, "The Nature of Mathematics and the Mathematics of Nature",
Elsevier Science (1998).
[7] M.W. Anderson, C.C. Egger, G.J.T. Tiddy and J.L. Casci, Stud. Surf. Sci. Catal., 142 (2002)
1149.
[8] G. Huo, D.I. Margolese, U. Ciesla, D.G. Demuth, P. Feng, T.E. Gier, P. Sieger, A. Firouzi, B.F.
Chmelka, F. Scht~th & G.D. Stucky, Chem. Mater., 6 (1994) 1176.
[9] M.W. Anderson, C.C. Egger, G.J.T. Tiddy, J.L. Casci and K.A. Brakke, Angew. Chem., 44
(2005) in press.
[10] Y.H. Sakamoto, M. Kaneda, O. Terasaki, D.Y. Zhao, J.M. Kim, G.D. Stucky, H.J. Shin & R.
Ryoo, Nature 408 (2000) 449.
[11] S. Ruthstein, V. Frydman, S. Kababya, M. Landau and D. Goldfarb D, J. Phys. Chem. B 107
(2003) 1739.
[12] S. Ruthstein, V. Frydman and D. Goldfarb D, J. Phys. Chem. B 108 (2004) 9016.
[13] C.C. Egger, M.W. Anderson, G.J.T. Tiddy and J.L. Casci, PCCP, 7 (2005) in press.
[14] K. Flodstrom, H. Wennerstrom, C.V. Teixeira, H. Amenitsch, M. Linden, and V. Alfredsson,
Langmuir, 20 (2004) 10311.
[15] K. Flodstrom, H. Wennerstrom and V. Alfredsson, Langmuir, 20 (2004) 680.
[16] K. Flodstrom, C.V. Teixeira, H. Amenitsch, V. Alfredsson and M. Linden, Langmuir, 20 (2004)
4885.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka, N. Zilkov~i and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 593
1. INTRODUCTION
Experimental studies [1-3] have clearly shown that proton-containing forms of ZSM-5
zeolites are much more active in catalytic transformations of alkanes to alkenes and aromatics
when extra-framework zinc cationic species are also present. However, neither the detailed
mechanism of these transformations is known nor is the role of Zn cations in it clear. In
addition, matters are complicated by the fact that various forms of Zn species in zeolites are
2+ + 2+
possible: Zn , ZnOH , [Zn-O-Zn] or Zn(H20) 2+. Several experimental studies dealt with
the identification and the transformations of such species for various preparation techniques
and sample treatments [1,4-7]. Berndt et al. [4] proposed that ZnOH + species located in the
vicinity of Bronsted acid sites of ZSM-5 zeolites are the reactive zinc species for propane
aromatization. Biscardi et al. [ 1], however, suggested that such ZnOH + species are thermally
unstable due to the formation of water or binuclear Zn-O-Zn 2+ moieties. Unfortunately, there
is no experimental technique which allows a direct determination of the state of Zn species in
zeolites. For example, IR spectroscopy cannot discriminate between the O-H vibrations of
bridging zeolite OH groups and of ZnOH + species. Some experimental techniques, e.g. X-ray
absorption spectroscopy, isotopic titration of Bronsted acid sites with D2, and temperature
programmed reduction, provide indirect information about the structure of Zn containing
zeolites. Only by combining such methods, one would be able to obtain a clearer picture about
the state of Zn species in zeolites [ 1--4].
As complementary approach, several theoretical model studies of Zn containing zeolites,
mainly of ZSM-5, have been carried out. Using density functional (DF) calculations for
594
isolated cluster models, Bell et al. [8,9] investigated the coordination of Zn 2+, ZnOH +, and
ZnOZn 2+ on different rings, located in the main and the sinusoidal channels of an MFI
framework, exhibited by zeolite ZSM-5. These authors found that Zn 2+ forms most stable
complexes in 5- and 6- membered rings of the main channels. Barbosa et al. [10,11] also
modeled Zn 2+ and ZnOZn 2+ species at zeolite rings in the context of H2 and CH4 adsorption.
However, thus far, a comprehensive study is lacking which attempts to estimate the relative
stability of possible Zn-containing species in a uniform approach. Moreover, the formation
and stability of Zn(H20) 2+ species has not yet been modeled. Comparing complexes of Zn 2+,
ZnOH +, [Zn-O-Zn] 2+, and Zn(H20) 2+ for two quantum cluster models, this theoretical study
aimed at identifying the most stable type of cationic zinc species in zeolite ZSM-5. In
addition, we also modeled the most stable structures at one of the clusters by applying the
sophisticated covEPE embedded cluster approach.
2. COMPUTATIONAL DETAILS
Initially, we optimized the MFI structure at the molecular mechanical (MM) level with the
program GULP [12], using a previously reported shell model force field (FF) [13,14]. From
this structure we choose two model clusters, denoted as M5 and M7, according to the number
of T atoms in them, to be treated at the quantum mechanical level (QM). These models
comprise a single (fig. l a,c) and two coupled (fig. l b,d) 5-membered zeolite rings, respective-
ly, which face the main channel of the MFI structure. Each cluster contained 2 A1 centers,
located at the positions T2 and T9 for model M5 and at the positions T1 and T7 for model M7
(Fig. l a,b). The calculations were performed with the parallel density functional program
PARAGAUSS [15,16] using two different modeling strategies: isolated cluster models and the
recently developed hybrid QM/MM method, referred to as covalent elastic polarizable
environment, covEPE [13,14]. The QM part of the systems was described by a gradient-
corrected exchange-correlation functional [17]. The Kohn-Sham orbitals were represented by
Gaussian-type basis sets, contracted in generalized fashion [ 18].
In the isolated cluster models, the dangling bonds were saturated by H atoms, and the
direction of these bonds at the cluster boundary was fixed as obtained in the MM model. The
lengths of the corresponding T-H or O-H bonds were pre-optimized, keeping all other atoms
in the cluster fixed, followed by an optimization of the cluster with fixed positions of all
saturating H atoms. In the subsequent geometry optimization steps, both in protonic and Zn-
containing models, only the positions of the O centers in the ring, the protons of the bridging
OH groups, and the guest species were relaxed. The most stable species obtained with isolated
M5 cluster models were used to construct the initial structures for the embedded cluster
models. The covEPE embedding scheme, employed here, is well suited to deal with zeolites
and. solid oxide systems that feature polar covalent bonds [13]. Instead of saturating dangling
bonds of the model QM cluster with artificial H atoms, suitably parameterized monovalent
pseudopotential centers O* are invoked which represent real framework O atoms and
minimize the perturbation of the electrostatic field at the cluster boundary. The method
completely and explicitly includes both mechanical and electrostatic interactions between the
QM cluster and its MM lattice environment, including also mutual polarization of both
subsystems. For a correct description of the embedded cluster environment, a specially
parameterized FF based on potential-derived charges was derived, both for pure-silica [13]
and aluminosilicate materials [14]. The structure of an entire system is variationally
determined by minimizing a total energy expression that includes all degrees of freedom of
595
both QM and MM regions (explicitly including about 700 centers). A detailed description of
the covEPE method can be found in Refs. [13,14].
3. RESULTS
Table 1
Selected interatomic distances (pm) in the model clusters and average Zn-O distances <R>
Model Oz-H Zn-Oz Zn-O a <R> O-H a Hw-Oz
M5/2H 99, 99
M7/2H 98, 98
M5/2H,H20 99, 102 100, 98 193
M7/2H,H20 98, 105 99, 98 210
M5/Zn 1908/188, 214 c 197
M7/Zn 215, 216, 236d/198, 227 e 218
M5/H,ZnOH(a) 98 195, 2118 180 195 98
M5/H,ZnOH(b) 98 225b/205,220 ~ 182 208 98
M7/H,ZnOH 110 197, 201 d 185 194 98
MS/ZnH20 199b/195, 225 c 202 205 100,104 199,164
M7/Zn(HzO)(a) 211,238d/200 e 206 214 99,99 214,202
M7/Zn(H20)(b) 226, 233d/200, 224 e 209 218 99,99 233,234
M5/ZnOZn 198, 2168/196, 215 ~ 181, 183 198
M7/ZnOZn 195,217d/197, 211 e 181, 181 197
M5*/Zn f 192, 2248/198, 200 ~ 204
M5*/Zn(H20) f 198b/200, 201 c 201 200 101,102 166,176
a Distances inside guest species; b distance to O bound to A1 at T2; c distance to O bound to A1 at T9;
a distance to O bound to A1 at T1; e distance to O bound to A1 at T7; f embedded cluster models
596
/)
T ~'T9 f~:
<.. "5;" .:/
7,
,,
As expected, Zn 2+ species in both M5 and M7 clusters (Figs. 1c,d) interact stronger with
zeolite O centers from A1-O-Si bridges, similarly to other metal cations [23]. However,
depending on the size of the model cluster, different numbers of such O centers are located
suitably to interact with Zn 2+ cation, 3 in M5 with an average Zn-O distance <R> = 197 pm
and 5 in the M7 model, but at longer Zn-Oz distances <R> = 218 pm (Table 1).
For the M5 model we found two structures of ZnOH + species, differing by the location
of zeolite OH group (Figs. l e,f). In structure M5/H,ZnOH(a) (Fig. l e), which is 38 kJ/mol
more stable than structure M5/H,ZnOH(b), ZnOH + is coordinated at the T2 A1 center with Zn-
Oz distances of 195 and 211 pm. In the alternative structure M5/H,ZnOH(b), the ZnOH +
moiety interacts with O centers bound to both A1 centers, but at longer Zn-Oz distances, 205-
225 pm. The stable structure of ZnOH + species at the M7 model, M7/H,ZnOH (Fig. l g), is
similar to the most stable structure at M5, i.e. ZnOH + is coordinated at one of the A1 centers.
In addition, the OH group of the guest species forms a strong hydrogen bond with H + of the
bridging OH group of the zeolite substrate, with R(Hz-Ozo) = 144 pm. In the second model
structure of ZnOH + species at the M7 cluster, the zeolite proton completely migrated on the
guest species and the complex M7/Zn(H20)(b) formed spontaneously (Fig. lj). That latter
structure is 15 kJ/mol more stable than the complex M7/H,ZnOH (Fig. l g).
The H20 molecule in the optimized structure M5/Zn(H20) (Fig. lh) forms two
hydrogen bonds with O atoms of the zeolite, a stronger one with O bound to A1, R(Hw-Oz) -
164 pro, and a weaker one to O bound to Si at 199 pm. In addition to the M7/Zn(H20)(b)
structure described above, we considered two further initial adsorption complexes of H20 at
the M7/Zn cluster. However, during geometry optimization, both yielded the same final
structure M7/Zn(H20)(a) (Fig. li) where the H20 molecule forms two weak hydrogen bonds
with O atoms, both bound to Si centers with Hw-Oz distances of 202 and 214 pm. At variance
with M7/Zn(H20)(a), H20 in the complex M7/Zn(H20)(b) mentioned above has no hydrogen
bonds with zeolite O centers (Fig. l j) and is 19 kJ/mol less stable. The latter value may be
considered as estimate of the hydrogen bonding contribution to the binding energy of H20.
We considered one structure of the binuclear species Zn-O-Zn 2+ on model M5 (Fig. lk)
and two structures on model M7 (Fig. 11). Initially, the two M7/ZnOZn structures differed by
the orientation of the ZnOZn species with respect to the A1-A1 direction: parallel and
perpendicular. Yet, in both cases optimization yielded the same structure with the ZnOZn 2+
moiety oriented parallel to the A1-A1 direction. The ZnOZn 2+ species bind similarly to M5 and
M7 rings: each Zn centers is coordinated to two O centers which, in turn, are bound to one of
the AI centers. The Zn-Zn distance is 279 and 277 pm in the M5 and M7 models, respectively;
this characteristic might be used to identify ZnOZn 2+ moieties in EXAFS studies.
The structures of the complexes M5*/Zn and M5*/Zn(H20) obtained with the embedded
cluster models using the covEPE approach (Fig. l m,n) are similar to those of the isolated
clusters, although the individual Zn-O distances differ to some extent (Table 1).
Table 2
Model exchange reaction energies a (kJ/mol) for Zn species at the clusters Mk (k - 5, 7)
No. Model exchange reactions k = 5a k = 7a
1 Mk/2H + Zn 2+ --+ Mk/Zn + 2H § 376 315
2 Mk/2H + ZnOH + ---+ Mk/H,ZnOH + H + 476 453
3 Mk/2H + Zn(H20) 2+ --+ Mk/Zn(H20) + 2H + 660 654
4 Mk/2H + ZnOZn 2+ --+ Mk/ZnOZn + 2H + 709 699
a A positive value corresponds to an endothermic reaction
Table 3
Energies (kJ/mol) for transformation reactions between different Zn-containing species
No. Reaction k=5 k=7
5 Mk/Zn + H 2 0 --+ Mk/Zn(H20) -184 -130
6 Mk/H,ZnOH --+ Mk/Zn(H20) -52 -34
7 2Mk/H,ZnOH --+ Mk/ZnOZn + Mk/2H + H2O 111 149
8 2Mk/Zn(H20) --+ Mk/ZnOZn + Mk/2H + H20 215 217
Due to the larger size of Zn-O-Zn moiety, one may expect that it will bind stronger to
the M7 cluster; however, the difference between the exchange energies of M5/ZnOZn and
M7/ZnOZn is only 10 kJ/mol. All Zn species have a stronger preference for exchange of
protons on the M7 than on the M5 model, likely due to the size of the model. The M7 cluster
offers more space for accommodating Zn species and better access to more O centers
compared to M5, as could be seen from the structures of the complexes described above.
The main goal of our study was to clarify which is the most stable type of Zn species in
ZSM-5 zeolite. The exchange energies reported above are not suitable for such a comparison
because the corresponding isolated species were assumed to be in the gas phase. Therefore,
we considered interconversion reactions between the zeolite-supported species studied (Table
3) that allow us to evaluate their relative stability.
From reaction (5) one concludes that H20 is strongly adsorbed at a Zn 2+ cation which is
embedded in a zeolite; the binding energy BE(H20) is 184 and 130 kJ/mol for M5 and M7,
respectively. The larger BE of H20 in the M5/Zn(H20) complex is likely due to the higher
stability of the M7/Zn species compared to M5/Zn. These values are significantly larger, by
106 and 53 kJ/mol, than the BE of H20 on the protonic forms of M5 and M7. To check this
result with a more accurate modeling approach to the zeolite structure, we also calculated the
BE of H20 on a M5/Zn cluster embedded using the covEPE embedding method [13]. In this
case, the BE is calculated 20 kJ/mol smaller than for the corresponding isolated cluster, likely
because the embedded model cluster is more flexible, which then allows a better coordination
of the bare Zn 2+ cation. Yet, the binding energy of H20 is still rather high, 164 kJ/mol.
As described in Section 3.1., in one of the models initially constructed as M7/H,ZnOH
complex, the optimization resulted in the migration of a zeolite proton towards the ZnOH +
moiety and the formation of a Zn(H20) 2+ species. Indeed, this transformation, reaction (6), is
calculated favorable both on M5 and M7 cluster models by 52 and 34 kJ/mol, respectively. A
previous theoretical study [24] considered a possible dissociation of H 2 0 on Z n "+ cations in
zeolites to ZnOH + and a bridging OH group. The reaction was calculated exothermic by 138
kJ/mol; the corresponding values from our calculations are 132 (M5) and 96 (M7) kJ/mol,
599
using reactions (5) and (6). However, one should keep in mind that Yakovlev et al. [24] did
not consider the formation of the most stable species, Zn(H20) 2+.
The relative stability of the ZnOZn 2+ species was estimated separately with respect to
two ZnOH + or two Zn(H20) 2+ species, i.e. we considered the reaction between two mono-
nuclear Zn-containing species located close to each other inside a zeolite channel, which leads
to the formation of one binuclear ZnOZn 2+ moiety, a separate cluster in protonic form, and a
water molecule, reactions (7) and (8). This dehydration process and the formation of ZnOZn 2+
species is endothermic by more than 100 and 200 kJ/mol per binuclear moiety, when the
reaction includes ZnOH + and Zn(H20) 2+, respectively (Table 3).
In summary, from our isolated cluster calculations we can conclude that the most stable
species both at M5 and M7 rings are Zn(H20) 2+ species with very high BE values of H20.
4. SUMMARY
existence of ZnOZn 2+ moieties in zeolites, we found in agreement with some previous theo-
retical and EXAFS studies that these moieties are less stable than other types of cationic Zn.
ACKNOWLEDGMENTS
We thank V. Nasluzov and E. Ivanova Shor for advice regarding the covEPE calculations.
This work was supported by Deutsche Forschungsgemeinschaft, National Science Found
(Bulgaria), Alexander yon Humboldt Foundation, and Fonds der Chemischen Industrie.
REFERENCES
[ 1] J.A. Biscardi, G.D. Meitzner and E. Iglesia, J. Catal., 179 (1998) 192.
[2] J.A. Biscardi and E. Iglesia, J. Catal., 182 (1999) 117.
[3] J.A. Biscardi and E. Iglesia, Phys. Chem. Chem. Phys., 1 (1999) 5753.
[4] H. Berndt, G. Lietz and J. V61ter, Appl. Catal., 146 (1996) 365.
[5] V.B. Kazansky, V.Yu. Borovkov, A.I. Serikh, R.A. van Santen and B.G. Anderson, Catal. Lett.,
66 (2000) 39.
[6] EI-M. E1-Malki, R.A.van Santen, and W.M.H. Sachtler, J. Phys. Chem. B, 103 (1999) 4611.
[7] V.B. Kazansky, A.I. Serykh, B.G. Anderson, and R.A. van Santen, Catal. Lett., 88 (2003) 211.
[8] A.T. Bell, in Catalysis by Unique Metal Ion Structures in Solid Matrixes, G. Centi, B.
Wichterlowa and A.T. Bell (eds), Kluwer Academic Publishers, Dordrecht, 2000, pp. 55-73.
[9] M.J. Rice, A.K. Chakraborty and A.T. Bell, J. Phys. Chem. B, 104 (2000) 9987.
[10] L.A.M.M. Barbosa, G.M. Zhidomirov, and R.A. van Santen, Catal. Lett., 77 (2001) 55.
[11] L.A.M.M. Barbosa and R.A. van Santen, J. Phys. Chem. B, 107 (2003) 14342.
[12] (a) J. Gale, GULP, 1.3 version, Imperial College, London. (b) J.D. Gale, J. Chem. Soc. Faraday
Trans., 93 (1997) 629.
[13] V.A. Nasluzov, E.A. Ivanova, A.M. Shor, G.N. Vayssilov, U. Birkenheuer, and N. R6sch, J.
Phys. Chem. B, 107 (2003) 2228.
[14] E.A Ivanova Shor, A.M. Shor, V.A. Nasluzov, G.N. Vayssilov and N. R6sch, to be published.
[15] T. Belling, T. Grauschopf, S. Krtiger, F. N6rtemann, M. Staufer, M. Mayer, V.A. Nasluzov, U.
Birkenheuer, A. Hu, A.V. Matveev, A.M. Shor, M.S.K. Fuchs-Rohr, K.M. Neyman, D.I.
Ganyushin, T. Kerdcharoen, A. Woiterski, A.B. Gordienko, S. Majumder, and N. R6sch,
PARAGAUSS version 3.0, Technische Universit~it Mtinchen, 2004.
[16] T. Belling, T. Grauschopf, S. Krfiger, M. Mayer, F. N6rtemann, M. Staufer, C. Zenger, and N.
R6sch, in: High Performance Scientific and Engineering Computing; ed. Bungartz, H.-J.; Durst,
F.; Zenger, C.; Lecture Notes in Computational Science and Engineering, Vol. 8, Springer,
Heidelberg, 1999, p. 439.
[17] (a) A. D. Becke, Phys. Rev. A, 38 (1988) 3098. (b) J. P. Perdew, Phys. Rev. B, 33 (1986) 8822.
[18] H: (6slp)--~[4slp], O: (9s5pld)--~[5s4pld], A1, Si: (12s9p2d)--~[6s4p2d], see Ref. [19]; Zn:
( 15s 11p6d)---~[6s5p3d], see Ref. [20].
[19] (a) F. B. Van Duijneveldt, IBM Res. Report (1971) RJ 945. (b) S. Huzinaga (ed.) Gaussian Basis
Sets for Molecular Calculations, Elsevier, Amsterdam, 1984. (c) A. Veillard, Theor. Chim. Acta,
12 (1968) 405. (d) M. R. B~ir, and J. Sauer, Chem. Phys. Lett. 226 (1994) 405.
[20] S. Huzinaga, J.Chem. Phys. 66 (1977) 4245.
[21] B. Hunger, M. Heuchel, S. Matysik, K. Beck, and W.D. Einicke, Thermochem. Acta, 269/270
(1995) 599.
[22] J. A. Ryder, A. K. Chakraborty and A. T. Bell, J. Phys. Chem. B, 104 (2000) 6998.
[23] a) N. R6sch, G.N. Vayssilov, and K.M. Neyman, in: Host-Guest-Systems Based on Nanoporous
Crystals, F. Laeri et al. (Eds.), Wiley-VCH, Weinheim, 2003, p. 339. b) G.N. Vayssilov and N.
R6sch, J. Phys. Chem. B, 105 (2001) 4277. c) K.I. Hadjiivanov, and G.N. Vayssilov, Adv. Catal.,
47 (2002) 307. d) D. Nachtigallova, P. Nachtigall, M. Sierka and J. Sauer, Phys. Chem. Chem.
Phys., 1 (1999) 2019.
[24] A.L. Yakovlev, A.A. Shubin, G.M. Zhidomirov and R.A. van Santen, Catal. Lett., 70 (2000) 175.
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 601
Institut ~ r Materialphysik and Center for Computational Material Science, Universit~it Wien,
Sensengasse 8, A- 1090 Wien, Austria
Periodic DFT calculations have been performed to characterize the acidic properties of
bridging and terminal OH groups and various defect sites present in zeolites. As a model
system mordenite with Si/A1-23 has been used. Binding energies and vibrational frequencies
of adsorption complexes of frequently used probe molecules such as CO and NH3 have been
calculated. Ammonia is protonated at bridging OH groups while it is H-bonded upon
adsorption at terminal silanol groups. A defect with a two-membered silica ring is a weak
Lewis site as indicated by very low interaction energies with probe molecules. The Bronsted
acidity of a defect with the bridging OH group in the two-membered ring is mid-way between
that of bridging OH groups and terminal silanol groups. A three-coordinated surface AI atom
is a strong Lewis site.
1. INTRODUCTION
Identification and characterization of catalytically active centers is one of the most important
tasks in the optimization of the performance of zeolite-based catalysts. Rather well
understood are the acidic properties of bridging OH groups belonging to the strongest
Bronsted acid sites in zeolites. Less well known are chemical properties of various defect sites
inevitably created during the synthesis or thermal treatment of the catalyst. One of the reasons
is the difficult identification of defect sites, due to their low concentration relative to other
sites like bridging and terminal OH groups. In this study we use state-of-art ab initio
simulation techniques to characterize chemical properties of active centers in mordenite. In
some sense we follow the strategy frequently used by experimentalists to characterize the
acidity of the active sites: we investigate the modification of the vibrational frequencies of
small probe molecules induced by interaction with the active center. As probe molecules we
have choosen carbon monoxide and ammonia which are frequently used in IR and TPD
experiments. Our attention is also directed to defect sites shown to be formed under realistic
conditions (see Ref.[1]), some of them being known experimentally: silanol nests, two-
membered rings (either pure Si-O rings or containing an A1 atom and a compensating proton),
and three-coordinated A1 atoms. In addition, we consider also the bridging and terminal OH
groups whose acidic properties are well known and documented. In this study, these two
functionalities serve as the reference systems for strong and weak Bronsted acid sites. More
detailed studies of the chemical properties of bridging and terminal OH groups can be found
in Re fs [2,3 ].
602
2. METHODOLOGY
I'X ....
"" "".... I""",.
i~.....~ X., ,. ~ ,
. I
"-"":\ I I".../"'!
........\ ~.....~i"..X.,..':
--'":"-.I .....
/"' Sn~afl :"i Main i:" ~nall :]
: cha~'rtel ! channel ~ chai Lnel ?'
f
"t
....
.- -:, ~.
. . . . . . . . "::i')
/" ,.!.
-
f :.......'. 'X
.-,.......r . )..j /.-,. ....... I
.....
/i ! ........! I
\,..j.---..,,...i \..x ......<,.....'/
Fig. 1. Conventional orthorhombic (solid line) and primitive monoclinic (dashed line) cell of
mordenite projected onto the (001) plane.
603
A1
~1.. L i
8 A1
a) b) c)
Fig. 2. Adsorption complexes of ammonia at bridging OH groups in mordenite: top view of the
monodentate complex created in the main channel (a), front (b) and side view (c) of the tetradentate
complex in the small channel of mordenite (cf. Fig. 1)
604
Table 1
Calculated harmonic vibrational frequencies (cm -1) of free ammonium ion and of NH4 +
created upon adsorption at bridging OH groups and at defect with silanol nest in mordenite.
... VI V2 V3 V4 V5 V6 V7 V8 V9
NH4 + 3403 3403 3403 3314 1709 1709 1453 1453 1453
Monodentate 3488 3475 3374 1705 1677 1650 1533 1477 1278
Tetradentate 3426 3331 3273 2556 1706 1675 1497 1457 1390
Silanol nest 3507 3438 2915 2635 1680 1620 1545 1447 1259
Carbon monoxide interacts with the H atom of the bridging OH groups via carbon atom
to form weakly bonded 1:1 adsorption complexes [16]. Out of 26 possible locations for
bridging OH groups which can be created in mordenite[8], in this study we consider three OH
groups in the main channel and the same number in the small channel of mordenite. The
average binding energy for sites in the main channel is 18 kJ/mol, whereas the adsorption in
the side channel is endothermic (Agbind=-3 kJ/mol). Due to a systematic error of DFT method
which does not account correctly for the van der Waals interactions, our calculations
underestimate experimental adsorption energy o f - 2 6 kJ/mol[ 16]. Even taking into account a
possible stabilization of the complexes by van der Waals interactions, our results show that
CO adsorbs predominantly in the main channel of mordenite.
The downshift of the OH mode and the upshift of the CO stretching frequencies reflect
the strength of the OH...CO interaction and in turn measure the acidity of OH group. Maache
et al. [17] measured a AvoH of 332 cm l and Avco of 34 cm -~ for CO in mordenite. The
calculated AvoH and Avco are 353 and 41 cm ~, and 482 cm "l and 64 cm ~ for sites in the
main and in the small channel, respectively. Our calculations predict unrealistically large
shifts of both AvoH and Avco for the complexes in the small channel. Combined with the
result that the adsorption in the small channel is endotermic, we conclude that CO adsorbs
only in the main channel.
TSil2
TSill
Fig. 3. Top view of the (001) surface of mordenite. Terminal silanol groups designed
TSil 1 and TSiI2 are considered in this study.
Table 2
Calculated harmonic vibrational frequencies of a free ammonia molecule and of NH3
physisorbed at terminal silanol groups and defect sites created at the (001) surface of
mordenite. For notation see Figs. 3 and 5. _
VI V2 V3 V4 V5 V6
NH3 3509 3509 3389 1623 1623 997
TSill 3521 3483 3388 1628 1611 1132
TSil2 3515 3491 3399 1630 1616 1132
2MR 3503 3497 3388 1611 1601 1195
2MR-AI 3509 3502 3395 1624 1614 1160
3AI 3487 3476 3382 1612 1608 1259
energy for CO at terminal silanol groups is 7 kJ/mol. Our calculations can be compared with
results of Beebe et al.[20] who measured AvOH--92, AvCO=I 5 cm -~ and an isosteric heat of
adsorption of 11 kJ/mol for CO adsorbed at terminal silanol groups on a SiO2 surface.
Fig. 4. Detailed view of the unperturbed defect with silanol nest (left) and the adsorption complex
upon chemisorption of ammonia (right).
connected in the silanol nest. Thus H-bonding is found to modify the reactivity of otherwise
chemically identical species (SiOH groups).
The vibrational frequencies of ammonia adsorbed at a silanol nest are collected in Table
2. As we have demonstrated in the previous discussion, frequencies in the region between
1250 to 1600 cm ~ are a clear evidence for the formation of an ammonium ion. Bordiga et al.
[22] characterized defective sites in silicalites. Upon adsorption of ammonia at the sample
with presumably high concentration of silanol nest defects, a peak a t - 1 5 2 0 cm ~ appeared.
Although authors did not identified this maximum, we believe that this measurement supports
our hypothesis of an enhanced acidity of OH groups in silanol nests.
H
Si
Fig. 5. Detailed view of surface defects: a two-membered ring with a bridging OH group
(2MR-AI), a two-membered ring silica (2MR), and a defect with a three-coordinated AI atom (3Al).
607
b) .- ~N
N
T \
78kJ/mol..~ "~
..........1.4.4..,/.re...............................
o!
Fig. 6. Opening of the two-membered silica ring upon interaction with ammonia: physisorbed
ammonia (a), transition state (b), and chemisorbed NH3 (c).
energy is < 1 kJ/mol), it is known to react with molecules containing slightly acidic atoms
[24,25] . In this respect it behaves like a weak Lewis acid or alternatively like a Bronsted
base. This behavior can be illustrated by interaction with ammonia. In the former case NH3 is
physisorbed to the silicon atom (Fig. 6(a)). The calculated interaction energy is 48 kJ/mol.
Another possibility is chemisorption of ammonia in which case one Si-O bond in the 2MR is
broken, one proton of NH3 is transferred to a dangling SiO group and a new bond is
established between the N atom and a coordinatively unsaturated Si atoms (see Fig. 6). This
process is activated, the calculated reaction barrier is 78 kJ/mol. The adsorption energy of
ammonia is in this case 92 kJ/mol, i.e. almost doubled compared to the physisorbed complex.
As in the case of the adsorption at terminal silanol groups, the vibrational frequencies of
NH3 physisorbed at a 2MR are derived from those of the free ammonia. Due to rather weak
interactions, both stretching and bending vibrational frequencies are modified only
moderately (see Table 2). The characteristic vibrational frequencies of the NH2 group created
in the chemisorbed configuration are 3620 (asymmetric stretching mode), 3529 (symmetric
stretching mode), and 1530 cm l (bending mode). Bordiga et al. [26] observed a broad band
centered a t - 1 4 8 0 cm -I upon dosing with ammonia. The band has been attributed to 6(NH)
vibrations of the NH2 group after opening the strained bonds in the 2MR ring.
4. C O N C L U S I O N S
We have used periodic DFT technique to characterize various sites in mordenite. Our analysis
showed that the acidity of silanol groups significantly increases upon formation of silanol nest
- whereas ammonia forms only H-bonded complexes with the isolated silanol group, it is
protonated upon interaction with the OH group in the silanol nest. This effect leads to
increased reactivity of silanol nests, as documented e.g. by decrease of activation energy for
608
the rate-determining step for Beckmann rearrangement of cyclohexanone oxime from 266
kJ/mol for isolated silanol group to 149 kJ/mol for silanol nest.
Defect with the bridging OH group and A1 atom in the two-membered ring is a Bronsted
acid site with acidity on mid-way between framework bridging and terminal OH groups.
Ammonia is only H-bonded in the adsorption complex but the shifts of the OH and CO
stretching frequencies measuring the acidity are definitely larger compared to those for
terminal silanol groups.
Lewis sites are represented by defects with three-coordinated A1 atom and two-membered
silica ring. The former is a strong Lewis site interacting strongly with both CO and NH3
molecules. Ammonia can be either physisorbed or chemisorbed at defect with two-membered
silica ring. In the latter case the ring is opened and proton transfer from ammonia to O atom of
the defect takes place. In this way the NH2 and OH groups are created.
ACKNOWLEGMENT
This work has been supported by the Austrian Science Funds under project no. P 17020.
REFERENCES
[1] T. Bucko, L. Benco and J. Hafner, J. Chem. Phys., 118 (2003) 8437.
[2] T. Bucko, L. Benco, Th. Demuth and J. Hafner, J. Chem. Phys., 117 (2002) 7295.
[3] T. Bucko, J. Hafner and L. Benco, J. Chem. Phys., 120 (2004) 10263.
[4] G. Kresse, Vienna Ab-initio Simulation Package, Vienna University, Austria.
[5] J.P. Perdew and Y. Wang, Phys. Rev. B 46 (1992) 6671.
[6] B. Hammer, L.B. Hansen and J.K. Norskov, Phys. Rev. B 59 (1999) 7413.
[7] T. Bucko, J. Hafner and L. Benco, J. Phys. Chem. B, submitted
[8] T. Demuth, J. Hafner, L. Benco and H. Toulhoat, J. Phys. Chem. B, 104 (2000) 4593.
[9] H.G. Karge and V. Dondur, J. Phys. Chem., 94 (1990) 765.
[10] M. Niwa, N. Katada, M. Sawa and Y. Murakami, Stud. Surf. Sci. Catal., 98 (1995) 101.
[11] D.J. Parillo and R. Gorte, J. Phys. Chem., 97 (1993) 8786.
[12] D.T. Chen, L. Zhang, C. Yi and A. Dumesic, J. Catal., 146 (1994) 257.
[13] H. Stach, J. J~inchen, H. G. Jarschkewitz, U. Lohse, B. Parlitz and B. Hunger, J. Phys. Chem.,
96 (1992) 8480.
[14] H. Stach, J. J~inchen, H. G. Jarschkewitz, U. Lohse, B. Parlitz, B. Zibrovius and B. Hunger, J.
Phys. Chem., 96 (1992) 8473.
[15] A. Zecchina, L. Marchese, S. Bordiga, C. Paze and E. Gianotti, J. Phys. Chem. B, 101 (1997)
10128.
[16] S. Savitz, A. L. Myers and R. J. Gorte, J. Phys. Chem. B, 103 (1999) 3687.
[ 17] M. Maache, A. Janin, J. C. Lavalley and E. Benazzi, Zeolites, 15 (1995) 507.
[18] B. Fubini, V. Bolis, A. Cavenago, E. Garrone and P. Ugliengo, Langmuir, 9 (1993) 2712.
[19] A.A. Tsyganenko and M. A. Babaeva, Opt. Spectrosc., 54 (1983) 665.
[20] T.P. Beebe, P. Gelin, J.T. Yates, Surf. Sci., 148 (1984) 526.
[21 ] T. Bucko, J. Hafner and L. Benco, J. Phys. Chem A, 108 (2004) 11388.
[22] S. Bordiga, P. Ugliengo, A. Damin, C. Lamberti, G. Spoto, A. Zecchina, G. Spano, R. Buzzoni,
L. Dalloro and F. Rivetti, Top. Catal., 15 (2001) 43.
[23] A.M. Ferrari, E. Garrone, G. Spoto, P. Ugliengo and A. Zecchina, Surf. Sci., 323 (1995) 151.
[24] B.A. Morrow and A. Devi, J. Chem. Soc., Faraday Trans., 68 (1972) 403.
[25] B.A. Morrow and I. A. Cody, J. Phys. Chem., 80 (1976) 1995.
[26] S. Bordiga, I. Roggero, P. Ugliengo, A. Zecchina, V. Bolis, G. Artioli, R. Buzzoni, G. Marra, F.
Rivetti, G. Spano and C. Lamberti, J. Chem. Soc., Dalton Trans., 21 (2000) 3921.
[27] J.A. van Bokhoven, A. M. J. van der Erden and D. C. Koningsber, J. Am. Chem. Soc., 125
(2003) 7435.
[28] J.A. Lercher, Ch. Grfindling and G. Eder-Mirth, Catal. Today, 27 (1996) 353.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 609
In the present study, we have investigated the formation of Cu(I)-NO complexes formed over
a variety of zeolite structures such as Cu-ZSM-5, Cu-MCM-22, Cu-MCM-58, Cu-ZSM-12,
and Cu-L by ESR spectroscopy at various microwave frequencies. The influence of the
zeolite framework and the NO loading on the formation and magnetic properties of these
Cu(I)-NO complexes as well as their thermal stability were studied to explore the major
parameters determining the geometrical and electronic structure of the adsorption complexes.
It is found that Cu(I)-NO complexes could only be observed for relatively low NO loadings
because of dimer formation at Pxo > 15 Pa and Pxo > 50 Pa for Cu-MCM-22 and Cu-ZSM-5,
respectively.
1. INTRODUCTION
The nitrogen oxides (NOx) are hazardous pollutants formed as by-products during the
combustion processes in industrial boilers and vehicle engines and are responsible for smog
formation, acid rain and global warming. Many metal ion exchanged zeolites have been
reported to be active for catalytic decomposition and reduction of nitrogen monoxide,
indicating the great affinity of the NO molecule to the zeolites. Among many metal ion
exchanged zeolites, copper catalysts have received considerable attention due to their activity
and selectivity towards the decomposition of nitrogen oxides (NOx). The catalytic decom-
position of NO is the simplest and most desirable method for the removal of NO from exhaust
streams and Cu-zeolites are the most suitable catalysts for this reaction among the many
catalysts reported [1]. Especially the high-silica zeolites such as ZSM-5 and MCM-22 have
been reported to be more promising catalysts for NO decomposition [2,3] than low-silica
zeolites, the reason being the stabilization of the reaction intermediates which may be
required for the further transformations. Though the fundamental reason why the low-silica
zeolites are less active is not yet understood completely, it has been reported that low-silica
zeolites such as zeolite Y release their excess oxygen, following an oxidation treatment, more
easily than the more active high-silica Cu-ZSM-5 zeolite [4]. Hence, clarification of the state
of the NO adsorbed, the interaction of NO with Cu(I) and the reaction mechanism are
particularly important for progress in the chemistry of NO decomposition and the design of
610
active decomposition catalysts. The Cu(I)-NO complexes are one of the most important
intermediates in the decomposition processes. The interaction of NO with Cu(I) is reported to
be a complex redox process, where the oxidation of the site is proposed to occur via the
elimination of N20 from dinitrosyl through the formation of Cu(I)-(NO)2 from monomeric
Cu(I)-NO complexes [5]. Both mononitrosylic Cu(I)-(NO) and dinitrosylic Cu(1)-(NO)2
complexes have been identified by IR spectroscopy [3]. However, the detailed structure of
these complexes is largely unknown. ESR spectroscopy is especially useful, because it
provides direct access to the determination of the electronic and geometrical structure of the
adsorption complexes, as well as the sites where the Cu(I)-NO species are formed in the
zeolites, which are essential for the proper understanding of the catalytic activity of the
zeolitic materials, as recently shown for Cu(I)-NO in Cu-ZSM-5 zeolites [6]. Most catalytic,
spectroscopic and theoretical studies of nitric oxide in copper exchanged zeolites have been
focused on ZSM-5 [7-10] because of its great potential for the NO decomposition process
even at higher temperatures such as 773 K.
2. EXPERIMENTAL SECTION
Zeolite L (ns/nAl- 3) was kindly supplied by Zeolyst International. ZSM-12 (nsi/nAl- 9),
ZSM-5 (nsi/nAl = 17), MCM-22 (nsi/nAl----22) and MCM-58 (nsi/nAl = 22) were synthesized as
described previously [11], calcined and ion-exchanged with Cu(CH3COO)2. Prior to the
adsorption of NO, the samples were activated at 400 ~ under vacuum (p < 3.10 - 7 Pa) in a
stainless steel vacuum line. This prevents NO from reacting with small amounts of residual
oxygen, which was, from our point of view, the most critical part in the sample preparation.
Subsequently, the samples were cooled to 25 ~ and exposed to a defined quantity of NO
(defined gas volume of ca. 4 ml and pressures OfPuo = 0.5 mbar for W-, Q- and X-band
samples, respectively). Thereafter, the samples were cooled to liquid nitrogen temperature and
sealed off. It is worth mentioning that the Cu(I)-NO adduct is unstable at room temperature
and fades away with time.
The ESR spectra were recorded at X-band ESP 380 (Vmw- 9.4 GHz), Q-band EMX 10-
40 (Vmw = 33.9GHz), and W-band E680 (Vmw = 94.2 GHz) Bruker spectrometers at
temperatures of 77 K (X-band) and 10 K (Q- and W-band). The magnetic flux density Bo of
the magnet coils used for the X- and Q- band ESR was calibrated by lH NMR magnetometer
MJll0R. A LiF standard with gLiF 2.002288(4) was used for the Bo calibration of the
=
cryomagnet (W-band). The Bo modulation amplitudes used were 0.7 mT (X-band) and 0.2
mT (Q-band) and the modulation frequencies were adjusted to Vmoa= 100 kHz at X- and W-
band, and 70 kHz at Q-band. The microwave power used was low enough to prevent
saturation of the spin systems. For X-, Q- and W- band measurements, ESR quartz tubes with
an outer diameter of about 3.8 mm, 1.8 mm and 0.8 mm, respectively, were used. The NO
pressure dependence of the Cu(I)-NO (formed over Cu-MCM-22 and Cu-ZSM-5) ESR signal
intensity was studied at X-band at 77 K at various NO pressures ranging from 1 Pa to 220 Pa.
Every time the same amount (0.01 g) of fresh copper-exchanged zeolite sample was subjected
to the sample treatment described above and then the desired amount of NO was introduced.
A DPPH sample was used as intensity standard in a TEl04 two-sample cavity for the ESR
measurements.
611
Fig. l a shows the low temperature (77 K) X-band ESR spectra of Cu-ZSM-12, Cu-L, Cu-
ZSM-5, Cu-MCM-22 and Cu-MCM-58 zeolites after activation at 673 K in vacuum and
subsequent adsorption of 0.5 mbar of NO at 298 K. Although autoreduction of Cu(II) to Cu(I)
takes place to a large extent, the signal of Cu(II) is still clearly visible. This signal
superimposes the spectra of the Cu(I)-NO complexes and, therefore, determination of the g
values and the Cu hf splitting of these adsorption complexes is hardly possible. In the Q-band
spectra (Fig. 3B), the Cu(II) signal is almost separated from the signals of two different Cu(I)-
NO complexes A and B. The g values and Cu hf splitting parameters of the two Cu(I)-NO
species A and B were obtained by simultaneous simulation of the X- Q- and W-band spectra
and are comparable to the recently published data on Cu-ZSM-5 [7-9] indicating the
formation of mononitrosylic complexes with Cu(I) cations in all zeolites under study. The
obtained Cu hf coupling parameters are in the order of 160.10 -4 cm -~ and 200.10 -4 c m l for
C"A_L and C"A//, respectively, while typical g-values for e.g. Cu-MCM-22 are 2.001 (g_L) and
1.861 (g/l). The principal values of the Cu hf coupling tensors as well as its orientations with
respect to the g tensor frame will be analyzed and discussed in a subsequent paper [12].
Furthermore, the typical nitrogen hf coupling constant of 29"10-4cm~ of NO adsorption
complexes was resolved in the g_z spectral region.
~ ~ Cu-ZSM-12
Cu-L
Cu-ZSM-5
Cu-MCM-58
~ Cu-MCM-22 Cu-MCM-22
Fig. 1. X-band (a) and Q-band (b) ESR spectra of Cu(I)-NO complexes in various zeolites
A close inspection of the gzz spectral region in the Q-band ESR spectra of Cu(I)-NO
complexes formed over Cu-ZSM-12, Cu-ZSM-5 and Cu-MCM-22 zeolites reveals the
formation of a minor species B in addition to the presence of the major species A for both
materials as illustrated in Fig. 2, while only one species is detected for zeolite Cu-L.
Obviously species A and B differ in their gzz and Azzc" parameters. However, no additional
spectral features due to the minor species B have been observed at the gxx/yy region and, hence,
the other spin Hamiltonian parameters must be assumed to be identical to those of species A
within the given experimental errors. In previous studies [6,10], from X-band spectra alone
the existence of only one species was proposed. However, the accuracy of the reported spin
612
Hamiltonian parameters is low as altogether at least ten parameters (three principal values of
g tensor, three principal values of A c" tensor together with three Euler angles describing the
orientation of the A c" principal axes frame with respect to g tensor frame and one nitrogen hf
coupling) were deduced from a single X-band powder pattern. While satisfactory simulation
of our X-band ESR spectra can also be achieved by using only a single species, the simulation
of the Q-band spectra using the same parameters failed terribly. Thus, we have to state
clearly, that two different Cu(I)-NO species are observed for the high-silica zeolites CuZSM-
5, CuMCM-22, CuMCM-58, CuZSM-12 whereas just a single species is formed over CuL
(Fig. 2).
/ '
A
I I I Cu-L
g I/ , g~ , A_ ~'<"' ,
] __ AC.,~,
.~= Cu-ZSM-12 I ' ~ A ,4Cu(A)
~ .7. - -
i I I -+ I
gt~io I
g.~.-
8
!
