High-speed experimentation techniques applied to the study of the synthesis of zeolites and silsesquioxanes

Article abstract of DOI:10.1016/S1381-1169(01)00494-0

High-speed experimentation (HSE) techniques are newly developed methods that, allowing the fast preparation and analysis of large numbers of parallel experiments, enable to screen an extensive parameter space. These techniques are particularly useful to investigate chemical systems for which a precise knowledge of the overall reaction or formation mechanism is not available. Here, we report the application of HSE techniques to the study of the synthesis of two families of siloxane compounds: zeolites and silsesquioxanes. Zeolites are well known crystalline microporous materials with broad applications as heterogeneous catalysts. Silsesquioxanes are small, discrete siloxane cages used both as homogeneous catalysts and as model compounds for silica surfaces and zeolites. With our HSE approach we were able to investigate the synthesis of aluminium-rich zeolite beta (Si/Al ratio from 2.5 to 5) and to identify a new and fast way to synthesise silsesquioxane precursors for Ti-catalysts active in the epoxidation of alkenes.

Full text of DOI:10.1016/S1381-1169(01)00494-0

Journal of Molecular Catalysis A: Chemical 182–183 (2002) 319–325
High-speed experimentation techniques applied to the study
of the synthesis of zeolites and silsesquioxanes
Paolo P. Pescarmonaa, Johannes J.T. Ropsa, Jan C. van der Waala,b
,
Jacobus C. Jansena, Thomas Maschmeyera,∗
a
Laboratory of Applied Organic Chemistry and Catalysis, DelftChemTech, TU Delft, Julianalaan 136, 2628 BL Delft, The Netherlands
b
Avantium Technologies BV, Zekeringstraat 29, 1014 BV, P.O. Box 2915, 1000 CX Amsterdam, The Netherlands
Received 5 July 2001; received in revised form 6 November 2001; accepted 6 November 2001
Abstract
High-speed experimentation (HSE) techniques are newly developed methods that, allowing the fast preparation and analysis
of large numbers of parallel experiments, enable to screen an extensive parameter space. These techniques are particularly
useful to investigate chemical systems for which a precise knowledge of the overall reaction or formation mechanism is
not available. Here, we report the application of HSE techniques to the study of the synthesis of two families of siloxane
compounds: zeolites and silsesquioxanes. Zeolites are well known crystalline microporous materials with broad applications
as heterogeneous catalysts. Silsesquioxanes are small, discrete siloxane cages used both as homogeneous catalysts and as
model compounds for silica surfaces and zeolites. With our HSE approach we were able to investigate the synthesis of
aluminium-rich zeolite beta (Si/Al ratio from 2.5 to 5) and to identify a new and fast way to synthesise silsesquioxane
precursors for Ti-catalysts active in the epoxidation of alkenes. © 2002 Elsevier Science B.V. All rights reserved.
Keywords: High-speed experimentation techniques; Zeolites; Silsesquioxanes; Heterogeneous catalysts; Homogeneous catalysts
1. Introduction
torial approach, in which thousands of compounds are
synthesised in a single pot [1]. Nevertheless, the paral-
lel approach used in HSE techniques allows correlat-
ing each sample with its properties (i.e. its activity as
catalyst), while this is not possible using mix and split
methods where the samples are screened as a whole.
In this sense, HSE techniques are more suitable when
the goal of identifying a new lead is coupled with
that of a fundamental understanding of the system
under study.
Here, we present the application of HSE tech-
niques to the study of the synthesis of two families
of siloxane compounds: zeolites and silsesquioxanes.
Zeolites are well-known crystalline microporous ma-
terials with broad applications as heterogeneous cat-
alysts. Silsesquioxanes are small, discrete siloxane
High-speed experimentation (HSE) techniques [1]
recently have found applications in the field of cataly-
sis [2]. These techniques are based on the use of auto-
mated workstations allowing the preparation and the
screening of large numbers of compounds. Synthesis
and analysis are performed in parallel wells, i.e. each
sample in a separate vessel. Typically, 100 samples can
be prepared and screened in a working week with these
parallel workstations. This is a large number compared
to a standard experimental approach on a lab scale, but
is still small if compared to a mix and split combina-
∗
Corresponding author. Fax: +31-1527-84-289.
