In situ NMR and XRD studies of the growth mechanism of SBA-1

Article abstract of DOI:10.1039/b419293c

In situ ¹⁷O, ¹⁴N and ²⁹Si nuclear magnetic resonance (NMR) coupled with in situ energy dispersive X-ray diffraction (EDXRD) have been used to investigate the growth of the siliceous mesoporous material, SBA-1, synthesised

Full text of DOI:10.1039/b419293c

View Online / Journal Homepage / Table of Contents for this issue
R E S E A R C H P A P E R
In situ NMR and XRD studies of the growth mechanism of
SBA-1w
a
C. C. Egger,z M. W. Anderson,*a G. J. T. Tiddyb and J. L. Cascic
a Department of Chemistry, UMIST, Manchester, UK. E-mail: m.anderson@manchester.ac.uk;
Fax: þ44 161 200 4559; Tel: þ44 161 200 4517
b Department of Chemical Engineering, UMIST, Manchester, UK
c Synetix, Johnson Matthey, Billingham, UK
Received 23rd December 2004, Accepted 16th February 2005
First published as an Advance Article on the web 17th March 2005
In situ 17O, 14N and 29Si nuclear magnetic resonance (NMR) coupled with in situ energy dispersive X-ray
diffraction (EDXRD) have been used to investigate the growth of the siliceous mesoporous material, SBA-1,
synthesised under acidic conditions from a micellar solution of the surfactant hexadecyltriethylammonium
bromide (HTEAB) and tetraethylorthosilicate (TEOS). For the last decade, the mechanism of growth of such
materials has been thought to be driven by electrostatic interactions described as a co-assembly process between
the silica species (I1) and the micelles (S1Xꢀ). However, this postulated model referred to as the ‘‘charge density
matching model’’ has never been fully supported by experimental data for the acidic syntheses. We have carried
out a detailed in situ study which challenges the so-called S1XꢀI1 pathway and instead suggests that a salting-
3
out effect coupled with a drastic change in the water activity are responsible for the composite I1 (SBA-1 space
group Pm3ꢀn) mesophase precipitation. Substantial reorganisation of the precipitated phase then results in the
final structure.
static interactions between surfactant micelles, counter-ions,
and charged inorganic species. It has had the great advantage
of explaining the formation of all mesoporous materials.
According to this mechanism, cationic inorganic species can
react directly with anionic surfactant (SꢀI1 pathway) or via the
micellar counteranions of the positively charged surfactants
(S1XꢀI1 pathway). A similar route starting with anionic
inorganic species is also proposed (the S1Iꢀ and SꢀM1Iꢀ
pathways). Different types of mesophases have been obtained
according to ref. 3.
Introduction
Following the emergence of MCM-type meso-materials,1,2 the
SBA (Santa Barbara) series was first reported in 1994.3,4 The
synthesis of such materials depends upon the inorganic/organic
interface which exists during synthesis, however, the mechan-
ism associated with this self-assembly process has not been
extensively studied. In particular, syntheses under highly acidic
conditions, often found in SBA-type syntheses has received
very little attention from a mechanistic viewpoint. Recently, we
have shown that the structure of SBA-1 is a very low curvature
surface formed from a high curvature surfactant micelle meso-
phase (see Fig. 1a and 1b).5 The consequences are that the
final structure is formed at the low curvature interface between
silica and adsorbed water rather than silica and surfactant,
thereby creating a structure with windows. In this work we
monitor the kinetics of the processes which leads to this final
structure.
The main synthetic difference between SBA mesoporous
materials compared with the MCM family is the acidity of
the medium. Changing pH changes the charge sign and nature
of the inorganic species and consequently different materials
are achieved. The SBA series is indeed different from the MCM
series both in terms of structure6 and robustness upon thermal
treatment.3 Both the MCM and SBA materials involve the
growth of a silica structure within the aqueous zones of a
surfactant mesophase. Because only moderate surfactant con-
centrations are employed, the mesophase must precipitate from
solution at some stage during the reaction. To explain this, a
model involving ‘‘charge density matching’’ has been generally
accepted since 1994.3,4 This model proposes that mesophases
are precipitated from aqueous solutions due to strong electro-
The S1XꢀI1 pathway (leading to SBA materials) has been
reported to be supported by experimental evidence, sum-
marised as follows:3,4 (i) the formation of positively charged
silica species at a pH below the isoelectric point (protonated
silicic acids); (ii) a constant proton concentration during the
synthesis; (iii) a 1 : 1 chloride-to-surfactant association ratio
(hence the intermediate role given to the counter-ions); (iv) the
easy removal of the template with ethanolic solution suggesting
a neutral final skeleton; (v) the importance of the silica source
in forming the meso-structure (if there is no hydrolysis occur-
ring there is no mesophase produced, hence the assumption
that I1 species are crucial). However, none of these 5 experi-
mental pieces of evidence unequivocally proves the model:4 (i)
the chemistry of silicas at negative pH has never been investi-
gated, meaning that only assumptions at this stage can be made
about speciation of silica at such low pH.7 Moreover, between
pH ¼ 0 and 2, even though protonated silanols are postulated
to explain the acid-catalysed hydrolysis of TEOS, they have
never been detected experimentally;7 (ii): there are no data to
support the statement that the concentration of protons in
solution remains constant. Moreover, the acidity of the med-
ium can be excessively high (4.4 M HCl for a typical SBA-1
synthesis, giving a theoretical pH of about ꢀ0.7) such that pH
measurements with a pH-meter no longer have any meaning.
So, unless other techniques such as titrations are used to
measure the concentration of protons, it is difficult to be sure
of the constancy of proton concentration in solution; (iii) there
w Electronic supplementary information (ESI) available: Experimental
data. See http://www.rsc.org/suppdata/cp/b4/b419293c/
z Current address: Institut de Science et d’Inge
´
nierie Supramole
´
cu-
´
laires, 8 allee Gaspard Monge, F-67083 Strasbourg, France.
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
1845

View Online
little in situ work has been published on SBA materials to
corroborate the proposed S1XꢀI1 and SꢀI1 pathways. Final-
ly, simple DLVO considerations (see supplementary material)w
would suggest that at such high salt concentrations where
Coulombic interactions are almost entirely screened then
charge matching pathways will be highly compromised.
Choice of experiments: in situ NMR (17O, 14N, 29Si) and
EDXRD
Fig. 1c gives a schematic of the processes which are targeted in
this work: the initial hydrolysis of silica species; ordering of the
surfactant phase and of the silica phase; condensation or cross-
linking of the silica phase. The first process, hydrolysis and
polymerisation of TEOS, is of interest as it is not reported in
the literature at such low pH.7 Thus, the acid-catalysed hydro-
lysis of silica plus the various degrees of polymerisation of
these species need to be investigated, with and without the
surfactant present, to account for the effect of the micelles
on the silica chemistry. This is targeted by 17O NMR as
17O-species in the liquid state are easily and rapidly detectable
using NMR. This is important as rapid hydrolysis cannot be
followed with 29Si MAS NMR since the relaxation time for
silicas is too large.