Fig. 2. Q-band spectra of Cu(I)-NO complexes in Cu-L and Cu-ZSM-12 (a) as well as Cu-ZSM-5 and
Cu-MCM-22 (b) showing two species A and B in the g= spectral region. The stick diagrams indicate
the Cu hf splitting of the Cu(I)-NO ESR signals of species A and B at the g= spectral region.
However, the reason for the observation of two different species at low temperature is
still under discussion [9,13]. The presence of two different species A and B of Cu(I)-NO
complexes is tentatively assigned to the formation of these complexes with Cu(I) ions located
at different cation sites of the zeolite frameworks. Nachtigall et al. [8] as well as Pietrzyk et
al. [ 14] have recently identified two different Cu(I) cation sites in Cu-ZSM-5, M5(7) and I2,
on the basis of quantum mechanical calculations. Likewise, two different Cu(I)-NO species
were found by IR spectroscopy in Cu-MCM-22 after NO exposure [3] in some agreement
with the existence of two Cu(II) cation sites in this zeolite [15]. An alternative explanation for
the existence of two different types of Cu(I)-NO complexes in these high-silica zeolites might
be the coordination to either two or three framework oxygen atoms [13]. The Euler angle ~,
which is the angle between the z-axes of the Cu hyperfine tensor and the g-tensor, was
estimated to 35 o. It was not possible to calculate the Euler angles a and y from the powder
ESR spectra. Surprisingly, the 14N (I = 1) hyperfine splitting was not observed for a loading of
1 NO molecule per 40 unit cells. A decrease of the NO loading to 1 molecule per 400 unit
cells revealed a nitrogen hyperfine splitting of ANy = 29 9 10-4 cm -1 in the respective Q-band
spectrum (not shown). This confirms clearly that the unpaired electron is localized in the ny*
NO molecular orbital with a probability of 50 %.
613
In addition to the multifrequency ESR studies of the Cu(I)-NO complex, the influence
of the NO loading pressure (viz. the amount of NO adsorbed) in the sample preparation on the
formation of Cu(I)-NO and possible Cu(II)-NO complexes was studied at X-band frequency.
We have to note that the situation for Cu(I)-NO and Cu(II)-NO is somewhat complementary.
The Cu(I)-NO adsorption complex is accessible by ESR and the Cu(I) adsorption site is ESR
silent, whereas the reverse situation holds for the adsorption of nitric oxide at Cu(II) cations.
The relative intensities of the Cu(I)-NO ESR signals as a measure of the concentration of the
adsorption complex in dependence on the NO pressure were determined at 77 K and are
plotted in Fig. 3 (a and b) for Cu-ZSM-5 and Cu-MCM-22, respectively. Since the ESR
spectra of the Cu(I)-NO species heavily overlap with those of the Cu(II) cations a double
integration was not feasible and the Cu(I)-NO signal intensity was determined by taking the
amplitude of the undisturbed Cu hf line at 360 mT with respect to the standard DPPH sample.
This approach is justified since no (Cu-MCM-22) or only moderate (Cu-ZSM-5) line
broadening of the Cu(I)-NO ESR signal was observed due to NO adsorption. In the case of
Cu-ZSM-5 the Cu(I)-NO signal intensity increases with increasing NO adsorption pressure
and attains a maximum at about 50 Pa and then it decreases rapidly and reaches zero at 220
Pa. In the case of Cu-MCM-22, the maximum intensity could was at a lower NO pressure of
approximately 15 Pa and then the intensity decreases rapidly and attains already zero at a
comparable lower pressure of 150 Pa.
1.4,. a 1.0
T
b T
S" 1.2 /~-----
S" 0.8 ~
1.o
o~ 0.8 0.6
E 0.6 E 0.4
>
~ 0.2 ~ 0.2
cr
o.o , , , . , .
! 0.0
0 50 100 150 2~0 2!~0 0 20 40 60 80 100 120 140
NO pressure (Pa) NO pressure (Pa)
Fig. 3. NO pressure dependence of the relative intensity at 77 K of the Cu(I)-NO ESR signal in (a) Cu-
ZSM-5 and (b) Cu-MCM-22 zeolites measured at X-band (77 K).
Cu(I) sites could be observed as seen from the gzz spectral region of the Q-band ESR spectra
of species A and B taken at different NO loadings (Fig. 4). When the NO pressure is increased
further, once all the Cu(I) sites are filled with monomers, ESR silent dimers, Cu(I)-(NO)2,
having a singlet (S = 0) ground state are formed. It is worth to note that though a triplet (S = 1)
ground state has been experimentally confirmed in the case of Na(I)-(NO)2 adsorption
complexes in Na-A zeolites [16] a singlet ground state has been predicted by quantum
chemical calculations [17] for NO dimer formation at Cu(I) cations. The dimers are formed
because of the adsorption of a second molecule of NO on the ESR active monomeric Cu(I)-
NO, which results in the decrease of the Cu(I)-NO signal intensity. After dosing 220 Pa of
NO, formation of the Cu(I)-NO signal is completely lost (Fig. 3a) and dimer formation is
completed. The formation of monomers and dimers has been likewise observed by IR [5] at
low and room temperature. Moreover it is also confirmed by IR studies, that the dimer is
formed at the expense of Cu(I)-NO monomers which are only formed at low NO adsorption
pressures. IR studies [5] have further proven that at higher NO pressure, Cu(II) and N20 are
formed because of the decomposition of Cu(I)-(NO)2 at room temperature. Hence Cu(I)-(NOh
has been confirmed as one of the essential intermediates during NO decomposition which
occurs over dehydrated copper exchanged zeolites. Because the oxidation of Cu(I) to Cu(II)
by NO occurs instantaneously only at room temperature or higher temperatures, the formation
of, for instance Cu(II)-NO species is not observed at low temperatures over the dehydrated
Cu-ZSM-5 zeolite by IR [5]. Since our studies have been carried out at low temperature (77
K), no such oxidation is expected in the present case. Indeed, the Cu(II) ESR signal intensity
was observed to be the same for dehydrated samples and samples loaded at higher NO
pressures (150 Pa and 200 Pa) in agreement with previous IR results [5].
0.5 Pa
0.5 Pa
5 Pa
5Pa
I g~ I I
I g~l I I
I
The formation of Cu(II)-NO, is reported to involve the transfer of the unpaired electron
from the 1-Iy MO of the NO to the partially filled 3d AOs of the Cu(II) ion, resulting in the
partial oxidation of NO and hence partial reduction of Cu(II) [5]. Therefore, the Cu(II)-NO
complex is ESR silent (S = 0) and its formation could not be checked through ESR directly,
but of course is expected to result in a decrease of the Cu(II) ESR signal intensity. But since
615
there is no significant decrease in the Cu(II) ESR signal intensity when compared to the
original dehydrated Cu(II) signal, the formation of the Cu(II)-NO complexes in our system at
low temperatures can be ruled out likewise. The only noticeable effect of NO adsorption on
the Cu(II) ESR spectrum is an obvious increase in its line width due to g and A strain at
higher NO loading [12]. We assume that NO molecules, which are adsorbed in close
proximity to the Cu(II) ions, slightly modify the Cu(II) spin Hamiltonian parameters and in
that way cause a distribution in the principal values of the g al : copper of coupling tensors.
The overall characteristic of NO adsorption over Cu-~ : ;M-22 zeolites is found to be
comparable to Cu-ZSM-5 but displays also some distinct characteristic features. The Cu(I)-
NO complexes formed in Cu-MCM-22 shows similar NO pressure dependence (Fig. 3b) as
that of Cu-ZSM-5, but the Cu(I)-NO signal intensity maximum is already obtained at NO
pressures of about 15 Pa and the Cu(I)-NO signal is completely lost at 150 Pa versus 220 Pa
observed for the Cu-ZSM-5 system. This indicates that complete formation of ESR silent
Cu(I)-(NO)2 dimers occurs at lower NO loadings in Cu-MCM-22 in comparison with Cu-
ZSM-5. The two adsorption steps, the filling of the Cu(I) adsorption sites initially by one NO
molecule and subsequently by two NO molecules, with increasing NO loadings seem to be
better separated from each other for the zeolite Cu-MCM-22. The difference in behavior of
Cu-MCM-22 compared to Cu-ZSM-5 may be attributed to the difference in the framework
structure of the zeolite. The formation of Cu(I)-NO monomers at low NO adsorption pressure
and Cu(I)-(NO)2 dimers at high NO pressures has also been observed by IR spectroscopy at
low temperature in Cu-MCM-22 [3]. However, a direct comparison of the NO doses, where
the interconversion of the monomer into the dimmer species occurs, between the IR
experiments reported by Frache at al. [3] and our work does not seem to be justified as the
experiments were conducted in different ways. A further distinct difference to the Cu-ZSM-5
system is the preference of one Cu(I) adsorption site at low NO loadings. It is obvious from
the Q-band spectra (Fig. 4) that there is a significant change in the relative population of two
Cu(I)-NO species A and B with increasing NO pressure. At 0.5 Pa, species A with the smaller
Aiso cu value is formed predominantly, whereas at higher NO pressure of about 5 Pa, species B
is also getting formed in higher amounts. Hence it can be inferred that at low NO pressures
there is a preference for NO to adsorb at Cu(I) cations, which are coordinated to three oxygen
atoms from the framework (species A), whereas at higher NO pressures complex formation at
Cu(l) cations coordinated to two framework oxygen atoms also occurs. The presence of two
distinct Cu(I)-NO species with different preference towards NO adsorption was also observed
by Frache et al. [3] in over-exchanged Cu-MCM-22 by infrared spectroscopy and is in line
with our ESR results.
4. CONCLUSIONS
The ESR spectra of the Cu(I)-NO complexes are subjected to temperature dependent changes
of their conformation or electronic structure which has to be clarified in future studies.
ACKNOWLEDGEMENTS
REFERENCES
[ll M. Iwamoto, S. Yokoo, K. Sakai, S. Kagawa, J. Chem. Soc. Faraday Trans. 1 77 (1981) 1629.
[2] S.K. Park, V. Kurshev, Z. Luan, C.W. Lee, L. Kevan, L. Microporous Mesoporous Mater. 38
(2000) 255.
[3] A. Frache, M. Cadoni, C. Bisio, L. Marchese, Langmuir 18 (2002) 6875.
[4] J. Valyon, K. Hall, J. Phys. Chem. 97 (1993) 1204.
[5] E. Giamello, D. Murphy, G. Magnacca, C. Morterra, Y. Shioya, T. Nomura, M. Anpo, J. Catal.
136 (1992) 510.
[6] Z. Sojka, M. Che, E. Giamello, J. Phys. Chem. 101 (1997) 4831.
[7] A. P6ppl, M. Hartmann, Stud. Surf. Sci.Catal. 142 (2002) 375.
[8] D. Nachtigallova, P. Nachtigall, J. Sauer, Phys. Chem. Chem. Phys. 3 (2001) 1552.
[9] M. Davidova, D. Nachtigallova, P. Nachtigall and J. Sauer, J. Phys. Chem. B 108 (2004) 13674.
[ 10] C. Prestipino, G. Berlier, F.X. Llabres i Xamena, G. Spoto, S. Bordiga, A. Zecchina, G. Turnes
Palomino, T. Yamamoto, C. Lamberti, Chem. Phys. Lett. 363 (2002) 389.
[11] V. Umamaheswari, M. Hartmann, A. P/~ppl, J. Mol. Catal. A. 223 (2004) 123.
[12] V. Umamaheswari, M. Hartmann, A. POppl, J. Phys. Chem. B 109 (2005) 1537.
[13] C. Freysoldt, A. P6ppl and J. Reinhold, J. Phys. Chem. A. 108 (2004) 1582.
[14] P. Pietrzyk, W. Piskorz, Z. Sojka, E. Broclawik, J. Phys. Chem. B 107 (2003) 6105.
[15] T. Wasowicz, A.M. Prakash, L. Kevan, Microporous Mesoporous Mater. 12 (1997) 107.
[16] H. Yahiro, A. Lund, R. Asa, N.P. Benetis, M. Shiotani, J. Phys. Chem. A 104 (2000) 7950.
[17] R. Ramprasad, K.C. Haas, W.F. Scheinder, J.B. Adams, J. Phys. Chem. B 101 (1997) 6903.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 617
The relation between the structure and magnetic properties of VO 2+ (d ~) species and nitrosyl
adducts of Cu § (d 1~ and Na + (d ~ ions exchanged in the ZSM-5 zeolite was studied by DFT
and EPR methods. Electronic g tensors were predicted using the spin-unrestricted zero order
relativistic approximation (ZORA) and the scalar relativistic Pauli Hamiltonian calculations.
The main purpose was to reproduce theoretically the g tensor components of the investigated
intrazeolite paramagnets and to interpret the observed Agii shifts in terms of the associated
molecular orbital contributions.
1. I N T R O D U C T I O N
Transition metals introduced to ZSM-5 zeolite as guest ions exhibit a number of geometrical
structures (speciation), depending on the nature of the hosting sites. In the case of
incompletely filled valence shell, their local environment can be unraveled from detailed
analysis of the corresponding EPR spectra. However, the connection between experimental
spectrum and structure is often unclear. As a result, a great deal of important molecular
information about the investigated system is hardly available without the help of the
supporting quantum chemical calculations. With the development of the density functional
theory and the methods for calculation of g and A tensors, an in-depth interpretation of the
magnetic parameters in terms of geometry and electronic structure has been advanced
substantially. Nowadays, owing to recent progress in the relativistic DFT methods, the
calculation of g values allows for nearly quantitative reproduction of the Ag;; shifts, even for
large transition-metal-containing systems, within reasonable computing costs.
The {VO}I/ZSM-5, {NaNO}I/ZSM-5 and {CuNO}II/ZSM-5 zeolites, although
spectroscopically well characterized [ 1 , 2 , 3 ], have attracted rather limited interest of
theoretical EPR studies as yet. This is particularly apparent in the case of the g tensor
calculations and its advanced molecular interpretation, since such data are practically missing.
Previous g tensor calculations of both nitrosyl adducts have been confined to minimal
{NaNO}! and {CuNO}ll models [4], whereas the calculations intentionally devoted to VO 2+-
zeolite system, have been restricted to aqua complexes only [5].
618
2. METHODS
The Kohn-Sham calculations were performed within the spin unrestricted scheme with DMol
[6] code using the local spin density VWN and the gradient corrected BPW91 functionals.
Double numerical basis set (DNP) was used, with the inner electron core frozen. The cage
complexes were optimized using the analytical gradients with constrains imposed by an
embedding scheme reported previously [7]. Following a previous paper [8] the two types of
clusters, [Si2A1Oz(OH)8]- and [SisA12Os(OH)I2]2-, applied to model the sites hosting the
investigated paramagnetic species, were labeled hereafter as 3-T and 7-T, respectively.
Single point calculations of g tensors were performed with the ADF package (2004.01)
[9] using either the local density (VWN) or the gradient-corrected (GGA) functionals and the
gauge-including atomic orbitals (GIAO). A standard uncontracted triple-~ Slater-type orbital
basis set was supplemented by polarization functions. Core electrons were explicitly included
in the ADF computations. Relativistic calculations of the g tensors were performed within the
spin unrestricted ZORA collinear approximation [10,11], and the scalar relativistic approach
(treated in the quasi-relativistic framework) in conjunction with the first order Pauli
Hamiltonian, developed by Schreckenbach and Ziegler (SZ) [12].
a b c
Fig. l. Optimized structures along with spin density contours of the (a) {VO}~/7-T, (b) {NaNO} ~/7-T
and (c) {CuNO} ll/3-T clusters. Bond lengths are given in angstroms, angles in degrees.
The magnetic structure of the most complicated {CuNO} ~ system results from the
interaction of the closed 3d l~ shell of copper with the open shell of NO. The bonding
mechanism consist in partial hybridization of the occupied 3dz2 and empty 4s orbitals of Cu
and tilting of the NO ligand to provide better overlap with its ~* and cy" in-plane orbitals [ 13].
619
The resultant opening of the copper 3d shell allows for direct ligand to metal spin transfer
(LMST), which in conjunction with the spin polarization of the out of plane 3d orbitals by the
paramagnetic ligand, produce a considerable spin density on the metal (Fig. l c). Such
complicated mechanism makes the {CuNO} ~1 unit very sensitive to the variation of the Cu-
N-O angle and the copper-NO distance, as well.
The common feature of the contours shown in Fig. 1 is large confinement of the spin
density to the distinct {VO} i, {NaNO} 1 and {CuNO} 11 units. Therefore hereafter, such
moiety bearing a dominant part of the magnetic properties of the whole system we name a
magnetophore [ 15 ].
The results of the relativistic DFT g tensor calculations of the vanadyl, sodium nitrosyl and
copper nitrosyl intrazeolite complexes are collected in Table 1. For further discussion of the
results, the g tensor can be factorized into two components, g = gel + Ag. Such a form reflects
better the anisotropy of the Ag shifts induced by various perturbations due to the molecular
structure, relative to the free electron ge =2.0023 value.
Table 1
Calculated g tensors for various {VO} i, {NaNO} 1 and {CuNO} 11 complexes
System symmetry method gxx gry g,_z gxxeXp gyyexp gzzeXp
{VO} 1(H20)5 a) C2 ZORA (U) 1.9828 1.9810 1.9502 1.982 1.982 1.943
Pauli (U) 1.9831 1.9813 1.9506
C2v Pauli (U) 1.9862 1.9863 1.9524
Pauli (R) 1.9857 1.9860 1.9429
{VO}I/7-T b) CI ZORA (U) 1.9839 1.9823 1.9474 1.987 1.987 1.925
Pauli (U) 1.9843 1.9827 1.9479
{NaNO} ic) C~ ZORA(U) 2.0058 1.9990 1.7331 1.996 1.995 1.862
Pauli (U) 2.0066 2.0019 1.7177
Pauli (R) 2.0072 2.0020 1.9485
{CuNO} ~ d) Cs ZORA (U) 2.0071 2.0109 1.9580 1.999 2.003 1.889
Pauli(U) 2.0075 2.0113 1.9585 2.005 2.005 1.917
Pauli (R) 2.0196 2.0213 1.9384
{CuNO}II/3-T C1 ZORA (U) 2.0110 2.0158 1.9156
Pauli(U) 2.0117 2.0171 1.9136
Experimental values: a) and b) ref. [5], c) ref. [1], d) ref. [2,31, U-unrestricted, R-restricted
The results obtained for the {VO} 1(H20)5 reference complex depend on the symmetry of
the molecule in a moderate way only. The unrestricted ZORA and Pauli g tensors obtained for
two conformations with the (72 and C2v symmetry are essentially the same, and remain in an
excellent agreement with experiment. For the {NaNO} ~ magnetophore the discrepancy
between the spin unrestricted and restricted scheme appears to be more sound, however, in
both cases the calculated values follow an experimental pattern gzz ( ge ~ gxx ~ gyy. The
{CuNO} I I species, because of the significantly different electronic structure, give rise to
distinctly different g tensor with gzz ( gxx .~ g~, ( ge, which is associated with the involvement
of the 3d orbitals. For the zeolitic model {CuNO} zl/3-T the calculated g tensor agrees very
well with the experimental one, and both calculation methods give very similar results.
However, some of the calculated components, e.g., gzz for the vanadyl, and gxx, gyy for the
nitrosyl complexes are systematically slightly overestimated. This can be traced back to the
overestimation of the covalent bonding by LDA and GGA exchange-correlation functionals
[16].
620
where Agrel 0 -- mgijKE + Ago.MV + Ag0"Darwincombines scalar relativistic corrections due to the
kinetic energy, mass-velocity and Darwin terms in the quasi-reletivistic Hamiltonian, whereas
the terms Agao9 and AgPo9 are dia- and paramagnetic contributions to Ag, respectively. The
paramagnetic term AgPo. dominates deviation of the g tensor from the free electron value and
contains contributions due to the frozen core (Ag0"p'c~ as well as from the coupling induced
by the magnetic field between the occupied orbitals (Ag0.p'~176 and the coupling between the
occupied and virtual orbitals (Ag0"p'~ the latter being usually the most important [12,14]
Contributions arising due to mixing of the occupied and virtual MOs account for about
90% of the total Ag shifts in the case of the {VO} i, {NaNO} l and {CuNO} II cage complexes,
and therefore they essentially determine all the observed qualitative trends. To facilitate the
discussion we focus our attention on the magneto~hore part of those complexes, namely
{NaNO} 1 and {CuNO} of C~. symmetry and {VO} (H20)5 of Cgv symmetry. Such approach
seems to be justified in the first approximation since the embedding in ZSM-5 does not lead
to any appreciable delocalization of the spin density onto the framework of the zeolite (see
Fig. 1). Indeed, the dominant couplings occur between those MOs that exhibit very small
contributions due to H20 (vide supra) or, in the case of {NaNO} 1 and {CuNO} 11, zeolite
lattice ligands [15]. Therefore, they preserve large similarity with the parent metal or NO
orbitals. A more detail discussion of influence of the zeolite framework on the g tensor
components is the subject of another paper [ 15].
Due to the C2v symmetry of the vanadyl complex, in the spin-restricted formulation only
couplings between the SOMO (a2) and the orbitals belonging to bl irreducible representation
may result in nonzero contributions to the gxx. Similarly only the mixing of the a2 SOMO with
bz MOs can contribute to the gyy, and with al MOs to the gzz. Additional terms are symmetry
allowed in the spin-unrestricted approach. But as long as the shape of the a and fl MOs along
with the corresponding orbital energy differences remain similar, the spin-unrestricted terms
remain small, and can be neglected for a qualitative discussion [16]. For the {VO} l species
the most important contributions to the g tensor are presented in Fig. 2a, along with the shapes
and energies of the relevant MOs. The gzz component is primarily determined by coupling
between the 40a2 ez-SOMO (dxy) and the virtual 43al antibonding MO derived from the dx2_y2
orbital. This term gives dominant negative contribution to the Ag value, which is opposite to
the smaller positive contributions due to the couplings of fl-SOMO with the cy-bonding fl-28a~
and the nonbonding fl-35al occupied MOs, both localized on the ligands. The overall shift is
negative giving Ag=z = - 0.0594. For the gxx and gyy two terms enter Ag;i to a nonequal degree.
Couplings of the a-40a2 SOMO with the vacant a-41 bl (dxz) and a-42b2 (dyz) MOs give rise
to the dominant negative contributions to Agxx and Agyy, respectively. The second and much
smaller term arises due to the couplings of the SOMO with the orthogonal pair of the V=O
bonding MOs. The resultant/3-37b2 +-~ fl-40a2 and fl-38bt 6-~ fl-40a2 orbital mixing contribute
in a positive way to Agxx and Agyy, respectively.
621
E/eV
- 10.0-
F_JeV
~ 3 (al) E/eV (a'
- 12.0-
43
41
A..
"o I 23 -.-.-~.
~.LT_ 'e"
(a'~" 14 ,, ~ ~ a
40 ~l_ 41( b 1 7 ~ ' ( a 2 )
- 14.0
iij- a,)~13 i - ~, (a'~
i i,--i
=: : O~4 (a')
( 12
I
- 16.0- I
!~ I
it--, t
37 _ - 16.0 I[
- 18.0- 35
~~(al) -19.0~ (a~ 10 ~+~ 19"0i ~ 16 IL
(a' [
- 20.0-
21.0 ( a+~
28 • ,. ~ ~ gyy
...................... 9 gzz
In agreement with the dominant nature of the a-40a2 ~ (41b~, 42b2 and 43al)
contributions, all g tensor components are below the free-electron value, as actually observed
in experiment (Table 1). The gz.- may be identified with the gll of vanadyl complexes for which
the g tensor is usually interpreted in terms of an axial symmetry, tantamount with the C4v
point group. Distortions caused by the aqua ligands are small. This can be inferred form slight
splitting of the 37b2 and 38bl as well as the 41bl and 42b2 MOs, which are responsible for the
g tensor anisotropy in the equatorial plane. Such small values are hardly resolvable
experimentally at conventional frequencies. As the result gxx and gyy are nearly equal and may
be identified with the g• A very similar picture was obtained for the {VO} I/7-T cluster.
Construction of an accurate model for calculation of the g tensor of the {CuNO} ~/ZSM-
5 system is not trivial. The low C~. symmetry of the adduct, inferred from EPR spectra and
DFT investigations [2,7], gives rise to a large number of magnetic field-induced couplings
with the nonzero contributions to the Ag;i shifts. Additionally, the orientation of the principal
axes of the g tensor does not coincide with the molecular frame. The principal g~ and gzz axes
are located in the Cu-N-O plane, whereas the gxx axis is perpendicular to the plane. Deviation
of the gzz axis from the direction of the N - O bond, equal to 26.7 ~ is caused by the
involvement of the 3d orbitals of copper, as it can be inferred from the comparison with its
electrostatic {NaNO}I analog, where the noncoincidence angle is almost negligable (0 = 1.8~
The main causes responsible for this deviation in {CuNO} 11 are the a-22a' ~ a-23a" and a-
19a ' ~--~ a-22a ' couplings.
622
The principal couplings contributing to the g tensors of the {NaNO} l and {CuNO}il
magnetophores along with the contours of the involved magnetic orbitals are presented in Fig.
2b,c. The diagrams show clearly different electronic nature of the g tensors of the {NaNO}l
and {CuNO} species, despite that the relevant gii values are quite similar (Table 1). The g
tensor of {NaNO} ~ is determined by the mixing within the MOs of the NO ligand, and the
coupling space is defined by the positions of the 10a' and 14a" orbitals. The largest a-13a'
(SOMO) ~ a-14a" (LUMO) term is responsible for the Agzz negative shift, characteristic of
all ligand-centered NO complexes with the spin doublet ground state. The other two in-plane
couplings between the occupied/7-12a' (or) and/3-SOMO together with a minute/3-10a'(rt b)
and fl-SOMO coupling, shift the gx~ component above the free-electron value. In this model
the gyy component remains unperturbed, but despite that its value is slightly altered by -
0.0003 due to the Agrel0 + Agd//contributions (Table 1).
Involvement of the copper d-orbitals in the Cu-NO bond results in much more
complicated picture of the g tensor (Fig 2c). The coupling space spreading from 14a' to 23a"
encompasses new orbitals such as 20a' and 19a', which have no analogs in the magnetic
orbitals of the {NaNO} ! counterpart. However, there is a clear correspondence between the
MOs that couple contributing to the gz= component in the {NaNO} 1 and {CuNO} II
magnetophores. In the case of {CuNO} ii the coupling a-22a' (SOMO) +-> a-23a" (LUMO)
exhibits still a largely intra-ligand character, but the involvement of the copper 3dz2 and 3dy=
orbitals is here quite significant. This effect influences the g tensor mainly by decreasing the
SOMO-LUMO gap (the Ag,,z shift is larger in {NaNO} l than in {CuNO} 11) due to covalent
interactions, and obviously has a tremendous impact on the copper hyperfine tensor [7,17].
The metal-based /3-17a' and fl-20a' orbitals mix with /3-SOMO and determine a slightly
positive shift of the g~x component. These couplings have no analogs in the {NaNO} i species,
indicating that the origin of the gx~ shift is entirely different in both cases.
The value of Agyy is determined by one stronger metal to ligand ,/3-19a' ~ ,B-22a' and
two weaker intra-ligand/5-14a' ~ fl-22a' and ,/3-16a' ~ / 3 - 2 2 a ' couplings. The 14a' and 16a'
orbitals are produced by mixing of the hybridized rt and cr(2p) MOs of the NO ligand with the
in-plane copper 3d orbitals, so they can be treated as rough analogs of the 10a' and 12a'
orbitals. On the contrary, the fl-19a' +-> fl-22a' coupling has no counterpart in {NaNO} ~. As a
result the electronic nature of the gyy values in {CuNO} l~ and {NaNO} ~ is clearly dissimilar,
though in both cases the unpaired electron is primarily localized on the NO moiety.
I I I
I I I
I I I
1,99 _
1,98
1,97
[] g33 ', ' ',
1,96 . . . . . . . . . . . . . 4 . . . . . . . . . . . . . . I. . . . . . . . . . . . . . + . . . . . .
1,95 -s D -~ -~ . . . . . ) i
i l
i i I
1,94
1,4 1,5 1,6 1,7
dv_-o/angstrem
Fig. 3. Dependence of the calculated VWN g-tensor components in {VO}~ aqua complex of C2
symmetry on the V=O bond length.
with dv=o quite substantially. On the other hand, the gzz component, resulting from the
coupling between the SOMO and al MOs, with negligible contributions due to oxygen
vanadyl atom, is inde/pendent on the variation of dv=o. A similar behavior has been observed
previously for other d LnM=O complexes [ 16].
The a l orbitals have large contributions due to the equatorial ligands, therefore the gzz
value is expected to be sensitive to the metal-ligand distance, caused by change in the ligand
nature or symmetry of the complex. The biggest difference in the g tensor for the aqua
complex and {VO} l/7-T is indeed observed in the case of the gzz component (Table 1).
2 _ _ _ ~ _ _ _ ~
1,95
1,5
i t t
1,85
0,5 _r ........ ,___x , ~ _ _ W _ ~ '_ 9- o - g x x \I
1,75 -~g. ! -\ ........
0 -0- g zz ~_~
• gzz (Na-N = 2.1 A) "~ ~
-0,5 I I I ' I
1,65
80 100 120 140 160 180 80 100 120 140 160 180
O~NaNO/deg. (XCuNO/deg.
Fig. 4. Dependence of the g tensor components on the 0MNObending angle for the (a) {NaNO }~and (b)
{CuNO }i i magnetophores.
The sensitivity of the g tensor was also tested against the M - N - O angle 0 for the
{NaNO} 1 and {CuNO} ~l magnetophores. Because the potential for such bending is rather flat
[13] meaningful comparisons with experiment can be made only for the EPR spectra
registered at low temperatures, which do not exhibit any thermal averaging of the g and A
tensors [2,3], otherwise the calculated data should be angularly averaged [18]. The results
presented in Fig. 4a,b show that while the values of gxx and gyy were practically insensitive to
changes in ~ o , for the gzz component its pronounced decrease with the increasing t~NO was
observed. The gzz component is solely determined by the a-SOMO(a') ~ a-LUMO(a")
coupling, the magnitude of which is gauged by the energy gap As, which strongly depends on
the bending angle. In the range of 90 ~ ( 0 ( 120 ~ the g~= changes slowly and declines rapidly
624
when 0 ) 160 ~ Further increase of the 0 value makes the systems almost degenerate, and as
the limit of the validity of both ZORA and Pauli methods is approached, the calculations
become unreliable.
In the case of {NaNO} I because the Na+-NO potential is very shallow, the calculations
of the g tensor dependence on the 0 angle (Fig. 4a) were performed for two distinct bond
length: dNa-N --- 2.1 A (an experimental value) and dNa-N = 2.5 A (a theoretical value). For the
shorter bond length the g,.z value is more sensitive to the angular distortion. Furthermore, from
the inspection of Fig. 4a it is clear that the distance of dNa-N = 2.1 A and the bending angle of
142 ~ estimated from EPR measurements of {NaNO}IA zeolite [19] cannot be reconciled
mutually.
4. C O N C L U S I O N S
The EPR g tensors of paramagnetic metal-centered vanadyl species and ligand centered
nitrosyls of copper(I) and sodium(I) in the ZSM-5 zeolite can be successfully calculated using
the spin unrestricted zero order relativistic approximation (ZORA) and scalar relativistic Pauli
Hamiltonian methods. The electronic nature of the principal contributions to the observed Ag
shifts was identified and discussed in terms of symmetry and energy of the involved magnetic
orbitals.
ACKNOWLEDGEMENTS
This work has been supported by the Committee for Scientific Research of Poland, KBN,
project number 3 T09A 147 26. The calculations have been carried out with the computer
facilities of CYFRONET-AGH under grant number KBN/SGI2800/UJ/018/2002.
REFERENCES
[1] H. Yahiro, A. Lund, M. Shiotani, Spectrochim. Acta A, 60 (2004) 1267, and references therein.
[2] Z. Sojka, M. Che, E. Giamello, J. Phys. Chem., 101 (1997) 4831.
[3] A. Poeppl, M. Hartmann, Stud. Surf. Sci. Catal. 142 (2002) 375.
[4] K.M. Neyman, D. I. Ganyushin, V. A. Nasluzov, N. Rosch, A. Poeppl, M. Hartmann, Phys.
Chem. Chem. Phys., 5 (2003) 2429.
[5] P.J. Carl, S. L. Isley, S. C. Larsen, J. Phys. Chem. A, 105 (2001) 4563.
[6] DMol, InsightII, Release 2000.1, San Diego: Accelrys Inc., 2000.
[7] P. Pietrzyk, W. Piskorz, Z. Sojka, E. Broclawik, J. Phys. Chem. B, 107 (2003) 6105.
[8] D. Nachtigallova, P. Nachtigall, M. Sierka, J. Sauer, Phys. Chem. Chem. Phys., 1 (1999) 2019.
[9] ADF2002.03, SCM, Theoretical Chemistry, Vrije Universiteit, Amsterdam, The Netherlands,
http://www.scm.com.
[10] E. van Lenthe, E. J. Baerends, J. G. Snijders, J. Chem. Phys., 99 (1993) 4597.
[11] H. Eschrig, V.D.P. Servedio, J. Comput. Chem., 20 (1999) 23.
[12] G. Schreckenbach, T. Ziegler, J. Phys. Chem. A, 101 (1997) 3388.
[13 ] M. Davidova, D. Nachtigallova, P. Nachtigall, J. Sauer, J. Phys. Chem. B, 108 (2004) 13674.
[14] G. Schreckenbach, Y. Ziegler, Theor. Chem. Acc., 99 (1998) 71.
[ 15] P. Pietrzyk, Z. Sojka, J. Phys. Chem. B, submitted.
[ 16] S. Patchkovskii, T. Ziegler, J. Chem. Phys., 111 (1999) 5730.
[17] Z. Sojka, P. Pietrzyk, Spectrochim. Acta A, 60 (2004) 1257.
[18] S. Patchkovskii, T. Ziegler, J. Am. Chem. Soc., 122 (2000) 3506
[19] A. Poeppl, T. Rudolf, P. Manikandan, D. Goldfarb, J. Am. Chem. Soc., 122 (2000) 10194.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 625
a Center for Complex Molecular Systems, Institute of Organic Chemistry and Biochemistry,
Academy of Sciences of the Czech Republic, Flemingovo n. 2, 166 10 Prague 6, Czech
Republic
b Department of Physical Chemistry, University ofPardubice, Nfim. Cs. Legii 565, 53010
Pardubice, Czech Republic
1. I N T R O D U C T I O N
theoretical and experimental techniques can bring information about the character of the
cationic sites in zeolites. In this contribution the interaction and vibrational dynamics of CO
probe on the Li +, Na +, and Cu + FER was investigated. Ferrierite was chosen in order to make
calculations with the periodic DFT level feasible (due to a large unit cell of MFI these
calculations are computationally very demanding).
CO/Li+/FER and CO/Na+/FER systems where investigated previously by FTIR [7]. The
band at 2190 cm -~ observed for Li+/FER was assigned to Li + sites at the main channel and the
shoulder observed at lower frequencies was suggested to be due to Li + sites in perpendicular
channel. Two distinct bands observed at 2175 and 2159 cm ~ for Na+/FER were assigned to
sites in the main channels and to sites where CO interacts with more than one cation,
respectively.
2. E X P E R I M E N T
The starting ammonium form of ferrierite zeolite !Si/A1 = 8.6) used in this study was suplied
by Research Institute of Inorganic Chemistry, Usd nad Labem. From parent zeolite, the
sodium- and lithium-exchanged samples (with 100% exchange level) were obtained by three
times repeated ion exchange in 0.5 M aqueous solution of NaC1 and LiC1, respectively. Cu
loaded sample was prepared from Na-FER matrix by the ion exchange in 0.01 M CuCI2
solution. Resulting material contained 3.5 wt. % of copper (Cu/A1 was 0.39). The Cu-FER
sample was firstly reduced under 80 Torr CO at 450~ overnight in order to achieve
quantitative conversion of Cu 2+ ions to Cu + form. Thin self-supported wafers of alkali metal
exchanged ferrierites were degassed under dynamic vacuum at 450 ~ The spectra were
collected with a resolution of 1 cm ~ on a Nicolet FTIR spectrometer Protege 460 equipped
with an MCT cryodetector.
0
t- 0.05 " "
O
t~
.Q
3. CALCULATIONS
The CO stretching frequencies are calculated from the vco[CCSD(T)]/rco[DFT] scaling [8],
using the following equation
vco = a rco[DFT] + b + A~o + Av.
In this paper we use the 1-T zeolite model for the A~o correction, we assume a constant
anharmonic correction Av (-29 cm -l) calculated previously for monocarbonyl species in
various molecular environments [8,9], and a and b scaling parameters are obtained by fitting
the linear dependence of CCSD(T) harmonic frequencies and DFT equilibrium distances for
small model systems [8].
Structures and equilibrium CO distances were obtained with the periodic DFT model,
using the VASP code.[10-13] The Kohn-Sham equations were solved variationally with a
plane-wave basis set using the projector-augmented-wave (PAW) of Bl6chl (Ref. [14]) as
recently adapted by Kresse and Joubert (Ref. [15]). The Perdew-Burke-Erzenhofer (PBE)
[ 16,17] functional was used together with the plane-wave basis set with the energy cutoff of
400 eV. Brillouin-zone sampling was restricted to the gamma-point.
The single unit cell of Ferrierite contains 36 T atoms (Si or A1) and 72 O atoms. The
equilibrium volume of the all-silica FER unit cell (cell parameters 19.1468, 14.3040 and
7.5763 A, volume 2076.70 As) fitted previously was used [18] in all calculations on
CO/M+/FER and also in the calculations on a set of small molecules used for the evaluation of
the o~co-rco correlation. One or two framework Si atoms were replaced by aluminum and the
charge was compensated by Li § or Na + cations. Calculations were carried out for each of four
distinguishable framework T-positions (Imm m space group, the numbering scheme of Ref. [19]
is adopted).[20]
i,,t!)
_4"... "
/l 0!
X' i\x' __~
i i " <~x \\. \ :7)'
.. ,:.., \ ",,'..9 /,
4. RESULTS
The spectra of CO adsorbed on alkali metal exchanged FER (Fig. 1A and 1B for Li + and Na +,
respectively) were taken at RT under 140 Torr CO pressure while spectra for CO/Cu+/FER
system (Fig. 1C) were taken at RT and vacuum conditions (following the CO adsorption at 25
Torr). Spectrum of CO/Li+/FER shows a band maximum and a band shoulder at 2188 and
2176 cm -~, respectively. Two distinctive bands are apparent in CO/Na+/FER spectra at 2173
and 2157 cm-, the latter being broader and less intense. A single band at 2156 cm ~ is
observed for CO/Cu+/FER.
CO stretching frequencies were calculated for CO adsorbed on Na + and Li + cations in
FER by means of vco[CCSD(T)]/rco[DFT] correlation method[8] using periodic DFT model
and PBE functional. Framework A1 atom was placed to various framework T-sites and CO
was adsorbed on various metal sites. Results are summarized in Table 1. The notation
introduced in Ref. [21 ] is adopted, the position of cations are shown in Fig. 2.
Table 1
Adsorption energies and CO stretching frequencies (in kcal/mol and cm -1, respectively)
calculated for CO adsorption complexes on Li+/FER and Na+/FER systems for various
extraframework cation sites in FER.
Metal Framework Metal CN c Eadsd
CO locationb Vco
A1 position sitea
Li + T2 I2 Intersection 2+ 1 -7.2 2183
T1 P6 P channel cavity 2+2 -4.9 2173
T1,T1 P6 P channel cavity 4+0 -4.0 2167
The coordination of alkali metal with zeolite framework is not changed upon the
adsorption of CO. For the Li+/FER system the CO interaction with intersection site (I2/T2) is
clearly stronger than for channel wall site (P6/T1). On the contrary, in Na+/FER there is
almost no difference in CO interaction energies at channel wall sites and intersection sites.
For both alkali metals the CO adsorption on the sites in the vicinity of two framework AIO4
tetrahedra is about 1 kcal/mol weaker than on sites in the vicinity of single framework A104
tetrahedron. Example of the coordination for bare Li + and Li+CO complex is shown in Fig. 3b
for P6 sites (AI in T2 position).
The vco primarily depends on the alkali metal coordination with the zeolite framework.
The intersection sites with lower metal-framework coordination show higher frequencies.
629
5. DISCUSSION
For isolated M+...CO species the CO stretching frequency depends on two factors: CO
polarization due to the electrostatic field of cation and formation of partial covalent bond
between metal ion and CO (G-donation and ~-backdonation) taking place for transition metal
cations. The CO stretching frequency shifts for isolated M+...CO species are relatively well
understood [22,23]. The situation is significantly more complicated when CO is adsorbed on
extraframework cationic sites in zeolites. Additional effects can influence the CO stretching
frequency: (i) the charge on the M + cation is lowered due to the cation interaction with the
zeolite framework oxygen atoms, (ii) in addition to the CO interaction with alkali metal cation
the CO interacts with the zeolite framework atoms on the surface of channel wall (this
interaction drives the orientation of CO molecule in zeolite channel system), and (iii) long-
range electrostatic interaction of CO with zeolite (Madelung potential). Individual effects has
been used for interpretation of experimental data and modeled theoretically. Based on the
electrostatic embedding model it was concluded that Madelung potential has a large effect on
CO vibrational frequency [24,25]. Effect of the electron density on Na + ion induced by the
increasing number of framework A1 in the six-member ring on the CO vibrations was
investigated by Vayssilov et al. [26]. The effects of the ion coordination, framework AI
localization, and topology of the zeolite channel was investigated for several systems using
combined QM-pot model [8,27] and periodic DFT model [18]. Excellent compilation of
experimental data discussing number of possible effects on CO stretching frequency appeared
recently [1 ].
A r'
Fig 3. Structure and coordination of bare metal cation (left part) and CO/M+/FERadsorption
complex (right part) for Cu+and Li+cations (parts A and B, respectively).
630
Despite all the progress, a consistent model allowing to interpret the spectral
characteristics of CO adsorbed on various monovalent cations in various zeolite matrices has
not been presented yet. In particular, it is not clear why CO stretching frequencies are site-
specific in some systems (e. g., CO/Li+/MFI [5,27]) and they are not site-specific in other
systems (e. g., CO/Cu+/MFI [4,8]). Recently, we have proposed a vco[CCSD(T)]/rco[DFT]
scaling method for precise calculations of stretching frequencies of adsorbed CO [8]. A very
good agreement between experimental and calculated frequencies was found for Cu+/MFI
[8], Cu+/FER [18], Li+/MFI [27], and Li+/FER and Na+/FER (this work). Based on a good
agreement between theory and experiment the IR spectra of adsorbed CO can be analyzed at
the atomic scale level.
The calculations on CO/Cu+/FER system reported in Ref. [18] clearly show that vco are
the same for all Cu + sites, in agreement with experimental observation. Upon the interaction
with CO the Cu + ion is coordinated only to two framework oxygen atoms of single AIO4
tetrahedra, regardless the Cu + coordination before CO adsorption complex formation. The
changes in Cu + coordination upon the CO adsorption are depicted in Fig. 3a for Cu + at P6 site
(framework AI at T2). Since the coordination of Cu + with the zeolite framework is the same
for all CO/Cu+/zeolite adsorption complexes the electron density on the Cu + ion is almost the
same in all adsorption complexes. One of the major effects on CO stretching frequency in
CO/Cu + complexes is the extent of charge transfer between Cu + and CO (electron donation
and electron backdonation) that, in turn, is determined by the charge density and orbital
populations on the Cu + ion. It is therefore not surprising that IR spectra of CO adsorbed on
Cu + zeolites show only one band. In other words, the CO stretching vibrations are not site-
specific for CO/Cu+/zeolites because the Cu + coordination is the same in all adsorption
complexes.
Rather different situation was found for CO adsorbed on extra-framework alkali metal
cations. It was found previously for CO/Li+/MFI system [27] and for CO/Li+/FER system
(Table 1) that the stretching frequencies of CO adsorbed on the intersection sites are larger
than CO frequencies on the channel wall sites. Contrary to the Cu + ions the coordination of
the Li + ions with the framework is almost unaffected by the CO adsorption (Fig. 3b).
Therefore, the Li + ions in the channel wall sites are coordinated to more framework oxygen
atoms than Li § in the intersection sites. As a result, the electron density on the Li + cation in
the intersection site is smaller (positive charge is larger) than in the channel wall sites and
electrostatic interaction between CO and cation is larger. It is apparent that the CO interaction
energy with the Li + intersection site is larger and vco is greater than for channel wall site.