E-mail address: th.maschmeyer@tnw.tudelft.nl (T. Maschmeyer).
1381-1169/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S1381-1169(01)00494-0

320
P.P. Pescarmona et al. / Journal of Molecular Catalysis A: Chemical 182–183 (2002) 319–325
cages of general formula (RSiO1.5)a(H2O)0.5b
[R: organic group; a + b = 2n; n = 1,2,3,. . . ;
b ≤ a + 2]. Completely condensed silsesquioxanes
(b = 0) can be seen as model compounds for building
blocks present in zeolite structures [3]. Incompletely
condensed silsesquioxanes (b = 0) can be used as
model compounds for silica surfaces and as homoge-
neous catalysts after the insertion of a metal centre
[4]. For both zeolites and silsesquioxanes, the synthe-
sis consists of a multiple-step reaction for which an
overall mechanism is not available; many parameters
influence the syntheses determining which structures
are formed and in which amounts. Therefore, an HSE
approach allowing the fast screening of broad param-
eter spaces can be suitable to study and optimise the
synthesis of both families of compounds.
at 1500 rpm for 1 h and dried in a vacuum centrifuge
(see Footnote 1). After the experiment, the teflon
inserts were cleaned overnight with NaOH (4 M)
at reaction temperature or with HF (10%) at room
temperature.
Powder diffraction data were collected on a Philips
PW1840 diffractometer generating Cu K␣ radiation.
Data were obtained from 5 to 50◦ 2θ with a step of
0.01◦ and a counting time of 0.5 s per step. The re-
ceiving slit had a width of 0.3 mm. The tube voltage
was 40 kV and the tube current was 50 mA. In order
to reduce the analysis time, an automatic sampler was
installed, thus allowing X-ray analysis of 24 sam-
ples in 15 h. Elemental analyses were performed on
a Perkin-Elmer 3000 DV ICP-OES. After dissolving
the zeolites in a 1% HF (40%)/1.3% H2SO4 solution,
the concentrations of silicon, aluminium and sodium
were measured with induced coupled plasma optical
emission spectroscopy (ICP-OES). The samples were
analysed twice as independent duplicates to get an
indication of the precision of the analysis. The ac-
curacy in the analysis of Si and Al are estimated at
about 5%, for Na at 10% (relative values). High-
resolution solid-state 27Al MAS NMR spectra were
recorded on a 400 MHz Varian Inova spectrometer.
2. Experimental
2.1. Synthesis of zeolite beta
Experiments were performed on an automated par-
allel synthesis workstation1 coupled with a personal
computer supplied with software enabling to program
the workstation. Reactants: Ludox-HS 40 colloidal
silica (40 wt.% suspension of SiO2 in water) as sil-
ica source; sodium aluminate (40–45 wt.% Na2O,
50–56 wt.% Al2O3) as alumina source; tetraethylam-
monium hydroxide (TEAOH) (35% solution in wa-
ter) as template source; deionised water. Appropriate
amounts of reactants for the zeolite synthesis were
dispensed by the pipette arm of the workstation in a
6 × 4 matrix of small 3 ml teflon-lined stainless steel
autoclaves. An equal amount of alumina source was
dispensed in each autoclave in order to set the pH of
all the samples at ∼14. In a typical experiment, after
dispensing the alumina, the template and the silica
sources, the autoclaves were filled up to 75% of their
2.2. Synthesis of incompletely
condensed silsesquioxanes
Experiments were performed on the same auto-
mated workstation used for the synthesis of zeolite
beta. Reagents: cyclohexyl, cyclopentyl, phenyl,
methyl, ethyl, tert-butyl, normal-octyl, and allyl trich-
lorosilanes (RSiCl3); acetone, acetonitrile, methanol
and tetrahydrofuran (THF) as solvents; deionised wa-
ter. In a typical experiment, 2 ml aliquots of each of
the four solvents were dispensed in a rack containing
4 ml glass tubes, followed by the addition of 340 ␮mol
of each of the organosilanes. The hydrolysis reaction
was started by adding 0.5 ml of deionised water to
each reaction tube and placing the rack on an orbital
shaker (see Footnote 1) at 50 ◦C for 18 h. By means
of a vacuum centrifuge (see Footnote 1), the solvents
and the excess water were removed from the sam-
ples, which were then stored under argon. Titanium
insertion was performed by dissolving each crude
silsesquioxane mixture in 2 ml of THF, followed by
the addition of 54 ␮mol of titanium isopropoxide,
volume with water. Next, the obtained aluminosilicate
∼
gels were placed in an ultrasonic bath (T
60 ◦C)
=
for 1 h. The autoclaves were then closed and heated
statically at 170 ◦C. After 48 h, heating was stopped
and the autoclaves were cooled down to room temper-
ature. Finally, the autoclaves with the formed solids
were washed twice with deionised water, centrifuged
1
Kindly provided by Avantium Technologies.