The rate of ordering within the mesophase is monitored by a
combination of energy-dispersive X-ray diffraction (EDXRD)
as well as 14N NMR. Finally, polymerisation and condensation
of silicas are assessed with 29Si MAS NMR. The range of
experiments performed are detailed in Table 1 with the com-
postion of reagents given in Table 2.
Fig. 1 (a) Arrangement of spherical and oblate ellipsoidal micelles
which make up the I1 cubic phase, space group 223. (b) Surface of the
structure of SBA-1 which wraps around these micelles. (c) Schematic
representation of the kinetics related to processes that may be involved
in forming SBA-1 and targeted in this in situ work. The time scale is
purposely arbitrary, it only gives an idea of the different processes
sought.
Experimental
Surfactant preparation
are no data to support the 1 : 1 Cl : surfactant association.
Moreover, the chloride anions are believed to act as inter-
mediates between the cationic micellar surface and the cationic
silica oligomers via H-bondings. Yet, it is known (ref, 8, p. 55)
that chloride anions are not well solvated in water (only one
molecule of water of solvation) and, consequently, it is difficult
to agree with the concept that they could act as H-bridges
between species; (iv) the easy removal of surfactant with
ethanol has been reported subsequently by Ryoo’s group in
1999.9 However, they use an HCl–EtOH mixture rather than
EtOH alone; (v) no experimental evidence is shown for the
existence of I1 species.
Nearly all other mechanistic studies on the growth of
mesoporous materials have employed basic conditions. This
is the case, for instance, of all the in situ NMR work by
Chmelka and co-workers10 and the in situ EPR work by
Galarneau et al.11,12 When attempts are made to study the
acidic mechanism (in film formation), it is under ‘‘soft’’ con-
ditions.13 A review published in 2002 gathering all the different
types of in situ techniques used for monitoring mesophase
growth has very little concerning work in acidic conditions.14
This is probably due to the fact that the chemistry of silica
solutions is not well known below pH ¼ 0 which makes any
mechanistic assumption really delicate. A recent paper from
Tiemann et al. has opened a new pathway for studying in situ
processes at low pH (3.9 M HCl) using in situ small-angle
X-ray scattering (SAXS)15 and a series of papers from
Alfredsson et al.16,17 use combined SAXS and 1H NMR to
follow the preparation of mesoporous silicas in the presence of
block-coploymer templates at low pH. There are some limita-
tions to the techniques available to study these processes owing
to the low concentration of reacting species. A typical surfac-
tant/water molar ratio is 1 : 3500 and TEOS/water is 5 : 3500
for SBA-1 which makes 14N NMR or 29Si MAS NMR for
instance, extremely challenging. For the same reason, the
appearance of ordering as a function of time is difficult to
access experimentally. This might be another reason why so
The surfactant hexadecyltriethylammonium bromide (HTEAB)
was prepared by mixing 1-bromohexadecane (98% Lancaster)
and triethylamine (99% Jansen Chimica) in absolute ethanol
under reflux conditions for 24 h. Ethanol is then removed with
a rotary evaporator until a white, viscous paste is obtained.
The resulting gel is recrystallised by a minimal addition of
chloroform, and then ethyl acetate until the whole solid pre-
cipitates. The detailed procedure is described elsewhere.9 Batch
purity was then checked with 1H NMR (solvent chloroform-d)
as well as microanalysis (C, H, N and Br). A purity of 97% is
normally achievable.20 Optical microscopy penetration scans
using a polarised light optical microscope were also performed
on HTEAB with water in order to have an approximation of
the minimum temperature at which lyotropic liquid crystal
phases occur. The solid surfactant placed between microscope
slide and cover-slip is contacted with water which allows the
observation of all the mesophases as distinct bands.18,19 We
report a detailed study of HTEAB phase behaviour else-
where.20
SBA-1 preparation
The silica-surfactant mesostructures were allowed to form
under various time and temperature conditions, using TEOS
(98% Aldrich) or TMOS (98% Aldrich) as a source of silica,
our lab-made HTEAB, distilled water and aqueous HCl
(33 wt% BDH). An optimum synthesis condition for SBA-1
is HTEAB:H2O : HCl : TEOS 1 : 3500 : 280 : 5. The details of
the protocol are given in ref. 9. In order to perform these in situ
experiments the ratios where sometimes modified (see Table 2)
in order to optimise the experiment. When swelling of the
SBA-1 phase was required for EDXRD experiments—as a tool
to track changes in d-spacings—a 10 wt% bromohexadecane
to HTEAB was added to the initial mixture (see Table 2).
However, in all instances SBA-1 was formed at the end of the
experiment. After a given ageing time at a given temperature
1846
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Online
Table 1 Range of experiments performed at various conditions of pH, temperature and composition
Experiment Experiments performed
code
pH
Temp./1C
HTEAB EDXRD
batch codec
tbatch code
t from
t from
present
from EDXRD/mind
17O/ mind
14N/ mind
1
2
EDXRDb
ꢀ0.7
ꢀ0.7
ꢀ0.7
ꢀ0.7
ꢀ0.7
0
4
25
25
60
60
25
25
60
60
25
60
Yes
Yes
No
Yes
No
Yes
No
Yes
No
Yes
Yes
Yes
Yes
No
No
F, H
B, D, G, I
35F, 30H
13B, 25D, 14G, 15I
EDXRD,b 17O, 14N
1.4
3.7
13
60
3
17O
4
17O, 14N
5
17O
6
EDXRD,b 17O, 14N, 29Sia
C
E
?
?
17
50
7
17O, 29Sia
0
8
EDXRD,b 17O, 14N
0
9
17O
0
10
11
12
13
14e
15e
a
EDXRD,b 17O, 14N
1
14N
17O
17O
17O
17O
1
1 þ EtOH 25
2
1
2
25
25
25
4500
4500
b
c
Both TEOS and TMOS used as source of silicon. Subsequently heated to 90 1C. F, B (H2O/TEOS) ¼ 140; H, D (H2O/TEOS) ¼ 700; G has
d
TMOS as source; I has additional swelling agent. See text for explanation of t-value. Question mark indicates particles settle in reactor and
e
consequently are not in X-ray beam. Timescale much longer than duration of experiment. Error for D and H is ꢁ 5 min and ꢁ 2 min for other
samples.
(high quality batches 1 week at 4 1C), the mixture is heated to
100 1C within 10 min for 1 h. The surfactant moiety is then
burned out by calcination at 550 1C overnight, with a slow
ramp rate (0.5 1C minꢀ1). Samples were then characterised by
powder X-ray diffraction.
cal positioning from one experiment to the next to avoid any
shimming error. 29Si NMR experiments were carried out in
a 7 mm Bruker probe using a 7 mm rotor. Magic-angle
spinning (MAS) in situ 29Si measurements were performed
on the silica-surfactant mixtures at 1 kHz. In all cases, single p/4
were applied with proton decoupling during detection and 30-s
recycle delays. 29Si spin–lattice (T1) relaxation measurements
NMR spectroscopy
were conducted using
a
saturation-recovery pulse
All NMR spectra were acquired at 9.4 T on a Bruker-MSL400
spectrometer. Measurements with H217O (Goss Scientific In-
struments LTF, 10.46 at%) were collected in a specially
designed vial (ca. 0.2 mL) positioned in a 10 mm Bruker static
probe using a single pulse sequence with a recycle delay of 2 s
and 128 scans. 14N NMR measurements were carried out in
home-made vials (ca. 1.5 mL) in a 10 mm Bruker static probe
using a quadrupolar-echo pulse sequence, 901 pulses and 30 ms
interpulse delay, with a recycle delay of 0.5 s. Specially
designed containers for both 17O and 14N allow identi-
sequence. T1 values of TEOS in water have been measured in
this work to be 50 s, whereas T1 values of typical gels formed
are 15 s for Q3 species, and 58 s for Q4 species. These
parameters were chosen as a compromise recognising the very
low concentration of silica and that there will be a slight
emphasis of Q3 over Q4. In all cases, there is an unavoidable
error when starting the reaction, as there is a time gap between
the addition of TEOS and recording the first spectrum. This
time delay is 120 s but is negligible for the detection timescale
of this experiment.