Charge distribution analysis supports this interpretation. In addition, the Li + cation located on
top of six-member ring containing two framework A104 tetrahedra (P6 site with AI atoms in
two T1 positions) has even larger electron density, therefore, the CO interaction is weaker and
frequency shift smaller than for other sites. The site specificity of stretching vibrations of CO
adsorbed on Li+/zeolite is due to the different coordination of Li + in intersection and channel
wall sites. The Na + ion coordination with the zeolite framework also remains unchanged upon
the fonaaation of CO adsorption complex. Sodium cation is coordinated to 2-3 framework
oxygen atoms at I2, P2, P7, and M5 sites while it is coordinated to 4 framework oxygen atoms
in P6 sites. However, in case of Na + cations there is no clear distinction between intersection
and channel wall sites.
It can be concluded that distinctive bands in IR spectra of adsorbed CO can be observed
for system where coordination of metal cation is not changed upon the interaction with CO.
Such situation can be expected for systems with weakly bounded probe molecule. On the
contrary, when the interaction of metal with CO is strong the coordination of metal cation
631
changes dramatically upon the CO adsorption and the resulting IR spectra show single band.
Based on this observation the consistent interpretation of IR spectra of CO adsorbed on extra-
framework metal cation can be offered. Experimental and theoretical data for CO/M+/FER
(M + = Li +, Na +, Cu +) are compared in Table 2 where also the interpretation of experimental
data is suggested.
Table 2
Spectral characteristics of IR spectr a 0 f CO adsorbed on Li +, Na +, and Cu + exchanged FER.
System vco a Ref. interpretation b
6. C O N C L U S I O N S
ACKNOWLEDGEMENT
This work was supported by Grants of the Ministry of Education of the Czech Republic No.
LC512 and No. MSM0021627501 (RB) and research project Z4 055 905 (OB, DN, and PN).
632
REFERENCES
[1] K.I. Hadjiivanov; G.N. Vayssilov. Characterization of oxide surfaces and zeolites by carbon
monoxide as an IR probe molecule. In Advances in Catalysis, Vol 47, 2002; Vol. 47; pp 307.
[2] P. Hollins, Surf. Sci. Rep., 16, (1992), 51.
[3] V. Bolis; A. Barbaglia; S. Bordiga; C. Lamberti; A. Zecchina, J. Phys. Chem. B, 108, (2004),
9970.
[4] R. Bulanek, Phys. Chem. Chem. Phys., 6, (2004), 4208.
[5] C.O. Arean; O.V. Manoilova; M.R. Delgado; A.A. Tsyganenko; E. Garrone, Phys. Chem.
Chem. Phys., 3, (2001 ), 4187.
[6] B. Bonelli; E. Garrone; B. Fubini; B. Onida; M.R. Delgado; C.O. Arean, Phys. Chem. Chem.
Phys., 5, (2003), 2900.
[7] S. Bordiga; G.T. Palomino; C. Paze; A. Zecchina, Microporous Mesoporous Mater., 34,
(2000), 67.
[8] O. Bludsky; M. Silhan; D. Nachtigallova; P. Nachtigall, J. Phys. Chem. A, 107, (2003),
10381.
[9] O. Bludsky; M. Silhan; P. Nachtigall, J. Chem. Phys., 117, (2002), 9298.
[lO] G. Kresse; J. Hafner, Phys. Rev. B, 48, (1993), 13115.
[11] G. Kresse; J. Hafner, Phys. Rev. B, 49, (1994), 14251.
[12] G. Kresse; J. Furthmuller, Phys. Rev. B, 54, (1996), 11169.
[13] G. Kresse; J. Furthmuller, Comput. Mater. Sci., 6, (1996), 15.
[14] P. E. Blochl, Phys. Rev. B, 50, (1994), 17953.
[15] G. Kresse; D. Joubert, Phys. Rev. B, 59, (1999), 1758.
[16] J.P. Perdew; K. Burke; M. Emzerhof, Phys. Rev. Lett., 77, (1996), 3865.
[17] J.P. Perdew; K. Burke; M. Ernzerhof, Phys. Rev. Lett., 78, (1997), 1396.
[~8] O. Bludsky; M. Silhan; P. Nachtigall; T. Bucko; L. Benco; J. Hafner, J. Phys. Chem. B,
accepted, (2005),
[19] P. A. Vaughan, Acta Crystallogr., 21, (1966), 983.
[203 It should be pointed out that this numbering scheme differs from that used in Database of
Zeolite Structures: http://www.iza-structure.org/databases/, where TI and T4 positions are
switched.
[21] P. Nachtigall; M. Davidova; D. Nachtigallova, J. Phys. Chem. B, 105, (2001), 3510.
[22] A.S. Goldman; K. KroghJespersen, J. Am. Chem. Soc., 118, (1996), 12159.
[23] M.F. Zhou; L. Andrews; C.W. Bauschlicher, Chem. Rev., 101, (2001), 1931.
[24] P. Treesukol; J. Limtrakul; T.N. Truong, J. Phys. Chem. B, 105, (2001), 2421.
[25] J. Limtrakul; P. Khongpracha; S. Jungsuttiwong; T.N. Truong, J. Mol. Catal. A., 153, (2000),
155.
[26] G.N. Vayssilov; M. Staufer; T. Belling; K.M. Neyman; H. Knozinger; N. Rosch, J. Phys.
Chem. B, 103, (1999), 7920.
[27] D. Nachtigallova; P. Nachtigall; O. Bludsky, Phys. Chem. Chem. Phys., 6, (2004), 5580.
[28] G.T. Palomino; S. Bordiga; C. Lamberti; A. Zecchina; C.O. Arean. Vibrational and optical
spectroscopic studies on copper-exchanged ferrierite. In Impact of Zeolites and Other Porous
Materials on the New Technologies at the Beginning of the New Millennium, Pts a and B,
2002; Vol. 142; pp 199.
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. Zilkov4and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 633
U. Ziese a, C.J. Gommes b, S. Blacher b, A.H. Janssen a*, A.J. Koster c and K.P. de Jong a@
*Present address: Shell Research & Technology Centre Amsterdam, The Netherlands
We have combined electron tomography with posterior image processing for a quantitative
morphological study of mesoporous catalysts at nanometer scale resolution. As a first
example for the possibilities of the approach, mesopore size distributions of single crystallites
of commercial USY and XVUSY zeolites were derived.
1. INTRODUCTION
Oil refining and (petro)-chemical industry heavily rely on the use of heterogeneous catalysis
to deliver high-quality transportation fuels and chemicals. Social concerns and environment
legislation for a sustainable future call upon continuous improvements of these catalysts and
concomitant deeper understanding of their properties. To this end structural characterization
of supported catalysts has been pursued extensively.
In general, the catalytically active phase in heterogeneous catalysts consists of nanometer
sized metal particles supported on micro- and mesoporous metal oxide carriers as alumina,
silica and titania. Size, shape and distribution of metal particles and pores in the carrier
influence reactivity, selectivity and accessibility of the catalyst. Bulk techniques such as
EXAFS, X-ray diffraction or physisorption studies, in particular when combined with electron
microscopy, are indispensable techniques to obtain this information.
Porosity characteristics like pore size distributions and surface areas, for instance, are
generally examined by physisorption or XRD. Unfortunately, these methods do not yield
information on the three-dimensional location, shape and connectivity of the pores. TEM and
STEM deliver a range of acknowledged techniques for structural characterization, e.g.
imaging, diffraction and element analysis, and can give an idea on mesopore morphology. A
single electron microscopy projection image, however, is only a two-dimensional projection
of the three-dimensional sample in nature and does not provide depth information and
unequivocal information on pore morphology.
634
Electron tomography, a form of 3D-TEM, on the other hand, has been shown to be a versatile
method for the non-destructive characterization of the interior of truly unique structures,
hundreds of nanometer in size, at nanometer-scale resolution [1]. The method, which is
conceptually related to computer tomography as applied in medical applications, utilizes a
series of electron microscopy projection images, acquired under different rotation angles of
the sample, to compute a 3D reconstruction of the material, representing the electron optical
density distribution of the sample.
Initial applications of electron tomography to mesoporous catalysts concentrated on a
qualitative interpretation of catalyst structure. Examples are the display of internal structural
changes with respect to the pores and cavities caused by different post-treatments in single
zeolite crystals [2-4] or during synthesis in carbon templated catalyst particles [5]. Metal
inclusions in mesoporous silica and carbon black were detected down to a size of a nanometer
[6-8]. For recent reviews on the role of 3D-TEM and electron tomography in catalysis
research we refer to Anderson [9], Ziese [ 10] and Thomas [ 11 ].
Here we have combined electron tomography with posterior image processing for the
quantitative three-dimensional structural investigation of mesopores inside single particles of
USY and XVUSY zeolites. Similar approaches have successfully been followed in the field of
X-ray tomography, where quantitative descriptions of the morphology of pores inside
microscopic samples were obtained [12]. Data evaluation involves denoising (i.e. enhancing
the signal to noise ratio of the 3D reconstruction), segmentation / binarization and the
morphological characterization of the 3D reconstruction. The approach is a step towards a
local quantitative description of the 3D morphology of heterogeneous catalysts at the
mesoscopic length scale.
2. EXPERIMENTAL SECTION
From a technical point of view, it has to be ensured that the morphological analysis is derived
from electron tomography reconstructions of the best achievable quality. Transmission
electron microscopy images suitable for tomographic data evaluation should conform to the
following requirements: The imaging mode, acquisition parameters (e.g. focus) and electron
dose have to be adjusted for optimal image contrast and resolution at minimal radiation
damage. While this is principally a prerequisite for electron microscopy, it is especially
demanding in electron tomography where tens to hundreds of images are acquired. The image
contrast should be a monotonic function of the mass density of the sample (projection
requirement). While amorphous or weakly diffracting materials of e.g. carbonaceous,
siliceous or zeolitic composition may thus be imaged in bright field mode, highly diffracting
materials as e.g. aluminas are favourably imaged in HAADF-STEM mode (high angular
annular dark-field scanning transmission electron microscopy, contrast- Z 2) [7]. For optimal
resolution in the reconstruction, very small and thin pieces of material, unobstructed by
embedding media or carrier films should be investigated (resolution = 3 * thickness / number
images) and tilt series should be acquired around two perpendicular tilt axis (problem of
missing wedge).
The image analysis applied in the present work aims at determining the 3D pore size
distribution of single crystallites of two types of mesoporous zeolites. Fig. 1 displays some
essential steps in the image processing involved in the quantitative morphological analysis.
Fig. l a shows one of the BF-TEM projection images of a XVUSY zeolite particle, from
which a tomogram was calculated via filtered backprojection. A tomogram (3D
reconstruction, map) is a volumetric description of the electron optical density throughout the
sample and is frequently displayed as a series of thin virtual cross-sections (slices) at different
heights. Fig. 1b displays one of the 2 nm thin slices.
Morphological analysis necessitates a separation of the area of interest, e.g. mesopores,
from the background in the image (binarization). For achieving this goal, the following
procedure was applied: (i) grey level image preprocessing for increasing the contrast between
the pores, skeleton and background, (ii) binarization in order to discriminate between pores
(white pixels) and the rest of the image (black pixels), (iii) measurement and quantization of
the occurrence of characteristics.
The preprocessing and binarization steps were performed automatically on each cross-
section independently. First, a mask was determined that allows discriminating the material
and its porosity from its outside (Fig. l c). On the basis of statistical properties of the image,
an optimal threshold was determined according to the maximum entropy method. With this
method, a binary image is obtained in which the pores are white and the background and
skeleton of the material are black (Fig. 1d). Finally, a morphological opening transformation,
which removes all image details that are smaller than a given probing element (e.g. a circle),
was applied to eliminate the spurious small white objects that are still present in the resulting
binary images (Fig. 1e).
In order to perform the measurement of the porosity, all the binarized slices were piled
up to recover the 3D binary images. Figs. 2a and 2b display the surface rendered
representations of the USY and XVUSY zeolite particles. A qualitative interpretation of the
images suggests that USY mainly consists of smaller cavities, while there are also larger
connected pores present in XVUSY. According to that two different approaches were
followed for the quantization of the pore sizes.
636
Fig. 1. Image series showing some steps involved in the quantitative morphological analysis of a
commercial XVUSY zeolite catalyst particle. (a) BF-TEM projection image, (b) 2 nm thin virtual slice
through the original reconstruction, (c) masked particle, (d) binarized pores, (e) denoised pores and
(f) enlarged view of the grey levels assigned to pore elements of different sizes.
(i) The diameter of compact cavities in the USY particle was identified with the
hydraulic diameter = 6 * volume / surface, which in the case of a sphere coincides with its
diameter [15]. For this approach each pore is automatically individualised and its volume and
surface area are determined. Fig. 2c (black line) displays the resulting pore size distribution.
(ii) The local diameter of elongated pores with a minimal thickness of 3 pixels = 6.4 nm
in the XVUSY particle was measured using an opening size distribution, which has been
recently called chamber size distribution [16-18]. According to this method, elongated pores
are divided into pieces with different diameter and each piece is assigned a size corresponding
to the size of the largest probing element that can be placed inside its volume. A two-
dimensional visualization of the used method, when applied to the 2D image of Fig. 1e with a
disk as a probing element, is given in Fig. 2f. Here, different grey levels were assigned to pore
elements of different sizes. For the actual 3D analysis, the spherical probing element was
approximated by an octahedron with diameters in the range of 6.4-27.8 nm. The pore size
distribution of the XVUSY particle is indicated by a dashed black line in Fig. 2c. For
comparison, the distribution calculated from the hydraulic size of the same particle (Fig. 2c,
grey line) is added.
The pore size distribution of the USY particle has a peak at a pore size of 7 nm, the
distribution of XVUSY has a maximum at a size of 11 nm. Moreover, with pore sizes up to
28 nm the distribution for XVUSY is broader than that of USY, indicating a higher amount of
larger pores in XVUSY zeolite. The findings are in good agreement with the visual
appearance of 2D TEM projection images and the 3D reconstructions of the particles (Figs.
l a, 2a and 2b) and also agree well with physisorption measurements of the bulk materials [4].
637
0 5 10 t5 20 25 30
pore size (nm)
The suggested approaches to use the hydraulic diameter for the characterization of
compact cavities and the opening size distribution for larger elongated pores are thus suitable
for the envisioned quantitative three-dimensional texture analysis of mesoporous zeolites at
nanometer scale resolution. As for any data derived from electron microscopy, care has to be
taken that the sample particles are representative for the bulk of the material and that the
resolution of the three-dimensional reconstruction is sufficient for the intended investigation.
A prerequisite for the comparison of results gained from the presented kind of analysis with
nitrogen physisorption measurements, is the accessibility of all pores (e.g. through
micropores) by the latter method. Otherwise the accessibility of the pores has to be taken
account for in the tomography derived pore size distribution. For detailed discussions on the
comparison between computationally and experimentally attained pore characteristics we
refer to Sweeney [19] and Garboczi [20].
4. CONCLUSION
We have combined electron tomography with posterior image processing for a quantitative
morphological study of mesoporous catalysts at nanometer scale resolution. As a first
example for the possibilities of the approach, mesopore size distributions of single crystallites
of commercial USY and XVUSY zeolites were derived and are in good agreement with the
visual appearance of the particles in TEM and physisorption measurements of the bulk
material.
The presented image processing path for the quantitative morphological analysis of
mesoporous catalysts should be seen as an example illustrating the information that can be
gained from electron tomography. Employment of other means of denoising (e.g. nonlinear
anisotropic diffusion [21]), binarization (e.g. eigenvector segmentation [22]) and
morphological analysis (e.g. spatial statistics and integral geometry [23,24]) might be helpful
for a quantitative 3D structural description of local radii, curvatures and connectivity of pores
in solid catalysts at the mesoscopic length scale. While bulk methods for texture analysis thus
prevail in the statistical characterization of the physical properties of mesoporous materials
(e.g. adsorption and diffusion), single particle analysis by electron tomography does not rely
on a priori assumptions on pore geometry. The quantization of the true 3D shape, will be
possible and can aid in a better interpretation and modelling of these properties.
638
ACKNOWLEDGEMENTS
We thank Hans Meeldijk (Utrecht University, The Netherlands) for continuous support with
electron microscopy and David Mastronarde (University of Colorado, Boulder, USA) for
software for electron tomography reconstructions. The research of U.Z. has been made
possible by support of the National Research School Combination Catalysis (NRSC-C).
C.J.G. acknowledges the National Funds for Scientific Research (FNRS, Belgium) for a Ph.D.
research fellow position.
REFERENCES
[1] K.P. de Jong, A.J. Koster. Chem. Phys. Chem. 3 (2002) 776.
[2] A.H. Janssen, A.J. Koster, K.P. de Jong, Stud. Surf. Sci. Catal. 135 (2001) 2144.
[3] A.H. Janssen, A.J. Koster, K.P. de Jong, Angew. Chem. Int. Ed. 40:6 (2001) 1102.
[4] A.H. Janssen, A.J. Koster, K.P. de Jong, J. Phys. Chem. B 106 (2002) 11905.
[5] A.H. Janssen, I. Schmidt, C.J.H. Jacobsen, A.J. Koster, K.P. de Jong, Microporous
Mesoporous Mater. 65 (2003) 59.
[6] A.H. Janssen, C.-M. Yang, Y. Wang, F. Schtith, A.J. Koster, K.P. de Jong, J. Phys. Chem. B.
107 (2003) 10552.
[7] P.A. Midgley, M. Weyland, J.M. Thomas, B.F.G. Johnson, Chem. Commun. (2001) 907.
[8] J.M. Thomas, P.A. Midgley, P.L. Gai, T.J.V. Yates, M. Weyland, Microsc. Microanal.
10 (Suppl 2), 2004, 34.
[91 M.W. Anderson, T. Ohsuna, Y. Sakamoto, Z. Liu, A. Carlsson, O. Terasaki, Chem. Commun.
8 (2004) 907.
[10] U. Ziese, K.P. de Jong, A.J. Koster, App. Cat. A 260 (2004) 71.
[ll] J.M. Thomas, P.M. Midgley, Chem. Commun (2004) 1253.
[12] S. Blacher, A. L6onard, B. Heinrichs, N. Tcherkassova, F. Ferauche, M. Crine, P. Marchot,
E. Loukine, J.-P. Pirard, Coll. Surf. A: Physicochem. Eng. Aspects, 241, (2004) 201.
[13] U. Ziese, A.H. Janssen, J.L. Murk, W.J.C. Geerts, T. Krift, A.J. Verkleij, A.J. Koster,
J. Microscopy 205:2 (2002) 187.
[14] D.N. Mastronarde, J. Struct. Biol. 120 (1997) 343.
G.W. Scherer, J. Am. Ceram. Soc. 73 (1990) 3.
[16] J. Serra, Image Analysis and Mathematical Morphology, Academic Press, Chap. 10 (1982).
[17] P. Soille, Morphological Image Analysis, Principles and Applications, Springer, Chap. 4
(1999).
[18] C.D. Tsakiroglou, A.C. Payatakes, Advances in Water Resources 23 (2000) 773.
[193 S. M. Sweeney and C. L. Martin, Acta Mater. 51:12, (2003), 3635.
[20] E.J. Garboczi, D.P. Bentz, Ceram. Trans. 16 (1991) 365.
[21] A.S. Frangakis, R. Hegerl, J. Struct. Biol. 135:3 (200), 239-250.
[22] A.S. Frangakis, R. Hegerl, J. Struct. Biol. 138:1-2 (2002) 105-13.
[23] K.R. Mecke, D. Stoyan, Statistical Physics and Spatial Statistics, Lecture Notes in Physics
554, Springer (2000).
[24] K.R. Mecke, D. Stoyan, Morphology of Condensed Matter, Lecture Notes in Physics 600,
Springer (2002).
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 639
The catalytic activity of mordenite can be substantially modified following treatments such as
acid leaching and steam treatment. In the present study the effect of both such treatments,
separately or sequentially, on the acidity and the structural and catalytic properties of
mordenite has been investigated. Steaming appeared to remove mainly external surface
framework AI whereas acid leaching also removed A1 from the bulk framework. Steaming
reduced the amount of framework AI more significantly than acid leaching as well as causing
a greater reduction in the number of acid sites. After steaming these sites were found to be
stronger. In the propylation of naphthalene the highest ratio of 2:6/2:7 DIPN was observed
when the sample was steamed followed by acid treatment. This corresponded to the lowest
external surface activity as indicated by cyclohexanol dehydration reaction.
1. INTRODUCTION
The dealumination by steaming and acid leaching of mordenite is important since its catalytic
activity is significantly influenced by such treatment [1-4]. From a structural point of view
such treatment results in loss of framework aluminium resulting in the formation of extra-
framework species such as A1 (OH)3 or, after dehydration, charge-balancing species such as
AI(OH)2 +, AI(OH) 2+ or AI 3+ all of which can act to balance the remaining framework
aluminium charge. Such treatment can also result in the formation of hydroxyl nests and an
enhanced mesoporous network. The extra-framework species, typically octahedrally
coordinated aluminium, can alter the catalytic activity of the zeolite by changing its acidic
characteristics or by inhibiting diffusion by blocking the intracrystalline pores. Specifically
acid leaching of steamed zeolites can remove extra-framework species formed during thermal
treatment but can also result in further framework dealumination [5,6]. Hence it is important
to optimize the combination of these treatments. The extra-framework aluminium species
formed from steaming are also thought to migrate to the external surface of the crystallite
where they may be able to cause in changes in the catalytic performance of the zeolite by
promoting non-shape selective reactions which occur on external surface sites. This has been
observed in the case of ZSM-5 [7].
The present paper presents a selection of results emanating from an extensive
investigation into the effects of steaming and acid leaching of H-Mordenite, separately or
sequentially, on its structural, acidic and catalytic properties. Ammonia TPD, 29 AI - and 27 Si-
NMR, pyridine FTIR and XRD were used to characterize the effect of the treatment on the
acidic and structural properties of the samples. Probe reactions were used to investigate the
effect of these treatments on the catalytic activity of the mordenite. Cyclohexanol dehydration
probed the external surface activity [8] and the propylation of naphthalene probed the relative
640
2. EXPERIMENTAL
The parent catalyst was the sodium form of the mordenite. The procedures to produce the H-
form have been described elsewhere [10]. The various samples obtained in this catalyst are
shown in Table 1. For steaming, the catalyst was loaded into a tubular reactor and heated to
550~ at 5~ 30kPa steam was then passed over the catalyst at 190ml/min. in a nitrogen
carrier gas for 5 hours at 550~ Acid leaching was carried out by slurrying 3.5 g of catalyst
in 100 ml nitric acid for 4 hours at 70~ with the concentrations as indicated in Table 1. Acid
was removed by vacuum filtration and the catalyst subsequently washed repeatedly with de-
ionised water before drying at 80~ for 12 hours.
Table 1
Description of treatment used on various samples
Sample Treatment
H550 Parent Na-M ion exchanged to the
ammonium form and calcined at 550~
H2 H550 sample subsequently washed with
2M HNO 3
HS H550 sample subjected to mild steaming
HS0.1 HS sample subjected to acid leaching in
0.1M HNO3
HSI HS sample subjected to acid leaching in
1M HNO3
Details of the procedures used for the % crystallinity determinations, 29A1 and 27Si-
NMR and ammonia TPD have been described elsewhere [10]. In the case of the 29Al -NMR,
framework aluminium (Al(f)) is assigned to the peak at 55ppm, non-framework octahedrally
coordinated aluminium (Al(nfo)) at 0ppm and a shoulder at 55ppm to a disturbed
tetrahedrally coordinated or a penta-coordinated state of the aluminium (Al(nft)), the nature of
this species not being fully understood [11 ]. The method described by Massiot et al. [12] was
used to obtain an estimate of the amount of various A1 species detected in the NMR studies.
Quadrupole coupling constants for trigonal and tetragonal sites were obtained from Brog et al.
[13] and the quadrupole frequency from Samoson et al.[14]. Pyridine FTIR studies were
carried out using a Nicolet 5ZDX FTIR spectrometer. A 15 mg sample was pressed into a
self-supporting wafer which was subsequently housed in a specially designed cell which
allows for controlled temperature programming of the sample without interfering with the
infrared absorption. The cell was evacuated to 10-5 bar and then the wafer dried at 400~
Pyridine adsorption took place at 100~ for 30 minutes. The cell was evacuated for 45
minutes before the spectrum was taken. Further spectra were taken after desorption at the
various temperatures for 30 minutes. The probe reactions were carried out in a standard
tubular stainless steel plug flow reactor. The cyclohexanol dehydration was performed at
atmospheric pressure and 120~ Nitrogen fed via a mass flow controller was saturated with
cyclohexanol at 80~ before being fed to the reactor at a WHSV- 1.58h ~. DME was used as
641
an internal standard for mass balance calculations. Propylation of naphthalene was carried out
in a 500 ml stirred batch autoclave at 160~ 5g of naphthalene was diluted in 50 ml of
undecane and then propene was introduced into the autoclave until the total pressure reached
10 bar. The reaction took place over 1 g of catalyst. The undecane was used as an internal
standard.
3. R E S U L T S AND DISCUSSION
Tables 2 and 3 show the results obtained in the NMR, infrared and XRD studies of the five
samples reported in this paper. Columns two and three in Table 2 show the values the bulk Si
and AI concentrations obtained using AAS. The last three columns show the concentrations
on a per unit cell basis from the NMR studies. There was no measurable loss of Si during
these treatments.
In the first instance it can be seen when comparing H550 and HS that steaming did not
cause any significant change in the amount of bulk A1 but did result in a significant decrease
in the amount of framework A1. The extra-framework A1 species reported here include both
octahedrally coordinated (nfo) and tetrahedrally coordinated (nft) species, the latter as
mentioned above being ascribed to a disturbed tetrahedrally coordinated or a penta-
coordinated state of the aluminium [ 11 ]. The steamed sample showed a high concentration of
both forms of extra-framework species.
Acid leaching is known to remove A1 species not contained in the framework [5]. This
can be seen to be the case in the present study by comparing the pair of samples H550 and H2
as well as the steamed samples, HS, HS0.1 and HS 1. In all these cases acid leaching resulted
in a decrease in bulk aluminium. However acid leaching did not cause any significant loss of
framework A1 particularly in the case of the steamed sample, HS. In the case of H550, acid
leaching did result in some loss of framework A1. These results indicate that the steam
treatment seemed to produce a more stable form of the mordenite, similar for example to that
observed in the case of the formation of USY.
The extra-framework species resulting from steam treatment can remain either within
the pores of the mordenite or migrate to the external surface. As already mentioned acid
leaching this sample did not cause any further loss of framework A1 but did, as expected,
remove AI from the bulk. Such treatment resulted in the concentration of A1 (nfo) decreasing
from 1.4 to 1.1 and AI (nft) from 1.5 to 0.5 while the amount of bulk A1 decreased
progressively from 6.5 to 5.6 to 5.0. The relatively high amount of A1 (nft) in HS is surprising
and this may support the proposal that it is not a genuine non-framework species but, as
already mentioned, rather a disturbed non-fully coordinated species. Katada et a1.[15], in a
recent study of dealuminated mordenite, have shown that the presence of even small amounts
of water vapour during calcination can result in dealumination and this is consistent with the
fact that the H550 sample had a lower Al(f) content (5.4) than that of the parent Na-MOR
sample which had an Al(f) content of 6.5 and that it already ~':ad a significant amount of
octahedrally coordinated extra-framework A1 present. The loss of framework aluminium as
determined by the NMR studies was accompanied by a decrease in the unit cell volume of up
to 1% relative to the parent material. This is consistent with similar findings made in the case
of, for example, ultrastable zeolite Y (USY). Although not shown in Table 2 a close
correlation was observed between the concentrations of framework aluminium and of Si(1A1).
642
Table 2
Effect of various steaming and acid leaching treatments on the A1 contained in the mordenite
structure
Table 3
Effect of various steaming and acid leaching treatments on the % crystallinity and acidity of
mordenite
Table 3 shows the results obtained in the study of the acidity of the various samples.
Steam treatment hardly influenced the relative % crystallinity but acid leaching always
resulted in a slight decrease in crystallinity. This is accompanied by a decrease in bulk A1 and
in the amount of especially A1 (nft). This may indicate that, when acid leaching removes the
A1 (nft) species, hydroxyl nests are formed accompanied by enhanced mesoporosity. The
TPD spectra showed two main features, viz. a sharp peak at temperatures up to 250~ and a
much broader peak in the temperature range 500~ to 550~ The number of acid sites
(Bronsted + Lewis) was determined from the integrated area under the spectra as well as the
titrated amounts. The low temperature peak does not arise from physisorbed ammonia as it
did not decrease even after 60 hours desorption in flowing helium at 100~ Insofar as the
peak desorption temperature of ammonia in TPD studies may be indicative of the strength of
sites it can be noted that the average peak desorption temperature for the two high
temperature peaks, viz. strong Bronsted and Lewis sites, for the steamed samples was about
100~ greater than that for the samples H550 and H2. There is thus clear evidence that,
although there are fewer sites after steaming, they are stronger. In the pyridine FTIR studies,
the amount of acid sites was determined using previously reported molar extinction
coefficients, viz. 1.67 and 2.22 cm/mmol for Bronsted and Lewis type sites respectively [ 16].
Both TPD and IR results showed that acid leaching or steaming of H-MOR decreased
the amount of acidity. Steaming caused a significant loss of acidity (H550 vs. HS) but acid
leaching of the steamed samples hardly changed the total amount of acidity. When comparing
H550 and HS it can be seen that even though the bulk A1 content is similar there is a
significant decrease in the concentration of acid sites measured and a decrease in the amount
643
of A1 (f). As is well known the amount of aluminium usually corresponds well with the
number of acid sites observed in TPD experiments. In order to explore this relationship
further calculations were carried out to see if there was any relationship between the number
of acid sites and the various measures of A1 such as framework A1, bulk A1, Al(nfo), Al(nft)
etc. Fig.1 shows the relationships between the concentration of acid sites and the amount of
bulk (total) A1 obtained from AAS measurements as well with the amount of framework A1
and the amount of [Al(f)-Al(nf0)] determined from NMR measurements. There is clearly no
1:1 correlation between the total or bulk concentration of A1 and the total number of acid
sites. Also no such correlation was observed between the concentration of acid sites and the
concentration of framework A1. There was however an indication of a 1:1 relationship
between number of acid sites and the amount of [Al(f)-Al(nfo)], the number of acid sites
being underestimated, when correlated with framework AI, to an amount roughly equal to the
amount of extra-framework octahedrally-coordinated AI. If it were assumed that the Al(nfo)
and Al(nft) species are weak sites i.e sites from which ammonia easily desorbs at low
temperatures, then this observation suggests that the extra-framework octahedrally-
coordinated species, Al(nfo), is blocking access by ammonia to certain acid sites.
2"5 I A 9 A
~1.5| -- J - II Framework AI
| [] 9 ~ ~ Al(f)-Al(nfo)
k AI
0.5
0 ~ ~
Fig. 1. Relationship between the total number of acid sites as measured by ammonia TPD and (i) bulk
AI (AAS), and from NMR measures (ii) framework AI and (iii) the difference between the amount of
framework A1 and the amount of non-framework octahedrally co-ordinated A1 atoms.
644
Table 4 shows the results obtained in the study of the catalytic behaviour of the modified
samples using propylation of naphthalene and cyclohexanol dehydration as probe reactions.
The former gives an indication of the extent to which the reaction occurs inside the porous
system as opposed to the external surface and the latter indicates the extent of activity of the
external surface sites. In the case of the propylation of naphthalene, selectivity in favour of
the 2:6 DIPN isomer as opposed to the 2:7 DIPN isomer will be indicative of the reaction
occurring mainly inside the porous system as opposed to at the external surface. It is useful in
the first instance to compare in Table 4 the results of the non-steamed samples (H550 and H2)
with the steamed samples (HS, HS0.1 and HS 1). For the naphthalene propylation reaction the
latter group generally show a higher propene conversion and a higher 2:6/2:7 ratio of DIPN
and a lower extent of cyclohexanol dehydration. This is all indicative of there being less
activity on the external surface of the mordenite after steaming. If samples H550 and HS are
specifically compared the latter has much less framework A1, more extra-framework A1 and a
lower concentration of stronger acid sites. It has often been proposed that extra-framework A1
species block pores and reduce diffusivity and this would have resulted most likely in an
increase in the selectivity to 2:7 DIPN. This was not observed and therefore these results seem
to indicate that the dominant effect of steaming is not to cause pore blockage but to reduce the
number of acid sites and that this resulted from a loss of framework A1 at the external surface.
Table 4
Probe reaction results: a. Propene conversion and selectivity ratios between 2:6/2:7 DIPN in
the propylation of naphthalene, b. Conversion in the cyclohexanol dehydration reaction.
Thus the reaction will be taking place mainly within the porous system where
transition state and/or product shape selectivity would dominate. Given that it is likely that the
reaction may take place near the entrance to the pores it is possible that the reaction is
controlled by transition state shape selectivity. Propene preferentially enters the pore and
adsorbs typically on a 13acid site whereafter the first propylation reaction takes place resulting
in the formation of 2:IPN. The second propylation reaction will thus sterically preferably
occur at the exposed 6 position. Extra-framework A1 species are often thought to migrate to
the external surfaces but, in the present instance, if they do they have not demonstrated any
catalytic behaviour otherwise an increase in cyclohexanol dehydration and in 2:7 DIPN
formation would have been observed. Migration of such species is probably only occurring
during the subsequent acid treatment. Finally although the TPD results showed that the
steamed samples had stronger acid sites this does not appear to be a critical factor in the case
of the two probe reactions used in this study.
Acid leaching of H550 sample results in a decrease in the concentration of acid sites,
an increase in the amount of extra-framework A1, and higher selectivity to the 2:6 DIPN and
645
an increase in cyclohexanol dehydration. It is also known that such treatment promotes the
formation of mesopores and surface hydroxyl nests. These results are consistent with the
steaming results except for the cyclohexanol dehydration reaction which shows an increase
for the H2 sample. One possibility is that the acid leaching caused extra-framework A1
species to migrate to the surface of the crystallite where they are able to catalyze the
dehydration reaction. Acid leaching of the steamed sample yields the predicted results. For
samples HS0.1 and HS1 the acid site concentrations are similar to HS but the 2:6 DIPN
selectivity significantly higher. This may show that for these samples the removal of species
from the pores enhances the opportunity for reaction to take place inside the pores or at
surface hydroxyl nests and increases the formation of the 2:6 isomer. The slightly anomalous
result of an enhanced cyclohexanol dehydration activity for the sample HS1 may reflect the
same situation as for H2. A stronger concentration of the leaching acid may have caused
further loss of framework A1 (cf. Table 2) and if these are near the entrances to the pores they
will reduce the yield of 2:6 DIPN. At the same time the extra-framework A1 species formed
can migrate to the surface with the acid leaching and there they will increase cyclohexanol
dehydration conversions. Interestingly however, as for sample H2, if these extra-framework
A1 species are present on the surface they are hardly able to catalyze the propylation of
naphthalene. The results of samples H550 and H2 are consistent with those of Nesterenko et
al. [17] who, in a recent study of the calcination and acid leaching of mordenite using 6M
methanesulphonic acid, found that the acid leaching removed all extra-framework A1 species
(as detected by 27A1 -NMR) and that this was accompanied by a significant increase in the
2:6/2:7 DIPN ratio. They ascribed this to acid leaching opening up the porosity of the
mordenite. They did not report any acid site measurements but did confirm the enhanced pore
volume of the acid treated sample.
4. CONCLUSIONS
REFERENCES
[1] K.L. Kranich, Y.H. Ma, L.B. Sand, A.H. Weiss, I. Zwiebel, Adv. Chem. Ser., 101 (1984) 502.
[2] C Mirodatos and D Barthomeuf, J Chem. Soc, Chem. Commun. (1981) 39.
[3] C. Mirodatos, B.H.Ha, D. Otsuka and D. Barthomeuf, (Ed. L. Rees) Proc. 5th Int. Zeolite Conf.,
Heyden, London, (1980) 382.
646
[4] A.K. Ghosh and G. Curthoys, J. Chem. Soc., Faraday Trans., 1, 79 (1983) 805.
[5] F. Mauge, A. Auroux, J.C. Courtelle, Ph. Engelhard, P. Gallezot, J. Grosmangin, (Eds. B Imelik
et al) Catalysis by Acids and Bases, Elsevier, Amsterdam, (1985) 91.
[6] J. Sherzer, J. Catal., 54, 285 (1978).
[7] N.Y. Topsoe, F. Joensen, E.G. Derouane, J. Catal., 110 (1988)404.
[8] H.G. Karge, H. Kosters, Y. Wada, (eds.D Olson, A Bisio) in: Proc. 6th Int. Zeolite Conf.,
Butterworths, London, (1984) 308.
[9] Y. Sugi, T. Matsuzaki, T. Hanaoaka, Y. Kubota, J-H, Kim, Catal. Letters, 26 (1994) 81.
[10] A.W. O'Donovan, C.T. O'Connor, K.R. Koch, Microporous Materials, 5 (1995) 185.
[ 11 ] J. Rocha, S.W. Carr and J. Klinowski, Chem. Phys. Letters, 187 (1991) 401.
[12] D. Massiot, C. Bessada, J.P. Coutures, F. Taulelle, J. Magn. Res., 90 (1990) 231.
[ 13] K.C. Brog, W.H. Jones, C.M. Verber, Phys. Lett., 20 (1966) 258.
[14] A. Samoson, E. Lipmaa, G. Engelhardt, U. Lohse, H.G. Jerschkewitz, Chem. Phys. Lett., 134
(1987) 589.
[15] N. Katada, T. Kanai, M. Niwa, Microporous Mesoporous Materials, 75 (2004) 61.
[16] C.A. Emeis, J. Catal., 141 (1993) 374.
[17] N.S. Nesterenko, S.E. Timoshin, A.S. Kuznetsov, V. Montouillout, F.Thibault-Starzyck, C.
Fernandez, J.P. Gilson, F. Fajula and I.I. Ivanova, Proc.14 th Int. Zeolite Conf., Cape Town,
(2004), 2163.
Studies in Surface Science and Catalysis, volume 158
J. t2ejka, N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 647
1. INTRODUCTION
Separation science and catalysis [1], protein and polymer chain folding [2,3], biological
membranes [4], colloids [5], and nanotube applications [6] are among the areas in which the
interaction of water with hydrophobic micro- and nanopores is of importance.
Hydrophobicity is usually associated with the lack of hydroxyl groups or of other specific
adsorption sites that can hold water molecules via hydrogen bonds and/or electrostatic forces
[7]. On the other hand, it has been well demonstrated that adsorption in d < 2 nm diameter
pores does not necessarily involve any surface sites because the typical micropore filling at
extremely low p/po < 0.02 relative vapor pressures is mainly due to attractive interactions
between adsorbate molecules which results in Type I adsorption isotherms [8-13]. One would
expect therefore that the small, highly polar H20 molecules that are prone to form hydrogen
bonds would saturate such small channels regardless of the hydroxyl coverage of their walls
hence, microporous materials would not exhibit significant hydrophobicity. In contrast, it is
well known that most highly siliceous zeolites are hydrophobic to various extents. Many
reports state that the increasing Si/Al ratio and the decreasing OH content play a major role in
increasing the hydrophobicity of zeolites [ 14,15].
While these parameters are important, we have shown recently that they are not a
satisfactory explanation for the high hydrophobicity and the associated unusual adsorption
properties of strongly aluminum deficient (Si/A1 > 20) H-Y zeolites [16-18]. Unlike non-
polar adsorbate molecules such as paraffins or Ar, polar adsorbates like methanol or water are
repulsed when trying to enter the enhanced, overlapping negative potential field of the large
oxygen atoms that constitute part of the pore walls of these microporous materials. This long
known [19] electrostatic force field seems to exert an as yet unexplored polarity sieving effect
(PSE) in Y zeolites on adsorbates having different dipole moments.
648
This paper reviews some aspects of the PSE. It has been demonstrated that molecular
spectroscopy is a useful tool for exploring fine structural differences between H-Y zeolites
having different hydrophobicities when their bulk Si/A1 ratios, crystallinity, and lattice
constants are virtually identical [16-18]. Here we present further spectroscopic observations
on various aluminum deficient H-Y structures and also some commercial siliceous
microporous silicates, H-ZSM-5 and TS-PQ.
2. E X P E R I M E N T A L
All adsorbents tested are commercial products from PQ Corporation and Zeolyst International
[20]. Some of their properties are summarized in Table 1. Adsorbed vapors were generated
from high purity liquids obtained from Fluka and Aldrich. Gases were 99.99%+ purity
products obtained from MG Industries. Some properties of these compounds are shown in
Table 2.
Adsorption experiments were performed on an RXM-100 instrument from ASDI on
samples pretreated at <10 .3 Pa and 500 ~ C for 2 hours. Transmission and attenuated total
reflectance (ATR) FTIR spectra were obtained on a Nicolet Magna 550 spectrometer. Self-
supported wafers dehydrated at 500 ~ C and 2 x 10 -6 Pa for 2 hours were used for OH-
measurements. The T-O vibrations were measured at atmospheric conditions on the same
spectrometer using a single bounce diamond ATR accessory from ASI and on a Kaiser
Optical Systems HoloProbe Raman spectrometer equipped with 200 mW Nd:YAG diode
pumped laser source (frequency doubled to 532 nm; laser power on sample - 8 0 roW) using a
Table 1
Adsorbents tested. Silicalite was prepared according to the same procedure as TS-PQ, but
without titanium [21]. The hydrophobicity index is the ratio of c-hexane/H20 adsorbed on
unit catalyst as detailed in reference [ 17]. Further data have been reported in [ 16-18, 20-24].
Table 2
Selected properties of adsorbate molecules. ~ and a are from [25], v and d are from [ 18].
fiber optic connected Mark II type probe head. Diffuse reflectance (DR) FTUV and FTIR
spectra were measured on a Bruker IFS 66/S spectrometer on samples dehydrated in situ at
-10 .5 Pa at various temperatures using a DR cell from Harrick Scientific. Further
experimental details have been reported elsewhere [ 16-18, 22, 23].
E o6 =3 CBV 901
{.-
O " CBV 780 J ,~J:] H-ZSM5 I
(Si/AI - 4 0 ) ~ ](Si/AI -4330)]
"(3
"50.
C~
I
3739
.O- O
-(3 I cBv
. .....
.I
......................... ~
I IV3701 3608
:r ...... [ (Si/AI ~42)~. / I "~;!/363//_3%%% 6
. . . . . . 9,~-:--~, ,
"o o
<
0.0 0.2 0.4 0.6
P/Po 0 , , , , , , ,
What force increasingly retards the adsorption of molecules as their polarity increases
and ultimately prevents the micropore condensation of highly polar H20? Why is this PSE
more pronounced in the aluminum deficient H-Y than in the siliceous ZSM-5 framework?
Molecular modeling experiments were able to provide possible answers [ 16]. The overlapping
negative potential field of large oxygen atoms that constitute part of the pore wails of
microporous silicates has been known for a long time [ 19]. We believe that this electrostatic
field repels the negative parts of polar adsorbate molecules but does not affect the non-polar
adsorbates. This action is separate from the well-known curvature effect that controls the van
der Waals interactions and diffusion of Lennard-Jones fluids in micropores and depends
mainly on the polarizability of adsorbates as well as their relative size compared to the
micropore diameter [9-13, 26]. According to our model calculations [ 16], more than 80% of
the overall potential is electrostatic in the microporous silicate channels, therefore the
electrostatic repulsion can override the van der Waals attractions or even hydrogen bonds
between the pore walls and strongly polar adsorbate molecules such as H20. Fig. 5 shows
that the negative potential field in the channels of an aluminum free Y zeolite was found to be
largely homogeneous while that in an aluminum-free ZSM-5 changes significantly due to the
relative proximity of atoms to the pore centers [16]. Thus, water molecules can enter the
pores of ZSM-5 through positive spots rendering this zeolite inherently less hydrophobic than
the Si-Y. In agreement with this water repelling function of the electrostatic potential, it was
reported recently [6, 27] that H 2 0 couldn't penetrate hydrophobicized quartz capillaries but
penetrates hydrophobic carbon nanotubes, which don't have a significant electrostatic field.
50 --3736
0 ~ o .~
i
CBV 901
CBV 780
--
.........
o -5o " "- i..~3744 CBV 390H ---
E
-100
~d
I
.~ -15o ~d
o
13_ -250
-300 ~.