P.P. Pescarmona et al. / Journal of Molecular Catalysis A: Chemical 182–183 (2002) 319–325
321
i
Ti(OPr )4. After 5 h at 60 ◦C, THF was removed with
the help of the vacuum centrifuge and the samples
were stored under argon.
3.1.1. The HSE approach
Various parameters influence the synthesis of zeo-
lites: the nature of the silica and alumina sources and
the ratio between the two, the nature and the concen-
tration of the template, the presence, type and con-
centration of alkali cations, the water content, the pH,
the temperature and the reaction time [9]. The HSE
workstation employed in the project can only handle
low-viscosity liquids: this influenced the choice of the
silica and alumina sources used in the experiments.
As silica source we chose the cheap, easy to handle
Ludox-HS 40 colloidal silica and as alumina source an
aqueous solution of sodium aluminate. Hydrothermal
synthesis using tertraethytlammonium ions (TEA+) as
templates in an alkaline environment was chosen from
among the methods commonly used for the prepara-
tion of aluminium-rich zeolite beta for it’s adaptabil-
ity to the HSE workstation. The method we applied
is similar to that described by Borade and Clearfield
[10], with a different silica source and a higher water
content in order to allow handling by the HSE work-
station. Given these restrictions, we selected the Si/Al
ratio and the TEA+/Al ratio in the starting reaction
mixture as the most relevant parameters determining
the formation of aluminium-rich zeolite beta. There-
fore, we screened the parameter space defined by the
combination of six Si/Al ratios and four TEA+/Al
ratios as a function of the presence of zeolite beta.
The range in which the two parameters were varied
had been chosen in order to investigate the possibility
of lowering the Si/Al in zeolite beta and the amount
of the expensive TEA+ employed with respect to the
synthesis method proposed by Borade and Clearfield
[10]. The screening was performed by means of an
X-ray powder diffractometer supplied with an auto-
mated sampler. The intensity of the strongest peak in
the diffractogram for each zeolite species present was
taken as a measure of crystallinity.
The activity of the Ti–silsesquioxanes in the epox-
idation of 1-octene with tert-butyl hydroperoxide
(TBHP) was determined by adding to each dried
sample ∼0.36 mmol of TBHP in cyclohexane solu-
tion (TBHP: catalyst ratio = 7.5) and 0.0144 mol of
1-octene, that, therefore, acted both as reactant and
solvent (1-octene: catalyst ratio = 300). The em-
ployed 1-octene contained 2% in volume of decane
as an internal standard for the following GC analysis.
Samples were taken after reacting for 5 h at 80 ◦C
and analysed on a UNICAM Pro GC (see Footnote
1) using a CP-Sil-5B column. The epoxidation re-
action did not proceed further at room temperature.
The reported activities were obtained by normalis-
ing the 1,2-epoxyoctane GC peak area by means
of the internal standard. Since 1-octene was used
both as reactant and solvent, it has not been possi-
ble to realise a mass balance. The reaction is com-
pletely selective towards 1,2-epoxyoctane: no other
products were detected. The conversion selectivity
of TBHP towards 1,2-epoxyoctane is between 80
and 94%.