Table 2 Molar quantities of components used for the various experiments performed. Some adjustment of molar quantities was made in order to
suit the experimental technique
Experiment Experiment EDXRD
code
pH
Temp./oC HTEAB H20a
HCl TEOS
TMOS H2O TEOS TEOS/HTEAB
or H2O/TMOS or TMOS/HTEAB
type
batch code
2
6
17O
ꢀ0.7 25
20
174
—
—
1
2876 280
165
952
—
—
—
—
17
18
8
5
17O
0
1
2
25
25
25
17223 280
155 680 280
1561 803 280
3500 280
3500 280
700 56
14
15
2
17O
17O
7504
76 390
21
20
—
—
5
EDXRD
EDXRD
EDXRD
EDXRD
EDXRD
EDXRD
EDXRD
EDXRD
14N
D
H
B
ꢀ0.7 25
ꢀ0.7
ꢀ0.7 25
5
5
5
5
700
700
140
140
140
140
140
140
101
140
169
140
140
140
140
1
4
1
5
2
1
5
1
F
ꢀ0.7
4
1
700 56
700 56
5
2
G
Ib
C
E
ꢀ0.7 25
ꢀ0.7 25
1
0.9
5
5
5.5
2
700 56
700 17
5
5
5
5
5
6
0
1
25
25
1
1
5
10
2
700
1
5
ꢀ0.7 25
1
505 53
700 18
5
6
14N
0
1
0
0
0
0
25
25
25
25
25
25
1
5
10
6
14N
1
812
1.5
4.8
4.8
5
29Si
1
700 13
700 13
700 13
700 13
5
—
5
6
29Si
1
5
5
5
—
7
29Si
—
—
b
7
29Si
—
—
10% enriched H217O was used for 17O NMR experiments. Additional 0.1 mol bromohexadecane swelling agent.
a
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
1847

View Online
suggest that: (i) 17O is present only in water and silica species;
(ii) ethanol molecules are formed throughout the process but
do not contain 17O; (iii) water molecules recovered at the end
of a hypothetical full condensation will contain less 17O than at
the beginning of the reaction.
For 17O NMR measurements, owing to the small amount of
TEOS in SBA-1 synthesis and to the potential difficulty to
monitor silica species, the molar ratio H2O/TEOS has been
decreased from 700 to about 20. However, the ratio TEOS/
HTEAB is maintained.
For 14N NMR measurement owing to the very dilute con-
centration of HTEAB in water in a typical SBA-1 synthesis and
as a consequence to the related difficulty to detect 14N signal,
the HTEAB concentration has been increased approximately
5 times along with that of TEOS.
For 29Si MAS NMR measurements concentrations were
such that the gels are roughly 10 times more concentrated in
TEOS (or TMOS) and HTEAB than in a typical SBA-1
synthesis. This is to enhance the signal to noise ratio of 29Si
species. As there is no evidence of Q0 species even at early
stages of reaction the spectra were added together on a time
scale corresponding to 4 h (480 scans with a repetition delay of
30 s) in order to enhance the signal-to-noise ratio. After 24 h,
the final gels left in the initial beaker were heated to 100 1C for
1 h, the resulting solids being filtered and dried at ambient
temperature. X-ray diffraction was run on the final powders.
Samples of SBA-1 have been prepared with and without the
surfactant, from pH 2 to ꢀ0.7, using the amounts of reactant
given in Table 2. Fig. 3 gives the results for the experiments
carried out at pH 0 (3a and 3b) and approx. ꢀ0.7 (3c and 3d) in
the form of a 2D intensity plot (left), combined with the peak
decay, both as a function time (right). At the end of the run
carried out at RT, batches are kept for two weeks in a sealed
vial before further study at higher temperature (60 1C).
Concerning the entire work carried out at RT, three ob-
servations can be made: (i) pH influences the chemical shift of
the water peak; the chemical shift becomes less shielded as the
pH value decreases. It typically varies from d ¼ 0 ppm (pH ¼ 7)
to d E 2 ppm (pH ¼ 0) and then to d E 10 ppm (pH ¼ ꢀ0.7);
(ii) as soon as TEOS is introduced in the vial, a decay in the 17
O
water peak is observed with time. It is this decay which
contains the kinetic information of processes involved; (iii)
no ‘‘17O-silica’’ peak is observed unless small silica species are
stabilised through addition of extra ethanol.22–24 Upon heat-
ing, two trends are observed: (i) the chemical shift becomes
more shielded at 60 1C, but the process is always reversible; (ii)
there is a slight decrease in intensity of the signal that can be
attributed to the Boltzman effect. A change in the shape of the
peak is also observed and will be discussed later. Thus,
intensity decay of the water peak is the only direct information
that can be extracted from the full set of data.
At pH 1 and 2 (data not shown), nothing happens to the
water signal, the system is stable on the experimental time
scale. Indeed, these two pH values are relatively high and too
close to the isoelectric point of Si(OH)4 (pKi ¼ 2) for the
hydrolysis and polymerisation to be rapidly acid catalysed.
Such a high pH also does not yield SBA-1 due to the lack of
reactivity of TEOS and hence no experiment has been carried
out in the presence of surfactant under these conditions.
At pH ¼ 0 a decay is observed both in the presence and
absence of HTEAB owing to the acid catalysed processes (Fig.
3a and 3b, respectively). However, the rate of the decay is
substantially faster in the presence of HTEAB. These decays
do not obey a simple exponential behaviour suggesting that
more than one process is involved. However, it is possible to
measure the shortest time at which the decay is almost com-
plete. Without HTEAB it requires about 3000 s (50 min) for
the water peak to decay to its minimum value, whereas with
HTEAB only 1000 s are needed (about 17 min). The transfor-
mations that 17O in water undergoes, if similar, are three times
faster in the presence of HTEAB. No 17O signal at higher
Energy dispersive X-ray diffraction (EDXRD)
The energy dispersive diffraction method uses a white beam of
X-rays; hence the wavelength, l, becomes a variable. The
detectors are positioned at desired angles, y, which are fixed.