I.d'"
y Zeolit e _
~ ZSM5; Straight Channel "t
8"*
~d
0,1
/'<'
razor";:%,
i
.8 t
'-' 1072~! ''1057 4,56 x,ogl - oj
507 c.v.o,
Silicalite . . . . . . . . .
<-
"
'A',
311 ; 1 1 .
"1
;i; ; 1126
4~ 1.7 9~t- 29 7
o 2 837 61 3~!i
in . ;ii-
~ -i : .,, ' ". ' ~ 9 ..
<
200 400 600 800 1000 1200
1200 1000 800 600 400
Wavenumbers [cm -1] Raman Shift [cm -1]
Fig. 7. FTIR ATR spectra of Al-deficient Fig. 8. Raman spectra of various
Y zeolites at ambient conditions. Al-deficient zeolites at ambient conditions.
The next question to be answered is: if the electrostatic field in the micropores of Y
zeolites does repel polar water molecules so strongly, then why aren't all Y zeolites
hydrophobic? The simple answer is that the repulsion is only strong enough if lattice defects
including Vacancies, aluminum atoms and associated cations don't substantially disturb the
homogeneous, strongly negative electrostatic field in the pores. At some point they will totally
eliminate hydrophobicity. According to the experimental results with CBV 901 (Fig. 1), high
hydrophobicity can be maintained up to about 5 A1/unit cell. A US-Y type zeolite, such as
Zeolyst CBV 500 with Si/A1 - 2.6 ratio is never hydrophobic [17, 20]. CBV 780, which has
similar Si/AI ratio to CBV 901 but a lower hydrophobicity index (Table 1), is also evidence
that a low number of vacancies or other structural defects is also important. This becomes even
more apparent in the case of the highly siliceous but only slightly hydrophobic CBV 390H
(Table 1). The increased number of vacancies manifests itself as an increase in the hydroxyl
contents of CBV 780 and CBV 390H as shown in Fig. 6. The extremely sharp T-O-T
vibrations in the FTIR and Raman spectra of CBV 901 (Figures 7 and 8) are indicative of the
low number of defects (polarizable terminal T-O bonds) as discussed before [18].
s3 ol ~d ....................... 3490 ].
0
O <
3700 3600 3500 3400 3300 3700 3600 3500 3400
Wavenumbers [cm -1] Wevenumbers [cm n]
Fig. 9. FTIR transmission spectra of Silicalite Fig. 10. FTIR transmission spectra of TS-PQ-1
treated (T) and untreated (U) with CH3COONH4; treated (T) and untreated (U) with CH3COONH4;
measured on in-situ dehydrated samples. measured on in-situ dehydrated samples.
652
~d
~d
1300 1200 11 O0 1000 900 ;oo ;so 300 3so
Wavenumbers [cm -1]
Wavelength [nm]
Fig. 11. FTIR ATR spectra of TS-PQ-1 Fig. 12. FTUV DR spectra of TS-PQ-2 treated (T)
treated (T) and untreated (U) with CHBCOONH4; and untreated (U) with CHBCOONH4; measured
measured at ambient conditions. after in situ dehydration at <10 -3 Pa and 25 ~
critical for these processes [22]. Although TS-PQ, a microporous MFI-type titanium silicate
made by a non-alkoxide based technology developed by the PQ corporation [20, 21], was
found to be an extremely active and selective catalyst for the oxidation of n-hexane, it did not
show such hydrophobic character [21-24]. In fact it preferably adsorbed the aqueous H202
phase even in the presence of a large excess of hydrocarbon. This allowed the dosing of this
valuable oxidant as the reaction proceeded thereby reducing the undesired spontaneous H202
decomposition normally seen.
Fig. 2 shows that modifying the synthesis conditions can substantially change the
hydroxyl content, i.e., defect structure of silicalite. The virtually aluminum free silicalites are
frequently highly defective structures, which might be beneficial for selective acid-base
catalysis [28]. Albeit to a lesser extent, similar defects exist in the TS-1 catalysts [29]. To
reduce the lattice site defects and improve the XRD signals of these pentasils, they are often
treated with ammonium acetate [29]. Skeels and Breck [30] used CHB-COONH4 to facilitate
the (NHa)2SiF6 based dealumination process of zeolites and noted that it could be detrimental
to the crystal structure that has been confirmed again very recently [31 ].
By probing the post-synthesis effect of CHB-COONH4 on TS-PQ related materials, we
found that it substantially reduced the silanol content of silicalite (Fig. 9) as Bolis et al. [29]
~'X 104
~'X 1~4
8 37.3 TSPQ-2 u 960
--
'- 12[""~
'". TSPQ-1 u - -
23
>, i\. TSPQ-1T .......... ~ 6 ]~ ~7TSPQ-2T
1 .........
i 374 963
< 4
<
.+..,
c-
i ~416 A
0~ 4- o
0 ~ -.......
>, 2-
c 0 C
Table 3
Deconvoluted hydroxyl vibrations in the FTIR spectra of various silicates. T = treated
U = untreated with ammonium acetate. Sili.- Silicalite; TS - TS-PQ (see Table 1)
3748 +1 1.5 . . . . . . . .
3744 +1 1.6 - - 2.6 2.6 - 1.4 2.9 1.1
3739 +3 4.1 5.2 8.6 4.4 1.8 0.3 1.9 5.3 1.7
3729 +4 2.9 2.4 9.1 9.2 2.7 0.8 2.4 9.4 3.1
3715 +8 5.7 6.9 10.2 11.7 - - 1.8 18.6 -
3702 ~-3 0.3 6.3 11.8 - 3.1 1.5 2.7 - 1.6
3665 • 12.0 11.7 10.0 13.0 - 1.3 12.9 18.4 1.5
3629 • 0.2 0.5 0.7 . . . . . .
3598 • 7.4 46.3 53.1 10.9 3.3 10.7 10.0 25.9 4.4
3564 • 0.1 0.3 0.3 - - - 3.7 - -
3533 • 22.1 33.9 29.7 9.9 - 2.1 4.6 18.8 0.7
3470 +3 - - - 63.9 37.3 32.1 16.8 74.5 14.0
reported before. A very similar result was obtained with a TS-PQ-2 sample with high
titanium content (Tables 1 and 3). However, Fig. l0 shows that the ammonium acetate
treatment generated lots of silanol groups on the relatively defect free TS-PQ-1. Interestingly
this does not seem to affect the 960 cm -I FTIR band (Fig. 11), which is a well-known
benchmark for the lattice connectivity of Ti 4+ ions [22]. Fig. 12 indicates that the acetate
treatment results in a shoulder near 220 nm in the UV spectrum of TS-PQ-2. Such band has
been frequently observed on TS-PQ catalysts [21-24]. Some higher coordinated Ti 4+ ions
near 260 nm also appear. Significant structural change can be observed in the Raman spectra
of both TS-PQ samples in Figures 13 and 14. According to Fig. 8 Raman can also distinguish
fine structural differences of TS-PQ and Silicalite. The broad lines of the MFI type silicates
compared to the sharp bands of CBV 901 suggest that the MFI materials are more defected.
Table 3 shows that the FTIR hydroxyl spectra (stretching modes) can be deconvoluted
with r 2 > 0.999 correlation into a handful of weakly perturbed (>3650 cm 1) and strongly
perturbed (internal hydrogen bonded) modes in the 3000 to 3800 cm -1 range. The structural
assignments of different vibrations have been discussed in detail before [ 18, 28, 32]. Data in
Table 3 suggest that the overall vibration energy of the hydrogen bonded hydroxyl groups is
weaker (vibration at lower frequencies) in the MFI materials than in the Y structures; that
might be associated with the smaller pore diameters having better proximity to form hydrogen
bonds. The ammonium acetate treatment had no preferred effect on any hydroxyl type in the
pentasil structures. None of the treated pentasil materials became strongly hydrophobic.
4. C O N C L U S I O N S
Experiments on Y zeolites indicate that PSE plays a more important role in the
hydrophobicity of zeolites than their Si/A1 ratio or hydroxyl content as long as the pores are
large enough to neglect the Coulombic field of the silicon ions in the pathway of the
adsorbate molecules and the exchangeable cations or lattice defects do not disturb the
654
negative electrostatic field of the large oxygen ions paving the narrow zeolite channels.
Consequently the usually pursued purely siliceous zeolite lattice and the total absence of
hydroxyl groups are not essential to attaining strong hydrophobicity. Note that the PSE is
associated only with the ionic and permanently polar character of pore walls and not their
elemental composition; hence, it likely plays a role in the penetration of H20 into
biomembranes or other small non-silica based ionic pores also.
In contrast to the common belief we found the aluminum-free pentasil silicates, with
or without lattice titanium content, only slightly hydrophobic. They adsorb substantially
more water than the CBV 901 type Y zeolite. Their hydrophobicity does not increase
measurably when treated with ammonium acetate as proposed in the literature, results
indicate that this post-synthesis ammonium acetate treatment is not a simple defect reduction
method. It can also increase the structural silanol content and harm the fine structure of these
microporous silicates.
REFERENCES
[1] N . R . E . N . Impens, P. van der Voort, E. Vasant, Microporous Mesoporous. Mater., :28 (1999) 217.
[2] Y. Zhu, X. Fu, T. Wang, A. Tamura, S. Takada, J. (3. Saven, F. Gai, Chem. Phys., 307 (2004) 99.
[3] L. Livadaru, A. Kovalenko, J. Chem. Phys., 121 (2004) 4449.
[4] O. Beckstein, P. C. Biggin, M. S. P. Sansom, J. Phys. Chem. B, 105 (2001) 12902.
[5] J. Dzubiella, J.-P. Hansen, J. Chem. Phys., 121 (2004) 5514.
[6] T. Ohba, H. Kanoh, K. Kaneko, J. Phys. Chem. B, 108 (2004) 14964.
[7] A.V. Kiselev, J. Coll. Interf. Sci., 28 (1968) 430.
[8] S. Brunauer, L. S. Deming, W. E. Deming, E. Teller, J. Am. Chem. Soc., 62 (1940) 1723.
[9] R. M. Barrer, F. R. S. Reucrofi, P. J. Reucroft, Proc. Roy. Soc., 258A (1960) 449.
[10] D. H. Everett, J. C. Powl, J. Chem. Soc., Faraday Trans. I., 72 (1976) 619.
[11] E. G. Deruane, Chem. Phys. Lett., 142 (1987) 200.
[12] P. J. M. Carrot, K. S. W. Sing, Stud. Surf. Sci. Catal., 39 (1988) 77.
[13] S. Yasonath, P. Santikary, J. Phys. Chem., 98 (1994) 6368.
[14] D. H. Olson, W. O. Haag, W. S. Borghard, Microporous Mesoporous Mater., 35-36 (2000) 435.
[15] C. K. W. Meininghaus, R. Prins, Microporous Mesoporous Mater., 35-36 (2000) 349.
[ 16] I. Halasz, S. Kim, B. Marcus, J. Phys. Chem. B, 105 (2001) 10788;
[17] I. Halasz, S. Kim, B. Marcus, Molecular Physics, 100 (2002) 3123.
[ 18] I. Halasz, M. Agarwal, B. Marcus, B. Cormier, Microporous Mesoporous Mater., submitted.
[19] D. W. Breck, Zeolite Molecular Sieves, Krieger Publ. Co., Malabar, FL., 1984.
[20] www.pqcorp.com; www.zeolyst.com.
[21 ] E. Senderov, R. Hinchey, I. Halasz, US Pat. Appl. Publ. 2003152510 (2003).
[22] I. Halasz, M. Agarwal, E. Senderov, B. Marcus, Appl. Catal. A, 241 (2003) 167.
[23] I. Halasz, M. Agarwal, E. Senderov, B. Marcus, Catal. Today, 81 (2003) 227.
[24] I. Halasz, M. Agarwal, E. Senderov, B. Marcus, Stud. Surf. Sci. Catal., 145 (2003) 435.
[25] CRC Handbook of Chemistry 77th Edition, Ed. by D. R. Linde, CRC Press, Boca Raton, 1996-97.
[26] A. Saito, H. C. Foley, AIChE journal, 37 (1991) 429.
[27] V. M. Starov, J. Coll. Interf. Sci., 270 (2004) 180.
[28] S. Bordiga, P. Ugliengo, A. Damin, C. Lamberti, G. Spoto, A. Zecchina, G. Spano, R. Buzzoni,
L. Dalloro, R. Rivetti, Top. Catal., 15 (2001) 43.
[29] V. Bolis, S. Bordiga, C. Lamberti, A. Zecchina, A. Carati, F. Rivetti, G. Spano, G. Petrini,
Langmuir, 15 (1999) 5753.
[30] G. W. Skeels, D. W. Breck, Proc. Sixth Int. Zeol. Conf., Reno; Ed. by D. H. Olson and A. Bisio,
Butterworths, Surrey, UK, 1983, p. 87.
[31 ] H-M. Kao, YC. Chen, C-C. Ting, P. T. Chen, J-C. Jiang, Catal. Today, 97 (2004) 13.
[32] A. Zecchina, S. Bordiga, G. Spoto, L. Marchese, G. Petrini, G. Leofanti, M. Padovan, J. Phys.
Chem., 96 (1992) 4991.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka, N. Zilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 655
A Born Oppenheimer molecular dynamic (BOMD) approach was chosen to study the
interaction of water molecules with Cu-FAU models. We have compared the reactivity of Cu ~
and Cu H with the zeolite and compared the results to those calculated for Na. This study
shows a decrease of the coordination with time for Cu ~ whereas there is not a significant
change for Cu ~l. BOMD shows also that, in the presence of water, Na + may lead to easier
cation exchange than for transition metal cations.
1. I N T R O D U C T I O N
Transition metal ions exchanged in zeolites constitute a class of very efficient catalysts for the
selective catalytic reduction (SCR) of NO• by NH3 or hydrocarbons [ 1]. Among them, Cu I~-
faujasite (CulI-FAU) is recently operating in industrial processes for stationary sources
applications [2]. It has also been shown that the catalysts exhibit different activity, depending
on the cocations incorporated in the dehydrated zeolite. For instance, when Na + is exchanged
for Ba 2+, the Cu-FAU is more efficient. This increase of activity was attributed to an increase
of accessible Cu in the presence of Ba 2+ instead of Na +, because Ba 2+ occupies the sodalite
and prism positions that would otherwise be occupied by Cu 2+. More generally, the activity of
the catalysts depends on the location, the degree of cation exchanged and on the nature of
cation in zeolites. Adsorption properties and accessibility to reagents in zeolites are also
affected by the nature and by the location of extraframework cations. Experiments showed
that depending on the amount of water added to the catalysts, an increase or a decrease of the
catalysts activity can be observed. In particular an activation of catalysts in the SCR of NO•
in FAU, has been attributed to the presence of a small amount of water [1]. In contrast, a
deactivation of catalysts in the SCR of NOx, in ZSM-5, has been attributed to the presence of
water molecules [3]. Different explanations for this deactivation have been proposed
(dealumination, formation of CuO species, etc ...). Recent experiments concluded that the
deactivation of the catalysts may be related to the migration of the metallic cations, in the
presence of water [4].
It has been proposed that in a fully hydrated Cu-ZSM5 Cu 2+ exhibits an hexahedral
coordination which may have partial bonding to framework oxygen atoms [5]. For ZSM-5, it
656
has been also suggested, from EPR and EXAFS experiments, that hydrated Cu t coordination
is decreased to 2-3 [6] and that hydrated Cu II has a 4-5 coordination [7], with part of the
coordination due to water. Other experiments showed that the addition of ligands may change
the cation position: for instance, it was shown that addition of CO in a dehydrated CuI-FAU
moves Cu + at site II' toward the supercage (Fig. 2.) [8]. These observations indicate that the
position of a cation may evolve during a reaction even if it is generally assumed that the
positions of the extraframework cations are related to the A1 positions in the framework [9-
11]. A detailed knowledge of the evolution in time of the [reactants(or products)-cations]
interactions during the catalytic process is thus of particular interest. A time-dependent
simulation method, ie. molecular dynamics, can be very useful for such dynamical
investigations at the molecular level given that the time scale of this evolution remains in the
limits of the dynamics simulation. A Born Oppenheimer Molecular Dynamic (BOMD)
approach was chosen to study the interaction of water molecules with Cu-FAU and to
investigate the behaviour of Cu r and Cu ~ cations in their presence. Indeed, a dynamical
approach has to be preferred to a static optimization when a large number of degrees of
freedom govern the molecular motion, although a large sampling of the space of
configurations is then necessary, requiring multiple trajectories.
2. C O M P U T A T I O N A L METHODS
2.1. Models
Four models (Fig. 1.) of increasing sizes and with different terminal atoms have been used:
the simplest model A includes 2 S i and 1 AI, i.e. 3 tetraedra; model B models one 4-member
ring with terminal OH; Models C and D are built from the 3 four-member ring pattern which
connects two 6-member rings (site II), characteristic of the faujasite supercage. When the
models include 2 framework AI, these 2 AI are compensated by Cu ~I, or one A1 is
compensated by Cu ~, and the other one is compensated by a proton. In this later case, they are
models of the CulH-FAU catalyst, after the reduction step in the SCR of NO [ 1]. Model D is
similar to model C, enlarged with OSiH3 chains instead of terminal H on the two A1 and on
one Si.
2.2. Methodology
The clusters were cut into a solid structure of FAU which was obtained by Molecular
Mechanics (MM) optimization, using periodic boundary conditions and a random distribution
of AI for a Si/AI ratio of 2.5 (Cerius2, [12] cvff-aug-ionic force field). The MM optimization
has been performed using Na + parameters. The clusters were then saturated with H atoms
used to saturate the dangling bonds. During the BOMD simulations, the terminal H atoms
have been kept fixed in order to take into account the constraints of the solid. The calculations
were performed using the DFT methodology and the deMon code [13] using the PBE
exchanged-correlation functional [ 14].
We used DZVP basis sets for all the atoms. In Huzinaga's notation their contraction
patterns are (63321/531/41 *), (6321/521/1"), (621/41/1") and (41/1 *) for Cu, Si and (Al=O)
and H, respectively. The associated auxiliary basis sets used to fit the density were A2* (up to
/-4) for Cu, AI and Si, and A2 (up to/--2) for other atoms.
The dynamics was run with time steps of 0.5 fs, using the velocity-Verlet algorithm
[15] and the Berendsen thermostat [16]. The temperature of the dynamics was fixed at 300K.
657
The results reported correspond to equilibrated trajectories of 1.0-3.0 ps, with the shorter time
for the C and D models and those containing Cu ~1.
,%
B
C D
Fig. 1. Cu-FAU models (small balls are H, medium balls are O, large balls are Si, Cu and A1 are
labeled)
3. R E S U L T S AND DISCUSSION
Table 1
Average Cu I ion - oxygen bond distances (fluctuations) (A) for models A-D in the presence
of water molecules
Model A Model B Model C Model D
(8 waters) (4 waters) (7 waters) (4 waters)
Cu - O(zeolite) 2.0(0.09) 2.2 (0.12) 2.0 (0.07) 2.1 (0.06)
2.6 (0.30) 3.0 (0.23) 2.7 (0.25) 2.9 (0.17)
r2
,,
Multiple starting geometries have been used for the dynamics. For all models, the
simulations led to equilibrated structures including two quite different bonds between Cu and
the zeolite oxygen atoms: one short bond of 2.0-2.2 A, with a small fluctuation (<0.12 A) and
a much longer bond of 2.6-3.0 A, with a much larger fluctuation of 0.25-0.30 A.
The role of the water molecules, whatever is the number 4-8, has thus been to reduce
the coordination of Cu ~ with the zeolite, dissymmetrizing the bonding with one strong bond
kept unchanged and the other one weakened. Remarkably, for the models containing 7 or 8
waters, i.e. models A and C, only one water molecule also bonds strongly to Cu I with a bond
length of 2.0-2.1 A and small fluctuations (0.07-0.13 A). All the other water molecules are
moving at a much larger distance from Cu, forming water clusters, with varying shapes but
with very comparable potential energies. The large amplitudes of the water movements are
ascertained by the high values of the calculated fluctuations (see Table 1). In the case of
models B and D, in which only 4 water molecules are incorporated, the description is slightly
different, with 2 water molecules in the first coordination shell of Cu I.
We can thus conclude from a comparison of the results shown in Table 1 that the
presence of water modifies substantially the Cu ~ - zeolite bonding, weakening it and thus
allowing the incorporation of one (two) water ligand(s). During the dynamics, the Cu position
fluctuates, but this fluctuation is not very large, since the Cu-A1 and Cu-Si distances with the
closest framework elements have average values of 2.70 and 2.76 A, respectively, with small
fluctuations of 0.02 and 0.025 A (model C).
These results are in agreement with XANES experimental results for Cu-ZSM5, which
conclude to coordination of 2 for Cu ~, in the hydrated zeolite, as well as static optimization of
ZSM5 clusters [21].
Since the bond energies involving Na + are lower than the bond energies involving Cu ~,
we have studied the BOMD of Na+-FAU with H20 in order to see if, in the same conditions,
Na + remains attached to the zeolite or if it can be considered as entirely solvated with water.
When Cu ~ is replaced for Na + in model A, the structures are quite different. Indeed, the
equilibrium distances Na+-O without the presence of water are already larger (2.25 A). We
have considered a model containing 12 H20 because the water sphere was to loose with only
8 H20. The MD simulation of the system, including now 12 water molecules, leads to an
average structure where the distances between Na + and the zeolite oxygens remain equivalent
but increased to 2.50 A, 3 water molecules being also incorporated in this first shell of weakly
coordinated ligands at 2.7, 2.9 and 2.9 A. All the other water molecules fluctuate at much
larger distances (between 3.5 and 5.6 A from Na+). Comparison with NaC1 solvated in 13
water molecules shows that the first shell of neighbors is at a similar distance (2.5-2.8 A) in
these two systems, with 4-5 waters in the first solvation shell for NaC1 in water or 2 zeolite
oxygens and 3 waters for Na-FAU in water. This comparison underlines the difference of
water effects between Cu-FAU and Na-FAU, where Na + appears to be much more mobile
than Cu ~.
This result is not incompatible with conclusions drawn from Monte Carlo simulations
of Na-FAU, which suggest, at high water loading, a correlated motion of water molecules
(from the supercage into the sodalite cages) and of the sodium ions from sites I to I', inducing
rearrangements of the cations in the other sites [22].
660
Cu ~ at site III is more stable when situated in an ideal position [20]. Model B has thus
been used to study hydration of Cu H at site III, with replacement of both Cu ~ and H + for Cu I~
and addition of 4 water molecules (Fig. 4.).
Without water, the equilibrium structure of this model includes 3 short bonds between copper
and the zeolite oxygens (2.0 A) and one longer bond (2.6 A).
In the presence of 4 waters, the average equilibrated structure contains still three short bonds
of 2.2 A, with small fluctuations (maximum 0.10 A) between Cu ~I and the zeolite oxygen
atoms. The interaction with the fourth zeolite oxygen atom has become very weak (Cu-O at
3.4 A), whereas 2 water molecules have entered the first shell of ligands at 2.0 and 2.1 A. The
fluctuations of these five close oxygen atoms are lower than 0.10 A. The two other water
molecules move at average distances of 3.3 and 3.8 A.
It is clear that, in this case, the strong metal-zeolite interaction is not substantially
modified by the addition of a few water adsorbates, although two of them enter into the Cu ~
coordination shell. Work is in progress for an ideal Cu H associated with a larger model of the
zeolite framework, i.e. model D. Preliminary results for this large model with 4 waters
corroborates the results reported above, i.e. a coordination of about 5 for Cu I~ (3 bonds with
the zeolite oxygens and 2 with waters). Further work including 8 water molecules will allow
us to analyze to which extent a water cluster can be more efficient than isolated waters in
increasing the mobility of the Cu ~ cation, within the dynamics time scale.
4. C O N C L U S I O N S
- there is a large difference between the behaviours of Cu I and Na + in the presence of water,
showing that the mobility of hydrated Na + may lead to easier cation exchange than with Cu I,
which forms stronger bonds with both the zeolite oxygens and waters.
- whereas the average coordination of Cu ~ with the zeolite framework decreases in the
The incorporation of water in the first shell of Cu ligands suggests that the mechanism
involving the oxydo-reduction of the Cu cations in zeolite may be half way between
homogeneous and heterogeneous catalysis, although the role of the zeolite framework has
also to be considered [20,23-24].
REFERENCES
JDepartemento Engenharia Quimica, Av. Rovisco Pais, Instituto Superior Tecnico, 1049-001
Lisboa, Portugal.
A new infrared spectroscopic method is proposed for the characterisation of zeolites basicity.
NO2 disproportionation on alkaline zeolites is used to generate nitrosonium (NO+) and nitrate
ions on the surface, and the infrared vibrations observed are very sensitive to the cation
chemical hardness and to the basicity of zeolitic oxygen atoms. The influence of residual
water is described. The method has been applied to X and Y zeolites with Li, Na, K, and Rb
cations. The method is very easy to use and gives a linear response with chemical hardness of
the cation.
1. INTRODUCTION
The characterisation of basicity in solids remains a problem. In alkaline zeolites, basicity has
been linked to the electron density on the oxygen atom, to the nature of the alkaline cation, to
its size, ionization potential, and more recently to its chemical hardness [ 1,2]. Measuring the
basicity by catalytic reactions is always hampered by additional parameters, such as water
scavenging properties or transformation of the formed intermediate in a more stable
compound, or by a reaction not directly linked to the basic character of the solid.
Spectroscopic methods are very often used, but the choice of the probe molecule remains
difficult [3]. We propose a new infrared method using NO2 disproportionation. NO2
disproportionation is known to lead to NO + (nitrosonium) ions in exchange position on acidic
zeolites [4,5], where it is in close interaction with the framework oxygen atom. The frequency
of the v(NO +) vibration band is uniquely sensitive to electron density on that basic oxygen
atom [6]. We show here that this concept can be used to characterise basic zeolites. The
frequency of the v(NO +) vibration is a unique and convenient tool to characterise the basicity
of alkali-cation exchanged zeolites.
664
2. EXPERIMENTAL PART
We performed this reaction on a series of cation exchanged X (Si/Al = 1.3) and Y zeolites
(Si/AI = 2.5), detailed in Table 1.
Infrared transmission spectra were recorded on self-supporting wafers (2 cm 2, 10 to 15 rag,
prepared by application of a pressure of 5 . 107 Pa during 10 rain). The wafers were placed
into a quartz cell allowing heating of the sample as well as introduction of known quantities
of gas. The catalysts were activated under vacuum (10 -6 torr) by slowly heating (1 K.min -~
with a 1:30 h dwell at 393 K) up to 723 K, staying at this temperature for 5 h. The spectra
were recorded under vacuum, at room temperature. The spectrometer used was a Nicolet
Magna 550 at 4 cm -1 optical resolution and one level zero filling.
Table 1
Zeolite samples (values obtained from elemental analysis).
All calculations were done using Gaussian98 [7] with B3LYP [8,9] as exchange-
correlation functional in combination with a SDD basis set, i.e. D95 basis set [10] or first row
elements and a Stuttgart-Dresden effective core potential [11] for the remainder of the
periodic table. The basis sets for the atoms of N204 and the atoms that coordinate with the
cation are augmented with d and p polarization functions. After geometry optimization
infrared spectra are calculated analytically and frequencies are scaled with the scaling factor
proposed by Scott and Radom [12].
665
3. N I T R O S O N I U M ION F O R M A T I O N
Fig. 1. spectra recorded after introducing increasing amounts of NO2 on KX sample (0.11, 0.36, 0.75,
1.41 and 2.75mmol/g). v(NO +) and nitrate bands (around 1900 cm -~ and around 1400cm -~, resp.)
appear immediately after introduction of the first dose. v(OH) vibration bands at 3700, 3660, 3630 and
3580 cm -~ are also formed.
The reaction is simply performed by introducing NO2 in the infrared cell containing the
activated sample. When the sample is dehydrated thoroughly, v(NO § vibration bands appear
immediately (Fig. 1). NO2 disproportionates to form the nitrosonium cation and a nitrate
anion, adsorbed on the surface. A complex is formed with the alkaline cation and the basic
oxygen atom of the framework [6].
2 NO2 + ZO -M+[ZO - NO + NO3- M +]
The v(NO +) vibration band, between 1900 and 2200 cm -~ is comparable to the one
detected when NO + is in exchange positions in an acidic zeolite [4,5]. The nitrosonium ion is
thus in strong interaction with the basic oxygen atom, and is very sensitive to its electron
density.
35
927 3O
Na ,~ /
,/
/
i
> 25
N ..,--- u
.I.
K
20 /
R b ~/
.,,~.
15
10 i
2250 2150 2050 1950 1850 1850 1900 1950 2000 2050 2100
Fig. 2. (Left) Infrared spectra after NO2 disproportionation at low coverage (3-5 molecules per u.c.)
over M+X zeolites in the v(NO +) vibration range. (Right) relation between the chemical hardness of
the cations [14] and the v(NO +) wavenumbers.
The basic strength of a zeolite has been linked to the electronic charge density on the
oxygen atoms (as determined by Sanderson's electronegativity equalisation principle [15] for
fully exchanged zeolites) and is expected to be in the following order: RbX > KX > NaX >
LiX due to the ionisation potential of the alkaline metal [1]. In our experiments, the v(NO +)
frequency is indeed correlated with that value, but it is even better correlated to the hardness
of the cation [2] (as shown in fig. 2).
Hard atoms or molecules are more difficult to polarize than soft atoms or molecules: in
fully exchanged zeolites, harder alkaline cations (Li + and Na +) induce a smaller electronic
charge on the framework oxygens than softer cations (K + and Rb +) [16]. Here, we observe a
linear correlation between the frequency of the v(NO +) vibration band and the hardness of the
cation [14]. As the charge on the framework oxygens of Rb-exchanged zeolite X is higher
than that on the oxygens of a Li-exchanged zeolite X, the NO + vibrational frequency is
inversely proportional to the charge and thus to the basic strength of the oxygen atoms.
1377 1365
r-q
RbXI 0.2 abs.u. ~ ~
9 Large split 40 z
v 120
35 p:,
100 .......... i .,.,-'"
KX
N a +....... ....... ao ~z
80 Rb §
,."IN" . , . / / "
,....... , / 2s ~
~ 6o I . ,.'"" , S =,._ 20 ~
NaX_ z
~ 40
, Narrow split 10 ~
. . . . . . . .
K9
3
I . . . . I . . . . m . . . . I . . . . 9 9 9 I
1550 1500 1450 1400 1350 10 15 20 25 30 35 40 45 50
Wavenum bers (crn -~) qt V e -~
Fig. 3. Infrared spectra after NO2 disproportionation over M + X zeolites in the v~NOS) vibration range
(left) and evolution of the nitrates split with the hardness of the cations
RbY
I 1 abs. u.
4O Li §
_ .d)
35
,--"
30
,._.,.'"- Na +
KY _ 25
. _..-"'" ii
tll
> 20 K + .---
o.,.-
N ='15
l-
10 Rb §
ki 5
I . . . . . . . . . ! . . . . . . . . . . I . . . . . . . . . I ...... 0
Fig. 4. (Left) Infra red spectra after NO2 disproportionation at high coverage over M+Y zeolites in the
v(NO +) vibration range. (Right) relationship between the cation chemical hardness [14] and the
v(NO +) wavenumbers.
668
In our experiments, the NaY sample could not be completely dehydrated, and still
contained traces of water. The corresponding point clearly does not fit the correlation line in
fig. 4. The v(NO +) frequency was measured at 2090 cm l while it should appear around
2020 cm -~. We think that this is a clear indication of the modifications and perturbations
brought about by the presence of traces of water.
0"2 abs.u.
//" , . // . . . . . .
Fig. 5. Spectra recorded during adsorption of increasing doses of NO2 on NaY containing traces of
water (the spectrum of the activated sample as been subtracted for clarity). On top are the spectra
recorded in a first step (at 0.11, 0.36, 0.75, 1.41, 2.75 mmol NO2/g). Decreasing intensities are noted
around 1630 cm -1 and 3698 cm ~, where water bands are expected on NaY [18] (although no
corresponding bands could be detected there before NO2 adsorption), and v(OH) vibration bands for
acidic OH are increasing progressively at 3650 cm -~. This shows that water was present. Bottom
spectra: introduction of further doses of NO,-, (3.41, 4.07, 4.73, 5.39 and 6.05 retool/g), showing the
formation of NO + species when all of water is consumed (the negative band at 1630-1640 cm -~ does
not increase anymore).
669
Our interpretation is that NO2 reacts with water to from nitric acid. Surface water is thus
consumed, and nitric acid forms alkaline nitrates with the alkali cations, while acidic OH are
formed on the zeolite framework. The sample is then considerably modified. Further doses of
N02 will react with surface cations following the above mentioned reaction, but also with
acidic sites following Hadjiivanov's reaction [4].
3.6. Modelling
Modelling the interaction of N204, NO + and NO3-with alkali metal ions was done using
Density Functional Theory (DFT) calculations. Influence of the zeolite lattice was studied by
cluster type calculations, using a double 4 ring model representing the sIII' site inside
Faujasite, terminated by OH bonds (one A1 atom: Si/A1 = 5, sodium in the cation position).
The terminating oxygen atoms were kept fixed at their crystallographic position during
geometry optimization. During the modelling, and thus confirming our interpretation, NO +
and NO3 ions are formed (d(NiOl)= 1.128 A~ in NO +, d(NlO2) = 1.850 ~ in the nitrate,
q(NO +) = 0.454 e, q(NO3-) - -0.407 e). The frequency of the v(NO +) vibration (1972 cm "l)
agrees well with the experiment.
These calculations suggest that NO + and NO3- ions, formed after N02 adsorption on the
basic site of a zeolite, can be stable species and remain closely together, thus interacting with
the alkaline cation and the framework oxygen atoms of the zeolite. The complex formed can
account for the observed v(NO +) vibration frequency, and shows the influence of the
cation on electron density of the framework oxygen atoms. Further work is in progress to
fully understand the concept of basicity of alkali-exchanged zeolites in the framework of
DFT.
4. CONCLUSIONS
Disproportionation of NO2 on alkali cation exchanged zeolite (and possibly on other oxides
[19]) is a powerful tool for the characterization of the basicity of the framework oxygen
atoms. A complex is formed between the cation, the anionic framework oxygen atom, NO +
and NO3-. The frequency of the v(NO +) vibration together with the ionicity of the nitrate
(measured by the splitting in the nitrate doublet) are indicative of the chemical hardness of the
cation and of the basicity of the oxygen. Water traces remaining on the sample or in the cell
can perturb the measurement. The technique is shown to be an easy and efficient method for
measuring electron density on oxygen atoms and basicity of alkali-exchanged zeolites.
ACKNOWLEDGMENT
A.V. acknowledges a postdoc grant of the FWO-Flanders. Financial support came from the
GOA and WOG-FWO research grants.
REFERENCES
Laser hyperpolarized (HP) 129Xe NMR technique has been used to study the porous structure
of ordered purely siliceous MCM-41, SBA-15 mesoporous materials and mesoporous SiO2
thin films. The use of hyperpolarized xenon allows us to measure spectra at very low
concentration of xenon where xenon reflects mainly interaction between the adsorbed xenon
atoms and the surface. Variable temperature measurements allowed us to obtain information
on the heat of adsorption of xenon on the surface and to evaluate the chemical shift of xenon
in interaction with the surface.
1. INTRODUCTION
129Xe NMR spectroscopy is a powerful technique for the characterisation of porous solids
[1,2]. The main advantage of 129Xe NMR is the high sensitivity of the chemical shift to the
local environment of the Xe atoms. However, application of thermally polarised 129Xe NMR
to materials is often hampered by a relatively weak signal due to low concentration of
adsorbed xenon and long relaxation times (this is particularly the case for mesoporous silica
[3]). An increase in sensitivity of several orders of magnitude (x 104) can be achieved by
using optical pumping techniques for the production ofhyperpolarised (HP) xenon [4]. Recent
advances including HP Xe delivery into MAS probes [5] and circulating continuous flow
systems [6] have made it very attractive for materials applications [7] and have allowed to
probe very small quantities of matter [8].
Since 1991, great interest has been focused on the new class of mesoporous materials,
named MCM-41[9] and SBA-15 [10] prepared as powder or as thin films silica [11] and non
silicate [12] oxide.
In this article, we focus on the use of hyperpolarized xenon NMR spectroscopy to
understand the organisation of these mesoporous materials.
2. EXPERIMENTAL
2.1. Materials
The synthesis and characterization of MCM-41 used in this work has been described
elsewhere [12]. The powders with template were calcined at 773 K in the flow of oxygen.
SBA-15 mesoporous solids were synthesized by using tetraethyl orthosilicate (TEOS), and
triblock poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) (EOz0PO70EO20)
672
Pluronic P-123 copolymers. The synthesis conditions were described elsewhere [14]. The
structural parameters of the samples under study are presented in Table 1.
Table 1
Structural parameters for the studied MCM-41 and SBA-15 samples.
Sample SBET(m2/g) Pore diameter(nm)
MCM-41 1065 3.9
SBA-100~ 993 6.7
SBA-60~ 938 4.2
The mesoporous silica thin films having 2D hexagonal, 3D hexagonal and 3D cubic
structures were prepared via the sol-gel chemistry process using CTAB as surfactants [8]. The
thin films were deposited on a glass substrate via the dip-coating method. While the substrate
was withdrawn, evaporation took place leading to a self assembly-condensation process.
2.3. NMR
All the spectra shown in this paper were recorded on a Bruker AMX300 spectrometer
operating at 83.02 MHz for 129Xe. Prior to the NMR measurements, MCM-41 and SBA-15
samples have been compressed at 380 MPa to pellets of ca. 0.2 mm thickness.. For flow
experiments, the pellets were broken into smaller pieces that would fit inside a 10 mm o.d
NMR tube equipped with two Young valves. The samples were subsequently evacuated at
573 K for 12 hours. For the supported films, six slices were cut into 0.3xl.5 cm 2 pieces,
which represent about 0.1 mg of SiO2 or TiO2 and placed into the NMR tube. 8-1024 FIDs
were accumulated with 10 laS (rt/2) pulses and 1-30 s delays. The chemical shifts are
referenced to the signal of xenon gas at 10 torr.
This term is proportional to xenon density, so it shifts to higher values (lower field) with
increasing the xenon concentration.
In the presence of strong adsorption sites (SAS), one has to add a further contribution to
the observed chemical shift. At very low xenon concentration, each xenon spends relatively
long time on these SAS, which would increase the observed chemical shift. At higher xenon
loadings the relative population of xenon adsorbed on SAS decreases and the chemical shift
will decrease to the values characteristic for the rest of the surface.
95
SBA-60~
90
85
80
E
Q..
Q..
MCM-41
70
65
60
SBA.100o C
Fig. 1. Variation of 129XeNMR chemical shift at 298 K vs. xenon pressure for MCM-41 and SBA-15
synthesized at 60 and 100~ respectively.
Fig. 1 shows the variation of the observed chemical shift with the partial pressure of
adsorbed xenon on MCM-41 and SBA-15. Two samples of SBA-15 synthesized respectively
at 60~ and 100~ Chemical shifts of xenon in mesoporous materials depend on their pore
size. The bigger the pore diameter, the smaller are the observed s[~ifts [ 14].
For MCM-41, the 129Xe chemical shift is practically pressure independent, in agreement
with published results of the studies of xenon adsorption in mesoporous systems [18-20]. In
the case of SBA-15 sample synthesized at 100~ Fig. 1 shows that the chemical shift moves
to high field as xenon pressures increase which is typical of the presence of a small amount of
SAS in the pores. The most probable candidates for such sites are the micropores in the walls
of SBA-15, which presence is well established [21]. The length of such micropores should be
of the order of the wall thickness which for SBA-15 does not exceed several nm, so xenon
will be in fast exchange between micro- and mesopores. Similar behavior of the chemical
674
shift on xenon partial pressure for xenon adsorbed on mesoporous silica has been observed
recently [7b,22]. The slope of the chemical shift vs pressure relation for 129Xe provides further
information on the micropore size. The increase of the chemical shift vs pressure for SBA-15
synthesized at 60~ indicates that each micropore is large enough to accomodate more than
one Xe atom. This suggests that the micropores in this sample are larger than twice the kinetic
diameter of xenon (dxe = 3.96 A) along at least one direction.
Such a size would be in fair agreement with the genesis of the micropores, formed by the
polymerization of silica around loops of the poly(oxyethylene) chains, normally producting
up to 9 A from the surface of micelle of nonionic surfactant [23].
Fig. 2. HP 129Xespectrum at 163 K of xenon adsorbed on silica films with different hexagonal (Hex)
or cubic structures
The second signal corresponds to xenon adsorbed into the film ~ores. The chemical shift
measured reflects the structure of the film. Unlike conventional 2 Xe NMR techniques, HP
129Xe renders a highly sensitive detection of xenon local structures.
below 253 K due to a higher adsorption of xenon. Chemical shifts of TP xenon coincide
within the experimental error with those of HP xenon which indicate that, in our case, the
spectra truly represent the properties of the entire sample, despite the fact that HP xenon does
not probe all its volume.
180
140
~', 100
6O
20
fdm
0
130 180 230 280 330 380 430
T0~)
Fig. 3. Effect of temperature on the chemical shift of xenon adsorbed on MCM-41, SBA-15 (100~
and silica thin film (2D hexagonal).
For the supported silica film, the signals observed at room temperature and just below are
very different from those observed in the case of mesoporous powders [8], which are typically
much narrower. The reasons for such broad signals at room temperature can be a distribution
of pore diameters, a poor organization of the mesoporous phase and/or the exchange process
between xenon adsorbed in the pores and on the outer surface of the film. The first two
explanations can be ruled out since TEM and XRD data show a quite narrow distribution of
pore sizes and a high degree of ordering of the mesoporous structure [8]. As for the exchange
phenomena, they are known to be responsible for broadening of NMR signals of Xe in
mesoporous powders in the case of small particle dimensions (<10pm) due to the formation
of an interparticle secondary pore system. In our case, such a severe broadening is not
possible, since this mesoporous continuous film is free of interparticle pores. The observed
broadening could thus be attributed to the presence of shallow mesopores, in which xenon
atoms are not confined, but spend essential part of their time in the outer space. When the
temperature decreases, exchange with the gas phase drastically diminishes and the observed
chemical shift becomes fimction of the chemical composition of the surface and of the Xe-Xe
interactions in the adsorbed layer. The compared evolutions of the xenon chemical shift
values with temperature in MCM-41, SBA-15 powders and in film (Fig. 3) illustrate these
ideas.
To obtain quantitative information on xenon adsorption, let us consider the expression of
the observed chemical shift 6obs. Under the conditions of fast exchange, ~obs can be expressed
676
a s ~obs = Z ~ N i where 6, is the chemical shift of xenon adsorbed on site i and Ni is the relative
xenon population on site i.
Taking the chemical shift of xenon in the gas phase 6 o - 0, and assuming that xenon
adsorbed in the pores is described by Henry's law, one can obtain:
- ~" (1)
1+
fk~ RT
where 6a is the chemical shift of adsorbed xenon on the surface, D the pore diameter, kads the
Henry constant at temperature T, R the ideal gas constant a n d f a pore geometry factor.
The Henry constant, which is temperature dependent, can be expressed as:
kads(T) =
( Qaa]
exp - RT ) (2)
where Qads is the effective heat of adsorption for the pores under consideration and/co is the
pre-exponent factor which does not depend on the temperature.
Combining Eq. 1 and 2, the observed chemical shift can be expressed as:
1t.
(3)
D
1+
fko RTl/2exp(--RT-
Qads/
- +- Ln T - L n - + (4)
2 ~Sa f k o R RT
In the high temperature region, where the adsorption of xenon and hence, xenon-xenon
interactions in the adsorbed layer are low, and for the case where D and 6a do not depend on
temperature, it is possible to determine the heats of xenon adsorption by plotting [Ln(1/c~-
1/6a) + 0.5 Ln T] vs. 1/T. The values of heats of adsorption (-11.53,-17.3 and 9.1 kJ,
respectively for MCM-41, SBA-100~ and SBA-60~ indicate that, in all cases, physical
adsorption takes place.
677
4. C O N C L U S I O N
This paper clearly shows that the porosity of silica thin films can be probed directly by using
HP 129XeNMR. The evolution of the 129Xe NMR chemical shift with pressure allows
pointing out the presence of microporosity in SBA-15 materials. Variable temperature
measurements allowed us to evaluate the heats of adsorption of xenon adsorbed on different
types of mesoporous solids.
REFERENCES
[1] T. Ito, J. Fraissard, J. Chem. Phys., 76 (1982), 5225. (b) J. Ripmeester, J. Am. Chem. Soc., 104
(1982), 289.