3. Results and discussion
3.1. Synthesis of zeolite beta
The synthesis of zeolite beta has been studied by
means of HSE techniques. Beta is a large pore zeo-
lite characterised by three families of mutually per-
pendicular channels with 12-membered ring apertures
[5], which make it suitable as heterogeneous catalyst
for numerous organic reactions [6]. Zeolite beta has
been prepared with a wide range of Si/Al ratios (from
5 to ∞) [7]. The control of the Si/Al ratio is partic-
ularly interesting since it allows tuning the number
of acid sites and the hydrophobicity of the zeolite. In
this work, we report the optimisation of the synthe-
sis of aluminium-rich zeolite beta in order to reach
the lowest possible Si/Al ratio. The synthesis of an
aluminium-rich zeolite beta is particularly interesting
also because Tschernichite, the natural form of beta,
has Si/Al = 3 [8].
3.1.2. The HSE results
The products of the 24 syntheses mainly consist of a
mixture of two zeolites: zeolite beta and zeolite NaP1
(Fig. 1).2 NaP1 is a gismondine type of zeolite, with
a Si/Al ratio of 1.5 and Na+ ions as template ions.
For this reason, NaP1 is found as a product at low
2
All the reported results are averages of the values obtained in
different sets of experiments.

322
P.P. Pescarmona et al. / Journal of Molecular Catalysis A: Chemical 182–183 (2002) 319–325
Fig. 2. 27Al NMR of the pure zeolite beta obtained with Si/Al = 5
and TEA+/Al = 1.25 in the starting reaction mixture.
pure zeolite beta with this hydrothermal synthesis and
using TEA+ as template ion. The corresponding sam-
ple has been further characterised to determine the
actual Si/Al in the zeolite framework. From ICP-OES
analysis a Si/Al ratio of 4.7 and an Al/Na ratio of
0.8 were determined. To check if all the aluminium
species in the sample are part of the zeolite frame-
work, the sample was characterised by 27Al NMR.
The 27Al NMR spectrum shows a high intensity peak
at 59 ppm corresponding to tetrahedral aluminium
species and a low intensity peak at 4 ppm correspond-
ing to octahedral aluminium species (Fig. 2). Tetrahe-
dral aluminium species are those present in the zeolite
structure, while octahedral species are likely to be due
to non-zeolitic aluminium phases. The ratios between
the integrals of the two peaks is 92/8. Considering
just the tetrahedral aluminium as part of the zeolite
framework, the Si/Al ratio has then to be corrected to
a value of 5.1. This value is slightly higher than the
Si/Al ratio of 4.5 reported by Borade and Clearfield
[10] and is obtained using a higher TEA+/Al ratio
(1.25 against 0.8). The highest TEA+/Al ratio needed
can be due to the higher water content used in our
method. A relevant advantage of our synthesis method
is the use of Ludox HS-40 as silica source instead of
fumed silica, since the former is cheaper, easier and
safer to handle and forms a less dense synthesis gel.
Fig. 1. Presence of zeolite beta and zeolite NaP1 in the screened
parameter space.
values of Si/Al ratio in the starting reaction mixture,
while beta is formed at higher values of Si/Al ratio.
Pure zeolite beta is obtained at a Si/Al ratio of 5 and
TEA+/Al ratio of 1.25 in the starting reaction mix-
ture, while pure NaP1 is obtained at a Si/Al ratio of
2.5 for the entire screened TEA+/Al ratio. In between,
mixtures of both zeolites were observed. Some sam-
ples also contain amorphous silica and small amounts
of crystalline impurities (consistent with analcime). A
higher value of TEA+/Al ratio causes a positive trend
towards the formation of zeolite beta, as competition
takes place between the two templates (Na+ for NaP1
and TEA+ for beta). This experiment suggests that a
Si/Al ratio of 5 and a TEA+/Al ratio of 1.25 in the
original reaction mixture are the limit values to obtain

P.P. Pescarmona et al. / Journal of Molecular Catalysis A: Chemical 182–183 (2002) 319–325
323
Scaling-up of the synthesis of zeolite beta with a
Si/Al ratio of 5 and a TEA+/Al ratio of 1.25 in the
original reaction mixture to a 50 ml autoclave lead
to a similar result even if small amounts of NaP1
were present as byproducts. Differences in results
using HSE 3 ml autoclaves and conventional larger
volume autoclaves are probably due to the higher
heating rate of the small autoclaves, which is mainly
caused by their lower heat capacity, and to the higher
surface/volume ratio of the small autoclaves, which
favours heterogeneous nucleation and, therefore,
speeds up the crystallisation process.
synthesis of silsesquioxanes has been optimised as a
function of the activity of the catalysts obtained after
complexation of a titanium centre to the silsesquiox-
ane structures. The epoxidation of 1-octene with
TBHP as oxidant was chosen as test reaction (activity
measured by means of an automated GC).