The measurements were carried out at Station 16.4 at CLRC
Daresbury Laboratories (UK), following a similar set-up to
that used and described in detail by O’Brien et al.21 The main
advantage of EDXRD is that there are no moving parts in the
set-up: dynamic data collection is then simple and extremely
rapid (resolution down to 1 s is achievable). Most importantly,
in the SBA-1 synthesis stirring can be achieved in a similar
manner to that used for normal SBA-1 preparations. The white
energetic X-ray beam emerging from a synchrotron (20–150
keV) is very penetrating so that truly bulk samples can be
analysed. However, the fast collection of data has its draw-
back: the quality of the diffraction pattern is intrinsically
limited by the performance of the X-ray detector; the diffrac-
tion peaks are much broader than in standard scattering
patterns. Consequently, if diffraction peaks are close together
they are likely to overlap in EDXRD data.
Typically, a volume of ca. 15 mL of an SBA-1 mixture is
placed in a simple stirred glass cell in front of the beam. The
lowest angular detector collecting the X-ray of interest (large
d-spacing) forms an angle 2y of 0.71 with the incident beam.
This angle is optimal for the type of work desired here, and
consequently only data collected at this detector are presented.
Patterns have typically been collected every 2 min. A heating
and cooling facility (ethylene glycol bath) was available at the
station and has been used to examine the effect of temperature
on the kinetics of mesophase growth (from RT to low tem-
perature for the ageing step) as well as in condensing the silicas
upon temperature increase (condensation step). The tempera-
ture was directly measured within the cell with the use of a
thermocouple. Using the ethylene glycol bath, 90 1C was the
maximum temperature achievable and the temperature was
never lower than 4 1C. When the concentration of TEOS is 5
times greater than it should be in a normal SBA-1 synthesis, the
concentration of HTEAB is increased by the same ratio.
Results and discussion
Hydrolysis and polymerisation of TEOS: in situ 17O NMR
Starting with the silicon alkoxide, 17O from H217O is intro-
duced into the silica species through hydrolysis. The location of
17O arises from the acid-catalysed hydrolysis and polymerisa-
tion mechanisms which are given in Fig. 2 for pH o 2, pH ¼ 2
being the isoelectric point of Si(OH)4.7 These two mechanisms
Fig. 2 Acid-catalysed mechanisms of hydrolysis and polymerisation
of TEOS at pH o 2, emphasising the location of 17O in water and silica
species.
1848
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Online
the loss is 57% [ꢁ5%] without HTEAB, and 37% with
HTEAB, whereas at pH ¼ 0 the losses are, respectively, 35%
and 51%. When H217O molecules are used to hydrolyse TEOS,
17O is incorporated in the silica species (signal losses from the
initial water signals are collected in Table 3). When these
species start polymerising, they can lose their motion on the
17O NMR time scale, which renders part of the 17O signal
‘‘invisible’’. Consequently, only a loss in the water signal is
detected. However, the observed loss is always greater than can
be accounted for by complete hydrolysis and consequently an
additional process is occurring.
Correlating signal intensity loss in 17O NMR with the
corresponding fate of the water molecules is not straightfor-
ward.17O is a quadrupolar nucleus with spin 5/2. Conse-
quently, there are 5 possible transitions all of which will be
observed if the species undergo rapid reorientation (the rota-
tional correlation time for free water is of the order of
picoseconds and the quadrupole splitting is of the order of
7 MHz25). However, loss of mobility is liable to render all but
the central transition (mI ¼ – 1/2 - 1/2) invisible. This could
relate to bound-water molecules that are in exchange with free
water at a rate slower than ca. 1/7 MHz, on the order of 1.4 ꢂ
10ꢀ7 sꢀ1 (in the same manner 17O-containing silica species with
motion slower than 1.4 ꢂ 10ꢀ7
s
ꢀ1), 7 MHz being a typical
quadrupolar interaction for 17O.25 This central transition
represents 9/35 (ca. 25%) of the entire signal26 and conse-
quently if all the 17O nuclei had motion severely restricted then
the NMR signal intensity would decrease to ca. 25% of its
original value. In other words, if the signal decays by 75% then
we must multiply this value by exactly 35/26 to determine the
percentage of 17O with severely restricted motion (75% ꢂ 35/
26 E 100%). As an example, at pH ¼ ꢀ0.7 in the absence of
HTEAB there is a 57% signal intensity loss which amounts to
77% of 17O with reduced mobility. If 23% of this is concerned
with water consumed in the hydrolysis process then the re-
maining 54% must be related to ‘‘bound’’ water. Table 3 gives
an analysis of the 17O results in terms of the amount of bound
water. It should also be noted that even the central transition
could be lost from second order broadening effects caused by
severely restricted motion in solid-like components. This,
however, seems unlikely as the changes to the 17O NMR
spectrum occur when the solution is still clear and there is no
evidence of solid precipitation.
A number of trends can be extracted from the 17O in situ
work: (i) pH influences drastically the rate at which TEOS
undergoes transformations; (ii) HTEAB increases the reaction
rate (about threefold); (iii) the intensity loss in the 17O reso-
nance will be associated with oxygen with reduced mobility a
large fraction of which is most likely associated with substan-
tial amounts of bound water.
No 17O-silica species have been detected on the time scale of
the NMR experiment at the concentrations of TEOS used,
whereas Babonneau et al.22–24 observed them at slightly higher
pH and in the presence of extra EtOH. Ethanol seems to
stabilise silica species such as silicic acids and oligomers formed
through hydrolysis and partial polymerisation. To confirm
this, we carried out an experiment at pH ¼ 1 in the presence
of ethanol (H2O/EtOH ¼ 5, otherwise the same as experiment
code 14 in Table 2). Silica species are detected at about 25 ppm,
Fig. 3 Time-resolved in situ, room temperature, 17O NMR carried out
at pH ¼ 0 (a without surfactant, b with surfactant) and pH ¼ ꢀ0.7 (c
without surfactant, d with surfactant) showing the evolution in time of
the water signal. A spectrum of 50 mL of H217O is inserted at the
beginning of the series of 2-D data (line width is typically 170 Hz), as
well as the spectrum of H217O þ HCl (and that of H217O þ HTEAB if
applicable) in order to be able to compare the first peak after addition
of TEOS with a reference.
chemical shift (in silica signal22–24) has been observed. At
negative pH, with or without HTEAB, a large and fast decay
is observed (Fig. 3c and 3d). This strongly supports the acid-
catalysed mechanism. However, whereas the experimental
decay without HTEAB is easily fitted with an exponential
function (Fig. 3c), the fit is poor in the presence of surfactant
(Fig. 3d). Again this suggests at least two different types of
processes going on depending whether HTEAB is present or
not. A time constant t characterising the transformations
occuring in the case of TEOS at pH ¼ ꢀ0.7 without HTEAB
is of an exponential function type: I ¼ I0 ꢂ eꢀt/t with I0 the
intensity at time 0 and t equals 3.7 minutes (ꢁ0.5 min, see
Table 1). The time constant from the poor fitting of the
experiment with HTEAB would be 1.4 minutes, about 2.5
times faster than without the organic moiety (it should be
noted that the point at zero time is taken in the absence of acid
to ensure that hydrolysis is not initiated).