[2] (a) D. Raftery, B. Chmelka, Nucl. Magn. Reson. Basic Principles Prog., 30 (1994), 111.
(b) J.L. Bonardet, J.Fraissard, A. G6d6on, M.A. Springuel-Huet, Catal. Rev., 41 (1999), 117.
[3] T. Pietrass, J. M. Kneller, R. A. Assink, M. T. Anderson, J. Phys. Chem. B, 103 (1999), 8837.
[4] (a) B.C. Grover, Phys. Rev. Lett., 40 (1978), 391. (b) W. Happer, E.Miron, S. Shaefer, Van
Wingaarden, X.Zeng, Phys. Rev. A, 29 (1984), 3092.
[5] D. Raftery, E. Mac Namara, G. Fisher, C.V. Rice, J. Smith, J. Am. Chem. Soc., 119 (1997),
8746.
[6] (a) M. Haake, A. Pines, J. Reiner, R. Seydoux, J. Am. Chem. Soc., 119 (1997) 11711.
(b) R. Seydoux, A. Pines, M. Haake, J. Reimer, J. Phys. Chem. B, 103 (1999), 4629.
[7] (a) I. Moudrakovski, V. Terskikh, C. Ratcliffe, J. Ripmeester, L.Q. Wang, Y. Shin, G. Exarhos,
J. Phys. Chem. B, (2002), 106, 5938.
(b) A. Nossov, E. Haddad, F. Guenneau, A. G6d6on, PCCP, (2005) in press.
[8] A. Nossov, E. Haddad, F. Guenneau, C. Mignon, A. G6d6on, D. Grosse, C. Bonhomme,
F. Babonneau and C. Sanchez, Chem. Com., 21 (2002), 2476.
(b) D. Grosso, F. Cagnol, G. J. de A. A. Soler-Illia, A. Bourjois, A. Brunet-Bruneau,C. Sanchez,
Adv. Funct. Mater. 14(2004), 309.
[9] (a) C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli and J.S. Beck, Nature, 359 (1992),
710.
(b) C.J. Brinker, Y. Lu, A. Sellinger and H. Fan, Adv. Mater., 11 (1999), 579.
[10] D. Zhao, P. Yang, N. Melosh, J. Feng, B.F. Chmelka and G.D. Stucky, Adv. Mater., 10 (1998)
1380.
[11] D. Grosso, G.J. de A. A. Soler-Illia, F. Babonneau, C. Sanchez, P.A. Albouy, A. Brunet-
Bruneau, A. R. Balkenende Adv. Mater. 13 (2001), 1085.
[ 12] D. Grosso, F. Babonneau, P.A. Albouy, H. Amenitsch, A.R. Balkenende, A. Brunet-Bruneau, J.
Rivory, Chem. Mater. 14 (2002), 931.
[13] V. Fenelonov, V. Romannikov and A. Derevyankin, Microporous Mesoporous Mater., 28
(1999) 57.
[14] A. Galarneau, J. Cambon, T. Martin, L.C. de Menorval, D. Brunel, F. Di Renzo, and Fajula,
Stud. Surf. Sci., Catal., 142 (2002), 395.
[15] E.L. Crepaldi, G.J. Soler-Ille, D. Grosso and C. Sanchez, New J. Chem., 27 (2003), 9.
[16] B.C. Grover, Phys. Rev. Lett., 40 (1978), 391. (b) W. Happer, E.Miron, S. Shaefer, Van
Wingaarden and X. Zeng, Phys. Rev. A, 29 (1984), 3092.
[17] (a) D. Raftery, H. Long, T. Meersmann, P.J. Grandinetti, L. Reven and A. Pines,
Phys. Rev. Lett., 66 (1991) 584.
(b) A. Nossov, E. Haddad, F. Guenneau, M.A. Springuel-Huet, E. Haddad, V. Montouillout, B.
Knott, F. Engelke, A. G6d6on, C. Fernandez, Phys. Chem. Chem. Phys., (2005) in press.
[18] F. Cros, J.-P. Korb and L.Malier, Langmuir, 16 (2000), 10193
[19] C.J. Jameson, A.K. Jameson, H.-M. Lim, B.I. Baello, J. Chem. Phys. 100 (1996), 5977.
678
[20] E. Haddad, J.-B. d'Espinose, A. Nossov, F. Guenneau, A. G6d6on, Stud. Surf. Sci. Catal, 142
(2002), 1173.
[21 ] A. Galameau, H. Cambon, F. Di Renzo, R. Ryoo, M. Choi, and F. Fajula, New J. Chem., 27
(2003), 73
[22] I. Moudrakovski, V.Terskikh, C.Ratkliffe and J. Ripmeester, J. Phys. Chem. B, 106 (2002),
5938.
[23] J.N. Israelchvili, H. WennerstrOm, J. Phys. Chem., 96 (1992) 520.
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. 2;ilkov~iand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 679
A. A b r a h a m , a S.B. H o n g , b R. P r i n s , a a n d J.A. v a n B o k h o v e n a
alnstitute for Chemical and Bioengineering, Swiss Federal Institute of Technology (ETH),
8093 Zurich, Switzerland
1. INTRODUCTION
Zeolite MCM-22 ~ is one of the most complex and interesting structures. Template free MCM-
22 has an unusual crystal structure consisting of two independent channel systems, both
accessible through ten-membered rings. One of these pore systems contains two-dimensional
sinusoidal channels with ten-membered rings. The other consists of large super cages defined
by twelve-membered rings. The template-free MCM-22 structure contains hexagonal sheets
constructed by interconnecting cages through shared four-membered rings. Previous studies
indicate that there can be eight 14 or 1 3124
" different T-sites in MCM-22 depending on the
symmetry of the framework structure.
The templated MCM-22 framework structure contains hexamethyleneimine (HMI) as
organic structure-directing agent (SDA)between the hexagonal sheets. Calcination of this
material burns the hexamethyleneimine template between the sheets, thus allowing fusion of
the T-sites on each hexagonal sheet through silanol condensation to form a pillar-like T-O-T
linkage. This type of linkage could give rise to a local distortion of the structure compared to
the templated MCM-22 structure.
680
Various 2-4 groups studied the framework structure of zeolite MCM-22 using
multinuclear solid-state NMR, but there is no clear evidence for the framework structure of
both as-synthesized and template-free MCM-22. The present study gives more insight into the
aluminum distribution and framework stability of as-synthesized and physico-chemically
treated MCM-22 with varying Si/AI ratios.
2. EXPERIMENTAL
2.1. Preparation
A series of templated MCM-22 samples were prepared 5 with Si/A1 ratios of 15, 30,
and 60. Template-free MCM-22 (H form) was obtained by calcination of the corresponding
as-synthesized MCM-22 samples in air at 773 K for 6 h. Template-free MCM-22 (NH4 form)
was obtained by a carefully controlled step-wise calcination procedure 6 in the presence of
ammonia (NH3) and oxygen (02). H-forms of these samples were prepared by an open air
calcination at 773 K for 6 h. These samples in their H-form were treated in a flow of ammonia
at 393 K for 6 h to obtain the NH3(H) form of the template-free MCM-22.
2.2. Methods
Crystallinity and phase purity of the solids were determined by powder X-ray
diffraction using a Rigaku Miniflex diffractometer with Cu Ks radiation.
27A1 MAS and 3Q-MAS NMR experiments were carried out on a Bruker AMX-400
FT-NMR spectrometer at a magnetic field of 9.4 Tesla using a 4 mm CP-MAS probe-head.
The 27A1 chemical shifts were referenced to (NH4)AI(SO4)2" 12H20. The rf pulse length for the
27A1 MAS experiments was 0.6 gs, equal to a rd8 pulse and spinning frequency of 12 kHz and
a recycle delay of 1 s were employed. 27A1 3Q-MAS v'8 NMR experiments were performed
using a three-pulse sequence incorporating a z-filter 9 at a spinning speed of 12 kHz. The z-
filter 3Q-MAS sequence was used to obtain a pure absorption mode two-dimensional (2D)
contour plot with negligible phase distortion, which was obtained by equal jumps in the echo
and anti-echo coherence transfer pathways. The phase-sensitive 2D experiments were
conducted using the hyper-complex States procedure. 1~An r.f. field of 106 kHz was used for
the creation (0Q ~ +3Q) and the first conversion (+3Q ---, 0Q) pulses. For the last conversion
step (0Q ---, -1Q), which is the central transition selective soft 90 ~ pulse, an rf field of 24 kHz
was used. A 2D Fourier transformation followed by a shearing transformation 8'~ gave a pure
absorption mode 2D contour plot. A one-dimensional orthogonal projection of the 2D contour
plot along the F1 axis gave isotropic (~so) spectra devoid of the anisotropic second-order
quadrupolar interaction. From the centre of gravity of the different resonances, the second-
order quadrupolar parameter PQ and the isotropic chemical shift 6zso were calculated
according to the equations:
where coo is the Larmor frequency and CQthe quadrupolar coupling constant. Quantification
of the 27A1 resonances was done by computer fitting of 1D MAS spectra using MATNMR.
298i MAS NMR experiments were also carried out on a Bruker AMX-400 FT-NMR
spectrometer using a 7 mm double resonance CP-MAS probe-head. A spinning speed of 5
kHz was employed for all the experiments. A high-power decoupling pulse sequence
(HPDEC) with an rf pulse length of 2.5 gs and decoupling power of 1.35 db was used. A
681
relaxation delay of 30 s was used. The 29Si chemical shifts were referenced to octakis-
(trimethyl siloxy)silsesquioxane (Q8MS).
3. RESULTS
3.1. XRD
All the materials are highly crystalline and no reflections other than those from MCM-
22 zeolite were observed. The X-ray powder diffraction patterns of the samples showed that
the framework structure of as-synthesized MCM-22 is different to that of calcined MCM-22.
Table 1
27A1 3QMAS NMR Isotropic Chemical Shifts* and Relative
Intensities (brackets) of As-synthesized MCM-22 (Fig. 1, left and
Fig. 2)
Si/A1 Tdl Td2 Td3 Td4
15 62.5 (19) 57.0 (32.5) 56.5 (20.5) 49.0 (28)
30 62.5 (18) 57.0 (33) 56.5 (21) 49.5 (28)
60 62.5 (18) ._56.5(32) 56.0 (22) 48.5 (28)
Rounded to halves.
682
Si/A! = 60
30
15
Fig. 2.27A1MAS NMR spectra of as-synthesized Fig. 3.27A1MAS NMR spectra of calcined
MCM-22 with different Si/AI ratios MCM-22 with different Si/A1 ratios
27A1 3Q-MAS NMR spectra and the isotropic projections of calcined MCM-22 samples
are shown in Fig. 1 (right). The spectra are resolved in three peaks in the tetrahedral region
called, Tdl, Td2, and Td4. The fourth peak Td3, is not resolved in this case. In addition to
these three peaks, there are two peaks in the octahedral region, one of which is narrow and the
other broad. The isotropic chemical shift values are given in Table 2. A graphical analysis of
the 3Q-MAS spectra gave the quadrupolar parameter (Po) values of all the peaks which are in
the range of 1.5 to 2.5 MHz.
27A1 MAS NMR spectra of the (air) calcined MCM-22 samples are shown in Fig. 3. The
spectral shape in the tetrahedral range is different from that of the as-synthesized samples.
Deconvolution of the 1D MAS NMR spectra of the samples gave the relative intensities of all
peaks (Table 3). The intensity of the octahedral peaks decreases with increasing Si/A1 ratio, as
observed in the literature for zeolite beta. 12
Table 2
27A1 3QMAS NMR Isotropic Chemical Shifts and Relative Intensities (brackets) of H-
MCM-22 (Fig. 1, right and Fig. 3)
Si/A1 Ydl Td2 Yd3 Yd4 Oh (narrow) Oh (broad)
15 62.0 (14) 56.0 (42) - 49.5 (20) -0.5 (12) 4.5 (12)
30 61.5 (16) 56.5 (44) - 50.0 (23) -2.0 (10) 3.5 (7)
60 62.0 (17) 56.5 (45) - 49.0 (25) -2.0 (8) 3.5 (5)
30
30
Fig. 4.29Si MAS NMR spectra of as-synthesized Fig. 5.29Si MAS NMR spectra of calcined
MCM-22 with different Si/A1 ratios. High power MCM-22 with different Si/A1 ratios. High
proton decoupling was used. power proton decoupling was used.
Table 3
" 8'0" iO" 40" 2'0" ; Ppm - 8 0 - 6 ' 0 'iO :20 t) Ppm
Table 4
Relative Intensities (+ 5%) ofZVA1 MAS NMR Spectra ofMCM-22 (Si/AI=30)
Sample Td 1 Td2 Td3 Td4 Oh (narrow) Oh (broad)
As-syn 18 33 21 28 -
NH4 form 18 42 20 20 -
H form 16 44 - 23 10 7
H-form(cc) 15.5 46 - 25 9 4.5
NH3-H form 18 37 25 20 -
4. DISCUSSION
Previous 27A1 MAS NMR studies 3'4'15 of calcined MCM-22 showed the presence of
three peaks in the tetrahedral region. Lawton e t al. 3 assigned them to three groups of T-sites
considering the correlation between the Td A1 chemical shifts and the A1-O-Si angles
established by Lippmaa et al. ~2 A 27A1 3Q MAS NMR study confirmed the presence of three
peaks in the calcined MCM-22 structure. ~6 However, these three peaks were not assigned to
three groups of T-sites. The peak around 50 ppm was assigned to aluminums located on the
external surface or in large cavities on the external surface of the crystals, which were
removed by acid leaching. 27A1 MAS NMR studies 3'4'17 on as-synthesized MCM-22 showed
the presence of two peaks. Kolodziejski et al. attributed these two lines to the aluminums
located in two different pore systems. The present study using 27A1MAS and 3Q MAS NMR
evidences the presence of four peaks (Td l, Td2, Td3 and Td4) in the tetrahedral region for the
templated MCM-22 sample series with varying Si/A1 ratios. These four peaks are assigned to
four groups of combinations ofT-sites. The relative intensities of the peaks are about 2:3:2:3
(Table 1) for all the samples. Therefore, the distribution of aluminum on the crystallographic
T-sites is independent of the Si/A1 ratio. The T-O-T angles of the T-sites which give rise to
these peaks could be different from that of calcined MCM-22. The 27A1 MAS NMR peak
shape for as-synthesized and calcined MCM-22 is different, which is evidence for the
different local environments. A detailed assignment to individual T-sites is questionable based
on 27A1NMR because the aluminum occupancy on different T-sites is unknown.
The 29Si MAS NMR spectra peak patterns of as-synthesized MCM-22 samples are
different from those of calcined MCM-22. This confirms that changes occurred in the
framework of as-synthesized MCM-22 during the process of calcination to form H-MCM-22.
In the present study, the spread in chemical shift that was predicted by the orthorhombic space
group matched, but the peak positions, and intensity did not agree. 298i NMR does not clearly
confirm any of the by XRD predicted space groups, making an assignment of the peaks to
individual T-sites impossible.
5. C O N C L U S I O N S
REFERENCES
Ill M.E. Leonowicz, J.A. Lawton, S.L. Lawton, M.K. Rubin, Science, 264 (1994) 1910.
[2] G.J. Kennedy, S.L. Lawton, M.K. Rubin, J. Am. Chem. Soc., 116 (1994) 11000.
[3] S.L. Lawton, A.S. Fung, G.J. Kennedy, L.B. Alemany, C.D. Chang, G.H. Hatzikos, D.N. Lissy,
M.K. Rubin, H-K.C. Timken, S. Steuernagel, D.E. Woessner, J. Phys. Chem., 100 (1996) 3788.
[4] G.J. Kennedy, S.L. Lawton, A.S. Fung, M.K. Rubin, S. Steuernagel, Catal. Today, 49 (1999)
385.
[5] A. Corma, C. Corell, J. Perez-Pariente, Zeolites, 15 (1995) 2.
[6] P.J. Kunkeler, B.J. Zuurdeeg, J.C. van der Waal, J.A. van Bokhoven, D.C. Koningsberger,
H. van Bekkum, J. Catal., 180 (1998) 234.
[7] L. Frydman, J.S. Hardwood, J. Am. Chem. Soc., 117 (1995) 5367.
[8] A. Medek, J.S. Hardwood, L. Frydman, J. Am. Chem. Soc., 117 (1995) 12779.
[9] J.P. Amoureux, C. Fernandez, S. Steuernagel, J. Magn. Reson., 123 (1996) 116.
[10] D. States, R. Haberkorn, D. Ruben, J. Magn. Reson,. 48 (1982) 286.
[11] R. Ernst, G. Bodenhausen, A. Wokaun, Principles of Nuclear Magnetic Resonance in One and
Two Dimensions; Oxford University Press, New York, 1987.
[12] A. Abraham, S.-H. Lee, C-H. Shin, S. B. Hong, R. Prins, J. A. van Bokhoven, Phys. Chem.
Chem. Phys., 6 (2004) 3031.
[13] E. Lippmaa, A. Samoson, M. M~gi, J. Am. Chem. Soc., 108 (1986) 1730.
[14] M.A. Camblor, A. Corma, M.J. Diaz-Cabanas, Ch. Baerlocher, J. Phys. Chem. B. 102 (1998)
44.
[ 15] M. Hunger, S. Ernst, J. Weitkamp, Zeolites, 15 (1995) 188.
[16] P. M6riaudeau, A. Tuel, T.T.H. Vu, Catal. Lett., 61 (1999) 89.
[17] W. Kolodziejski, C. Zicovich-Wilson, A. Corma, J. Phys. Chem., 99 (1995) 7002.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. NachtigaU(Editors)
9 2005 ElsevierB.V. All rights reserved. 687
P. S a r v , M. Derewmskl , I. Heinmaa*
The aim of our study was to follow the changes in the state of boron and aluminum atoms in
B,A1-Beta (BEA) zeolites, caused by the postsynthesis treatments. 1D and 2D MQMAS NMR
were used to study the state of A1 and B. In the B,A1-Beta zeolite octahedral aluminum is
more easily reverted to the tetrahedral state by ion-exchange compared to the A1-Beta. The
applied treatments resulted in almost complete (80-90%) removal of boron atoms from the
framework. Results obtained indicate, that some of the T sites freed by the boron atoms could
be reoccupied by aluminum atoms. The extent of the process is related to the initial Si/(AI+B)
ratio. Aluminum is reinserted only into AI(IV)a positions. The concentration of AI(IV)b sites
decreases during the treatment. In the aluminum analog, the reinsertion process is
considerably lower.
1. INTRODUCTION
spectroscopy can resolve only two inequivalent T positions [10]. Based on the empirical
relationship between T-O-T angles and chemical shift, these lines are attributed to aluminum
sites Al(IV)a and AI(IV)b, comprising T-positions T3-T9 and T1,T2 respectively [10] .
Previously, a quantitative 27A1 MQMAS NMR and in-situ A1 K edge X-ray Absorption
Spectroscopy study on zeolite Beta, treated at different temperatures and steam content,
showed the quantitative reversal of framework aluminum from octahedral coordination
[10,11]. Bortnovsky et al. attribute this phenomenon to paired aluminum sites[5]. Van
Bokhoven et al. have shown, that only Al(IV)a sites are altered, claiming, that sites AI(IV)b do
not convert to octahedral aluminum and resist dealumination by steam at even 550~ [10].
In addition to aluminum, boron is one of the most frequent element incorporated in to
the zeolites. However, because of the low stability of the boron containing zeolites, the boron
atoms are nearly quantitatively removed from the framework during the thermal/hydrothermal
treatment followed by ion-exchange [12,13]. Zeolite A1-Beta can be boronated in an alkaline
solution, but boron atoms insert into the framework with more difficulty than aluminum
atoms [ 14]. Zones et al. have shown, that if using the same structure-directing agent, but
different trivalent metal atoms in the same concentrations, then the borosilicate system
produces SSZ-42 (IFR), while the aluminosilicate system yields Beta zeolite [15]. liB MAS
NMR spectrum of Beta zeolite shows only one narrow line for tetrahedral boron, which
indicates a presence of one type of sites [14]. At the same time SSZ-42 has two tetrahedral
boron sites [15]. Considering that Beta zeolite has two types of tetrahedral aluminum sites, so
one would expect two sites also for boron.
All these findings predict a very different behaviour of B,A1-Beta and A1-Beta zeolites
during thermal/hydrothermal treatment and the following ion-exchange. In our contribution
we would like to find out if boron atoms removal could influence the state of the aluminum
atoms and the process of the aluminum reinsertion and siting.
2. E X P E R I M E N T A L
B,Al-Beta and A1-Beta with different Si/B and Si/A1 ratios were synthesized hydrothermally
at 150~ using tetraethylammonium hydroxide (TEAOH) as a template [12]. The as-made
samples were calcined at 500~ and ion-exchanged with NH4NO3 solution at 90~ for 24
hours. The self-steaming was done by immersing the hydrated samples to an oven preheated
at 600~ and keeping them there for 2 hours. The samples were characterized with XRD,
SEM, FTIR and MAS NMR spectroscopy. Prior to NMR measurements all samples were kept
at 75% humidity for 48 hours.
~B, 27A1 and 29Si MAS NMR was applied to study the state of B, A1 and Si atoms in
the materials under study. The 29Si MAS NMR spectra were measured on Bruker AMX360
spectrometer (magnet 360MHz, 1H NMR frequency) using custom built MAS probe with 7
mm o.d. rotor (spinning speed approximately 5 kHz). 298i MAS NMR single pulse spectra
were recorded at 71.5 MHz, with a 3.5 (90 ~ pulse and a 30 s relaxation delay. 29Si spectra
show how the deboronation and dealumination affect the the silicon backbone of the zeolite.
The 27A1 MAS NMR spectra were measured in a 500 MHz frequency magnet (Bruker
AMX500 spectromemer) with a 10~ pulse, 0.2 s relaxation time and a 12 kHz rotation speed
using custom-made probehead with 3.5 mm o.d. rotor. 10000-20000 scans were accumulated.
27A1 frequency was and 130.3 MHz. KAl(SO4)2"12H20 was used as an external reference for
27A1 NMR experiments. A1 spectra were recorded to find out how aluminum is distributed
between framework (tetrahedral) and nonframework (octahedral) positions. The liB MAS
NMR single pulse spectra were measured with Bruker AMX500 spectrometer with a 15~
689
pulse, 1 s relaxation time and 12 kHz rotation speed using custom-made made probehead with
3.5 mm o.d. rotor. 1000-5000 scans were accumulated, lIB frequency was 160.46 MHz. The
external reference for l~B was NazB4Ov'10H20. From l~B spectra we got information about
the distribution of B sites between trigonal and tetrahedral coordination. 27A1 MQMAS NMR
spectra were measured with Bruker AMX600 spectrometer. The 3Q MAS NMR experiments
were performed with a three-pulse sequence using symmetrical coherence transfer pathways -
0>(+3)>0>(-1) [16]. Relaxation delays of 0. l s were used. Rotation speed was reduced to 11
kHz in order to get better excitation of multiple quantum coherences [ 17]. The first pulse had
the duration of 3.0 gs, the second pulse was 1.5 laS and the third pulse was 20 laS. The rf
strength of the third pulse was reduced to approximately 10kHz. 2D NMR signal was aquired
(1200 scans) with the States-TPPI method and a shearing transformation was performed by
employing a tl dependent first order phase correction in the mixed (ti, v2)-domain, after
constructing echo and antiecho signals [18]. From the 2D spectra the projections to the
isotropic chemical shift axis (v l) were extracted and subsequently simulated with two
gaussian lines to get the relative intensity of AI(IV)a and AI(IV)b sites. This approach is
justified, because the quadrupolar interaction parameters for both sites are practically the
same. So also the multiquantum excitation and detection efficiency is the same for both sites.
Isotropic projections of 2D MQMAS spectra give better resolution compared to the 1D MAS
spectra and therefore also the relative intensities of lines are more accurate. Quadrupolar shift
and isotropic chemical shift of these sites were determined from the 2D spectra using the well
known formulas [3]. The aim was to find out if there is any change in the local surrounding of
framework aluminum.
3. R E S U L T S
Table 1
The concentration of framework tetrahedral AI atoms before and after hydrothermal treatments
in aluminum Beta (Al-Beta) and boron-aluminum Beta (BA1-Beta) samples
sample tetrahedral A1 [x1016 atoms/mg] reinserted A1 [%]
as-made calcined ion-exch, ss+ion-exch.
BA1-Beta(7*) 6.5 6.8 7.4 7.6 12
BA1-Beta(10) 9.8 8.2 10.1 - 23
BA1-Beta(16) 16.2 13.5 - 17.6 30
BAl-Beta(27) 27.2 21.7 25.5 - 18
AI-Beta(10) 10.5 12.8 12.2 - -5
Al-Beta(22) 22.4 18.7 19.4 - 4
A1-Beta(49) 48.7 45.1 44.8 - -1
Note: the as-made material comprises also about 10 wt% of template.
* - number of tetrahedral AI atoms in as-made samples
We studied four B,AI-Beta and three Al-Beta samples with comparable aluminum content
(Table 1: as-made). Three B,Al-Beta samples had similar boron content, while BA1-Beta(27)
had two times less boron that the others (Table 2: as-made). In the as-made material the boron
was exclusively in tetrahedral environment giving one narrow line a t - 3 . 4 ppm (Fig. 1).
During the calcination most of the boron became three coordinated (broad line a t - 1 8 ppm)
and was washed out during the ion-exchange treatment. The remaining tetrahedral boron was
futher converted to trigonal coordination by self-steaming and washed out by ion-exchange.
690
The whole process can be monitored with ~B MAS NMR as for sample BA1-Beta(7) on Fig.
1. One may conclude that hydrothermal treatment and subsequent ion-exchange removes
boron almost quantitatively from the framework. The same holds for all boron-containing
samples (see Table 2).
self-steamed+ion-exchanged
self-steamed _ _ ....,'L
ion-exchanged
calcined ~ ~__
as-made U
(intensity divided by 4)
i i I I ~ l l f ~ l l l l l l I J I r l I l l f f l l ~ I r I f r I I I i l l , f ,
30 20 10 0 -10
(ppm)
Fig. 1. ~B MAS NMR spectra of sample BA1-Beta(7) before and after treatments.
Table 2
The concentration of framework tetrahedral B atoms before and after treatments
Sample tetrahedral B [x 1016 atoms/mg]
as-made calcined ion-exch, ss+ion-exch.
BAl-Beta(7) 7.9 3.1 2.3 0.6
BA1-Beta(10) 8.7 1.4 2.5 -
BA1-Beta(16) 7.1 1.2 - 0.1
BA1-BETA(27) 4.5 1.7 0.4
The 29Si MAS NMR spectrum of as-made BA1-Beta(7) gives two lines at -110 and-114 ppm
corresponding to Si(4Si) sites plus one line a t - 1 0 3 ppm corresponding to overlapping
Si(1A13Si) and SiOH sites (Fig. 2). Boron removal decreases the number of defect sites,
which the boron tetrahedra effectively are, and relaxes the framework. This manifests itself in
the narrowing of the lines corresponding to Si(3Si) sites- framework gets more ordered. At
the same time the intensity of the broad line at-103 ppm remains the same. Therefore we
may say, that aluminum is not removed from the framework. Si tetrahedra around it are all
randomly turned to accommodate the slightly bigger aluminum tetrahedra. Therefore the line
stays broadened and is not narrowed as the rest of the spectrum.
Fig. 3 gives the comparison between the 27A1 MAS NMR spectra of samples with low (A1-
Beta(10), BA1-Beta(10) and high (A1-Beta(49), BA1-Beta(27) aluminum content. One can
recognize two characteristic regions in the spectrum. Two lines around 55 ppm corresponding
to framework tetrahedral aluminum sites AI(IV)a and AI(IV)b, located in T3-T9 and T1-T2
positions respectively [10]. Plus two lines around 0 ppm corresponding to octahedral
aluminum. This octahedral aluminum is usually assigned to nonframework aluminum. In the
case of Beta zeolites, these octahedral aluminum sites can be inserted back into tetrahedral
positions by ammonium ion-exchange, which is seen also from the Fig. 3. The line at-55 ppm
691
corresponds to the first order spinnig-sideband from the satellite transition. Similar zero order
spinning sideband lies also under the main line at 55 ppm. The intensity of this sideband has
to be subtracted during the quantitative calculations, since it's intensity belongs to the satellite
transitions and not to the central transition. The results of the quantitative calculations are
gathered in Table 1 and Fig. 5. One can immediately recognize, that in the case of boron
containing samples aluminum is more easily reinserted (Table 1), and this reinsertion takes
place in site AI(IV)a in as seen also in previous studies [10]. For the low-aluminum A1-
Beta(10) sample there is even no octahedral aluminum formed after the calcination.
self-steamed+ion-exchanged
self-steamed
_•• ion-exchanged
.•• calcined
de
-80 -90 -100 -110 -120 -130 -140
(ppm)
Fig. 2.29Si MAS NMR spectra of sample BA1-Beta(7) before and after treatments.
ion-exchanged
. . . . card
as-made
j,-, . . . . . .
Fig. 3.27A1 MAS NMR spectra of samples A1-Beta(10), BAl-Beta(10), Al-Beta(49) and BA1-Beta(27)
before and after treatments.
692
/ ~ ion
_..] k. -exchanged
ned
j/ a~ade
65 60 55 50 65 60 55 50 65 60 55 50 65 60 55 50
(ppm) (ppm) (ppm) (ppm)
Fig. 4. Isotropic projections (vl)of the 2D 27A1 MQMAS NMR spectra of samples AI-Beta(10), BA1-
Beta(10), A1-Beta(49) and BAl-Beta(27) before and after treatments.
Further 2D MQMAS NMR studies gave spectra similar to the ones published by van
Bokhoven et al. [ 10]. For the sake of brevity we show only the isotropic projections of the 2D
spectra on Fig. 4. From there it is clear, that site AI(IV)a in BA1-Beta samples is preferentially
reinserted. The boron removal does not change the isotropic chemical shift of the lines: 58.5
ppm for the site AI(IV)a and 55.0 ppm for the site AI(IV)b. Therefore we may conclude that
the T-O-T angles around aluminum atom remain the same. The main relaxation and ordering
of the framework takes place in the aluminum free regions as confirmed by the 298i MAS
NMR (Fig. 2). The quadrupolar shift is about the same for both sites and changes from-2.0
ppm for as-made material to-0.6 ppm for ion-exchanged material. This allows us to say, that
both sites are detected to the same extent. Simulation with two gaussian lines gives us the
relative concentration of the sites. These numbers together with the absolute concentration
from Table 1 give the concentration changes for both sites during the hydrothermal treatments
(Fig. 5).
4. DISCUSSION
From Table 1 it is clear that the tetrahedral aluminum, converted to octahedral aluminum by
the calcination, is reinserted by the ion-exchange treatment. This takes place for boron-
containing BAl-Beta samples and for much greater extent than for A1-Beta samples. Although
as-made material shows lower concentration for the tetrahedral sites compared to the ion-
exchanged material, then one should bear in mind, that the template-free concentrations for
as-made material are about 10% higher. Results presented on Fig. 5 show that in BA1-Beta
samples the site AI(IV)a is preferentially reinserted while the site AI(IV)b loses its
concentration as a result of the treatment. In A1-Beta(49) and A1-Beta(10) samples the overall
concentration on tetrahedral aluminum decreases, while site AI(IV)b being more stable that
site AI(IV)a in complete consistence with previous studies by van Bokhoven et al. [10].
Sample A1-Beta(22) behaves similar to BA1-Beta samples, but to considerably lesser extent.
We may conclude, that boron removal during calcination and subsequent ion-exchange
facilitate the reinsertion of the AI(IV)a sites and hinder the reinsertion of AI(IV)b sites. Since
the concentration of AI(IV)a sites in ion-exchanged material is higher than in as-made
material and the tendency is even more pronounced by following self-steaming/ion-exchange
(Fig. 5 sample BAI-Beta(7); Table 1 sample BA1-Beta(16)), we may say that some of the
AI(IV)b sites are also converted to AI(IV)a sites. Our explanation to the reinsertion
phenomenon is, that boron removal creates defect sites ready to accept aluminum atoms and
693
25.0
• B,AI-Beta(27) • B,AI-Beta(27)
10.0 ~ : ~
B,AI-Bet a(10) < )1( B,AI-Beta(10)
O B,AI-Beta(7) o 5.0 O B,AI-Beta(7)
z
0.0 . . . .
Fig. 5. Selective reinsertion of the aluminum into A1 (IV)a framework positions, effected by boron
removal.
also makes the framework more flexible. Therefore aluminum is reinserted more easily. The
extent of the process is related to the Si/(B+A1) ratio and severity of the treatment, where
more metal atoms present in the initial material produce more pronounced reinsertion (Table
1 and 2). The selectivity of the reinsertion process may be related to the fact, that boron has
only one type o f T position in BA1-Beta zeolite, giving one narrow line in the ~lB MAS NMR
spectrum, and therefore creating one type of defect sites during removal (Fig. 1). For
comparison, in SSZ-42 zeolite boron has two nonequivalent T positions [15]. One should also
bear in mind, that the aluminum distribution is not homogeneous, because the relative
intensity of AI(IV)a and AI(IV)b peaks is far from 7:2, as expected and seen in the case of 29Si
MAS NMR s p e c t r a - the line a t - 1 1 0 ppm and the shoulder at -115 ppm. The T1-T2
positions are more populated by aluminum atoms, than the T3-T9 positions. In general, the
synthesis procedure may influence the distribution of aluminum and boron. Our goal was to
get the as-synthesized materials with similar aluminum distribution and to monitor the
changes after the postsynthesis treatment (Fig. 3 and 4). The differences between the boron
and aluminum distribution certainly deserve further studies.
5. CONCLUSIONS
We have shown that by diluting aluminum atoms with boron atoms in the framework of Beta
zeolite one can selectively increase the concentration of AI(IV)~ type aluminum sites in the
ion-exchanged material and decrease the concentration of AI(IV)b type sites, previously
believed to be more stable. During the hydrothermal treatment (calcination/ion-exchange/self-
steaming/ion-exchange) the boron atoms are almost completely removed from the sample.
ACKNOWLEDGEMENTS
This work has been supported by Estonian Science Foundation through grant No. 5936.
Authors also wish to thank Polish Academy of Sciences and Estonian Academy of Sciences
for support.
694
REFERENCES
[1] C.A. Fyfe, J.H. O'Brien and H. Strobl, Nature, 363 (1987) 6110.
[2] L. Frydman, J.S. Harwood, J. Am. Chem. Soc., 117 (1995) 5367.
[3] C. Fernandez, J.P. Amoureux, Chem. Phys. Lett., 242 (1995) 449.
[4] P. Sarv, C. Fernandez, J.-P. Amoureux, K. Keskinen, J. Phys. Chem., 100 (1996) 19223.
[5] O. Bortnovsky, Z. Sobalik, B. Wichterlovfi, Microporous Mesoporous Mater., 46 (2001) 265.
[6] E. Bourgeat-Lami, P. Massiani, F. Di Renzo, P. Espiau, F. Fajula, T. Des Courieres, Appl.
Catal., 72 (1991) 139.
[7] L.W. Beck, J.F. Haw, J. Phys. Chem., 99 (1995) 1076.
[8] C.A. Fyfe, H. Strobl, G.T. Kokotailo, C.T. Pastzto, G.E. Barlow, S. Bradley, Zeolites, 8 (1988)
132.
{9] J. Stelzer, M. Paulus, M. Hunger, J. Weitkamp, Microporous Mesoporous Mater., 22 (1998) 1.
[10] J.A. van Bokhoven, D.C. Koningsberger, P. Kunkeler, P. van Bekkum and A.P.M. Kentgens, J.
Am. Chem. Soc., 122 (2000) 12842.
[11] J.A. van Bokhoven, A.M.J. van der Eerden and D.C. Konigsberger, Stud. Surf. Sci. Catal., 142
(2002) 1885.
[12] M. Derewinski, F.D. Renzo, P. Espiau, F. Fajula, M.-A. Nicolle, Stud. Surf. Sci. Catal., 69
(1991) 127.
[13] M. Derewinski and F. Fajula, Appl. Catal. A, 108 (1994) 53.
[ 14] C. Yang, Q. Xu, C. Hu, Mat. Chem. Phys., 63 (2000) 55.
[15] S.I. Zones, S-J. Hwang, Microporous Mesoporous Mater., 58 (2003) 263.
[16] J.P. Amoureux, C. Fernandez and S. Steuemagel, J. Magn. Reson., A 123 (1966) 116.
[17] D. Massiot, J. Magn. Reson., A 122 (1996) 240.
[18] D. Massiot, B. Touzo, D. Trumeau, J.P. Coutures, J. Virlet, P. Florian and P. Grandinetti,
J. Solid State NMR, 6 (1996) 73.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~iand P. NachtigaU(Editors)
9 2005 ElsevierB.V. All rights reserved. 695
K. Morishige
1. INTRODUCTION
a first-order capillary condensation. To examine the effects of pore shape and connectivity on
capillary condensation and pore criticality, we performed measurements of a series of
adsorption-desorption isotherms of nitrogen and other simple gases onto MCM-41, SBA-15,
MCM-48, KIT-6, and SBA-16 templated nanoporous materials over a wide temperature range
between a bulk triple point (Tt) and To.
2. EXPERIMENT
All materials were prepared according to the literatures [5-7]. Adsorption isotherms were
measured volumetrically on a homemade semiautomated instrument equipped with a Baratron
capacitance manometer (Model 690A) with a full scale of 25000 Torr [4].
Hysteresis loops are often used in determination of pore size distribution of the conventional
mesoporous materials under several assumptions concerning the pore structure and the nature
of adsorption and desorption branches. In most cases, it is assumed that the material is made
up of a collection of unconnected cylindrical pores.
~20
~15
10
0
1 1.5 2 2.5 3 3.5 4 4.5
logP(Torr)
Fig. 1. Temperature dependence of the adsorption-desorption isotherm of nitrogen onto MCM-41.
697
The nature of the adsorption and desorption branches can be examined by plotting the
condensation/evaporation pressure in a form of T ln(P/Po) against temperature over a wide
temperature range including both regions of irreversible and reversible isotherms [9]. The
principle of this method relies on the simple idea that the equilibrium phase-transition
pressures in the hysteretic isotherms would be obtained by the extrapolation of the plot for
reversible capillary condensation to lower temperatures. Fig. 1 shows the
adsorption-desorption isotherms of nitrogen on MCM-41 with unconnected cylindrical pores
of radius 2.2 nm. At lower temperatures, the isotherms exhibited hysteresis loops of type H 1
in the IUPAC classification [10]. When the temperature was increased, the hysteresis loop
shrank and eventually disappeared a t - 7 9 K (Th). The adsorption steps were still nearly
vertical above Th. This indicates that Th is lower than Tcp.
Fig. 2 shows the plots of the capillary condensation and evaporation pressures against
temperature for nitrogen on MCM-41. Here, the condensation and evaporation pressures were
determined at the midpoint of the adsorption and desorption branches, respectively, and P0 is
the saturated vapor pressure of the bulk liquid. At lower temperatures, T ln(P/Po) is the
difference of the chemical potential with respect to the bulk liquid. The plot for capillary
condensation forms an almost linear relationship over a wide temperature range spanning Th,
whereas the same plot for capillary evaporation breaks at Th. We examined further the
temperature dependence of the capillary condensation and evaporation pressures of argon,
oxygen, and carbon dioxide onto MCM-41, nitrogen onto SBA-15 with larger cylindrical
pores, and nitrogen, oxygen, argon, and carbon dioxide onto SBA-16 with cagelike pores. The
plot for capillary condensation always connected smoothly into that for reversible capillary
condensation measured above Th. On the other hand, the same plot for capillary evaporation
did not connect smoothly into that for reversible condensation above Th. Therefore, the
temperature dependence of capillary condensation and evaporation pressures of simple gases
indicates that the adsorption branch of the hysteretic isotherm corresponds approximately to
the true equilibrium, irrespective of the type of adsorbents. On the other hand, capillary
, ,,
-10 N2/MCM-41
-20 Ads(. )
~-30 Oes(o )
~' -40
-60 L Th . .
-80
-90 I I I I I
evaporation from the cylindrical pores and the cagelike pores is always delayed. Use of the
adsorption branch in pore size analysis is more appropriate than the desorption branch.
40
79.3K 84.4K 89.2K 93.8K
35 N2/KIT-6 99 l K 1 0
[ ~ I 109.2K 117.7K
30
25 120.8K
20
15
10
0
2.5 3 3.5 4 4.5
LogP(Torr)
Fig. 3. Temperature dependence of the adsorption-desorption isotherm of nitrogen onto KIT-6.
699
o
-5 N2/KIT-6
-lO
.-~ -15
-20
E- -25
-30 .• 0 (O)Des
-35
-40 I I I ,1 ' I
of KIT-6 with that confined to the unconnected pores of SBA-15, we measured the adsorption
and desorption scanning curves of nitrogen onto KIT-6 and SBA-15 at a liquid nitrogen
temperature. In the independent domain theory of adsorption hysteresis, a "pore domain" is
assumed as a region of porous space in which capillary condensation and evaporation occurs
at well-defined vapor pressure values designated by the notations Xl2 and x2~, respectively
[11 ]. Adsorption hysteresis arises by the fact that x~2>x2~ for every one of these domains. The
main idea of the theory consists in presuming that pore domains constitute autonomous
entities for which the filling with either liquid (condensation) or vapor (evaporation) takes
place in accordance to their particular x12 and x2~ values, and irrespectively of the state
(liquid- or vapor-filled) of their neighboring pore entities. Capillary condensation and
evaporation in truly independent domain porous solids develops progressively according to a
pore size distribution of the solid. Pore size distribution analysis by a gas adsorption method
is usually based on this theory. Figs. 5 and 6 show primary desorption scanning curves of
800 -
700 - N2/KIT-6
600
~0
":'. 500
~- 400
E 300
200
100 [
I I I
700
600 N2/SBA-15
~-500
~.400
[..
~-300
> 200
100
0 t I I I
0.5 0.6 0.7 0.8 0.9
P/Po
Fig. 6. Primary desorption scanning curves for N2 on SBA-15 at 77 K.
nitrogen on KIT-6 and SBA-15, respectively. In both solids, we obtained the results that were
just expected from the independent domain theory. When the maximum amount of adsorption
was decreased, the shape of the desorption scanning curve remained almost unchanged. In the
case of the unconnected cylindrical pores of SBA-15, an ascending process will consist in the
sequential filling of pore domains according to their diameters. When starting from a
liquid-saturated porous structure, condensate desorption will occur successively from larger to
smaller cylinders as the vapor pressure is decreased. In the case of the interconnected
channels of KIT-6, half the channels in a particle that are interconnected with each other
constitute a single domain. When the boundary scanning curves of both solids are compared,
however, it is found out that the adsorption and desorption branches of KIT-6 are steeper than
those of SBA-15. Such a difference may come from the fact that the size distribution of pores
in a particle of SBA-15 leads to a finite range of condensation and evaporation pressures,
whereas the size distribution of channels in a particle of KIT-6 is not recognized in
condensation and evaporation processes. All the results for KIT-6 and SBA-15 indicate that
interconnections among pores of almost the same size do not have a significant effect on the
adsorption hysteresis and pore criticality.
as porous glasses and silica gels giving rise to the hysteresis loop of type H2 consist of an
interconnected network of pores of varying shape, curvature and size. As the relative pressure
at which a pore empties is thought to depend on the size of the necks, the connectivity of the
network, and the state of neighboring pores, the full understanding of the hysteresis
phenomenon within such porous materials will be a very difficult task. The SBA-16 ordered
mesoporous silica is one of the most suitable model adsorbents currently available for
verification of the theoretical predictions for an ink-bottle pore. The structure of SBA-16
material consists of 9.5 nm spherical cavities arranged in a body-centered cubic array and
connected through 2.3 nm openings [12].
Fig. 7 shows the adsorption-desorption isotherms of nitrogen on SBA-16. At lower
temperatures, the isotherms exhibited large hysteresis loops of type H2 in the IUPAC
classification. When the temperature was raised, the hysteresis loop shrank and eventually
disappeared at ~98 K (Th). Fig. 8 shows the relative pressures p/po of the capillary
condensation and evaporation as a function of reduced temperature T/Tc for Ar, N2, 02, and
CO2 on SBA-16. At lower temperatures the relative pressures of At, N2 and CO2 are taken
with respect to the bulk liquid. All data of the condensation and evaporation are represented
by a common curve. A plot of p/p0 against T/Tc for capillary condensation connected smoothly
into that for reversible capillary condensation measured above Th, whereas the same plot for
capillary evaporation broke at Th. This clearly indicates that capillary condensation in the
hysteretic isotherms takes place near the equilibrium while desorption from the large cavities
is delayed. This delayed desorption is not concerned with emptying of the small channels.