3.2.1. The HSE approach
To decide which parameter space had to be
screened to optimise the synthesis of silsesquiox-
anes, we considered all the parameters influencing
the hydrolytic condensation: the nature of the organic
group R and of the group X in the organosilane and
its concentration in the reaction mixture, the sol-
vent in which the synthesis is carried out, the rate
of addition and quantity of H2O, the pH, the tem-
perature and the reaction time [13]. Among these
parameters, we selected the organic group R and the
solvent as the most influential in determining which
silsesquioxane structures are formed and in which
quantities. Then, we defined a parameter space as
a combination of eight organosilanes with differ-
ent organic groups (cyclohexyl, cyclopentyl, phenyl,
methyl, ethyl, tert-butyl, normal-octyl and allyl) and
four water-miscible solvents (acetone, acetonitrile,
methanol, THF) and we screened it as a function of
the activity of the catalyst obtained by insertion of a
titanium centre in the silsesquioxane structures. Since
the goal is to identify a more efficient way to syn-
thesise silsesquioxane precursors, the reaction time
was set at 18 h, a much shorter time than that com-
monly required for the synthesis of silsesquioxane
1 [12].
3.2. Synthesis of incompletely condensed
silsesquioxanes
The synthesis of silsesquioxane precursors for
Ti-catalysts has been optimised by means of HSE
techniques. Incompletely condensed silsesquioxane
1, R7Si7O12H3 (a = 7, b = 3) is known to provide
an active homogeneous catalyst for the epoxidation
of alkenes after insertion of a titanium centre [11]
(Fig. 3). Silsesquioxane 1 is obtained by the hydroly-
tic condensation of the corresponding organosilane
RSiX3 (X = Cl, OMe, OEt) [12]. The synthe-
sis is rather time-consuming and can yield other
silsesquioxane structures as byproducts. Among these
compounds there might be incompletely silsesquiox-
anes other than 1 that, after insertion of a titanium
centre, generate active epoxidation catalysts. On the
basis of these considerations, we used HSE tech-
niques to identify a new and efficient synthetic route
towards silsesquioxane precursors for Ti-catalysts: the
Fig. 3. Insertion of Ti(OPri )4 on the incompletely condensed silsesquioxane R7Si7O12H3 (R: cyclohexyl, cyclopentyl).

324
P.P. Pescarmona et al. / Journal of Molecular Catalysis A: Chemical 182–183 (2002) 319–325
Fig. 4. Catalytic activity in the epoxidation of 1-octene of the Ti–silsesquioxanes in the screened parameter space.
3.2.2. The HSE results
of the R-group apart from ethyl and tert-butyl for
which silsesquioxanes synthesised in THF generate
more active catalysts than those synthesised in ace-
tone and methanol. The fact that acetonitrile gives the
best results can be explained on the basis of its high
polarity: it has been proposed that the presence of a
polar molecule reduces the activation barrier for the
condensation reactions that lead to the formation of
silsesquioxanes [14].
The activity in the epoxidation of 1-octene of the
Ti-catalysts as a function of the parameters varied
in the synthesis of the silsesquioxanes precursors is
reported in Fig. 4 (see Footnote 2). The values are
normalised to the activity of the catalyst obtained by
i
insertion of Ti(OPr )4 on the pure silsesquioxane 1
(the activity of which is set at 1). The results show
some general trends:
Among the R-groups, the order of activity is
cyclopentyl > cyclohexyl > phenyl > tert-butyl >
ethyl > methyl > allyl ∼ normal-octyl ∼ 0, for
all solvents apart from THF for which tert-butyl
and ethyl organosilanes yield more active catalysts
than phenyl organosilanes. This trend seems to be
related to the size of the organic substituent: bulky
groups probably hinder the formation of completely
condensed silsesquioxanes, which are not be able to
complexate titanium centres and therefore to generate
active catalysts.