The kinetics results shown at pH 0 and ꢀ0.7 are based only
on the decay of the height of the signal as the line width is
nearly constant. However, percentage losses are calculated
more accurately from integration of the first (reference H217O)
and last peaks and give the following results: at pH ꢀ0.7,
Table 3 Possible fate of water to account for loss of 17O NMR signal intensity. The signal loss is multiplied by 35/26 to determine the amount of
17O with restricted motion. This is then divided between water involved in hydrolysis and bound water
pH ¼ ꢀ0.7
pH ¼ ꢀ0.7
pH ¼ 0
pH ¼ 0
no HTEAB
with HTEAB
no HTEAB
with HTEAB
%
%
17O signal loss
57
77
23
54
38
51
23
28
35
47
23
24
51
67
23
44
17O with restricted motion
% Maximum loss for hydrolysis
% Minimum associated with bound species
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
1849

View Online
particles formed. They are extremely ‘‘sticky’’ and readily settle
at early stages of the reaction, as a very weak hump beside the
main water peak. At this chemical shift these species are
expected to be silanols.24 This simple experiment only proves
that if silicas are not seen in the in situ work detailed above, it
may only be related to their lack of mobility during the
reaction instead of an experimental limitation, such as TEOS
concentration (H2O/TEOS E 20).
at the bottom of the cell despite the stirring. Hence the particles
are not in the X-ray beam. Fig. 5 presents the growth of the
diffracted intensity as just defined as well as the evolution of the
d-spacing as a function of time, for batch B, synthesised at
pH ¼ ꢀ0.7. Final batches of SBA-1 have been kept in their
liquor solution during the time at Daresbury and then filtered
and dried upon returning to the laboratory. XRD of the final
powders have been recorded (data shown only for Batch B, as
in inset in Fig. 5, other data are contained in the supplementary
informationw).
A number of observations come from these results: (i) at
pH ¼ ꢀ0.7 ordering occurs within 2 min and is already almost
complete on the time scale of a few tens of minutes; (ii)
diffraction intensity increases substantially with heating; (iii)
d-spacings decrease slightly (ca. 2 A lower) with increased
temperature, due to an expected shrinkage of the structure
upon heating.
In order to extract kinetic information from these results,
growth curves for the two d-spacings tracked from time 0 to a
time just before heating have been fitted with exponential
functions. The result of the fits and the calculated exponential
time constants are given in Fig. 6 for batch B. Errors on the
heights of diffracted peak measurements have been estimated
and are explicitly shown with the use of error bars. Accuracy is
necessarily less in the case of normal concentration batches due
to the smaller amount of matter used. From the full set of data
for each batch, (Fig. 6 for batch B, other data shown in
supplementary informationw) the following information can
be determined:
Mesophase growth: in situ EDXRD
Energy dispersive X-ray diffraction experiments were run
varying pH, concentration, temperature of the initial mixture,
as well as the presence or not of a swelling agent. Tables 1 and
2 summarise the set of 8 in situ experiments carried out,
labelled batches B to I.
Fig. 4 shows the evolution of an EDXRD pattern (batch B)
with time. From such a set of patterns, the growth is measured
as the height of two diffracted peaks that grow with time
(initially 38 and 42 A). They are compared against the max-
imum height of the background around 26 A. The d-spacings
vary slightly with time and this variation has been followed
during the mesophase growth. Batch SBA-1 E at pH ¼ 1 has
led to an amorphous powder (data not shown). No growth has
been observed with EDXRD for this batch. At pH ¼ 0,
although no growth was easily observed, despite repetitive
attempts, the XRD pattern of the final material SBA-1 C is
characteristic of SBA-1 structure. The difficulty in monitoring
the appearance of ordering at pH ¼ 0 lies in the nature of the
(i) Growth occurs roughly on the scale of 15 min at RT.
(ii) The accuracy of the time constant, t, is different accord-
ing to the concentration of constituents within the cell. (batches
D and H ꢁ 5 min, higher concentrations, ꢁ2 min).
(iii) A 4 1C ageing step slows down the kinetics of mesophase
formation by ca. a factor of 3 (compare t from B (ca. 13 min)
with t from F (ca. 40 min)).
Fig. 4 In situ EDXRD patterns of the SBA-1 batch B at initial, middle
and final stages of the reaction. Growth is measured as the height of
peaks emerging from the background at d ¼ 38 and 42 A. These heights
are compared to the maximum background height around d ¼ 26 A.
Fig. 5 Above: Growth of two diffracted peaks with time, at RT and
then upon heating to 90 1C for SBA-1 batch B. Inset: final XRD
pattern of the resulting powder. Below: Evolution of the d-spacings
with time at RT and then at 90 1C. Initial pH is ꢀ0.7.
1850
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Online
Fig. 6 Time constants of growth extracted from exponential fits with
experimental data before heating, for SBA-1 batch B, for two different
d-spacings (38 and 42 A).
Fig. 7 Decay of 14N NMR signal as a function of time for pH ¼ 0
(above) and ꢀ0.7 (below). Insets: XRD patterns of the final materials.
(iv) The use of TMOS instead of TEOS, or of a swelling
agent, does not seem to affect greatly the rate of SBA-1
ordering (14 min for G, 15 min for I vs. 13 min for B).
A summary of the final time constants determined from the
raw data along with their uncertainty is given in Table 1. Two of
the messages coming out of EDXRD results have their echo in
the literature with the work from Tiemann et al.15 which details
the in situ SAXS study of meso-silica under acidic conditions:
detection of the mesophase growth within 2 min and contraction
of the mesophase with time and /or temperature.
formed at pH 1, this material remains particularly disordered
(data not shown). However, XRD patterns of materials made
at pH 0 and ꢀ0.7 are typical of SBA-1 structure. To avoid
confusion about the decay of the peak and the amount of real
loss in the signal, integrations of the first and last peak of the
two sets of data (at pH ¼ 1 the data are not shown) have been
calculated to be: 2% at pH ¼ 1, 67% at pH ¼ 0, and 50% at
pH ¼ ꢀ0.7 (error ꢁ5%).
The main information from the in situ 14N NMR work is
that no obvious process is observed at pH ¼ 1 after 4000 s,
whereas a decay of the peak height is observed on this time
scale at pH ¼ 0, and a transformation occurs within 1000 s at
pH ¼ ꢀ0.7. These time scales, strongly connected with the pH
values, are on the same order of magnitude as the ordering
monitored with EDXRD experiments.
Mesophase growth: in situ 14N NMR
In situ 14N NMR gives information about the local organisa-
tion of surfactant molecules as the reaction proceeds and can
consequently help track mesophase formation. However, in
this particular case, the 14N NMR tool is limited as the initial
stage of the reaction contains an isotropic solution L1 (single
line of width 50 ꢁ 10 Hz in 14N NMR) and the final product of
Both the 17O NMR and 14N NMR results show a similar
trend although they cannot account for the same processes as
they occur on different time scales: at pH ¼ ꢀ0.7 a transforma-
tion is visible in the 17O signal within 2 min whereas a change is
observed with the 14N NMR within 15 min. Yet, they both
intrinsically contain information about mobility of species on
their respective NMR time scale.
3
the reaction is the I1 cubic phase, isotropic by nature (single
line, 100 ꢁ 10 Hz, see supplementary information for 14N
NMR spectra of water/HTEAB mixturesw).20,27 Nevertheless,
this line-width change is sufficient to use the maximum signal
height as a kinetic probe of the onset of order. Tables 1 and 2
gather the different types of batches prepared for in situ 14N
NMR work with three different conditions of pH studied.