We calculated the temperature dependence of the capillary condensation and evaporation
pressures of nitrogen under the assumptions that adsorption and desorption in an ink-bottle
pore may be regarded as the process of the disappearance and formation of a gas bubble in a
19
N2/SBA-16 57.2K
17 61.5K
6 7 . 5 K 77"6K87"6K
~15 72.1K 93.2K
m 83.6K 96.2K
~13
E
>11
0 1 2 3 4
logP(Torr)
Fig. 7. Temperature dependence of the adsorption hysteresis of nitrogen on SBA-16.
702
0.9
0.8 - SBA-16 o~
0.7 -
0.6
~7 0.5
0.4 ~176176
02
0.2 ~oo ~ [] Ar
0.1 9 • CO2
0 I I I I I I
liquid droplet confined to the spherical pore [13]. A fit between the observed and calculated
transition pressures in a wide temperature range was reasonable in light of several
assumptions and approximations used. This clearly indicates that the energy barrier for the
formation and disappearance of vapor bubbles in the liquid confined to the pores is
responsible for the appearance of the adsorption hysteresis and the hysteresis temperature is
not concerned with the so-called capillary criticality.
REFERENCES
[1] L.D. Gelb, K.E. Gubbins, R. Radhakrishnan, and M. Sliwinska-Bartkowiak, Rep. Prog. Phys.,
62 (1999) 1573.
[2] K. Morishige, H. Fujii, M. Uga, and D. Kinukawa, Langmuir, 13 (1997) 3494.
[3] K. Morishige and M. Shikimi, J. Chem. Phys. 108 (1998) 7821.
[4] K. Morishige and M. Ito, J. Chem. Phys., 117 (2002) 8036.
[5] J.S. Beck et al., J. Am. Chem. Soc., 114(1992) 10834.
[6] D. Zhao, Q. Huo, J. Feng, B.F. Chmelka, and G.D. Stucky, J. Am. Chem. Soc., 120(1998)6024.
[7] F. Kleitz, S.H. Choi, and R. Ryoo, Chem. Commun., (2003) 2136.
[8] L.H. Cohan, J. Am. Chem. Soc., 60(1938)433.
[9] K. Morishige and Y. Nakamura, Langmuir, 20 (2004) 4503.
[ 10] K.S.W.Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T.
Siemieniewska, Pure Appl. Chem., 57 (1985) 603.
[11 ] D.H. Everett, in: E.A. Flood (eds.), The Solid-Gas Interface, vol.2,Marcel Dekker, New York,
1967, Chapter 36, pp.10550-1113.
[12] Y. Sakamoto, M. Kaneda, O. Terasaki, D.Y. Zhao, J.M. Kim, G.D. Stucky, H.J. Shin, and R.
Ryoo, Nature(London), 408 (2000) 449.
[ 13] K. Morishige and N. Tateishi, J.Chem.Phys., 119 (2003) 2301.
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 703
aFaculty of Chemistry, Nicolaus Copernicus University, 7 Gagarin St., 87-100 Torun, Poland
1. INTRODUCTION
Synthesis of M41S molecular sieves of the pore diameters from 1.5 to 10 nm [1,2] became a
breakthrough in the quest for materials intermediate between microporous and macroporous
solids. Since then, the preparation methods and properties of these products have extensively
been developed [3-7].
MCM-48, a member of the M41S family, exhibits a system of highly branched,
cylindrical pores, which belongs to the cubic space group (Ia3d). Relatively high thermal
stability, specific surface area, and specific pore volume are the most interesting physical
properties of MCM-48 for adsorption and catalytic applications. Non-modified material is
catalytically inactive. Active sites are generated by incorporation of heteroatoms into the
electrically neutral, purely siliceous skeleton. Modification of MCM-48 with A1 or other (e.g.,
transition) metals and metal oxides has already been reported [5,8-12]. Introduction of A1 into
the MCM-48 skeleton is particularly important since it may give rise to the Bronsted acid
sites in the form of bridging hydroxyl groups, SiOHA1, associated with (A104/2)-tetrahedra
[ 13]. These centres should primarily be responsible for the catalytic activity. The expected
influence of the content and/or distribution of A1 on the strength of such acid sites (similarly
as in zeolites) has been studied in an earlier work [ 14].
The MCM-48 sieves with the appropriate structure preserved after introduction of
relatively large amounts of A1 have been prepared by a post-synthesis grafting. The method
has been presented elsewhere [15]. Some products have partially been characterised with the
X-ray diffraction (XRD) and adsorption techniques as well as with catalytic tests [14,15].
The present work shows an extensive characterisation of a wider variety of the Al-grafted
MCM-48 materials with the AAS, XRD, NMR, and IR techniques as well as with the
adsorption analysis and a catalytic test of cumene cracking. The latter has often been chosen
to specify the acidity of solid catalysts [ 16,17].
704
2. E X P E R I M E N T A L
Table 1
Sample preparation conditions
Sample a Aluminium Solvent Si/A1 of
no. designation compound substr, product b
1 MCM-48 n.a. n.a. n.a. n.a.
2 A1-MCM-48(44) tri-sec-butoxide n-butanol 30 43.6
3 A1-MCM-48(35) isopropoxide n-hexane 32 34.5
4 AI-MCM-48(13) isopropoxide n-hexane 15 12.7
5 AI-MCM-48(7) tri-sec-butoxide n-butanol 5 6.9
6 AI-MCM-48(4) isopropoxide n-hexane 5 3.8
7 A1-MCM-48(3) isopropoxide n-hexane 2 3.5
~Samples 2-7 were prepared by grafting A1 onto sample 1. bValues determined with AAS. n.a. = not
applicable.
Table 2
Textural properties of the studied materials
Parameter . . . .
Sample a
1 2 3 4 5 6 7
dzll [rim] 3.71 3.46 3.40 3.40 3.42 3.32 3.34
a0 [nm] 9.09 8.48 8.33 8.33 8.38 8.13 8.18
S~EV [m 2 g-l] 1315 1197 1243 1186 1283 1077 1045
St [m 2 g-~] 1294 1160 1213 1164 1223 1010 995
Sext [m 2 g-~] 284 220 259 136 223 216 215
So [m 2 g-l] 1010 940 954 1028 1000 794 780
Vp [cm 3 g-~] 0.718 0.700 0.662 0.673 0.686 0.538 0.513
Vt [cm 3 g-l] 0.958 0.856 0.884 0.794 0.841 0.713 0.711
day [nm] 2.41 2.33 2.34 2.28 2.04 2.20 2.19
wth [rim] 1.74 1.57 1.52 1.55 1.69 1.53 1.55
aSample numbers as in Table 1. d2~ is the (211) interplanar spacing, a0 - unit cell constant (a0 =
61/2d21~), SBEV- the BET specific surface area, St- total surface area, Sext- external surface area, Sp-
surface area of primary mesopores, Vp- volume of primary mesopores, Vt - total pore volume, dav -
average pore diameter from the maximum of the BJH pore size distribution, and Wth-- wall thickness
(Wth -- a0/3.0919 - dav/2).
calculated from the low-angle X-ray diffraction (XRD) powder patterns (Siemens D 5000 and
Philips X'Pert PRO diffractometers). The other parameters (Table 2) were derived from
adsorption and desorption of nitrogen (77 K, Micromeritics ASAP 2010). The detailed
description of the measurements has been presented elsewhere [21,22]. The 27A1MAS NMR
spectra (Bruker AMX 300 WB) were recorded at a resonance frequency of 78.2 MHz.
0.20
c
133
'~ 0.15 1
E
o ' 3
"~ 2
E
0.10 -
0
>
O
'-' 0.05-
(I)
~-
~.- 0.00 I - , -~ I I I I I :.---.v-,. I v'"~l
O 1.5 2.0 2.5 3.0 3.5 4.0 4.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Pore diameter/nm
Fig. 2. BJH pore size distributions obtained from nitrogen desorption for parent and Al-grafled
MCM-48; sample numbers as in Table 1.
However, no agglomerates of A1 species have been detected by XRD up to 2 0 = 110 ~
Noteworthy is that the concentrations of the Bronsted centres in the materials grafted with use
of aluminium tri-sec-butoxide (Table 3, samples 2 and 5), despite smaller total amounts of AI,
are larger than those in the materials synthesized from aluminium isopropoxide (samples 3
and 6). In the former case, the dispersion of A1 is probably better from the cata|ytical point of
view, i.e., less AI is located inside the agglomerates. Thus, aluminium tri-sec-butoxide is
more efficient in creation of the Bronsted acid sites that are important for catalytic
applications of A1-MCM-48.
In accordance with Kim et al. [28], who have proposed a mechanism for incorporation of
Ga into MCM-41, formation of the acidic centres by the grafting of A1 may be explained as
follows:
I i i i
0.25 1.5
1455 3742
1622
/A''3z42 I
0.20 -
1.0-
::5 1491
"~ 0.15 4
O
I
r
0.5-
~
x._ .13
i,..
o 0.10-
o0
6391 O
.13
< E
0.05
-0.5 , , ,37441 , , ,
0.00 ............... ,.
1650 1600 1550 1500 1450 3770 3760 3750 3740 3730 3720 3710
Wavenumber/cm -1 Wavenumber/cm -1
Table 3
Fractions of tetrahedral A1 (fAl(t)), concentrations of total A1 and of the Bronsted and Lewis
acid sites as well as fractions of AI that yielded the Bronsted (fB) and Lewis (f~) centres"
sample numbers as in Table 1
Sample f A l ( t ) Concentration (Mmol/g) fB fL fB +fc
no. (%) A1 Bronsted Lewis (%) (%) (%)
1 0 0 0 32 0.0 0 0.0
2 77 330 50 153 15.1 46 61.1
3 79 413 32 150 8.0 36 44.0
4 67 1060 76 257 7.4 24 31.4
5 83 1900 130 442 6.8 24 30.8
6 68 3100 110 430 3.5 14 17.5
7 40 3307 163 554 4.9 17 21.9
3.3. Catalysis
As expected, the conversion of cumene over A1-MCM-48 increases with the concentra-
tion of the Bronsted centres of the catalysts as well as with the reaction temperature.
However, this increase tends to reach a plateau at ca 200 pmol g-1. Benzene and propene,
formed on the Bronsted centres, are the main products of the process while Gt-methylstyrene,
formed on the Lewis centres, is another product found in notable amounts (Figs. 6 and 7).
709
60 60
9
9 "'
50 5O
,~ ,~
AA
4O A
4O
O 0
E 30 E
"O
"O
30
(D (1)
>.. >.-
20 20
10 10
. . . . . . i
! i , I , I , I I i I ~ | I i
4. C O N C L U S I O N S
Grafting of aluminium onto the MCM-48 molecular sieve yields materials with Al
incorporated into the skeleton and retaining to a great extent a highly ordered mesoporous
structure characteristic of the parent material. The products exhibit large both the surface area
and the total pore volume. The surface area decreases with increasing AI content of the
materials. The observed reduction in both the volume and the average diameter of the primary
mesopores may result from the fact that the amount of AI incorporated at the surface of pore
walls is most probably rising with the total A1 content of the materials.
The incorporation of A1 results in formation of the Bronsted and Lewis acid sites. The
former ones can be related to the skeleton tetrahedral A1, the latter to the extra-skeleton
octahedral A1. The Lewis acid sites are formed in larger amounts as compared to the Bronsted
ones. The total contents of sites of both types are significantly lower (not exceeding ca. 60%)
than the amounts of the grafted AI.
710
Cumene cracking over A1-MCM-48 yields benzene, propene, and a-methylstyrene. The
first two products form on strong Bronsted acid sites whereas the last one on the Lewis acid
centres. The conversion increases with the total A1 content and, in particular, with the
concentration of the Bronsted centres of the catalysts, although it reaches a plateau. The
conversion increases also with the reaction temperature while it decreases with the number of
injections.
ACKNOWLEDGEMENT
The work was supported in part by the State Committee for Scientific Research (KBN) with a
grant no. 4 T09A 147 23.
REFERENCES
[1] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, and J.S. Beck, Nature (London), 359
(1992) 710.
[2] J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, J.B. Higgins, and J.L. Schlenker, J. Am.
Chem. Soc., 114 (1992) 10834.
[3] A. Corma, V. Formes, M.T. Navarro, and J. Perez-Pariente, J. Catal., 148 (1994) 569.
[4] S. Biz and M. Occelli, Catal. Rev. Sci. Eng., 40 (1998) 329.
[5] S. Jun and R. Ryoo, J. Catal., 195 (2000) 237.
[6] Y. Liu and T.J. Pinnavaia, J. Mater.Chem., 12 (2002) 3179.
[7] Y. Xia and R. Mokaya, Microporous Mesoporous Mater., 68 (2004) 1.
[8] R. Schmidt, H. Junggreen, and M. Stoecker, Chem. Commun. (Cambridge), (1996) 875.
[9] W. Zhang and T.J. Pinnavaia, Catal. Lett., 38 (1996) 261.
[10] M. Morey, A. Davidson, H. Eckert, and G. Stucky, Chem. Mater., 8 (1996) 486.
[ 11] A.A. Romero, M.D. Alba, and J. Klinowski, J. Phys. Chem. B, 102 (1998) 123.
[12] W.S. Ahn, D.H. Lee, T.J. Kim, J.H. Kim, G. Seo, and R. Ryoo, Appl. Catal. A, 181 (1999) 39.
[13] R. Mokaya, W. Jones, Z. Luan, M.D. Alba, and J. Klinowski, Catal. Lett., 37 (1996) 113.
[ 14] M. Rozwadowski, M. Lezanska, J. Wloch, K. Erdmann, and J. Kornatowski, Stud. Surf. Sci.
Catal., 142 (2002) 755.
[15] M. Rozwadowski, M. Lezanska, R. Golembiewski, K. Erdmann, and J. Kornatowski, Stud.
Surf. Sci. Catal., 142 (2002) 1631.
[16] J.W. Ward, J. Catal., 11 (1968) 259.
[17] R. Mokaya and W. Jones, J. Catal., 153 (1995) 76.
[18] C.T. Kresge, M.E. Leonowicz, W.J. Roth, and J.C. Vartuli, US Patent No. 5098684 (1992).
[19] L.Y. Chen, Z. Ping, G.K. Chuah, S. Jaenicke, and G. Simon, Microporous Mesoporous Mater.,
27 (1999) 231.
[20] Y. Xia and R. Mokaya, J. Phys. Chem. B, 107 (2003) 6954.
[21] M. Rozwadowski, M. Lezanska, J. Wloch, K. Erdmann, R. Golembiewski and J. Kornatowski,
Phys. Chem. Chem. Phys., 2 (2000) 5510.
[22] M. Kruk, M. Jaroniec, and A. Sayari, Langmuir, 13 (1997) 6267.
[23] D.B. Dadyburjor and A. Bellare, J. Catal., 126 (1990) 261.
[24] M. Rozwadowski, M. Lezanska, J. Wloch, K. Erdmann, R. Golembiewski, and J. Kornatowski,
Langmuir, 17 (2001) 2112.
[25] S.J. Gregg and K.S.W. Sing, Adsorption, Surface Area and Porosity, Academic Press, London,
1982.
[26] H.S. Cerqueira, P. Ayrault, J. Datka, and M. Guisnet, Microporous Mesoporous Mater., 38
(2000) 197.
[27] H.S. Cerqueira, P. Ayrault, J. Datka, P. Magnoux, and M. Guisnet, J. Catal., 196 (2000) 149.
[28] J.Y. Kim, S.J. Lee, and J.-S. Yu, Bull. Korean Chem. Soc., 21 (2000) 544.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka, N. Zilkov~i and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 711
a) Laboratoire des Mat6riaux, Surfaces et Proc6d6s pour la Catalyse (LMSPC, UMR 7515), ECPM-
ULP, 25, rue Becquerel, F-67087 Strasbourg Cedex 2, France. E-mail: blouis@chimie.u-strasbg.fr.
Part of the ELCASS (European Laboratory of Catalysis and Surface Science)
b) Laboratoire de Physico-Chimie des Hydrocarbures (UMR 7513 CNRS), Universit6 Louis Pasteur, 4
rue Blaise Pascal, F-67070 Strasbourg Cedex, France.
An original quantitative method based on H/D exchange between H20/D20 molecules and the
OH groups of different zeolites has been developed for the titration of the Br6nsted acid sites
present on the solid surface. The measured Br6nsted acid sites density appears to be in good
agreement with the theoretical amount estimated by the Si/AI ratio. Therefore, from the
decrease in the total number of Br6nsted acid sites, we have found an innovative approach to
determine both the amount and the localisation of Mo atoms inside the ZSM-5 zeolite
framework. A dependance was found between the decrease of acidic OH groups and the Mo
content. Depending on the Si/A1 ratio of the zeolite, i.e, the proximity between two Br6nsted
acid sites, the Mo-atoms substitute a different number of OH groups.
1. INTRODUCTION
Solid acids and particularly zeolites have found widespread applications in the refinery
industrial processes and in fine organic chemistry as an eco-friendly catalyst [1,2]. Since the
pioneering work done by Haag et al. [3], where a linear dependence between cracking activity
and A1 content (or H content) has been observed, there has been a growing interest to
determine the concentration of Br6nsted acid sites [4]. So, methods providing informations
about acid sites are of primary importance. The most frequently used methods: colorimetric
titrations, IR, TPD of basic molecules or calorimetry, based on the adsorption of basic
compounds, suffer from various drawbacks [5].
Zeolites have a well-defined three-dimensional porous structure arising from a framework of
[SiO44-] and [A1045-] tetrahedra linked together by all their comers. Since the trivalent A1 is
bonded to four shared O-atoms, the residual charge on each A104 unit is-1. Therefore, each
alumina tetrahedron requires a +1 charge from a cation in order to maintain electrical
neutrality throughout the structure: typically H + cations in acidic zeolites. The chemical
composition of a protonic zeolite can hence be represented by a formula of the type:
In relation to recent works done in our group on H/D exchange occurring between small
alkanes and solid acids [7], a quantitative method based on tt/D exchange between H20/D20
molecules and the OH groups of different zeolites has been developed for the titration of the
whole quantity of Br6nsted acid sites [8].
2. EXPERIMENTAL
D20
ZeoI-OH ~ ZeoI-OD 4" H20 4" HDO 4" D20
excess
H20
ZeoI-OD -- ZeoI-OH 4" H20 4" HDO 4" D20
excess
(CF3CO)20
HxODy = x CF3COOH 4" y CF3COOD
excess
Scheme 1. Set of H/D exchange steps between H20 / D20 and the OH groups of the zeolite
The partially exchanged water HxODy, composed by H20, HDO and D20, was collected in a
cold U-tube a t - 1 1 7 ~ (ethanol melting point). The recovered HxODy was weighed and
chemically trapped by trifluoroacetic anhydride (used in a two-fold excess). The acid solution
thus obtained was transferred under argon to a NMR tube for analysis (Bruker AM400
spectrometer). An internal standard solution of CHC13/CDCI3 was added. The integration of
CF3COOH(D) and CH(D)C13 on both 1H and 2H spectra allows an accurate quantification of
713
H/D content of the sample. The acid site density was then calculated based on the H/D ratio
measured and the mass of HxODy collected.
Fig. 1 shows the comparison between the measured amount of Br6nsted acid sites by this
method and the theoretical amount estimated by means of the Si/A1 ratio [6]. The Br6nsted
acid sites density measured on all zeolite samples: MOR, MFI, SAPO, EMT, BEA and HUSY
is in good agreement with the theoretical amount estimated by the Si/A1 ratio (Fig. 1).
Concentration of Br6nsted acid sites is equal to the framework A1 content. By means of
microcalorimetry, and the measurement of heats of ammonia adsorption, Gorte et al. have
already observed the same trend for unsteamed zeolites [9]. So, this anhydride method can be
used as a powerful tool to determine the whole quantity of Br6nsted acid sites on zeolitic
materials. Thus, by knowing the crystalline structure of any material, one is able to determine
the number n of H atoms (regarding Eq. 1) and therefore to estimate the Si/A1 ratio which is a
key parameter for the study of the catalysis by zeolites [10,11]. As an example, 1.17 mmol
H/gzeolite could be measured on a ZSM-5 sample (Si/AI=12.5). This value corresponds to
approximately 7 H atoms in the unit cell. By taking x - 96 and y - 192 from the Atlas of
zeolites for the MFI structure [6], a Si/AI = 12.7 is obtained corresponding to the value given
by the manufacturer. Fig. 1 also presents the results for H3PW12040 heteropolyacid, which
normally possess 0.89 mmol H/g [12]. By applying the D20/H20 isotopic exchange
(activation of the catalyst was performed during lh at 300~ under air to prevent from the
decomposition of the Keggin structure), about 0.91 mmol H/g were measured, which further
demonstrates the reliability and the versatility of the anhydride method to other crystalline
materials.
Gorte et al. have recently developed a method based on propylamine decomposition (via
Hofmann elimination into propene and ammonia) for measuring Br6nsted acid sites density at
200~ [ 13]. The results obtained with the two methods for different zeolites has already been
published elsewhere [8]. The general trend observed is a lower amount of Br6nsted acid sites
titrated with the propylamine method (via measuring the propene released from the solid)
compared to the anhydride method. While the former is based on a protonation of the
alkylamine, which probably involves acid sites of sufficient strength, the latter does not
discriminate between the H-atoms.
714
J
E] Measured value
[] Theoretical value
.~ 3-
~ 2-
"0
.o
II1
0
o
~ o ,,, .. = o
Solid Acids =
Fig. 1. Measured and theoretical amount of BrOnsted acid sites of different solid acids
The main difference is observed for the HUSY zeolite which presents theoretically 3.0 mmol
H / g zeolite; only 0.69 mmol H/g were measured (each experiment has been repeated twice).
In the important case of FAU-type zeolites, it is known that a great part of the charge-
balancing hydroxyls are inaccessible to reactant molecules [4], so that the number of bridging
OH groups is under-estimated by base adsorption. Moreover, almost no propene could be
detected over the heteropolyacid indicating a stronger adsorption; H3PW12040 is known to
be more acid than zeolites [14,15].
This work describes an innovative way for determining the chemical composition of both
zeolitic materials and other crystalline solid acids like heteropolyacids. While other methods
are restricted to the determination of Br0nsted sites of sufficient acidity to protonate amines,
this "anhydride method" based on D 2 0 / H 2 0 exchange steps opens the field of until now
unreported determination of the whole Br0nsted acid sites, and hence the chemical
composition of zeolites.
4. A P P L I C A T I O N TO Mo-ZSM-5 Z E O L I T E S
Fig. 2 shows the number of Br6nsted acid sites measured by the anhydride method for Mo-
ZSM-5 zeolites with Si/A1 = 15 and 40. For the sample having Si/AI = 15, a linear
dependence was observed between the increase in Mo-content and the decrease of the
number of O-H groups.
By comparing the mole ratio of Mo incorporated to the decrease of the number of acid sites, it
appears that two bridging hydroxyl groups are replaced by one Mo-atom. For the zeolite
having a Si/A1 = 40, the introduction of 2 wt% Mo also led to the consumption of two O-H
groups per Mo-atom. Therefore, we can assume that one Mo-atom is attached to two
neighbouring lattice oxygens, forming a metal complex on the surface which has an OF-
MoO2-OF stoechiometry (OF : framework oxygen). So, this method relates the amount of
Mo-introduced in the zeolite channels with the number of Br6nsted acid sites, and thus
determines the localisation and the geometry of the Mo-complex.
715
1.2
~ Si/AI= 15
1
-=- S i / A I = 40
z 0.8
O
E
.E. 0.6
.=.,,
"O 014
c
=.o 0.2
OQ
0
0 1 2 3 4
Mo content [wt%]
Fig. 2. Dependance between the amount of number of Br6nsted acid sites and the Mo content
However, while increasing the amount of Mo to 4 wt%, almost no further decrease in the
amount of Br6nsted acid sites has been observed. This indicates that the amount of anchoring
sites becomes too low to allow the Mo-complex to be introduced in the same geometry. Solid-
state NMR has confirmed the presence of a high amount of extra-framework species for the
sample having at least 4 wt% of Mo [ 16]. It seems that the type of geometry adopted by the
Mo-complex inside the zeolite channels depends on the Si/A1 ratio, thus the number of
Br6nsted acid sites accessible. Previous studies devoted to the aromatisation of methane over
Mo-ZSM-5 have also confirmed these observations [17,18].
5. CONCLUSION
This work reports a non-destructive method, based on H20 / D20 exchange steps, for
characterising the surface of solid acids. Unlike other methods which seek to discriminate
between acidic and non-acidic protons, this technique counts the total number of Br6nsted
acid sites in zeolitic materials. Furthermore, this method determines the localisation and hence
gives insights about the geometry of metal complexes and can serve as a basis to design new
single-site catalysts. Moreover, this way of titration is easy, rather cheap, and essentially does
not lead to the poisoning or destruction of the sample.
ACKNOWLEDGEMENT
Financial support of the Loker Hydrocarbon Institute, U.S.C., Los Angeles (S.W.) is
gratefully acknowledged. The authors also thank the NMR service, and especially M. J.D.
Sauer for his technical assistance.
716
REFERENCES
[ 1] H. Pines, "The Chemistry of Catalytic Hydrocarbon Conversion", Academic Press, New York,
1981.
[2] A. Corma and H. Garcia, Catal. Today, 38 (1997) 257.
[3] W.O. Haag, R.M. Lago and P.B. Weisz, Nature 309 (1984) 589.
[4] R.J. Gorte, Catal. Lett., 62 (1999) 1.
[5] A. Corma, Chem. Rev., 95 (1995) 559.
[6] C. Baerlocher, W. M. Meier and D.H. Olson, Atlas of Zeolite Framework Types (5 th ed.,
Elsevier, Amsterdam, 2001).
[7] a) J. Sommer, D. Habermacher, R. Jost, A. Sassi, A.G. Stepanov, M.V. Luzgin, D. Freude, H.
Ernst, J. Martens, J. Catal., 181 (1999) 265. (b) R. Olindo, A. Goeppert, D. Habermacher, J.
Sommer, F. Pinna, J. Catal., 197 (2001) 344.
[8] B. Louis, S. Walspurger and J. Sommer, Catal. Lett., 93 (2004) 81.
[9] D.J. Parrillo, C. Lee, R.J. Gorte, Appl. Catal. A 110 (1994) 67.
[ l 0] D. Barthomeuf, Mater. Chem. Phys. 17 (1987) 49.
[11] M. Mtiller, G. Harvey and R. Prins, Microporous Mesoporous Mater. 34 (2000) 135.
[12] N. Essayem, G. Coudurier, J. C. Vedrine, D. Habermacher and J. Sommer, J. Catal. 183 (1999)
292.
[13] O. Kresnawahjuesa, R.J. Gorte, D. de Oliveira and L. Y. Lau, Catal. Lett. 82 (2002) 155.
[14] T. Okuhara, N. Mizuno and M. Misono, Appl. Catal. A 222 (2001) 63.
[15] G.A. Olah, G.K.S. Prakash and J. Sommer, Superacids, John Wiley & Sons, New York, 1985.
[16] B. Louis, J.P. Tessonnier, S. Walspurger, C. Pham-Huu, J. Sommer, M.J. Ledoux, Microporous
Mesoporous Mater., submitted.
[17] R.W. Borry, Y.H. Kim, A. Huffsmith, J.A. Reimer, E. Iglesia, J. Phys. Chem. B 103 (1999)
5787.
[18] H. Minming, R.F. Howe, J. Catal. 108 (1987) 283.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkov~and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 717
PQ Corporation, R&D Center, 280 Cedar Grove Road, Conshohocken, PA 19428, USA
1. INTRODUCTION
The common practice of using silicon-ethoxide in synthesis of both types of Ti-silicalites, TS-
1 and TS-2, can cause certain economical and technological disadvantages when it is applied
to industrial-scale production. This is why a new process of making Ti-silicalites (of MFI and
MEL structural types) under the registered trade mark TS-PQ| was developed in PQ
Corporation [2]. Rather than the ethoxide, a high purity silica gel is employed. Various
commercial organic titanates also substitute the more frequently used Ti-alkoxides as Ti
sources.
Control of the quality of this potential catalyst and the possibility to predict its properties
have required development of an understanding of the chemical and structural state of the Ti
species that are associated with TS-PQ. Establishment of this understanding is demonstrated
in this paper with the MFI-type silicalite as an example.
2. EXPERIMENTAL
The colloidal mixture was treated in Parr autoclaves equipped with a propeller stirrer. The
crystallization was completed in the temperature range of 1 5 0 - 180~ for 2 to 72 hours. A
milky slurry-product usually was observed, which did not settle. To recover the nano-size
crystals two procedures were applied. They were different in either separation or inclusion
mineral components of the mother liquor into the final product. Separation of mother liquor
was provided by centrifugation at 15 000 rpm for 1 hour. In the second procedure the solid in
the slurry together with residual titania and silica, that could be dissolved in the mother liquor
at high pH, was coagulated by adjustment slurry pH to about 9.5. The precipitate was
separated by filtration. Precipitates resulted from the both recovery procedures were dried at
110~ and then calcined in air at 550~
Several post-synthesis procedures were applied to modify the product properties. To
remove non-crystalline non-framework Ti from the external surface of the zeolite, the as-
synthesized material (i.e. material which channels were still plugged up by TPA-cations) was
treated with strong mineral acids at about 80~ To cure structural defects and eliminate
crystal strains, ammonium acetate treatment or calcinations with steam were applied. The
ammonium acetate treatment was done on organic-free (i.e. calcined) samples. 10%
ammonium acetate solution was used also at 80~ and the treated sample was calcined for the
second time. Steaming was conducted at 550~ with 100% steam partial pressure.
The intensity ratio of the 960 cm-1 and 800 cm-1 peaks on FTIR spectra (hereafter known as
the 960/800 peak ratio) has been used repeatedly as evidence of Ti incorporation and as an
indicator of the relative abundance of single Ti atoms surrounded by Si atoms in both
crystalline and amorphous titanium silicates [5 - 9]. Values of that indicator measured in the
spectra of 250 calcined (organic-free) samples of MFI type TS-PQ are plotted in Fig. 1
against the Ti atom percent (Ti/(Ti+Si), %) in products. A wide array of preparations is
displayed. They differ in starting mixture formulation and conditions of pre-synthesis,
synthesis and post-synthesis treatment. In particular, various Ti-sources, mentioned in Section
2.1, were used. Ti-triethanolaminatoisopropoxide was used in nearly all of the high-Ti
formulations in Fig. 1. Among the Ti reagent sources this is the most stable to hydrolysis.
Products with the same bulk Ti concentration show a range of the 960/800 ratios. The highest
value in that range implies the limit for the number of isolated Ti atoms in the framework. A
line connecting these limiting values represents the position of products with "the maximum
Ti dispersion". Although in Ti-free silicalite a weak 960 cm -1 peak can show up due to the
presence of silanol-nest defects [ 10], its intensity is negligibly low in comparison to the effect
of Ti in the silica matrix and was assumed to be zero. It is seen in Fig. 1 that the line
demonstrates monotonous growth with Ti loading up to 2.8 Ti atom percent.
3.5
3.0 8 9
I~ 7 10
2.5 6. 14~
a. 5 9
2.0 41
'E 3 ~ 0
o
o 1.5 O TBT
9 TEP
O)
1.0 2~ ~
ev A TLA
~.. 0.5 A
M. [] Ti - free
0.0 , , , ,
Raman spectroscopy also shows that samples with the maximum 960/800 ratios are free from
non-fi'amework Ti in the form of crystalline TiOz, anatase. Only at the upper limit of Ti
incorporation, sample 10, was a small amount of anatase (about 0.03%) found. That amount is
less than the experimental error in Ti atom percent results (+0.1%), calculated from XRF
analyses data.
|~o " 9
<,,, .j=<,.~3 \ b
_ --__
, , i
Unlike the samples characterized by the maximum 960/800 ratios, those below the border
line (samples 11 - 14 in Fig. 1) reveal through their FTUV DR spectra (Fig. 2b) the presence
of other Ti-centers. The red side shifts of bands indicate expansion of Ti coordination and/or
an increase in the degree of Ti-oxide polymerization. In addition to IFT, three other types of
Ti site can be distinguished. The band that appears above 3 l0 nm is attributed to condensed
crystalline TiO2, anatase (AN), nano- and micro-particles [5, 12, 16]. The broad "hump" in
the 2 3 0 - 310 nm range is assigned to non-framework, oligomeric, octahedral Ti (NFO)
present as amorphous mixed TiO2 - SiO2 oxides [11, 13, 17-19]. The superimposed peak at
around 290 - 300 nm is assumed to relate to one-dimensional Ti-oxide nano-clusters (NC)
[20]. The band in the 215 - 230 nm range is assigned to isolated monomeric Ti with
coordination greater than tetrahedral [14]. It is referred to in this paper as "dispersed" (D) Ti,
and probably is responsible for surprising catalytic properties of TS-PQ [3].
Anatase is formed in the TS-PQ process at some point following the onset of Ti-silicalite
crystallization. The TiO2 in the mother liquor is consumed making anatase. Its formation
results in a decrease of Ti on silicalite surface, as discussed later in the paper. The anatase
admixture resists acid dissolution in as-synthesized samples.
Unlike crystalline anatase, non-framework, oligomeric TiO2 admixtures are precipitated
from the mother liquor on the external crystal surfaces during the post-synthesis cooling and
recovery stages. They enrich the Ti-silicalite surfaces with Ti that can be leached by acid
post-treatment of the as-synthesized samples.
Ti in "dispersed" centers is similar in its behavior to that of oligomeric species: it can also
be removed by acid post-synthesis treatment of as-synthesized TS-PQ. In addition, this type
of Ti is calcination sensitive and disappears after severe or repeated calcinations.
A response of unit cell dimension to increase of Ti loading in silicalite was considered the
main evidence of Ti isomorphous substitution in framework tetrahedral positions [1, 8, 21 -
23]. It was recommended that special treatment with ammonium acetate precede the unit cell
refinement [1 ]. In the present study, in order to clarify the treatment effect, refinements were
721
performed for both untreated and treated samples. The data are presented in Table 1. The
samples chosen for the unit cell study were those free of non-framework Ti (Samples 1-10).
Samples treated with ammonium acetate are designated with an "A" suffix, while that treated
by recalcination with steam is noted with an "S" suffix.
Table 1
Unit cell parameters of TS-P Q
Ti atom FTIR
Sample*
percent 960/800 a b c a .... Vol (:k)
Untreated
1 0.00 - 20.0821 19.8975 13.3841 90 5348.1 (0.9)
2 0.50 0.80 20.0868 19.9161 13.3881 90 5355.9 (0.8)
3 0.89 1.10 20.0982 19.9191 13.3984 90 5363.9 (0.9)
4 0.99 1.47 20.0856 19.9183 13.3916 90 5357.6 (0.9)
5 1.38 1.87 20.0997 19.9216 13.4012 90 5366.1 (0.9)
6 1.96 2.24 20.1029 19.9278 13.4020 90 5368.9 (1.0)
7 2.06 2.36 20.1062 19.9339 13.4051 90 5372.7 (0.8)
8 2.15 2.54 20.1047 19.9362 13.4101 90 5374.9 (1.1)
9 2.30 2.56 20.0995 19.9267 13.3986 90 5366.4 (0.7)
10 2.82 3.08 20.1082 19.9341 13.4091 90 5374.9 (0.7)
Ammonium acetate treated and steamed
1A 0.00 - 20.1166 19.8974 13.3823 90.60 5356.2 (1.6)
2A 0.50 n.d.** 20.1185 19.9151 13.3929 90.36 5365.9 (3.1)
3A 0.89 1.07 20.1236 19.9207 13.3951 90.29 5369.7 (4.4)
5A 1.38 1.65 20.1227 19.9351 13.4129 90 5380.6 (0.6)
6S 1.96 2.34 20.1273 19.9420 13.4137 90 5384.0 (0.8)
7A 2.06 n.d. 20.1223 19.9399 13.4161 90 5383.0 (0.7)
8A 2.15 n.d. 20.1283 19.9467 13.418 90 5387.2 (1.1)
9A 2.30 2.33 20.1287 19.9447 13.4167 90 5386.3 (0.8)
10A 2.59 3.00 20.1335 19.9537 13.4236 90 5392.8 (0.5)
* Numbers correspond to Fig. 1 ** Not determined
It can be seen from data in Table 1 that the ammonium acetate, as well as steaming, does
not affect significantly status of Ti in the zeolite. The 960/800 peak ratios undergo minor
changes with no evident pattern. The same is true for FTUV DRS results: only small
transformations with Ti sites may be indicated on the spectra. Ti concentrations also do not
change with an exception perhaps for the high Ti material, Sample 10, where the treatment
slightly reduced the Ti concentration, probably due to some solubilization of its components.
However, a consistent increase in the unit cell volume is observed after the post-synthesis
treatment. This increase is coincident with some change in defect structure of the materials.
Particularly, hydroxyl group concentration in Ti- as well as in Ti-free silicalite is generally
reduced, reflecting the healing of hydroxy-nest and condensation of isolated OH-defects [24,
25]. Comparison of FTIR spectra in the OH stretching mode region for samples before (Fig.
3a) and after (Fig. 3b) treatment illustrates these phenomena. Results point also to differences
in behavior of silanols and titanols. Relative intensity of the 3740 cm ~ component, assigned
722
to free internal OH-groups [26, 27], is increased or, at least, is decreased to less degree in Ti-
bearing sample than in pure silicalite (Fig. 3b). This can be explained by lower stability of Ti-
O-Si bridges in comparison to more stable Si-O-Si bridges. As a result Ti-O-Si bridges are
easier to break and the titanol-silanol pair is more difficult to condense than two silanols.
However, hydrogen-bonded hydroxyls of nest- and nanovoid- defects (3650 - 3000 cm-1) are
better preserved in the pure material.
"~" 0 . 5
r
::3
a
L_
e~
L
~ --
ID
~ I. \
C
e~
L_
0
.-\ \
,~ 0-_
f'-- w
3800 3600 3400 3200 ' 3800 ' 3600' :3400' 3200 3000
Wavenumbers c m -1
Fig. 3. FTIR spectra of untreated (a) and post-treated (b) samples. Solid l i n e - Ti-free sample 1. dotted
line - sample 9, dashed line - sample 10 (Table 1).
5390.0 -
.J
E (kl .
o
~ 5370.0,
C
Post-treatment leads to a separate set of samples, demonstrating a unit cell variation trend
shifted up from that of untreated s a m p l e s - Table 1. In Fig. 4, the unit cell volume
dependence on Ti concentration is presented for untreated and treated products. The plot
shows that the expansion of unit cell is not only a function of the Ti fraction in tetrahedral
positions, but also of a change in defect structure. Annealing of defects reduces scatter of
points around a regression line for treated TS-PQ. The point for the steamed sample, 6S, falls
exactly on the line for ammonium acetate products, evidencing that calcination with steam
affects crystals similarly to the ammonium acetate procedure. The range of expansion of the
unit cell with increasing Ti concentration, at least to 2.5 atom percent, is in agreement with
reliable evaluations of Ti capacity for TS-1 [1, 21 - 23]. Estimations of unit cell size in high-
723
Ti TS-PQ practically coincide with high-Ti TS-1 [1 ], but the difference is significant at low Ti
concentrations. It was reported that the cell volume of silicalite varies widely with change of
synthesis method and starting material sources [24]. Minor admixture level may be
responsible for the difference in silicalites. However, with Ti loading the difference has been
eroded.
Despite the possibility of accommodation of more than two and a half atom percent Ti in
the MF! framework, anatase admixture (determined by Raman spectroscopy analysis) starts to
contaminate products when Ti concentration in solid exceeds approximately 1.5 percent. This
is demonstrated in Fig. 5 by statistics of anatase amounts in the same two hundred fifty TS-
PQ samples that were plotted in Fig. 1. An earlier, similar observation was made in [28].
O O x •
4000
O x
E
O.
3 o c
Ill
=
L
6
X
X
X
|
et X
0 ,,,/
0 E
o
4 Surface = Bulk
~X
/"
~-
e-
2000 r-~ Anatase )N. ~,>,
I __ ~ (~ t.J Detection Limit
I o 04 O= 0..rn O
0
0 1 2 3 4 5 6 0 1 2 3
Ti Atom Percent Bulk"11 Atom Percent, XRF
Fig. 5 (left). Anatase amounts in TS-PQ for the samples shown also in Fig. 1. Symbols correspond to
legend for Fig. 1.
Fig. 6 (right). Correlation between surface and bulk Ti concentrations for 125 randomly chosen TS-PQ
samples.
4. C O N C L U S I O N S
TS-PQ process produces a family of materials with variable framework Ti amounts and
controlled presence of non-framework species, yet avoids the use of highly flammable
components. Isomorphous substitution of tetrahedral Ti into the silicalite framework of TS-
PQ is comparable with TS-1 and reaches at least two and a half atom percent. The high Ti
incorporation is possibly due to its over-saturation in the mother liquor when kinetic
conditions are unfavorable for the formation of a Ti-oxide crystalline phase. In equilibrium
with anatase, the Ti capacity of the silicalite framework does not exceed 1.5 atom percent. In
addition to isolated Ti framework sites, Ti can populate non-framework sites as amorphous
titania-silica monomers, oligomers and crystalline anatase. The amorphous phase is deposited
from the mother liquor during the post-crystallization stages of the process. Anatase is formed
from the mother liquor during the crystallization stage.
Internal surface of TS-PQ is characterized by silanol defect structure that can be similarly
modified by either ammonium acetate solution post-treatment or steaming.
ACKNOWLEDGEMENT
The authors thank management of PQ Corporation for permission to present the paper, Dr.
Randall C. Yeates, Schell Westhollow Technology Center, for performing XPS analyses.
REFERENCES
[1] R. Millini et al., J. Catal. 137 (1992) 497.
[2] E. Senderov et al., US Pat. Appl. Publ. 2003152510 A1 (2003).
[3] I. Halasz et al., Catal. Today 81 (2003) 227.
[4] H. Liu et al., J. Catal 218 (2003) 155.
[5] D. Huybrechts et al., J. Mol. Catal. 71 (1992) 129.
[6] D. Dutoit et al., J. Catal. 153 (1995) 165.
[7] M. Uguina et al., Appl. Catal. A 124 (1995) 391.
[8] A. van der Pol andJ. van Hooff, Appl. Catal. A 92 (1992) 93.
[9] G. Ricchiardi et al., J. Am. Chem. Soc. 123 (2001) 11409.
[10] A. Zecchinaetal.,J. Phys. Chem. 96(1992)4985.
[11 ] F. Geobaldo et al., Catal. Letters 16 (1992) 109.
[12] S. Bordigaetal.,J. Phys. Chem. 98(1994)4125.
[13] L. Le Noc et al., Stud. Surf. Sci. Catal. 101 (1996) 611.
[14] J. Klaas et al., J. Phys. Chem. B101 (1997) 1305.
[15] C. Lamberti et al., J. Phys. Chem. B I02 (1998), 6382.
[16] D. Trong On et al., J. Catal. 157 (1992) 235.
[17] S. Klein et al., J. Catal. 163 (1996) 489.
[18] G. Li et al., J. Phys. Chem. B 105 (2001) 2993.
[ 19] P. Wu et al., J. Phys. Chem. B 105 (2001) 2897
[20] R. Howe and P. Southon, 13th IZA Conf, Montpellier, France: Recent Res. Rep. 13-R-06
(2001 ).
[21] G. Perego et al., Stud. Sure Sci. Catal. 28 (1986) 129.
[22] C. Lamberti et al., J. Catal. 183 (1999) 222.
[23] C. Cundy and J. Forrest, Microporous Mesoporous Mater. 72 (2004) 67.
[24] G. Marra et al., Stud. Sure Sci. Catal. 84 (1994) 559.
[25] V. Bolls et al., Langmuir, 15 (1999) 5753.
[26] A. Zecchina et al., J. Phys. Chem. 96 (1992) 4991.
[27] S. Bordiga et al., Top. Catal., 15 (2001) 43.
[28] G. Deo et al., Zeolites 13 (1993) 365
Studies in Surface Science and Catalysis, volume 158
J. t2ejka, N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 725
Influences of various additives on the direct synthesis of high-silica mordenite (MOR) zeolite
with tetraethylammonium hydroxide (TEAOH) were studied in detail. The addition of NaF as
a fluoride source enhanced the nucleation and crystal growth of MOR zeolite. The addition of
NH4NO3 as a mineralizer was also effective for the synthesis of high-silica MOR zeolite. The
MOR zeolites synthesized with NaF and NH4NO3 were found to have much less framework
defects as compared to those synthesized without additives. Under well-optimized conditions,
the highly crystalline and pure MOR zeolite with a Si/A1 ratio of ~ 34 was synthesized by
adding seed crystals as well as NaF and NH4NO3.