The Ti–silsesquioxane obtained by the hydrolytic
condensation of cyclopentyltrichlorosilane in ace-
tonitrile presents the highest catalytic activity in the
screened parameter space (Fig. 4). This activity is
87% of that of the Ti–silsesquioxane obtained using
pure silsesquioxane 1, with the advantage that the syn-
thesis of our catalyst does not require any purification
process and is much less time-consuming than the
synthesis of compound 1. The experiment has been
scaled-up from 2 to 100 ml scale yielding to a similar
catalytic activity, thus proving that for this system the
use of small-scale HSE equipment produces results
comparable to those from conventional experiments.
Among the solvents, the order of activity is
acetonitrile > acetone > methanol > THF, for any

P.P. Pescarmona et al. / Journal of Molecular Catalysis A: Chemical 182–183 (2002) 319–325
325
4. Conclusions
[4] T. Maschmeyer, J.M. Thomas, A.F. Masters, NATO ASI Ser.
Ser. C 498 (1997) 461.
[5] M.M. Tracey, J.M. Newsam, Nature 332 (1988) 249.
[6] P.B. Venuto, Micropor. Mater. 2 (1994) 297.
[7] R.L. Wadlinger, G.T. Kerr, E.J Rosinski, US Patent Appl.
3 308 069 (1967).;
J.C. van der Waal, M.S. Rigutto, H. van Bekkum, J. Chem.
Soc., Chem. Commun. (1994) 1241.
[8] J.V. Smith, J.J. Pluth, R.C. Boggs, D.G. Howard, J. Chem.
Soc., Chem. Commun. (1991) 363.
We showed that HSE techniques can be success-
fully applied to study the synthesis of both hetero-
geneous and homogeneous catalysts. In the case
of the synthesis of zeolites, we have been able to
identify a facile and cheap method to produce pure
zeolite beta and pure zeolite NaP1. Moreover, we
studied the effect of the Si/Al and TEA+/Al ratios
on the formation of these two zeolites. In a sim-
ilar way, we identified a fast and straightforward
method to synthesise silsesquioxane precursors for
Ti-catalysts. At the same time, we gained some knowl-
edge on the effects of the solvent and of the organic
group R on the synthesis of incompletely condensed
silsesquioxanes.
[9] J. Pérez-Pariente, J.A. Martens, P.A. Jacobs, Appl. Catal. 31
(1987) 35;
M.J. Eapen, K.S.N. Reddy, V.P. Shiralkar, Zeolites 14 (1994)
294.
[10] R.B. Borade, A. Clearfield, Micropor. Mater. 5 (1996) 289;
R.B. Borade, A. Clearfield, Chem. Commun. (1996) 625.
[11] T. Maschmeyer, M.C. Klunduk, C.M. Martin, D.S. Shephard,
J.M. Thomas, B.F.G. Johnson, Chem. Commun. (1997) 1847.;
M. Crocker, R.H.M. Herold, A. Guy Orpen, Chem. Commun.
(1997) 241.
[12] F.J. Feher, D.A. Newman, J.F. Walzer, J. Am. Chem. Soc.
111 (1989) 1741;
References
F.J. Feher, T.A. Budzichowski, R.L. Blanski, K.J. Weller, J.W.
Ziller, Organometallics 10 (1991) 2526.
[1] P.P. Pescarmona, J.C. van der Waal, I.E. Maxwell, T.
Maschmeyer, Catal. Lett. 63 (1999) 1.
[13] M.G. Voronkov, V.I. Lavrent’yev, Top. Curr. Chem. 102
(1982) 199;
[2] M.T. Reetz, Angew. Chem. Int. Ed. 40 (2001) 284;
S. Senkan, Angew. Chem. Int. Ed. 40 (2001) 312.
[3] C.W. Earley, J. Phys. Chem. 98 (1994) 8693.
P.G. Harrison, J. Organomet. Chem. 542 (1997) 141.
[14] T. Kudo, M.S. Gordon, J. Am. Chem. Soc. 120 (1998) 11432;
T. Kudo, M.S. Gordon, J. Phys. Chem. A 104 (2000) 4058.