Owing to the very dilute concentration of HTEAB in water
in a typical SBA-1 synthesis and as a consequence of the
related difficulty to detect 14N signal, the HTEAB concentra-
tion has been increased ca. 5 times along with that of TEOS. It
also renders in situ EDXRD and 14NMR data directly compar-
able.
Condensation: in situ 29Si NMR
As SBA-1 forms, changes occur in the silica species which can
be monitored with in situ 29Si MAS NMR with high power
proton decoupling. Gels of different compositions have been
studied at RT at pH ¼ 0 with tetramethylorthosilicate (TMOS)
or tetraethylorthosilicate (TEOS) as silica source. The idea is to
compare their relative behaviour as their solubility in water is
slightly different (TMOS is more soluble than TEOS). The
experimental detail of the gels are summarised in Table 1.
Results with TMOS or TEOS, either with or without the
surfactant HTEAB are gathered in Fig. 8. Both resulting gels
were dried and checked by X-ray diffraction, data not shown,
revealing the SBA-1 structure. Owing to the low concentration
of silica species, low sensitivity of 29Si and long relaxation times
the spectra can only be collected over long time intervals—and
Only a single line is observed in these sets of in situ 14N
NMR experiments. The height of the peak is plotted as a
function of time for the three sets of spectra, recorded at pH ¼
1, pH ¼ 0 and pH ¼ ꢀ0.7 (Fig. 7 for the results at pH ¼ 0 and
pH ¼ ꢀ0.7, pH ¼ 1 is shown in supplementary informationw).
Insets with XRD patterns of the final materials are also
included. Similar to the in situ 17O NMR and EDXRD
experiments, transformations occurring are faster the lower
the pH. Although a material with the meso-scale definition is
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
1851

View Online
solutes drastically decreases and the aqueous solutes can pre-
cipitate out from the dilute aqueous phase. This explanation
would be extremely useful if it could be applied to the case of
SBA-1. Indeed, in a typical SBA-1 synthesis the concentration
of HCl is 4.4 M making salting out effects extremely likely:
although Clꢀ ions are poorly hydrated (one water molecule per
anion), H3O1 has three water molecules of hydration, ref. 8,
p. 55. Hence, 3 ꢂ 4.4 moles of water are used to hydrate the
protons from the 55 available (in one litre of solution). So 13.2
moles of water are used to solvate protons alone, to which 4.4
must be added to account for solvation of Clꢀ. Then, 1/3 of the
water (in moles) is used for hydration of HCl and is no longer
available for solubilising the micelles. This will change a great
deal the micellar solubility so that the micelles arrange them-
selves in a concentrated phase where less water is required. It is
known28 that alkyl(C12ꢀ16)tributylammonium bromides show
this type of partial miscibility above a certain temperature
when added inorganic electrolytes increase the phase separa-
tion. Thus, this family of tetra-alkyl ammonium surfactants is
prone to such behaviour. It is an indication that only a small
change in the interactions between the surfactant micelles can
result in phase separation. However, attempts to precipitate the
I1 phase from a dilute solution of surfactant only with high
concentrations of HCl have always been unsuccessful. Yet,
since a mesophase does precipitate when a source of silicon
alkoxyde is present it means that silica species play a crucial
role in the overall solubility of the system.
Fig. 8 In situ 29Si HP MAS NMR spectra of TEOS/TMOS in water at
pH ¼ 0 with and without HTEAB, recorded at RT.
even then the spectra display considerable noise. Nevertheless,
some observations are made as follows:
(i) At this pH, independent of the gel composition, there is
relatively little Q4 species within the gel, indicating that the
silica network is poorly condensed;
(ii) there is a decrease with time of Q2 and also Q3 species
within the gels; however Q3 species remain predominant,
whatever the gel composition;
The precipitation of an ordered mesophase occurs on the
same time scale whether monitored with EDXRD or 14N
NMR. An I1 phase is formed, to the detriment of other phases
which can be described with the concept of water activity.
When the I1 phase forms it must have the same water activity
(or chemical potential) as that of the dilute aqueous phase
within which it is dispersed. The water activity depends on the
‘‘free’’ water concentration. Water activity changes throughout
a phase diagram of a surfactant in water, decreasing mono-
tonically as the surfactant concentration is increased. Phases at
low water-content have lower water activity than those at high
water content (a lamellar phase has a lower water activity than
a hexagonal phase). A micellar solution has high water activity
since it has the lowest surfactant concentration. A schematic
representation of this effect throughout the partial phase
diagram of HTEAC is given in Fig. 9a. The reduction in water
activity with concentration can be represented as arising from
repulsions between the micelles.28–32 When a concentrated
phase coexists with a dilute one, the water activity is the same
in both—hence the water activity is constant in the two-phase
region. The concentrated phase has smaller repulsive interac-
tions between the micelles than would normally be expected—
i.e. there is an additional inter-micellar attractive force. Thus,
in the L1 þ I1 region, where the initial stages of SBA-1
formation occur, there must be an additional attractive inter-
action between the micelles in the concentrated phase. We
suggest that this arises from the presence of the silica oligomers
(as illustrated in Fig. 9b). In some way these must ‘‘cross-link’’
adjacent micelles to provide the attractive interaction (see
below). The partition of water in the system can be observed
by 17O NMR: the chemical shift changes with pH—data not
shown—suggesting a strong solvation of the protons by the
water molecules, unshielding the magnetic field that 17O now
experiences. Moreover, at 60 1C, as the overall system becomes
more mobile, two types of signal are detected strongly suggest-
ing two types of water, silicate-bound water and H1-solvation
water. When the partition is such that the system equilibrates,
the system is precipitated out: it has reached a point at which
the water activity is equal to that on an I1 phase. The I1 phase
forms as the silicate polymerises.
(iii) there is no obvious difference between gels with TMOS
as a silica source and those with TEOS;
(iv) there seems to be an influence from the HTEAB
molecules to the degree of linkage within the gel: there are
more Q3 species in the presence of HTEAB than without on the
same time scale; this may simply be a result of the increased
local concentration of silica species at the micelle surface,
giving a more rapid reaction rate.
The 29Si NMR results confirm the trend reported in the
literature that condensational cross-linking is not promoted at
very low pH despite increased rates of hydrolysis. The degree
of condensation is poor due to a preferential polymerisation
between chain ends and monomers. As pH decreases the
amount of Q4 species within the gel becomes less whereas Q3
species become predominant. From ref. 7 it is known that
oligomers formed by an acid-catalysed mechanism remain
small: up to seven silicon atoms only at 0.05 M HCl (pH E
1.3). At this stage, the detailed speciation of silica is required in
order to have a clearer idea of what is actually interacting as
the phase grows and in this respect mass spectrometry might be
useful in the future.