1. INTRODUCTION
Barter et al. reported first the hydrothermal synthesis of mordenite (MOR) zeolite [1,2]. Due
to its high thermal and acid stabilities, MOR zeolite is an industrially important zeolite used in
the adsorptive separation of gas or liquid mixtures and catalysis [3-5]. Recently, MOR zeolite
has been proposed to apply for hosts of semiconductor materials, chemical sensors and
nonlinear optical materials [6]. Therefore, there is still an incentive to develop the effective
method for synthesis of MOR zeolite, and especially the direct hydrothermal synthesis of
high-silica MOR zeolite using various organic structure-directing agents such as
benzyltrimethylammonium hydroxide, tetraethylammonium hydroxide (TEAOH),
hexamethyleneimine and benzene-l,2-diol have been widely studied because of its high
thermal stability [7-12]. To our knowledge, however, the maximum Si/A1 ratio reported in
literature so far is only 17.5 [9].
In general, in the hydrothermal synthesis of zeolitic materials, an addition of a so-called
mineralizer is very effective, which transfers the framework constituent elements from the
amorphous gel phase via the solution phase to the surface of the growing crystals. It is well
known that the hydroxide and fluoride ions play this part, resulting in forming soluble
aluminate and silicate complexes that enable the phase transfer. It is also well known that the
addition of fluoride into the starting synthesis gel favors crystallization of high-silica zeolites
[13-20], and fluoride ions have been considered to act as a mineralizing agent and/or a
structure-directing agent [21-26]. In addition, it is recognized that the addition of seed crystals
to crystallization system can enhance the crystallization rate and improve the purity of crystal
product [27-28].
From such viewpoints, we have now studied the direct hydrothermal synthesis of
high-silica MOR zeolite by various methods. Very recently, the highly crystalline MOR
726
zeolite with a bulk Si/A1 ratio of ~ 30 was successfully prepared using both TEAOH as a
structure-directing agent and NaF as a fluoride source, when AI(NO3)3 was employed as an
aluminum source [29]. The addition of seed crystals was also found to be very effective for
direct synthesis of high-silica MOR zeolite [30]. However, characteristics of the
as-synthesized MOR zeolites such as textural property, thermal stability and A1 distribution
were not clarified. In this paper, we analyzed the as-syntheszied MOR zeolites in detail and
tried to synthesize highly crystalline MOR zeolite with a higher Si/A1 ratio of more than 30.
2. EXPERIMENTAL
2.2. Characterization
The crystal phase and crystallinity of the solid products were identified by XRD (Rigaku
RINT 2000). The bulk chemical compositions were measured by XRF (Philips PW 2400) and
ICP (Seiko, SPS 7700). Fluorine contents were determined with an ion-selective electrode.
The crystal morphology was measured by SEM (Hitachi S-4000). 27A1 MAS NMR spectra of
MOR zeolites calcined at 500 ~ were recorded using a zirconia rotor of 7 mm diameter on a
Varian VXP-400 at 104.2 MHz and 29Si MAS NMR at 79.5 MHz. Prior to the 27A1 MAS
NMR measurements, the sample was moisture-equilibrated over a saturated solution of
NH4CI for 24 h. Nitrogen adsorption isotherms a t - 1 9 6 ~ were measured using a
conventional volumetric apparatus (BELSORP 28SA). To clarify the A1 distribution in the
framework of MOR zeolite, the FT-IR spectra of adsorbed CD3CN were measured at room
temperature (JEOL JIR-7000).
Table 1
Synthesis of high-silica MOR zeolites in the presence of NaF a
Starting gel Product (by-product)
Run Time Si/A1 F
Si NaOH NaF (days) Phase SBET VMic content
no. /A1 /AI /SiO2 XRF ICP (m2g"l) (cm3g -l)
(ppm)
1 15 3 0 3 MOR 15.3 441 0.16
2 20 4 0 3 MOR(am)
..... 3_ ....... _20_ . . . . . . . . . _4 . . . . . . . . . . . . . . o_ . . . . . . . . . . . . ......... _ M O ......................................................................................
4 15 3 0.8 3 MOR 15.6 283 0.17 4,500
5 20 4 0.8 3 MOR 19.1
6 25 5 0.8 3 MOR 20.1 6,000
7 30 6 0.8 3 MOR 26.1
8 35 5 0.8 5 MOR 31.0 28.8 298 0.13 2,700
9 40 7 0.8 3 MOR(BEA)
10 40 9 0.8 3 MOR 30.1 28.3 302 0.14 2,000
11 45 9 0.8 ...... 3 BEA(MFI) . . . . . . .
aSynthesis conditions" TEAOH/SiO2 = 0.23, H20/SiO2 - 7.4. SBET:BET specific surface area,
VMic"Micropore volume.
zeolite. Although the crystallinity and the Si/Al ratio of as-synthesized MOR zeolite were
independent of the NaF addition time, the fluoride content in the obtained MOR zeolite
crystals was markedly decreased.
From all above results, it was found that the addition of NaF as a fluoride source into the
synthesis gel depresses the nucleation of BEA zeolite, and enhances the nucleation and
crystallization of MOR zeolite.
than
synthesized
Co-crystallization
length 45,
of ~ 7.5
however,
even/am. For
at MOR
gels
the with
higher
zeolite
a Si/A1
Ncould
aOH/A
ratio 1of
nOtratio,
more be ~,~i
: '~ I'~
Table 3
Synthesis of high-silica MOR zeol!tes.in the presence ofNH4NO3 a
Starting gel Product (by-product)
Run Time
no. Si NH4 + NaOH (days) Phase Si/AI SMic SExt VMic
/A1 /SiO2 /AI XRF ICP (m2g-1) (m2g-l) (cm3g "t)
18 20 0.046 4 3 MOR 20.4 493 23 0.19
19 25 0.074 5 3 MOR 24.3 22.5 468 32 0.19
20 30 0.096 6 5 MOR(Quartz)
aSynthesis conditions: H20/SiO2 = 7.4, TEAOH/SiO2 = 0.23. SMic: Micropore surface area, SE•
External surface area, VM~c:Micropore volume.
Table 4
Synthesis of high-silica MOR zeolites using NH4NO3, seed crystals and NaF a
.... Starting gel Product (by-product)
Run A sol. B soi. F
Si SBEv VMic
no. Si NH4 + NaOH NaOH Phase /A1 (m2g l ) (cm 3g-l) content
/A1 /SiO2 /A1 /A1 (ppm)
21 b 45 0.123 9 0 am(MFI,MOR) ......
22 45 0.123 9 0 MOR(MFI)
23 45 0.096 6 3 MOR 34.2 c 486 0.18 60
24 45 0.074 5 4.9 MOR 29.2 c 511 0.18 210
25 50 0.096 6 4 .... MOR(MFI)
aSynthesis conditions" A solutioni H20/SiO2 = 7.4, NaF/ SiO2 - 0.8; B solution: TEAOH/SiO2 -
0.23; Seed crystals (Run no. 1) - 4 wt%, Time = 3 days. bwithout seed crystals. CDeterminated by
XRF. SBEv: BET specific surface area, VMic"Micropore volume.
730
4. CONCLUSIONS
Characterization of high-silica mordenites synthesized by various direct hydrothermal
synthesis methods was conducted in detail9 The addition of NaF and NH4NO3 was effective
732
for the synthesis of high-silica MOR zeolite. The MOR zeolites obtained were found to have
much less framework defects as compared to those synthesized without additives. Under
well-optimized conditions, the highly crystalline and pure MOR zeolite with a Si/AI ratio of
34 was synthesized by adding seed crystals as well as NaF and NH4NO3.
REFERENCES
b Institute of Catalysis and Surface Chemistry, CRC, Budapest P.O B. 17, H-1525, Hungary
c Environmental Group, Rey Juan Carlos University, 28933 Mostoles, Madrid, Spain
d Applied and Environmental Chemistry Dept., University of Szeged, 6720 Szeged, Hungary
Iron containing micro-and mesoporous substances were exposed to various treatments and
changes in coordination and oxidation states of iron are analysed. ZSM-5 (Si/Fe-- 200) and
MCM-22 (Si/Fe = 13.5) are applied as microporous hosts. Most of iron is stabilized in
tetrahedral framework positions in the first structure, whereas profound rearrangements may
proceed on iron with simultaneous redox processes in the second host. SBA-15 (Si/Fe = 76)
and MCM-41 (Si/Fe = 50) were selected as mesoporous matrices. The partly amorphous
feature of the pore walls is reflected in coordinations occupied by iron. Fe 3+ ++ Fe 2+ reversible
redox transformations may proceed in large part of iron. Further, novel types of coordinations
appear which cannot be detected in microporous hosts.
1. INTRODUCTION
Insertion of iron into micro- and mesoporous structures may result in the modification of the
acidic and redox properties - both of them playing primary role in catalytic processes. Iron
can be introduced either already at the first stage of synthesis by isomorphous substitution or
in a secondary step as ion exchange or grafting. In regular microporous systems the structure
is strictly crystalline in three dimensions, hence, the framework-substituted (FW) and extra-
framework (EFW) positions are clearly defined [1 ].
Mesoporous materials consist of less ordered structure, however, their application is
also preferred due to the large pore sizes and the variety of the possible modifications. These
substances are not crystalline in strict sense of three-dimensionality (as reflected e.g. with the
absence of XRD peaks in the 3 deg < 2 0 region in their XRD diffractograms) [2].
Nevertheless, mesoporous matrices are also able to accommodate iron which may modify the
original acid-base and redox properties.
M6ssbauer spectroscopy has been advantageously used for analyzing the coordination
and oxidation states of iron accommodated in porous structures. Isomer shift (IS) and
quadrupole splitting (QS) data can be interpreted e.g. in analogy to iron bearing minerals. As
tbr the oxidation state, the value of IS decreases with the increase of the valency. As for
coordination, in case of Fe 3+, the increase of symmetry results in the decrease of QS, and the
734
dependence is opposite for Fe z+ [3]. For example, in ZSM-5 the FW and EFW siting is clearly
distinguished [4], the Bronsted and Lewis acidic centres are identified under reaction
conditions [5]. In ferrisilicates of low iron content an additional feature may appear, namely
paramagnetic relaxation originated from separate, distant ions located in tetrahedral
framework sites. This component is represented in the spectra by a broad, unresolved line
[6,7]. Octahedral coordination for Fe 3+ usually gives rise the appearance of doublet (with QS
values exceeding 0.8 mm/s). This latter type of coordination is prevailing for iron in
mesoporous structures as reflected in their in situ spectra [8,9].
In the present study in situ M6ssbauer measurements performed on four structures are
reported. Comparison of data reveals that Fe 3+ is easily stabilized in FW positions in ZSM-5,
whereas significant portion may be removed to EFW sites in MCM-22. Further, stabilization
of iron in the pore wall of mesoporous structures is shown to depend on the synthesis (SBA-
15 and MCM-41). Finally, the partly amorphous character of pore walls in mesoporous
structures is manifested in the appearance of novel Fe z+ states (not detected in microporous
systems).
2. E X P E R I M E N T A L
2.1. Materials
Low iron content Fe,A1-ZSM-5 samples were synthesized by applying 57Fe as iron source.
Both the Si/Fe and Si/AI ratios were 200 in the synthesis gel. The hydrothermal crystallisation
was conducted at 430 K using tetrapropylammonium bromide as template. Two ZSM-5
samples are characterized. The first was in the calcined (Na +) form. An additional removal of
EFW iron species was attempted on the second sample by applying a reductive deferration by
HC1 + NH2-OH treatment on the calcined ZSM-5. Further details of the synthesis and
subsequent treatments are reported in [10].
Fe-MCM-22 sample was prepared from amorphous silica (Cab-O-Sil) in basic conditions
with hexamethyleneimine template at 410 K. For synthesis, composition of Si/Fe = 13.5 was
used. The product was calcined at 850 K in its Na + form, as described in [11 ].
Fe-SBA-15 was prepared by using Pluronic 123 surfactant dissolved in 1.9 N HC1.
Thereafter, FeCI3 and TEOS were added under stirring. Afterwards the synthesis took place
under static conditions in the original acidic media at 373 K. After filtration and drying, the
solid product was calcined at 770 K. The composition expressed in Si/Fe ratio is 76. The
average pore diameter in the particles is 8.5 nm with wall thickness of 2.3 rim. Further details
are described in [12].
Spherical Fe-MCM-41 particles were synthesized in a fast, room temperature procedure,
in alkaline media. C16TMABr, water, ethanol and TEOS were mixed then Fe(NO3)3 was
added to the solution (nominal Si/Fe ratio is 50). In the next stage gel was formed by
ammonia addition. Template was removed by heating in nitrogen to 773 K, thereafter
calcination followed in air at this temperature. Spherical particles were formed, with average
diameter of 250 nm, 2.5 nm pore diameter and 1.9 nm wall thickness. Further details and
characterisation are described in [13].
however, in some cases at a modestly elevated temperature ( ~- 500 K), to avoid sorption of
water occasionally present in traces in the system. (It is worth reminding here that the IS value
depends on the temperature of measurement; c.a. - 0.1 mm/s shift can be accounted for A T =
200 K).
3. R E S U L T S AND DISCUSSION
3.1. (Fe,AI)-ZSM-5
Iron containing ZSM-5 catalysts are usually applied in their H + form, but they are used in
their preceeding Na + form sometimes. The behaviours of the two forms were studied in low
iron-content samples (Si/Fe = 200). (See Fig. 1 for M6ssbauer spectra and Table 1 for data of
the Na + form, and Fig. 2 and Table 2 for those of the H + variant, respectively.)
Both similar and different features are apparent from comparison of the Na+and the H +
(HC1 + NH2-OH treated) forms of (Fe,A1)-ZSM-5. The dominant presence of a broad
component of small quadrupole splitting is characteristic for both samples (QS < 0.5 mm/s).
This attests for the presence of the Fe 3+ species located in highly symmetric, tetrahedral
framework positions [6,7]. The second common feature is that only a small portion of Fe 3+
can be reduced to Fe 2+ state in both samples (even in hydrogen above 600 K). This
observations provides a further support for the FW location of iron in high proportion.
The differences between the Na + and H + forms are reflected only on smaller portions of
spectral area. For example, autoreduction proceeds on the Na + form upon evacuation in a
lesser extent than on the H + form (4 % R / v a l u e vs. 14 % RL respectively). Or, smaller QS
value is found for the Lewis acidic (Si-O)3---Fe-O(Na)-Si site (1.44 mm/s) than shown for the
Bronsted acidic (Si-O)3-Fe-OH-Si group in the H + form (1.56 ram/s). (For the proposed
assignment of Lewis and Bronsted acidic sites see eg. [5].)
Table 1
Data from selected M6ssbauer spectra in sequential
treatments on Na + form of (Fe,A1)-ZSM-5 sample
(Tm is the temperature of measurement; IS isomer shift,
ram/s; QS, quadrupole splitting, ram/s; RI relative
intensity of the species in % of spectral area)
Treatm./(Tm) Comp IS QS RI
As rec (300 K) Fe 3+ 0.27 - 100 evac.
750 K ox, evac. Fe 3+ 0.23 1.44 41
(Tin: 300 K) Fe 3+ 0.32 0.46 55
Fe 2+ 1.30 1.54 4
670 K, H2 Fe 3§ 0.02 1.65 3
(Tm: 520 K) Fe 3+ 0.23 - 71
Fe 2+ 0.61 0.73 18
Fe 2+ 0.99 1.04 9
evac ~~~t #
' i , i , r . . . . 9
Table 2
Data extracted from in situ M6ssbauer spectra of the
HCI + NH2-OH treated (Fe,A1)-ZSM-5 (for caption,
see Table 1)
f/
.
Treatm. / (Tin) Comp IS QS RI evac %
v f
660K "~
Calc, & deferr. Fe 3+ 0.27 100
660 K, evac. Fe 3+ 0.23 1.56 ll
(7,,,: 300 K) Fe 3+ 0.05 - 75
Fe 2+ 1.16 2.10 14
610 K, H2 Fe 3+ 0.07 1.72 11
(Tin: 540 K) Fe 3+ -0.08 - 82
Fe 2+ 0.94 1.25 5
300 K, air Fe 3+ 0.26 100
-4 ' -2 ' 0 ' 89 4
It is worth also noting that the starting sate can seemingly be restored at the end of series
of the treatments as the reappearance of the broad central line shows (Fig. 2, bottom).
Namely, the dominance of the tetrahedral framework Fe 3+ siting is restored regardless to the
transient appearance of the octahedral EFW Fe z+ species. Thus, it seems that Fe 3+ ~ Fe 2+
reversible process may proceed in a small part of FW ions as well.
3.2. Fe-MCM-22
The MCM-22 framework is rather specific, consisting of a 10 member ring channel
system along with large supercages (0.7 nm in diameter and 1.8 nm in height). In the present
study this feature of the structure is correlated with the stability of iron species as reflected in
the in situ spectra obtained on the Na + form of the Si/Fe = 13.5 sample. Spectra are shown in
Fig. 3, data are collected in Table 3.
The first treatment, evacuation results already in a characteristic spectrum: presence of
(Si-O)3-Fe(OH)-Si Bronsted acidic centres is clearly manifested, i.e. charge compensating
Na + ions are removed from the vicinity of FW iron in a large portion. (The evacuation was
performed on sample stored on air previously, hence containing adsorbed water in a
significant amount). Simultaneously, autoreduction takes also place with removal of FW iron
and forming Fe 2~ in 16 % portion. A large part of the FW iron is still retained in hydrogen at
620 K - the proportion of Fe 2+ increases to 42 %. The structure of the original framework can
partially be restored by a calcination - the second evacuation exhibits FW siting for iron still
in c.a. 50 % proportion in relative spectral intensity. Among these conditions, formation of
Fe3+vw -(O,OH)- Fe2+EFWcan also be expected, since iron removed in the first reduction may
form neutral Fe2+-oxo species which is able to migrate inside the channels. In the following
oxidation this species may be transformed to Fe 3+ again. Hence, it is demonstrated that the
repetition of oxidation/evacuation/reduction cycle results in Fe 2+ ~ Fe 3§ reversible redox
process, connected with profound rearrangenment of iron siting in one part, together with
preserving the FW positions for the other portion of Fe 3+.
737
e
t calc. ~, .z
'-; i
-.a f 620 K
evac.
v V
:.~ .S
620 K s t~ lr
#
'-'i
-4 -2 0 2 4 -4 -2 0 2 4
mm/s
3.39 Fe-SBA-15
In contrast to the previously discussed two cases, SBA-15 is a mesoporous system with
average pore diameter of 8.5 nm. As mentioned in the Introduction, the crystallinity is less
expressed, and the pore wall is amorphous in an extent. Furthermore, the synthesis is
performed in strongly acidic media, resulting in preponderance of single ionic iron species in
the gel. Thus, incorporation of iron into the pore wall can be mostly expected in single ionic
form. The in situ M6ssbauer spectra obtained on the sample with a Si/Fe ratio of 76 in the
series of specified treatments are shown in Fig. 4, the respective data are extracted in Table 4.
Table 3
Data extracted from M6ssbauer spectra of Fe-MCM-22 (for caption see Table l)
1st cycle Repeated cycle
Treatm/(7,,). Comp. IS QS RI IS QS RI
Calc. Fe 3+ 0.29 0.58 27
(Tm: 300 K) Fe 3+ 0.36 0.73 35
Fe 3+ 0.36 1.02 73 0.36 1.26 65
Evacuation Fe 3§ 0.24 1.92 56 0.26 2.05 51
at 690 Fe 3+ 0.33 1.22 28 0.32 1.27 47
(Tin: 300 K) Fe 2+ 1.14 1.86 16 1.14 2.26 2
H2 at 620 K Fe 3+ 0.06 0.23 8
(Tm: 490 K) Fe 3+ 0.13 1.73 39 0.12 1.76 30
Fe 3+ 0.34 0.50 17
Fe 2+ 0.92 1.57 44 0.95 1.40 62
738
Table 4
Data extracted from sequential spectra of Fe-SBA- 15
(for caption, see Table 1)
Treatm/(Tm) Comp. a IS QS RI
as rec. Fe 3+ 0.33 1.38 58
Calc 770 K Fe 3+ 0.36 0.80 38
(Tin" 300 K) (Ve4+) - 0.18 - 5
Evac / Fe 3+ 0.28 1.94 41
650 K Fe 3+ 0.34 1.16 51
(Tin" 300 K) Fe 2+ 1.03 2.24 8
CO / Fe 3+ 0.17 1.94 45
620 K Fe 3+ 0.24 1.09 51
(Tm" 490 K) Fe z+ 0.90 1.91 4
H2 / Fe 3+ 0.13 1.78 23
620 K Fe 3+ 0.33 0.67 38
(Tm: 490 K) Fe 2+ 0.79 1.79 39
The partial crystallinity of the structure can be firstly derived from the spectrum
recorded after evacuation: Fe 3+ ions are distributed ca. evenly among distorted and more
symmetric environments (with QS values 1.9 and 1.2 mm/s, respectively). Simultaneously,
autoreduction is hardly noticeable, Fe 2+ is present only in 8 % spectral portion. It is probably
also connected to the lack of strict crystallinity- the removal from the framework with
simultaneous reduction cannot be interpreted for the case of amorphous pore walls. CO
reduction does not result in significant change in the shape of spectrum, i.e. Fe-O-Fe pairs are
absent. The reduction with hydrogen may result in more expressed reduction since single Fe 3+
ions may be reduced by forming hydroxyl bridges or silanolic OH groups in their vicinity.
3.4. Fe-MCM-41
Iron located in MCM-41 of spherical shape synthesized with smaller pores (av. diameter
2.5 nm) exhibits both similar and distinct features, as compared to the previous case of SBA-
15. In situ M6ssbauer spectra for the Si/Fe = 50 sample are shown in Fig. 5, data are extracted
in Table 5.
As for coordination states and iron types, MCM-41 shows similar features as found in
SBA-15. Evacuation results in close to even distribution among the distorted and more
symmetric environments of Fe 3+ (QS values are slightly smaller than previously, 1.7, and 1.0
mm/s respectively). Reduction with CO is more efficient, in particular for the Fe 3+ of distorted
symmetry. This feature is even more expressed in the case of reduction with hydrogen. This
preferred reducibility of the distorted component can probably be related to the pore wall
structure.
It is worth to make a short comparison here with Fe-MCM-41 of ,,conventional"
hydrothermal preparation (Si/Fe = 137, synthesized using the same C~6TMABr template at
373 K). In this latter case the Fe 3+ ~ Fe 2+ reduction was complete in hydrogen [9], in contrast
with the remaining Fe 3+ found in 66 % spectral area recently. This remarkable difference can
739
Table 5
Date extracted from M6ssbauer spectra of Fe-MCM-41
(tbr caption, see Table 1)
Treat/(Tm) Comp. IS QS RI
as rec. F e3+ 0.33 1.33 53
N2 770 K Fe 3+ 0.34 0.82 44
(Tin: 300 K) (Fe 4+) -0.06 0.29 2
Evac / Fe > 0.30 1.71 48
660 K Fe 3+ 0.34 1.02 49
(Tin: 300 K) Fe 2+ 1.13 2.19 4
CO / Fe 3+ 0.18 1.80 32
620 K Fe 3+ 0.29 0.98 57
(Tin 490 K) Fe z+ 0.66 2.21 11
H2 / Fe 3+ 0.18 1.73 14
620 K Fe 3+ 0.37 0.80 52 -4 -? 6 ' 4
(Tin 490 K) Fe 2§ 0.74 2.00 33 mm/s
Fig. 5. M6ssbauer spectra
of Fe-MCM-41
probably be attributed to the difference in the pore wall structure: more abundant presence of
silanolic-OH groups can be suggested in the sample of 373 K preparation with their active
participation in the Fe 3+ ++ Fe2+redox processes as described in [9].
4. C O N C L U S I O N S
Existence of a variety of coordination and oxidation states of iron and their interconversions
are demonstrated under various conditions in porous ferrisilicates. In crystalline microporous
systems, dominance of tetrahedral FW emplacement of iron is demonstrated in low iron
content ZSM-5 samples (Si/Fe = 200). Most of the FW Fe 3+ ions retain their coordination and
valency upon evacuation and treatment in hydrogen. Slight differences can be detected
between the Na + and H + forms, the former is even more stabilized.
In the more open MCM-22 framework the migration of various extra-framework
species is less restricted, rearrangement of ions may take place as was revealed from sudies
performed on a sample, more abundant in iron (Si/Fe = 13.5). Namely, sodium may leave the
original charge compensating sites. These sites can be occupied by protons or Fe2+-oxo
species formed upon evacuation and partial removal of iron from framework sites. By
repeating the oxidation/reduction cycles formation of Fe3+rw- (O,OH) -FeZ+E~w dinuclear
species can also be assumed. In these pairs the reversible redox Fe 3+ ~ Fe z+ transformations
may easily proceed.
In mcsoporous structures of SBA-15 and MCM-41 the F W - EFW distinction of iron
ions is not relevant since the structure of pore walls is partially amorphous. The stabilization
is provided primarily by the flexibility of structure in the pore walls, in which Fe 2+ ions can be
accommodated as well (in contrast to microporous ferrisilicates in which Fe 2+ occupies EFW
sites). In the pore walls various coordinations are available for both Fe 3+ and Fe 2+, enabling
thereby reversible Fe 3+ ~ Fe 2+ processes to take place in mesoporous structures as well.
740
ACKNOWLEDGEMENT
The authors are grateful to Jifi 0ejka for preparation and providing the Fe-MCM-22 samples.
REFERENCES
[1] G. Centi, B. Wichterlov/~ and A.T. Bell (eds.), Catalysis by Unique Metal Ion Structures, NATO
Sci. Series II. Vol. 13. Kluwer, Dordrecht. 2001.
[2] S. Schacht, M. Janicke and F. Schtith, Microporous Mesoporous Mater., 22 (1998) 485.
[3] R.G. Bums, Hyperfine Interact., 91 (1994) 739.
[4] K. Lfizfir, G. Pfil-Borbdly and H. Beyer, Zeolites, 11 (1991) 214.
[5] K. Lfizfir, A. M.-Szeleczky, G. Vorbeck, R. Fricke, A. Vondrovfi and J. 12ejka, J. Radioanal.
Nucl. Chem., Articles 190 (1995) 407.
[6] A. Meagher, V. Nair and R. Szostak, Zeolites, 8 (1988) 3.
[7] R. Kumar, S.K. Date, E. Bill, and A. Trautwein, Zeolites, 11 (1991) 211.
[8] A. Tuel, I. Arcon., and J.-M. Millet, J. Chem. Soc., Faraday Trans., 94 (1998) 3501.
[9] K. Lfizfir, G. Pfil-Borbdly, A. Szegedi, and H.K. Beyer, Stud. Surf. Sci. Catal., 142 (2002) 1347.
[10] P. Fejes, K. Lfizfir, I. Marsi, A. Rockenbauer, L. Korecz, J.B. Nagy, S. Perathoner, and G. Centi,
Appl. Catal., A, 252 (2003) 75.
[11] J. 12ejka, J. D6de6ek, J. Kotrla, M. Tudor, N. 2;ilkovfi and S. Ernst, Stud. Surf. Sci. Catal., 135
(2001) 352.
[12] K. Lfizfir, G. Calleja, J.A. Melero, F. Martinez and R. Molina, Stud. Surf. Sci. Catal., 154(2004)
805.
[13] /i,. Szegedi, Z. K6nya,.D. M6hn, E. Solymfir, G. Pfil-Borbdly, Zs.E. Horvfith, L.P. Bir6 and I.
Kiricsi, Appl. Catal., A, 272 (2004) 257.
Studies in Surface Science and Catalysis, volume 158
J. 12ejka,N. 2;ilkovfiand P. Nachtigall (Editors)
9 2005 ElsevierB.V. All rights reserved. 741
a Institute of Material Chemistry, VSB - Technical University of Ostrava, 17. listopadu 15,
708 33 Ostrava- Poruba
b Institute of Inorganic Chemistry, Academy of Sciences of the Czech Republic, 250 68 l~e~
near Prague, Czech Republic
cj. Heyrovsk~ Institute of Physical Chemistry, Academy of Sciences of the Czech Republic,
Dolejgkova 3,182 23 Praha 8, Czech Republic, brabec@jh-inst.cas.cz
Silicalite-1 was grown hydrothermally in two batches with crystals 2-3 and 200-250 lam in
length. Small crystals are penetrating overgrowths with a cruciform cross section, large
crystals are typical columns known from the literature. X-ray and electron diffraction show
the material to be orthorhombic, with the Pnrna space group and unit cell a = 20.0801(2) A, b
= 19.9239(2) A, c = 13.4167(1) A. Silicalite-1 is ubiquitously twinned on (110), and thus all
tour prominent vertical crystal faces on the large columnar crystals belong to the form {100}.
In early stages of their growth, the untwinned crystals are tabular on (010) and elongated
along [001], with prominent crystal forms {010}, {100}, and { 101 }. Etching of large crystals
with diluted HF and optical examination of the material in polarized light offer an explanation
of the hourglass structure and provide new clues about the growth process.
1. INTRODUCTION
There is a continuing discussion in the literature concerning various aspects of the internal
structure of large silicalite-1 crystals. Of primary importance is the understanding of the
growth of silicalite-1 that might provide answers to questions about intracrystalline mass
transport. Important papers on the topic were published by Hay et al. [1], Geus et al. [2],
Weidenthaler et al. [3] and recently by Agger et al. [4] who used AFM microscopy to analyze
terraces on crystal faces and FTIR to study individual crystal zones. In these papers is also
mentioned an internal hourglass optical effect, well visible in polarized light. Ko6ifik et al.
[5,6] observed an analogous effect on vertical faces of the crystals using sorption of iodine
from the liquid phase. However, the development of the coloring patterns has not been
explained satisfactorily.
The present research was undertaken to ascertain whether single-crystal X-ray and
electron diffraction applied to two different size categories of silicalite-1 can provide
additional clues that might shed new light on some of the above problems.
742
2. EXPERIMENTAL
3. RESULTS
C ~ C ~
9
t
9 9
t
eQ,
.o+ , . . .
. . .
Q . . . . . 9
, O" 9 .e.o- -O 9
9 9 9 - + 9
9 -e.g.. 9 O.e . . . .
9 + 9 . 9 , ,. e e e . , . . . . .
I .... I 9O . o , . O . . . . .
I ....9 9 ~
I 9 .
O . o . - O
. , l e e +
. .
. . . . .
. .
9 9 9 - 9 9 9 -e-O. .O.- . . . .
9 9 9 . e . e . 9 9 9
9 9 9 9 9
9 . . . . . 9
~ ~
a ~ . 9 + -Oe U
Fig. 1. Precession photographs: a - reciprocal net Okl (simulated); b- reciprocal net hOl (simulated);
c - reciprocal nets Okl and hOl superimposed (large crystal, Mo-K rac ~ation, precession angle la = 15~
744
Fig. 2. SAED patterns of chips obtained by crushing of large crystals: a - along [100]" b - along [010].
B A
Fig. 3. SAED patterns (beam stopper blocked the central beam): a - from area A; b - from area B;
c - bright field image of a twin with marked circular areas A, B.
745
Fig. 5. Small calcined crystals: left photo - not etched crystals; right photo - crystals etched with 4%
HF acid for 3 min
be expected (Fig. 4, left image). Although all four vertical faces are crystallographically
equivalent (100), they exhibit a non-uniform resistance to etching. So, on faces (|00)A
(labeling from Fig. 6) there often are acute-angle triangular areas that appear less damaged by
etching than the adjacent obtuse-angle triangles (Fig. 4, right image). Admittedly, the variable
resistance to HF may be due to a variable density of structural defects and/or to a somewhat
variable chemical composition [4]. Interestingly, on calcined large crystals (i.e. without TPA
in pores) no similar etching patterns developed, the crystals dissolved to a shapeless mass.
The small crystals have a different morphology (Fig. 5, left image), and they developed
no etching patterns on crystal faces even though they were calcined. Thus, a very low amount
of surface defects is assumed. Etching of small crystals sometimes resulted in disconnecting
one individual from the other, leaving behind a pyramid-shaped pit (Fig. 5, right image).
4. DISCUSSION
[~--IN-j
0oo). (~oo).
1
0 'I 1
1
Fig. 6. A model of the development of a penetrating twin into a column. Left drawings show a
perspective, right pictures are sections parallel to [001] of both orientations.
systematic extinctions except along all axes. The twinning also gives the hkO net its tetragonal
appearance, compounded by the very small difference between a and b, which cannot produce
visible spot splitting in precession photographs (splitting would only be seen in high-angle
Weissenberg photographs, see Price et al. [8]. Weidenthaler et al. [3] observed the same
effects in their crystals of silicalite-1.
The appearance of penetrating twins with a cruciform cross-section offers a mechanism
of explaining several observations, assuming that one orientation (Fig. 6, "A", shaded)
nucleated before the other ("B", unshaded). The contact surface between orientations A, B in
sections parallel to (001) is defined by the velocity of growth along [010] of both orientations.
These velocities must be the same, and the contact surface will make an angle of ~45 ~ with
the original planes (010). However, inasmuch as the growth of orientation B is lagging
behind, the contact surface parallel to [100]A (and [010]B) will be governed by the velocity of
growth along [010]A (low) and [00lIB (high), and will make a small angle with the (010)A
crystal face. Therefore, the horizontal lines shown in cross-sections in Fig. 6 will be
progressively wider apart as the section approaches either of the terminations of the column.
This model explains why the hourglass effect is better visible when viewing the
columnar twin down the [|00]A direction than along [010]A, as noticed or documented by
Weidenthaler et al. [3] (their Figs. 10 and 12) and Agger et al. [4] (their Fig. 5b) and as we
could ascertain in oil immersion under a polarizing microscope. Generally, an orthorhombic
crystal viewed down [100] and [010] will display different birefringences and, depending on
the thickness of the crystal, different interference colors. In a twin, both orientations
superimposed will be seen if the crystal is viewed along [010]A and their share on the net
optical effect will vary with the percentage of each orientations, changing only mildly from
the center of the twin outwards. However, when viewed along [100]A, orientation A will
748
represent 100% of the thickness of the twin in its central part, and its volume will widen
towards both ends, whereas orientation B will dominate the rest of the twin's volume. In this
view, the two orientations do not appear on top of each other but next to one another,
producing a well-visible hourglass effect. Another consequence of the model is that
orientation B will constitute clearly less than 50 volume % of the overgrowth, and that it
ultimately will form two flat pyramidal shapes as shown by Price et al. [8] or by Weidenthaler
et al. [3]. Incidentally, this division of volume between the two orientations is well visible on
some crystals etched with HF.
The initial morphology of silicalite-1 apparently was like that of the small crystals, and
later in the growth process evolved into the morphology of large crystals. During this
development, the velocities of growth along [100] and [010] must have changed, while the
velocity of growth along [001] remained highest throughout. The velocity along [100] must
have come to a virtual halt and, conversely, the velocity along [010] must have increased.
Interestingly, the areas on (100) faces of large crystals more damaged by etching with HF are
those that formed by late-stage growth along [010]. Let us mention that chemical zoning of
silicalite-1 crystals has been reported by Agger et al. [4]. Such effects might be associated
with changes in growth velocities.
5. CONCLUSIONS
X-ray single-crystal diffraction shows that the presently grown silicalite-1 is ubiquitously
twinned on (110). SAED on chips of large crystals and on small penetrating twins confirms
that the untwinned material is orthorhombic and corresponds to the space group and unit cell
reported in the literature. X-ray powder data support at the same conclusion. SAED confirms
that the crystals are tabular on (010) and elongated along [001]. Prominent crystal forms are
{010}, {100}, and { 101 }. The twin law is rigorously a reflection in (110) rather than a 90 ~
rotation about the e axis. Etching of silicalite-1 twins with diluted HF acid and optical
microscopy permitted us to propose a model explaining why the internal hourglass structure is
better visible in one direction than at right angles to it. The availability of two batches of
crystals, 2-3 and 200-250 ~tm in length, and their different morphology indicate that the
growth rates along axial directions evolved during the growth process.
ACKNOWLEDGEMENT
This research was supported by the grant of the Academy of Sciences K4040111, by the grant
GAAV A400400501 and by the grant GA(~R 203/05/0846.
REFERENCES
[1 ] D.G. Hay, H. Jaeger, K.G. Wilshier, Zeolites 10 (1990) 571.
[2] E.R. Geus, J.C. Jansen, H. van Bekkum, Zeolites 14 (1994) 82.
[3] C. Weidenthaler, R.X. Fischer, R.D. Shannon, O. Medenbach, J. Phys. Chem. 98 (1994) 12687.
[4] J.R. Agger, N. Hanif, C.S. Cundy, A.P. Wade, S. Dennison, P.A. Rawlinson, M.W. Anderson, J.
Am. Chem. Soc 125 (2003) 830.
[5] M. Ko~ifik, J. Kornatowski, V. Masafik, P. Nov~tk, A. Zikfinovfi, J. Maixner, Microporous
Mesoporous Mater. 23 (1998) 295.
[6] V. Masafik, P. Novfik, A. Zikfinovfi, J. Kornatowski, J. Maixner, M. Ko6ifik, Collect. Czech.
Chem. Commun. 63 (1998) 321.
[7] D.H. Olson, G.T. Kokotailo, S.L. Lawton, J. Phys. Chem. 85 (1981) 2238
[8] G.D. Price, J.J. Pluth, J.V. Smith, J.M. Bennett, R.L. Patton, J.Am. Chem. Soc. 104 (1982) 5971.
[9] M. Schneider, Program for Rietveld analysis of X-ray and neutron powder diffraction patterns,
EDV-Vertrieb, 1994.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka, N. Zilkovfi and P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 749
2Laboratoire de Rdactivit6 de Surface, UMR 7609 CNRS, Universit6 Pierre et Marie Curie
(UPMC), 4 place Jussieu, 75252 Paris Cedex 05, France. + Institut Universitaire de France.
3lnstitut de Recherches sur la Catalyse, UPR 5401 CNRS, 2 avenue Albert Einstein, 69626
Villeurbanne Cedex, France.
1. I N T R O D U C T I O N
In the last decades, zeolites exchanged with transition metal ions have received increasing
interest as catalysts for a variety of important reactions [1]. Cobalt loaded in pentasil zeolites
was found to be highly active in the decomposition of nitrous oxide, selective catalytic
reduction of NO• [2,3] or, more recently, in the ammoxidation of light alkanes [4-6]. Cobalt
ions can be introduced into zeolites by different methods, e.g., incipient wetness
impregnation, decomposition of organometallic complexes, hydrothermal synthesis or ion
exchange, carried out in liquid solution, by sublimation or by a solid-state procedure. This
work deals with the preparation of Co-ZSM-5 catalysts by the solid state ion-exchange
procedure, in an effort to minimize the metal loss during the synthesis, and is aimed at
studying the influence of the Co/A1 ratio on the nature of cobalt species loaded in ZSM-5
(stoichiometric exchange ratio for Co/A1 - 1/2 and overstoichiometric exchange ratio for
Co/AI = 1 or 3/2) and the catalytic performance of these species in ammoxidation reactions.
750
2. EXPERIMENTAL
two chromatographic units, one operated with a flame ionization detector while the other was
equipped with a thermal conductivity detector.
3. R E S U L T S AND DISCUSSION
Table 1
BET surface area and porosity measurements.
Microporous Porous volume Average pore
Catalysts SeEr (m2/g)
volume (cm3/g) (cm3/g) d i a m e t e r (A)
H-ZSM-5 545 0.15 0.81 36
Co-Z-A1/2 360 0.10 0.53 36
Co-Z-A1 350 0.10 0.52 37
Co-Z-A3/2 360 0.10 0.54 37
acidic Si-OH-AI groups during the exchange process; new Lewis acid sites appear in the
catalysts, possibly due to the presence of accessible cobalt ions [14].
1.4 ..............................................................................................................................................................
1.2 14911
1622 1546 II
~0.8
== A/X, /~ I[ 1455
@
.~0.4
z 1610 ]11450 i
0.2
0 "~ ' ......... I ~" ..... 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ions in the zeolite. It can be calculated that the molar ratio H2/Co is lower than 1 for Co-Z-A 1
and Co-Z-A3/2; some cobalt species cannot be reduced due to their inaccessibility [ 19].
On Co-Z-A 1/2, the major reduction peak corresponds to the reduction of Co304 which
is the main species in which cobalt is present, in line with XRD results. Co 2+ ions, either
dispersed or in the phyllosilicate phase, are the major species on Co-Z-A1 and Co-Z-A3/2,
oxide particles being present as traces o n l y - also in line with XRD. However it is difficult to
go any further in the quantification of Co 2+ ions in or out of the phyllosilicate layers due to the
fact that only one peak of reduction is observed for the two species and that reduction of
cobalt is not complete in the conditions of TPR. It can just be noticed that since the maximum
temperature of the high-temperature peak is shifted to higher temperatures when the
proportion of phyllosilicate increases [7], a higher proportion of phyllosilicate is expected in
Co-Z-A1 and Co-Z-A3/2 than in Co-Z-A1/2.
Table 2
H2-TPR results on the three Co-Z-A catalysts
T Co304 H2 C o 3 0 4 T C o 2+ H2 Co 2+ Molar ratio
Catalyst (~ (pmol/gcat) (~ (lamol/gcat) Co 2+ / Co304
Co-Z-A1/2 338 170 680 70 0.6
Co-Z-A1 358 25 783 250 13.3
Co-Z-A3/2 360 10 775 480 64
The EXAFS signal of Co-Z-A1 does not correspond to that of a single compound,
neither to a combination of EXAFS signals from two reference compounds, phyllosilicate +
isolated ions, or phyllosilicate + Co304. The best fit is obtained from the parameters of
phyllosilicate but with N values lower than those expected in a well-crystallized compound
(Table 3). On the FT, the peaks corresponding to the different shells are actually less intense
than for Co-Z-A3/2 (Fig. 3). It is supposed that cobalt phyllosilicate is present on Co-Z-A1 as
an ill-crystallized and defective phase; the EXAFS spectrum is the result of the superposition
of the oscillations from this phase and from the environment of isolated ions in a proportion
that cannot be determined, with Co304 as a non-interferring minor species.
754
Co
7 I
0.4-
/• experimental b
6
0 0.2-
5
4
o-
3
-0.2 -
1 I
0.6 signal (Co30 4 ) + 0.4 signal (LEN)
0 I I . .I . . I I
-0.4 I I I I I
2 3 4 5 6 4 6 8 10 12
R(A) k(A q)
Table 3
Best parameters for the fits of Co-Z-A1 and Co-Z-A3/2 1st and 2 nd shells
Sample Atom N or(A) R (A) AEo (eV) /9 (%)
Co-Z-A1 O 5.3 0.10 2.06 -3 0.2
Co 3.5 0.10 3.14 -7
Si 3.6 0.10 3.22 -7
Co-Z-A3/2 O 6.0 0.10 2.08 -3 0.3
Co 6.0 0.10 3.13 -6
Si 4.0 0.10 3.20 -8
N = number of neighbours, cy = Debye-Waller factor, R = distance between Co and a backscatterer,
AE0 = energy shift, p - agreement factor (2[kZth(k)-kZe•215
6 a b
4 4
2 2
~" o ~" o
-2 -2
-4 -4
I I I I I I I I I I I I I ! I I
0 1 2 3 4 0 1 2 3 4
R (A) R (A)
Fig. 3. k3-weighted Fourier transforms of a) Co-Z-A3/2 and b) Co-Z-A1 EXAFS signals limited to the
first two shells ( ....... experimental'--' simulation from parameters in Table 5).
755
3.8. Catalysis
The results compiled in Tables 4 and 5 illustrate the catalytic performance in ethylene and
ethane ammoxidation of the three catalysts prepared by solid-state exchange and of low-
loaded LEN prepared from cobalt nitrate by aqueous exchange. For all the catalysts, a steady
state is reached after a transitory period of 180 minutes, characterized by an increase of the
hydrocarbon conversion. During this period, the selectivity into acetonitrile (CH3CNCN)
increases while that relative to the deep oxidation product (CO2) decreases.