The mesoporous material formation mechanism
It is known from surfactant chemistry that upon addition of
electrolyte to a solution the solubility of the solute becomes
affected such that a concentrated surfactant phase (containing
ca. 50 wt% water) precipitates from the aqueous phase. This is
known as the salting-out effect. One simple way to explain this
effect is to consider the solvation of the introduced ions as each
ion requires a certain number of molecules of water as a
solvation sheath. Because the ions are usually small they are
easily and preferentially solvated. At high electrolyte concen-
tration, there are so many ions in the system that a large
number of water molecules serve to hydrate them. Conse-
quently, these are no longer available for solvating other
species present in solution, particularly those that are only
weakly solvated. Hence, the solubility of weakly solvated
The loss of signal intensity of both 17O and 14N demon-
strates that relatively slow molecular motions of both water
and surfactant develop as the reaction proceeds, but that they
1852
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Online
the I1 phase. After the initial hydrolysis of the TEOS (or
TMOS) the oligomeric silica species bind to the micelle surface,
probably being located between the head groups close to the
hydrophobic interior (in the ‘‘palisade layer’’). Here they can
become cross-linked, and eventually provide the inter-micellar
attraction leading to precipitation of a concentrated surfactant
phase. Thus the concentration of silica solubilised in the
precipitate is likely to be at least half that of the surfactant.
Initially, this precipitate might be a disordered micellar phase,
rather than an I1 structure. Note that compared to the material
formed at pH ¼ ꢀ0.7, the precipitate formed (more slowly) at
pH ¼ 0 is far less ordered. This may be because the slow
formation leads to a more polydisperse mixture of silica
oligomers, with differing stages of polymerisation and con-
densation. The precipitate phase at pH ¼ ꢀ0.7 rapidly becomes
more condensed and loses water. The water loss, possibly
resulting from the reduced water-binding of silica polymers
because the SiOH groups are removed, leads to an increased
surfactant concentration in the phase, resulting in the forma-
tion of the I1 structure. Hence, the I1 formation is probably a
two-stage process. [Precipitation of a concentrated amorphous
phase, structuring within the amorphous phase to form I1].
From the 29Si NMR data, we see that the silicas formed are not
highly condensed, thus there is access for water binding which
is reduced as the polymerisation becomes complete. The higher
concentration of Q3 species in the presence of surfactant
probably arises from the enhanced concentration of the silica
around micelles, leading to a faster reaction.
Recently, we have shown that the final structure of SBA-1 is
a very low curvature surface formed from a high curvature
surfactant micelle mesophase.5,33 Consequently, as the order of
the silica develops the silica must retract from the intermicellar
region in order to create windows. The final interface will be
the low curvature interface between the silica and the water
rather than between the silica and the surfactant. This is not
surprising as the silica becomes more hydrophobic upon poly-
merisation and also water is produced as the silica condenses.
Both these facts will serve to preferentially solvate the surfac-
tant with water in the later stages of synthesis producing a
bound water layer on the surfactant. A similar process must be
in operation for other cage-like mesoporous materials where
windows are formed such as SBA-6, SBA-12 and SBA-16.
Thus the mechanism is:
Fig. 9 (a) Simplistic representation of the change of water activity
within the schematic phase diagram of HTEAC. (b) Partition of water
due to a salting-out of the composite silica–surfactant mesophase from
the highly concentrated solution of HCl (4.4 M), coupled with a loss of
water mobility from binding to the silica species.
occur on different time scales. For the water, this slow motion
must arise from binding to the silica polymerising species
produced in the first seconds/minutes of the reaction. Thus,
the 17O signal loss is due to a broadening of the satellite
quadrupolar transitions for the bound water; this then ex-
changes rapidly on the NMR time scale with free water. The
signal loss is not from 100% of the water, indicating that the
silica species are too dilute for all the water to contact the silica
species on the time scale concerned. The fraction of 17O signal
lost is smaller in the presence of HTEAB at pH ¼ ꢀ0.7,
indicating that less water can access the silica species. Thus,
these species are present in larger units, since the surfactant is
unlikely to influence directly the free-bound water exchange
process. The surfactant micelles bind silica polymerising spe-
cies, long before the loss of 14N intensity is observed.
The 14N signal loss is due to a broadening of the quad-
rupolar transitions because of slow molecular motions; in this
case the motions are those that result in the 3D averaging of
the net electric field gradient at the 14N nucleus. The motions
that do this are the overall rotation around the surfactant
micelles and the tumbling of the micelles themselves. We
observe that the loss of the 14N intensity occurs on the same
timescale as the growth of X-ray peaks at 38 and 42 A. These
peaks reflect the growth of the ordered mesophase precipitate,
containing both silica and surfactant. Within this precipitate at
least some of the surfactant head group motions within
micelles become slow on the NMR time scale. Note that this
precipitate does not have a fixed structure, but must be labile—
at least in the early stages. When the micelles come together we
do not know their size. In the final SBA-1 solid there are
micelles of two different sizes. These must form in the first
stages of the precipitate because when polymerisation is well
advanced the structure will be too rigid to allow the necessary
holes within the network. When considering the loss of 14N
signal intensity we must consider the two types of micelle
separately (see Fig. 1). The disc micelles can not undergo
isotropic rotation because they are ‘‘locked’’ in the cubic
mesophase lattice, thus will give a quadrupole splitting. The
spherical micelles could give a sharp NMR line, provided the
diffusion of monomers around the micelle is fast enough (oca.
10ꢀ5 s). Assuming a micelle radius of ca. 2.3 nm, this requires a
self-diffusion coefficient D 4 3.4 ꢂ 10ꢀ12 m2 sꢀ1 [D ¼ l2/6t].
The normal self-diffusion coefficient for surfactant within a
micelle is expected to be ca. 10ꢀ10 m2 sꢀ1. Given that the silica
polymer does not have free rotation (D ¼ 0), it seems reason-
able that the micelles separated from this surface by only a few
water layers could have this reduced diffusion rate. The frac-
tion of signal lost (ca. 60%) is consistent with slow motions for
both types of micelles.
1. Fast formation of silica hydrolysed species—fast binding
to micelles.
2. Slower aggregation of these to form an amorphous
precipitated phase.
3. Further development of order within the precipitate as the
polymerisation continues to form the I1 structure.
4. Movement of the silica network away from the micellar
surface to be fully located in the water region between micelles
with the consequent formation of windows.
A general and final representation of the succession of events
in the mechanism accompanied with their time scale is dis-
played in Fig. 10.
This raises the question of the molecular mechanism by
which the initial silica species bind to the micelles. From
DLVO considerations a charge interaction model seems un-
likely. A related question is: why do the micelles catalyse the
polymerisation reaction? Presumably, the micelles also bind
H3O1 because this is the only way to catalyse the polymerisa-
tion. Thus the concentration of silica and acid are both higher
at the micelle surface than in the bulk. Because of the speed of
hydrolysis it is unlikely that any TEOS is solubilised within
micelles. Further polymerisation of the silica leads to cross-
linking of micelles, followed by development of the X-ray
patterns. At this stage the two sizes of micelles are present.
Further development of the polymerisation process will
reduce the silica water binding—and the location/strength of
silica binding to the micelles. When this happens, the structure
In the initial stages of the precipitation it is unlikely that the
L1 phase contains two different micelles in the necessary
proportions and of precisely just the size required to fit into
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
1853

View Online
timescale of about 2 min and is accompanied by a binding of
water to small oligomers. This reduction in water activity
results in salting out of an ordered silica/water/surfactant
mesophase on a timescale of around 15 min. The silica remains
highly fluid which permits substantial rearrangement of the
silica, retracting from the micelle interface and forming win-
dows in the final SBA-1 structure.