Table 4
Ammoxidation of ethylene on Co-ZSM5 catalysts: catalytic measurements
Total conversion Act. CH3CN (10 .4 m ol Selectivity CH3CN
Catalysts (%) g-i s-l) (%)
T(~ 450 475 500 450 475 500 450 475 500
LEN 2.0 3.6 7.95 0.98 1.90 3.90 96 97 98
Co-Z-A1/2 0.4 1.5 19.0 0.12 0.57 5.87 61 84 98
Co-Z-A1 4.5 11.6 28.0 2.07 6.50 13.09 98 99 99
Co-Z-A3/2 3.9 9.5 24.2 1.83 4.81 11.8 97 98 99
Table 5
Ammoxidation of ethane on Co-ZSM5 catalysts: catalytic measurements
Act. CH3CN (10 -4 mol Selectivity CH3CN
Catalysts Total conversion (%) g-I s-l) (%)
T(~ 450 475 500 450 475 500 450 475 500
LEN 4.0 7.1 9.8 1.00 2.10 2.90 67 80 79
Co-Z-A 1/2 5.9 7.7 10.3 1.91 2.74 3.59 74 83 82
Co-Z-A 1 8.5 13.2 20.6 3.24 5.28 8.23 84 84 83
Co-Z-A3/2 9.5 11.8 14.7 3.60 4.36 5.22 84 79 76
The results obtained show that the catalyst prepared with Co/A1-1 exhibits the highest
activities (Act CH3CN) and selectivities toward acetonitrile from ethylene and ethane.
Co-Z-A3/2 is less active than Co-Z-A1 despite its higher cobalt content; this lowers
activity can be explained by the presence of inaccessible cobalt ions inside well-crystallized
phyllosilicate. On the other hand, sample LEN contains 7 times less cobalt than Co-Z-A1 but
is only 2.5 to 3.5 times less active in C2H4 ammoxidation, with similar selectivities;
accessible isolated ions seem to be active sites for ethylene ammoxidation.
When the reactant is C2H4,the selectivity to acetonitrile is close to 100% on all the
samples except Co-Z-A1/2 for which the conversion is very low at 450 - 475~ and CO2 is
detected. The production of CO2 is linked to the presence of Co304 particles able to catalyze
ethylene combustion, but ammoxidation is favored if temperature and conversion increase.
When the reactant is C2H6, the selectivity to acetonitrile is close to 80% on all the
samples. Two other products are detected, ethylene and CO2. The selectivity in C2H4
increases with temperature and conversion, while the selectivity to CO2 decreases, as
observed for ethylene ammoxidation (on Co-Z-A1, the selectivities to C2H4and CO2 are 8%
and 8% respectively at 450~ 14.5 and 2.5% at 500~ The catalysts that give the higher
756
selectivities to CO2 are also the least active, LEN and Co-Z-A1/2 (at 450~ their selectivities
to CO2 are respectively 29 and 24%). The catalytic behavior of Co-Z-A1/2 recalls that
observed for ethylene ammoxidation and linked to Co304 particles, unlike LEN that did not
form CO2 when ethylene was the reactant; the transformation of ethane to acetonitrile needs
other types of sites than just isolated cobalt ions. This leads us to assume that the high activity
of Co-Z-A 1 for ethane and ethylene ammoxidation is due not only to the presence of isolated
cobalt ions, but also of unsaturated sites of the ill-cristallized cobalt phyllosilicate phase.
4. CONCLUSION
Using cobalt acetate as a precursor and for a Co/A1 ratio corresponding to the stoechiometry
of exchange (Co/A1 = 1/2), Co 2§ ions are only partially exchanged in the zeolite by a solid-
state procedure, the majority of cobalt being present in C0304 particles that catalyse the total
oxidation of the hydrocarbons. For overstoichiometric Co/AI ratios, thus when more acetic
acid is evolved during the heat treatment, a cobalt phyllosilicate phase is formed outside the
grains after dealumination of the zeolite. Isolated cobalt ions and unsaturated Co 2+ ions on the
phyllosilicate seem to be the active sites for hydrocarbon ammoxidation. While total
exchange is impossible to reach in solution, solid state exchange provides a simple way to
prepare an active and selective catalyst, but it must be acknowledged that a nominal
overstoichiometric Co/AI ratio is needed to reach this goal.
REFERENCES
The series of titania modified SBA-15 supports were prepared in post synthesis applying
titanium isopropoxide solutions and appropriate amounts of SBA-15. These supports were
used for preparation of molybdenum based catalysts. Ammonium heptamolybdate and
molydenum carbonyl were applied as the source of Mo. Results confirm that this method of
molydenum catalysts synthesis do not cause significant changes in symmetry of SBA-15. The
decrease of surface area and channel dimensions are the main effects of SBA-15
modifications both with TiO2 and Mo species. Only Lewis acid sites were found upon
incorporation of titania on SBA-15. The use of different sources of Mo leads towards
different oxidized molybdenum species which can be attractive in generation of their sulfided
or nitrided forms.
1. INTRODUCTION
Supported catalysts containing group VI and VIII elements are widely used in hydrotreatment
(HDS,HDN,HYD) processes. It is well known that nature of the support affects properties of
the active phase formed and consequently catalytic activity. The degree of interaction of
transition metal sulfides (TMS) or transition metal nitrides(TMN) with the support is quantita-
tively related to the active groups content on the support [1,2]. This interaction influences
mainly the dispersion of the supported TMS or TMN. Moreover, alteration of the support by
different modifiers [3] changes these interactions and finally catalytic activity.
Alumina, silica, aluminosilcates as well as titania [4-6] or zirconia[5,7] are the most
often oxides used as the catalytic supports. Although the role of the support in the
hydrotreatment catalysts is still controversial [8,9], the stabilization effect of supported TMS
and TMN is rather well documented [ 10-12].
The invention of the new family of the siliceous mesoporous materials by Mobil
[13,,14] designated as M41S family, with subsequent modifications, dramatically expanded
the the range of uniform pore size of ordered mesoporous materials and their possible
applications. Insertion of e.g. A1 atoms into the siliceous MCM-41 structures allows creation
of the acidic sites and formation of catalysts interacting e.g. with large organic molecules
from heavy feedstocks in rafineries. Moreover, incorporation of such transition metal sulfides
like molybdenum, tungsten and nickel or cobalt into the uniform channels of MCM-41 [15] or
SBA-15 can lead to the generation of a new class ol-HDS or HDN catalysts.
Works by Howe [16] and Gao [17] presented the first attempts to incorporate Mo or Ni-Mo
respectively into the MCM-41 materials. These early works documented that MoCI5 grafting
technique, impregnation with molybdate [16] or mixing of molybdenum oxide with MCM-41
at low concentrations generates materials with high dispersion of Mo species. In this work
758
2. E X P E R I M E N T A L
A series of molybdenum catalysts were prepared applying titania modified SBA-15 as the
support. SBA-15 was synthesized according to the procedure described by the group of
Stucky [18,19] applying Pluronic 123 block coplymer as the template in acidic conditions and
TEOS as the source of silicon. After crystallization of SBA-15 at 368 K for 24 hours, samples
were next washed with distlled water, filtered and dried in air. Template removal was
performed via calcination in air at 773 K for six hours and the final temperature was reached
in 8 hours.Titanium dioxide was supported on siliceous material via impregnation of titanium
isopropoxide in isopropyl alcohol solution, followed by hydrolysis with water. Usually 2
grams of template-free SBA-15 was suspended in 100 cm 3 of either 1% or 5 % solution of
titanium isopropoxide. Hydrolysis was performed at room temperature by adding the excess
of water (H20 : iso-PrOH - 5) while vigorously stirring. Samples with supported, hydrated
titanium dioxide were dried and calcined at 675 for 2 hours. Similar technique was applied by
Tuel [20] applying hexanuclear titanium oxo cluster as a source of Ti instead of Ti-iso-PrOH.
Molybdenum heptamolybdate or molybdenum hexacarbonyl were used as the sources of Mo,
while supporting on SBA-15-titania supports. The incipient wetness method was applied in
the case molybdenum heptamolybdate, whereas sublimation at room temperature in stream of
H2 in the case of molybdenum hexacarbonyl.
Both pure supports as well as those containing molybdenum were characterized with
XRD, XRF, thermogravimetry (Setaram SetSys TG-DSC 15), surface area and porosity
measurements (ASAP-2000), measurements of acidity applying pyridine as a probe molecule
(FTIR), The TPR experiments were performed with ASAP-2007 (from Micromeritics) in
which mixture of hydrogen (10 vol.%) and argon was used in all experiments.
Pure siliceous sample in this paper is designated as SBA-15, whereas pure titanium
dioxide as TiO2. Samples prepared by hydrolysis of titanium isopropoxide from 1% and 5%
solutions are designated as Si-Ti-1/0 and Si-Ti-5/0, respectively. Impregnation of the
calcined, titanium dioxide cantaining samples, with appropriately adjusted concentrated
solutions of ammonium heptamolybdate give samples Si-Ti-1/2, Si-Ti-I/4, Si-Ti-5/2 and Si-
Ti-5/4. The last number in these symbols represents the amount of supported Mo in wt. %.
4000
3000
~ 2000
SBA-15
(/1
C 9 Si-Ti-l/0
,=., 1000
_c
1 2 3 4 5 6 7 8
2O[ ~ ]
Fig. I. Low-angle X-ray diffraction patterns of SBA-15 and SBA-I5 with deposited titania.
(Si-Ti-5/0) respectively, the characteristic diffraction signal (d~00) is missing, indicating that
probaly internal walls of siliceous SBA-15 became covered with rather amorphous phase of
titania. The XRD diffraction patterns of titania containing samples, recorded at higher angles
(10-50 ~ of 2| did not show any reflexes originating from anatase. There was no reflexes
after deposition of Mo from ammonium heptamolybdate, as well. Similar effects were
observed when molybdenum hexacarbonyl was applied as the source of Mo.
Impregnation of titania modified samples with ammonium heptamolybdate, followed by
subsequent calcination at 675 K resulted in deposition of Mo in two different concentrations:
2 and 4 wt. %. The results of chemical analysis (1.92 wt.% and 3.83 wt.%) confirmed these
values. In the case of saturation of pure SBA-15 with Mo(CO)6 at room temperature more
than 90 % of weakly adsorbed carbonyl was removed during attempts of decarbonylation
Table 1
Characteristics of the supports and Mo loaded catalysts in oxidized state.
0,6
800 ~^ . ~ 3 ~
0,5 . . . . SBA-I~ ~ I
60o + Si-Ti-1/0 / / ~J
"T,
E " oo - - - ~ - Si-Ti'5/0 / / ~ ~ ; ~
e
.~, 0,4
o,4 4oo ~:.~/ . ~
%
~1 loo
0,2 o - - ~ ~ ~ -
0,0 0,2 13,4 0,6 0,8 1,0
~/. Relative Pressure [plpo]
0,1
0,0 !
0 5 10 15 20 25 30
Fig. 2. Pore size distribution and N2 isotherms (77 K) of titania modified SBA-15.
0,7
700
0,6 I~ 6eO ----0--- Si-Yi-llO
"T, I/ ~-- + si-ri-Mo-1/2
E
c 0,5
"T,
0,4 II i ....
0,3
O
> 0,2 I ~ oo 02 0,4 0,6 0,8 1,0
o
a_
0,1
0,0 i i | i |
5 10 15 20 25 30
Pore Diameter [nm]
Fig. 3. Pore size distribution and N2 isotherms (77 K) of Mo-loaded titania modified SBA-15.
process in flowing inert gas at higher temperatures. Practically, less than 0.5 wt. %(XRF
determination) of Mo was attached into siliceous surface after decarbonylation at 425 K,
whereas after decarbonylation at 575 K less than 0.1 wt. % remained on the surface. In the
case of sublimation of Mo(CO)6 onto the titania modified suppports the amounts of Mo after
decarbonylation at 425 K were 3.2 and 5.4 wt.%. Higher values are related to the supports
with higher content of titania. Decarbonylation of identical samples at 575 K gave the
concentration o f M o 2.3 and 4.5, respectively. These values are very close to those obtained
with impregnation procedure, and indicate that certain part of supported Mo carbonyl or
subcarbonyl (eg. Mo(eO)3ads) can be removed at higher temperatures of decarbonylation.
Table 1 presents BET surface area, pore volume and average pore radius of the studied
samples. Microporosity data were obtained from calculations based on cts plot method [21].
761
Pure SBA-15 with average pore diameter of 6.2 nm after deposition with TiO2 indicate the
significant decrease in BET surface area and porosity. However, no simple realtionship on the
decrease of channel dimensions with the amount of deposited titania can be observed. For
example, a deposition of 8 wt.% of TiO2 causes the decrease of surface area by 23%,
whereas deposition of 40 wt% of TiO2 changes initial surface area only by 40%. Similarily,
the pore volume after titania deposition decreases respectively by 26 and 46 %. Assuming that
titania attached to the silica channels indicate only microporosity one could expect decrease of
average pore diameter by a factor o f - 50 % in the case of 40 wt.% of deposited TiO2.
Comparing data in Table 1 and the results presented in Fig. 2, it become clear that BJH
method of calculation of average pore diameter is not very precise. The amounts of the
adsorbed nitrogen (see Fig. 2) as well as the shape of hysteresis loop show that after
deposition of titania, even in large quantities, the unique mesoporous structure of SBA-15 is
very well preserved. Moreover, pore size distribution shows that their size decrease with
increasing amount of titania, what is in agreement with appropriate decrease of pore volume.
Pure SBA-15 with micropores of 0.067 cm3g-~, similar to those already reported in literature
[21], after deposition of titania indicate decrease of micropores volume (see Table 1),
indicating that at least part ot titania is located in microporous system.
Deposition of molybdenum oxidized species (via thermal decomposition of ammonium
heptamolybdate) further decreases the dimensions of channels what is clearly exemplified in
nitrogen isotherms as well as in pore size distribution (see Fig. 3) for the support containing 8
wt.% of TiO2. Similar effects were observed for samples with higher content of titania (Si-Ti-
Mo-5/2 and 5/4) but hysteresis loop was slightly different and close to this presented in Fig. 4
(Si-Ti-5/Mo/300). These effects also finds the confirmation in values of BET surface area.
Deposition of molybdenum on titania modified SBA-15 supports results also in further
decrease of the micropores capacity. Support containing - 40 wt. % of TiO2 with 4 % of Mo
decreased BET surface area almost 50%. In both cases attachment of molybdenum to titania
resulted in narrowing of mesopores. However, even at high loadings the dimensions of
mesopores remained much bigger than those in MCM-41 structures [22]. An application of
Mo(CO)6 as a source of molybdenum resulted in adsorption of this compound over the basic
sites located on the titania surface. Subsequent decarbonylation at 425 K or 575 K
formed molybdenum species in which Mo ions at different oxidation state can be formed.
800
700
SBA-15
Si-Ti-l/Mo/300
e.-,
"7 Si-Ti-51Mol300
m 600
03
E
. o 500
.lo
.,o
400
"0
300
E
"~ 200
>
100
By analogy to alumina as the support, one can expect the presence of surface Mo 2+
ions after decarbonylation in He at 575 K [23]. Similarily, decarbonylation at 425 K in He or
H2 should lead towards surface Mo(CO)3ads. like species on the surface of titania. An
exposition of the samples prepared from Mo(CO)6 to the air, independently of the
decarbonylation temperature, resulted in oxidation of molybdenum surface species in which
Mo 6+ is predominant oxidation state. Although the concentrations of supported Mo via
decarbonylation technique are very close to those obtained in this paper from ammonium
heptamolybdate, the final surface species for samples non-exposed to air are completely
different and this behaviour will be the subject of another paper [24].
Samples containing similar content of Mo in oxidized state indicate, indpendently of
the applied technique, very similar shape of nitrogen adsorption-desorption isotherms (Fig. 4),
as well as very similar surface area (compare data in Table 1). Slightly higher surface area of
the samples containing even higher concentration of Mo (decarbonylation technique) than for
those obtained from (NH4)6Mo7024 can be an indication that different surface species can be
formed. TEM micrographs presented in Fig. 5 indicate that deposition of titania on siliceous
Table 2
Concentration of the Lewis acid sites [a.u.] after deposition of TiO2 on SBA-15.
**
R.T. - room temperature
time elapsed after pyridine injection into the helium stream
763
Si-Ti-1/Mol50l / ~ ~
_. 5,0
a
O
I- 4,5
T e m p e r a t u r e [K]
material even at large quantities do not change characteristic hexagonal array assigned to the
SBA-15. Moreover, the precise mesurements of pore dimensions confirm the results obtained
from adsorption of N2 at 77 K. Similar TEM micrographs were obtained after deposition of
molybdenum. All these results, as well as those obtained during nitrogen adsorption
measurements confirm that both titania and oxidized molybdenum species are located inside
the mesoporous structure. These results show additionally that thickness of deposited titania
can influence surface properties of generated supports and catalysts.
The FTIR acidity measurements with pyridine as probe molecule (see Table 2)
additionally confirm this assumption. The experiments performed in flow system in which
helium was applied as a carrier gas confirmed complete lack of acidity in SBA-15, although
large amounts of pyridine are physically adsorbed (large amounts of pyridine initially
adsorbed at room temperature) in very spacious porous system. As it can be seen even ti'om
few data presented in Table 2, the amount of the deposited titania can regulate overall Lewis
acidity (no Br6nsted acidity was found) of the prepared supports and in consequence further
deposition of molybdenum. The results presented in Table 2 show that at high (- 40 wt.%)
loadings of titania, there exist certain relationship between concentration of acid sites and the
content of TiO2, because sample Si-Ti-5/0 after 30 min. treatment at 375 K indicate half of
those observed for pure titania (obtained the same way as supported one). Although there is
no available data about the acidity of the supported molybdenum species our preliminary
results show that these properties can be tailored by appropriate amounts of titania and
molybdenum[24].
TPR spectra of the selected samples presented on Fig. 6 indicate that depending on
the amount of the supported molybdenum and the method of Mo deposition a large variety
of molybdenum species can be formed. This is demonstrated by different maxima obtained
during TPR experiments. If in the case of Si-Ti-1/2 sample (low Mo content) the presence of
characteristic two maxima at 695 and 1075 K can be assigned to the reduction of surface
Mo 6+ subsequently towards M o 4§ and Mo o [25] the assignement of small maximum at 875 K
is still unresolved. Similar effects were observed for SBA-15 supports containing 40 wt % of
TiO2. If the titania containing supports were loaded with higher amounts of Mo only two
maxima (825 and 1225 K) were present on TPR spectra. It is assumed that due to higher
concentration of Mo species located inside the mesopor~us structure of SBA-15, the
764
formation of aggregates of MoO3 with low homogeneity can occur. This in consequence
generate surface species with different affinity to reduction. Such effects were observed by
Zaki [26] while comparing surface Mo species deposited on titania and zirconia. Samples
obtained via decarbonylation of supported molybdenum hexacarbonyl show another type of
TPR spectra. Here, practically only one maximum a t - 795 K is observed which probaly can
be assigned to the reduction of more or less decarbonylated M o species [23] towards surface
Mo2+ ions.
Presented results indicate that SBA-15 is an appropriate support for deposition of
titania and molybdenum and can be used for preparation of specifically tailored
molybdenum catalysts. Sulfidation or nitridation of these catalysts can lead to formation of
the unique bifunctional catalysts for HDS, HDN or other hydrocracking reactions.
ACKNOWLEDGEMENT
This work was partially supported by Polish Ministry od Science and Informatization within
the project 7T09B 027 21.
REFERENCES
[1] H. Kn6zinger, Proc. 9 th Int. Congr. Catalysis, Calgary 1988 (M. J. Phillips and M. Teman Eds.)
Chem. Inst. Can. Ottawa, vol.5, 1988, p. 20.
[2] H. Topsoe, B.S. Clausen, N. Buriescu, R. Candia, S. Morup, Stud. Surf. Sci. Catal., 3 (1979) 479.
[3] J. Miciukiewicz, M.Laniecki, F. Domka, Catal. Letters, 51 (1998) 65.
[4] S.S. Chan, I.E. Wachs, L.L. Murrell, L. Wang, W.K. Hall, J. Phys. Chem. ,88 (1984) 5831.
[5] M.I. Zaki, B. Vielhaber, H. Kn6zinger, J. Phys. Chem., 90 (1986) 3176.
[6] S.W. Swanson, B. J. Streusand, G. A. Tsigdinos, Proc. 4th Int.Conf.Chemistry and Uses of
Molybdenum, (H. F. Barry and P. C. H. Mitchell Eds.) Climax Molybdenum Co, Ann Arbor
1982, p. 323.
[7] K.C. Pratt, J.V. Sanders, V. Christov, J. Catal. 124 (1990) 416.
[8] V.H.J. de Beer, G. C. A. Schuit, Stud. Surf. Sci. Catal.,1 (1976) 343.
[9] P. Grange, Catal. Rev.- Sci. Eng. 21 (1980) 135.
[10] R. R. Chianelli, M. Daage, M. J. Ledoux, Adv.Catal. 40 (1994) 177.
[11] H. Topsoe, B.S. Clausen, Catal. Rev.-Sci. Eng. 26 (1984)395.
[12] B. C. Wiegand, C. M. Friend, Chem. Rev. 92 (1992) 491.
[13] C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J.C. Beck, Nature, 359 (1992) 710
[14] J. C. Beck, .C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmidt, C.T.W. Chu,
D.H. Olson, E.W. Sheppard, S.B. McCullen, J.B. Higgins, J.L. Schlenker, JACS, 114 (1992)
10834.
[ 15] A. Corma, V. Forn6s, M.T. Navarro, J. P6rez-Pariente, J. Catal, 148 (1994).
[16] S. D. Djajanti, R.F. Howe, Stud. Surf. Sci. Catal., 105 (1997) 2067.
[17], J. Cui, Y.-H. Yue, W.-Y.Dong, Z. Gao, Stud. Surf. Sci. Catal. 105 (1997) 687.
[18] D. Zhao, J. Feng, Q. Huo., N. Melosh, G.H. Fredrickson, B. Chmelka, G.D. Stucky, Science,
279 (1998) 548.
[ 19] D. Zhao, Q. Huo, J. Feng, B.F. Chmelka, G.D. Stucky, J.Am.Chem.Soc., 120 (1998) 6025.
[20] A. Yuel, L.G.Hubert-Pfalzgraf, J. Catal. 217 (2003) 343.
[21 ] S. H. Yoo, R. Ryoo, M. Kruk, M. Jaroniec, J. Phys. Chem.B, 106 (2002) 4640.
[22] T.Tatsumi, K. A. Kayona, N. Igarashi, Chem. Comm. (1998)325.
[23] A. Brenner, R.L. Burwell Jr., J. Catal. 52 (1978) 364.
[24] M. Laniecki, M. Wojtowski, in preparation.
[25] R. Nakamura, R. G. Bowman, R.L. Burwell Jr., J.Am.Chem.Soc., 103 (1981) 673.
[26] M. I. Zaki, B. Vielbahr, H. Kn6zinger, J. Phys. Chem. 90 (1986) 3176.
Studies in Surface Science and Catalysis, volume 158
J. (~ejka,N. Zilkovfiand P. Nachtigall (Editors)
9 2005 Elsevier B.V. All rights reserved. 765
Commercial and dealuminated mordenites (MOR) as well as home made and cobalt modified
beta (BEA) zeolites have been characterized by 29Si and 27A1 solid state Magic Angle
Spinning Nuclear Magnetic Resonance (MAS-NMR) spectroscopy. The quantitative
contributions of Si(nA1) and Si(OH)x sites to the NMR signal intensities were calculated from
the various Si/A1 ratios and relative 29Si and 27A1 NMR signal intensities assuming twin, lone
and no A1 containing periodical building units of the zeolite framework. More lone compared
to twin and silicalite units were suggested for BEA than for MOR zeolites. Some octahedral
A1 transformed back to tetrahedral coordination due to ion-exchange of HBEA to Co/BEA
zeolites. We conclude that we are able to distinguish the Si(OH)• groups which are original
defect sites or produced in a dealumination or calcination process.
1. INTRODUCTION
The silicon-aluminium ordering in zeolites and the state of aluminium incorporated either in
the framework or out-of-lattice (extraframework) positions can be obtained by 29Si and 27A1
NMR spectroscopy. The Si/AI ratio of zeolites and the number of crystallographically distinct
sites for the five different Si(nA1) (n=0-4) configurations can be determined by 29Si NMR.
The ratio of tetrahedral silicon and aluminium in the zeolite framework can be directly
calculated from the line intensities in a 29Si MAS NMR spectrum by the following equation
assuming that the A1-A1 avoidance rule of Loewenstein is obeyed and Si(OH)• signals are not
included in the bands [1 ]:
The resulting Si/A1 ratio may strongly underestimate the actual Si/A1 ratio as defect sites
(Si(OH)x groups) are generally present in the zeolite framework.
IH-29Si cross-polarization (CP) makes possible the detection of silicon atoms to which
one or more hydroxyl groups are attached. In the CP spectra the line intensities of silicon
766
atoms bearing OH groups are selectively and strongly enhanced. 27A1 NMR characterizes
aluminium species of different (tetra- or octahedral) coordination. The line (at ca 55 ppm) of
tetrahedral aluminium in the zeolitic lattice is well separated from the line (at 0 ppm) of out-
of-lattice octahedral aluminium, following appropriate calibration their quantities can be
determined quantitatively [ 1].
The silicon-aluminium ordering in mordenites and in their dealuminated forms has been
determined by Debras et al. [2]. Jia et al. [3] claimed that the appearance of octahedral A1 in
the NMR spectra of beta zeolites does not necessarily involve framework aluminium
extraction. Kiricsi et al. [4] assigned octahedral A1 NMR resonance to transient-state
aluminium species which is leaving the framework. Bokhoven et al. [5] identified three-
coordinate aluminium in MOR and BEA zeolites with in situ X-ray absorption near-edge
spectroscopy (XANES).
Bodart el al. [6] computed the number of SiOH groups following dealumination of
mordenites by combination of their bulk Si/AI ratios with 29Si and 27A1 NMR results. They
demonstrated that aluminium atoms preferentially occupy tetrahedral positions in the four-
membered rings of the mordenite structure. They suggested that during the dealumination
process AI atoms are removing two by two from the four-membered rings, parallely four
silanol groups per extracted A1 atom are generated and this number gradually decreases to
two through a structural reorganization.
The aim of this work is to develop the method of Bodart et al. [6] further in order to
understand the dealumination process of MOR and BEA zeolites in a deeper level. We try to
calculate the number of Si(OH)2 species assuming different dealumination mechanisms in
order to reveal the origin of Si(OH)• groups.
2. EXPERIMENTAL
Various commercial mordenites (Norton MORNor and Degussa MORDeg) and home-made beta
zeolites with different global or bulk Si/AI ratios were studied by 29$1- and 27A1 MAS-NMR
9
spectroscopy. Part of the MORDeg sample was dealuminated by steaming for 20 min at 873 K
(MORDea0. The home-made HBEA was modified by cobalt chloride (2.6 wt% metal loading)
either by solid-state ion exchange (Co/BEAssm) or by incipient wetness impregnation
(Co/BEAIMp) [7, 8]. The amount of acid sites was determined by thermogravimetric (TG)
analysis of NH3 desorption (NETZSCH STA 409 Luxx). The bulk Si/A1 ratios were
calculated from the results of TG and PIGE [9] analysis. The NMR spectra were recorded
either on a Bruker MSL 400, or Avarice 500 spectrometer. For 29Si (79.4 MHz), a 6 ~ts (| =
rt/6) pulse was used with a repetition time of 6.0 s. For 27A1 (130.3 MHz), a 1 gs ( | rt/12)
pulse was used with a repetition time of 0.1 s. The infrared spectra were recorded on a Bruker
IFS 28 FT-IR spectrometer equipped with an MCT detector.
3. RESULTS
The >Si NMR spectrum of mordenites with a typical Si/A1 ratio of 5 displays three broad
signals for Si(2AI), Si(1A1) and Si(0A1) at about-100, -106 and-113 ppm, respectively, the
relative intensities of which are 1:2:2. Highly siliceous zeolites exhibit narrower lines with
distinctly different intensity distributions of the three Si(nAl) lines. Splitting of the Si(0A1)
signal reflects Si atoms in crystallographycally non-equivalent T-sites [1].
The 298i NMR spectrum of MORNor sample (Fig. l a) shows broad signals for Si(2Al),
Si(1A1) and Si(0A1) at about the above mentioned chemical shifts, but at different relative
767
intensities (Table 1). The lines assigned to Si(2A1) and Si(1AI) configurations (at-99 and
-105 ppm in Fig. l a) involve relatively few defect sites (Si(OH)• groups) in the spectrum,
because the shape of MORNor CP spectrum (not shown) is similar to the shape of its normal
spectrum (Fig. l a). However the lower Si/A1NMRratio of MORNor (7.0, Table 1) than its global
or bulk Si/AI ratio (8.4) indicates the presence of some defect silanol groups in this
mordenite.
The 29Si NMR spectrum of HBEA zeolite (Fig. 1d) shows three resonances at -104,- 111
and -115 ppm, which can be ascribed to Si(1AI), Si(0AI)A and Si(0A1)B sites, respectively
(Table 1). Splitting of the signal assigned to Si(0A1) configuration is due to two groups of
different crystallographic sites [10]. The CP spectrum of HBEA (not shown) confirms the
presence of SiOH groups in the line at-104 ppm assigned to Si(1A1) configuration, as its
intensity compared to the Si(0A1) line is higher in the CP than in the normal spectrum (Fig
l d). The difference in the framework (24.7) and NMR (16.5) Si/A1 ratio of HBEA shows also
the presence of defect sites (Table 1).
The effect of dealumination is clearly seen in the 29Si NMR spectra of Degussa
mordenite samples (Fig. 1). The broad lines of MORDeg (Fig. l b) are resolved to Si(2A1),
Si(1A1), and Si(0A1) configurations (Table 1), but following dealumination (MORDeal) the
relative intensity
a , ,7 d ;
HBE
-90 -100 -110 -120 (ppm) -90 -100 -110 -120 (ppm)
b ' e '
o , C
|
-90 -100 -110 -12o (ppm) -90 -100 -110 -120 (ppm)
r
MOR~D~~ ~ Co
-90 -100 -110 -120 (ppm) -90 -100 -110 -120 (ppm)
Fig. 1. 298i NMR spectra ofMORNor (a), MORI)eg (b), MORDeal (c), HBEA (d), Co/BEAssm (e)
and Co/BE&Me (f) zeolites
768
Table l
Bulk (mean of TG and PIGE), lattice or framework (Si/Alfram.=Si/Albuik / A1T(fraction)) and
NMR (Si/AINMR calculated by Eq. (1)) Si/AI ratios, relative tetrahedral A1T and Si(nAl) (n --
2, 1, 0) coordinations (%) of MOR and BEA zeolites calculated from 27A1 and 29Si NMR
spectra. Sum of 2 Si(OH)2 + SiOH concentrations were calculated by Eq. (3).
Si/A1 27A1 Si/A1 29SiNMR 2Si(OH)2
Zeolite
bulk A1T fram. Si/A1NMR Si(2A1) Si(1A1) Si(0A1)A Si(0A1)B + SiOH
MORNor 8.4 100 8.4 7.0 8.0 41.3 16.3 34.4 9.4
MORDeg 15.9 100 15.9 8.5 7.6 31.9 27.7 32.8 21.9
MORD~al 18.0 82.4 21.8 20.6 2.8 13.8 60.3 23.1 1.1
HBEA 18.2 73.8 24.7 16.5 0 24.2 65.2 10.6 8.0
Co/BEAs 19.5 8 7 . 1 22.4 13.8 1.1 26.9 62.0 10.0 11.1
Co/BEAI 1 9 . 5 98.7 19.8 18.9 0.6 20.0 65.7 13.7 0.9
of signals assigned to the Si(2A1) and Si(1Al) configurations at-104 and-106 ppm decreases
compared to the Si(0A1) line (Fig. lc). The signals assigned to Si(2A1) and Si(1A1)
configurations also contain defect hydroxyl groups at a high extent in the MORDeg sample, as
their relative intensities are much higher in the CP (not shown) than in the normal (Figs lb)
spectra (Table 1). The big difference in the NMR (8.5 and 20.6) and framework (15.9 and
21.8) Si/A1 ratios (Table 1) clearly indicates that Equation (1) is not applicable for these
Degussa mordenites due to the presence of defect silanol sites.
The presence of silanol (SiOH) and strong Broensted acidic (SiOHA1) sites are clearly
seen in the infrared spectra of mordenites (Fig. 2). The bands around 3740 and 3600 cm -I
wavenumbers can be ascribed to the former and latter sites, respectively. The SiOHA1 band
areas of MORNor and MORDeg samples are 3-4 times higher than that of MORDea~ zeolite
indicating loss of acidity due to dealumination in accordance with the 29Si NMR spectra.
0,1
0
:"'-, 3600
; ,
615-35 ORDeg
113
,r,,-
3727 ,-. . . . 9 "-
0
~" ,~ i ~. "'",
~a t i ~. "",
t i t "'-..
I
# .i \% "', ",...
! ] '.. MORNor ""-..
' i "" ................... "~. . . . . '7-", .....
l i "'" ....
.,-'"3,743,-" 36 O0
,...,~ i -"r "
..-"" j MORDeai
0
The 298i NMR spectra of Co/BEA zeolites (Fig. l) shows four resonances, which can be
ascribed to Si(2A1), Si(1A1), Si(0A1)A and Si(0A1)B sites, respectively (Table 1). The Si(2AI)
and Si(1A1) lines of Co/BEA zeolites include some Si(OH)x defect sites confirmed by the
enhancement of the corresponding lines in the CP spectra (not shown).
27A1 NMR spectra give information on the A1 distribution in structurally distinct sites of
the lattice. The 27A1NMR spectra of the mordenites and beta zeolites are shown in Fig. 3. The
sharp signals at 55 ppm are attributed to aluminium in the zeolitic framework at tetrahedral
coordination. The NMR spectra - except MORNor and MORDeg samples - exhibit an additional
line at 0 ppm, which is assigned to extraframework aluminium in octahedral out-of-lattice
positions. The relative concentration of A1 in tetrahedral positions (Table 1) were calculated
from the relative integrated line areas of the 27A1NMR spectra (Fig. 3). Due to dealumination
a rather high amount of octahedral aluminium was produced in MORDeal and the beta zeolite
samples also contain extraframework aluminium.
MORNor HBEA
J
MORDeg _j Co/BEAssIE
Co/BEAIMP
4. DISCUSSION
The periodic building unit (PBU) of MOR is a 12-membered unit composed of two finite
zigzag chains and a 4-membered ring, the one of BEA is a tetragonal beta layer composed of
16-membered units connected through 4-membered rings [ 11 ]. Bodart et al. [6] suggested that
two aluminium atoms are preferentially sitting in the 4-membered rings of the mordenite
structure in diagonal positions, the configurations with only one A1 atom are excluded. The
two silicon atoms in the 4-membered rings have Si(2AI) configurations, the four Si atoms
connected to the two A1 atoms from outside the 4-membered ring have Si(1A1)
configurations. Due to energetics reasons A1 should occupy the same positions in the 4-
membered rings of the beta zeolites. The relative abundance of these sites is 1/8 in the BEA-
and 1/3 in the MOR structure, consequently all aluminium atoms should sit in the 4-
membered rings of the MOR and BEA zeolites in two by two "twin" positions.
We develop the method of Bodart et al. [6] further taking some new assumptions. As the
intensity ratios of Si(1AI) to Si(2A1) lines is much higher than 2 in the 298i NMR spectra
(Table 1), beside the above mentioned "twin" positions PBUs with one "lone" AI atom sitting
in the 4-membered ring surrounded with 4 Si(1A1) positions are also allowed. Two kinds of
770
"twin" PBUs are distinguished in the BEA zeolites: 2 ("twin2") or 1 ("twinl") A1 atoms may
sit in the 16-membered PBUs. In addition to "twin" and "lone" at highly siliceous zeolites
"silicalite" PBUs are also assumed with Si(OH)2, SiOH and Si(0A1) positions only. The
"silicalite" PBUs may decompose to "original" Si(OH)x groups during preparation (e.g.
calcinations) of the zeolites, but irrespectively of the dealumination process. We take into
account the possible production of Si(OH)• groups from Si(XA1) configurations during
dealumination and preparation (silicalite):
Assuming that "Line(2)" around -100 ppm includes Si(2A1) and Si(OH)2 lines, "Line(l)" at
-105 ppm involves Si(1A1) and SiOH lines, and i f ~ Isi(nA0 is normalized to 1, then Eq. (1) is
modified to the following Eq. (3):
The 2Si(OH)2 + SiOH concentrations calculated by Eq. (3) are shown in Table 1. The sum of
silanol concentrations are all around 10%, except MORDeg, MORDeal and Co/BEAIMp samples.
An assumption should be taken to the ratios of SiOH to Si(OH)2 concentrations. It is well
seen from Eq. (2), that this ratio should be higher than 2 if not only "twin" units are present.
We assume that during dealumination the defect Si(OH)x groups are originated with equal
probability from the Si(2AI) and Si(1AI) configurations, which means that their ratios should
not change remarkably during the process. The following ratios are supposed:
A similar assumption is not necessary for the "original" defect silanol groups, because the
concentration of "silicalite" PBUs and "original" Si(OH)• groups are calculated as the
difference of all Si(0A1) and all Si(OH)• groups and those which are included in the "twin"
and "lone" PBUs or produced during their dealumination.
The computed results with the above assumptions are shown in Table 2. As there are 2-2
A1T and Si(2Al) atoms as well as 4-4 Si(1A1) and Si(0A1) species in one "twin" PBU of
mordenites, the following concentration relations are valid: A1T = Si(2A1) - Si(1A1) / 2 and
Alo = Si(OH)2 = SiOH / 2. The "lone" PBUs (1 A1T, 4 Si(1AI) and 7 Si(0A1)) include the
excess Si(1AI) with the appropriate other atoms. The "silicalite" units are formed from the
remaining Si(0AI) and "original" silanol groups.
The ratio of "lone" PBUs always exceeds the concentration of "twin" units, therefore the
"twin only" assumption of Bodart et al. [6] must be modified. We are able to distinguish the
"original" and "dealuminated" silanol groups and the ratio of SiOH to Si(OH)2 is always
much higher than two in accordance with the assumption taken in Eq. (2). "Silicalite" units do
not exist in the MORNor sample, because the concentration of Si(0A1) is practically zero in
this PBU. Therefore presumably the "original" silanol groups are sitting preferentially in the
"lone" units, far from the A1 atoms in the MORNor zeolite.
771
Table 2
MOR and BEA zeolite compositions (%) assuming perfect, theoretical twin (BEA: twin2 or
twin 1), lone and silicalite Periodical Building Units (PBU)
PBU s i site contribution
Zeolite PBU
A1 deal Alv Alo Si(2A1) Si(OH)2 Si(1A1) SiOH Si(0A1)
twin 35.6 0 5.9 0 5.9 0 11.9 0 11.9
MORNor lone 64.0 0 4.7 0 18.9 0 33.2
silicalit 0.3 1.2 1' 6.0 T 0.3
twin 22.8 0 3.8 0 3.8 0 7.6 0 7.6
MORDeg lone 25.3 0 2.1 0 8.4 0 14.8
silicalit 51.9 3.3 14.0 34.6
twin 15.0 0.9 2.5 0.1 2.5 0.1 5.0 0.3 5.3
MORDeal lone 22.0 1.3 1.8 0.1 7.3 0.4 13.6
silicalit 60.1 0 60.1
twin2/1 0 0 0 0 0 0 0 0 0
HBEA lone 61.4 20.7 3.8 1.3 15.4 5.2 56.5
silicalit 17.8 0 2.4 15.4
twin2/1 4.9/9.8 0.7/1.4 0.6 0.1 0.6 0.1 1.2 0.2 2.8/8.4
Co/BEAss~E lone 58.1 8.2 3.6 0.5 14.5 2.1 45.6
silicalit 28.0/22.4 0.3 7.6 20.1/14.5
twin2/1 4.2/8.3 0.1/0.1 0.5 0 0.5 0 1.0 0 2.1/6.3
Co/BEAIMv lone 68.8 0.8 4.3 0.1 17.2 0.2 47.9
silicalit 26.2/22.0 0.6 25.6/21.4
The ratio of "twin" to "lone" units is similar in the MORDeg (0.9) and MORDeal (0.7)
zeolites. During dealumination of the Degussa mordenite samples both A1 containing PBUs
decompose in a similar rate without preference. All (3.3 + 14.0 %) "original" silanol
decomposes and parallely new "dealuminated" silanol groups form (0.1 + 0.7 %).
As was already discussed in the Introduction, framework-related threefold-coordinated
A1-OH intermediate species may exist in dealuminated zeolites [5], which means a one-to-one
SiOH : AI(OSi)3 ratio. Indeed, the shoulder at ca. 30-40 ppm in the 27A1NMR spectrum of
beta zeolites (Fig. 4) suggests the presence of deformed tetrahedral atoms, i.e; (SiO)3A1OH, in
the structure. In the hydrated state water molecules are coordinatively bound to these
AI(OSi)3 species leading to an equilibrium between tetrahedral and octahedral configuration,
which is reflected in the 27A1NMR spectra. The extraordinary sharpness of the octahedral line
in the spectrum of HBEA (Fig. 4) indicates a very high symmetry of this configuration.
Exchange of the acidic protons of HBEA by cobalt ions results in an ionized tetrahedral
configuration, which explains the conversion of octahedrally into tetrahedrally coordinated
aluminium observed in the 27A1 NMR spectra (Fig. 4 and Table 2) in accordance with the
literature [12].
Based on the data shown in Table 2 it is nut possible to suggest if the "twin2" or "twinl"
units are more preferable in the structure of beta zeolites. Nevertheless a definite difference
exists between MOR and BEA zeolites that only slight excess of "lone" over "twin" units is
observed in the former, while a marked excess of "lone" over "twin" PBUs is present in the
latter structures (Table 2).
772
5. C O N C L U S I O N S
A detailed theoretical structure analysis of MOR and BEA lattices enabled us to a deeper
understanding of the origin of defect sites, the mechanism of dealumination and ion-exchange
through the 27A1 and 29Si NMR analysis of A1 and Si site contributions. We conclude that we
are able to distinguish the Si(OH)x groups which are original defect sites or produced in a
dealumination or calcination process. The dealumination of mordenite resulted in the
decomposition of original defect sites and the production of new silanol groups. The ion-
exchange of protons to cobalt ions effected the conversion of octahedrally into tetrahedrally
coordinated aluminium in BEA zeolites. All of the studied zeolites contain lone A1 sites in
remarkable amount contrary to the exclusive two by two A1 sitting in 4-membered rings
theoretical structure. The main difference between mordenites and beta zeolites is the
presence of the highly symmetric hydrated and hydroxyl group containing octahedral
aluminium in the lattice of HBEA which transforms to A1T due to ion-exchange or
impregnation and the relatively high amount of defect sites in mordenites compared to beta
zeolites.
ACKNOWLEDGEMENT
REFERENCES
[1] G. Engelhardt and D. Michel, High-Resolution Solid-State NMR of Silicates and Zeolites,
Wiley, New York, 1987.
[2] G. Debras, J. B.Nagy, Z. Gabelica, P. Bodart and P.A. Jacobs, Chem. Lett., (1983) 199.
[3] C. Jia, P. Massiani and D. Barthomeuf, J. Chem. Soc. Faraday Trans., 89 (1993) 3659.
[4] I. Kiricsi, C. Flego, G. Pazzuconi, W.O. Parker, R. Millini, C. Perego and G. Bellussi, J.
Phys. Chem., 98 (1994) 4627.
[5] J.A. van Bokhoven, A.M.J. van der Eerden and D.C. Koningsberger, J. Am. Chem. Soc., 125
(2003) 7435.
[6] P. Bodart, J. B.Nagy, G. Debras, Z. Gabelica and P.A. Jacobs, J. Phys. Chem., 90 (1986)
5183.
[7] J. Wittayakun, B. Nuntaitawegon, N. Grisdanurak, G. Kinger, H. Vinek, in Proc. Regional
Symp. on Chem. Eng. 2003, Metro Manila, Philippines, Dec 1-3, 2003.
[s] J. Wittayakun, N. Grisdanurak, G. Kinger and H. Vinek, submitted for publication.
[9] G. Debras, E.G. Derouane, J.-P. Gilson, Z. Gabelica and G. Demortier, Zeolites, 3 (1983) 37.
[lO] C.B. Dam and M.E. Davis, Catal. Today, 19 (1994) 151.
[ll] W.M. Meier and D.H. Olson, Atlas of Zeolite Structure Types, Elsevier, London, 1996.
[12] Y. Neinska, V. Mavrodinova, Ch. Minchev and R.M. Mihfilyi, Stud. Surf. Sci. Catal., 125
(1999) 37.