Acknowledgements
The authors would like to acknowledge technical support at
Station 16.4 at CLRC Daresbury for help to set up the
EDXRD experiments. Phil Hughes is also warmly thanked
for his dedicated time at the station. Thanks to John Ridland
at ICI Synetix for the original idea to perform 17O NMR
experiments. Thanks also to ICI Synetix for financial support.
Fig. 10 General kinetic picture representing the different steps in-
volved in forming SBA-1. TEOS is rapidly hydrolysed (complete within
2 min) to silicic acid species that start polymerising, preferentially
linearly (t E 1.4 min). As they polymerise, water molecules lose
mobility and bind to their surfaces. This effect coupled to the salting-
out effect due to the presence of 4.4 M HCl induces the composite
mesophase precipitation (t E 15 min). An ordered soft solid is formed,
which then rearranges to create windows between cages. Further
condensation occurs to produce a hard solid upon heating to 90 1C.
References
1
C. T. Kresge, M. E. Leonowicz, W. J. Roth, J. C. Vartuli and
J. S. Beck, Nature, 1992, 359, 710–712.
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. Mc Cullen, J. B. Higgins and J. L. Schlenker,
J. Am. Chem. Soc., 1992, 114, 10834–10843.
3
4
Q. Huo, D. I. Margolese, U. Ciesla, D. G. Demuth, T. E. Gier, P.
Sieger, A. Firouzi, B. Chmelka, F. Schuth and G. D. Stucky,
¨
Chem. Mater., 1994, 6, 1176–1191.
Q. Huo, D. I. Margolese, U. Ciesla, P. Feng, T. E. Gier, P. Sieger,
is so far developed that it is impossible for the micelles to
escape from the silica framework. The restricted molecular
mobility and the physical presence of the silica will reduce the
micelle exchange kinetics.
R. Leon, P. M. Petroff, F. Schuth and G. D. Stucky, Nature, 1994,
368, 317–321.
¨
5
6
7
8
9
M. W. Anderson, C. C. Egger, G. J. T. Tiddy, J. L. Casci and
K. A. Brakke, Angew. Chem., 2005, 44, in press.
M. Kruk, M. Jaroniec, R. Ryoo and J. M. Kim, Chem. Mater.,
1999, 11, 2568–2572.
C. J. Brinker and G. W. Scherer, Sol-Gel Science, Academic Press,
New York, 1990.
J. Israelachvili, Intermolecular and Surface Forces, Academic
Press, New York, 1992.
M. J. Kim and R. Ryoo, Chem. Mater., 1999, 11, 487.
At this stage the silica becomes too condensed to interact
with the surfactant head groups. So the head groups are
solvated by water (and H3O1) hence the silica moves as far
from them as it can—to the interstices between the micelles—
hence the ‘‘windows’’ between micelles develop at the points of
closest micellar contact. The development of windows and
their dimensions must be determined in part by the volume
fraction of silica in the precipitated phase. It is unlikely that
significant concentrations of silica can diffuse into the precipi-
tated phase after its formation. From previous modelling
studies5 we estimate the volume fraction of silica in the non-
calcined material to be 42%. In typical surfactant systems the
I1 cubic phase in equilibrium with a micellar solution contains
of order 40–50% surfactant (by volume or weight). This is
likely to hold here. Thus the I1 precipitate must contain (at
most) 15–20% water [including electrolyte/acid]. This will be
distributed between the surfactant head groups and silica
species, filling the volume between the micelles. From these
considerations we see that the micelles must be completely
covered with silica oligomers, hence it will be easy for cross-
linking of micelles to occur. Additionally, as the condensation
proceeds, it seems likely that silica binding to the micelles is
reduced because water and surfactant are excluded from the
condensed silica. As the silica shrinks it occupies the spaces
furthest from the micellar surfaces. Finally, we comment on the
fact that regular windows do not occur in other materials such
as MCM 41 where the rod micelles are fully encased in the
surrounding silica structure. Either there is a much larger
fraction of silica in the precipitated hexagonal phase in the
case of the MCM 41, or the (intermicellar) surface area of the
SBA-1 structure is too large to be covered by silica.
10 A. Firouzi, F. Atef, A. G. Oertli, G. D. Stucky and B. F. Chmelka,
J. Am. Chem. Soc., 1997, 119, 3596.
11 A. Galarneau, F. Di Renzo, F. Fajula, L. Mollo, B. Fubini and
F. M. Ottaviani, J. Colloid Interface Sci., 1998, 201, 105.
12 F. M. Ottaviani, A. Galarneau, D. Desplantier-Giscard, F. Di
Renzo and F. Fajula, Microporous Mesoporous Mater., 2001
44–45, 1.
13 K. Edler, A. Goldar, A. V. Hughes, S. J. Roser and S. Mann,
Microporous Mesoporous Mater., 2001, 44–45, 661.
14 J. Patarin, B. Lebeau and R. Zana, Curr. Opin. Colloid Interface
Sci., 2002, 7, 107.
15 M. Tiemann, V. Goletto, R. Blum, F. Babonneau, H. Amenitsch
and M. Linden, Langmuir, 2002, 18(26), 10053.
16 K. Flodstrom, H. Wennerstrom and V. Alfredsson, Langmuir,
¨
2004, 20, 680–688.
¨
17 K. Flodstrom, C. V. Teixeira, H. Amenitsch, V. Alfredsson and
¨
M. Linden, Langmuir, 2004, 20, 10311–10316.
´
18 (a) G. Friedel, Colloid Chemistry, The Chemical Catalog Co., New
York, 1926, vol. 1, pp. 102–125; (b) F. B. Rosevear, J. Am. Oil
Chem. Soc., 1954, 31, 628–639.
19 (a) C. E. Fairhust, S. Fuller, J. Gray, M. C. Holmes and G. J. T.
Tiddy, Handbook of Liquid Crystals, 1980, 3(7), 341; (b) S. Hassan,
W. Rowe and G. J. T. Tiddy, in Handbookof Applied Surface and
Colloid Chemistry, ed. K. Holmberg, John Wiley & Sons, New
York, 2001, pp. 465–508.
20 C. C. Egger, G. J. T. Tiddy, J. L. Casci and M. W. Anderson, in
preparation.
21 S. O’Brien, R. J. Francis, A. Fogg, D. O’Hare, N. Okazaki and K.
Kuroda, Chem. Mater., 1999, 11, 1822.
22 F. Babonneau, C. Toutou and S. Gaveriaux, J. Sol-Gel Sci.
Technol., 1997, 8, 553.
23 V. Gualandris, F. Babonneau, M. T. Janicke and B. Chmelka,
J. Sol-Gel Sci. Technol., 1998, 12, 75.
24 F. Babonneau and J. Maquet, Polyhedron, 2000, 19, 315.
Conclusion
In conclusion, we establish the timescale of some of the key
processes in the formation of SBA-1 under highly acidic
conditions. Initial hydrolysis of the silica source occurs on a
25 B. Halle and H. Wennerstrom, J. Chem. Phys., 1981, 75(4), 1928.
¨
1854
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5

View Online
30 L. J. Lis, W. T. Lis, V. A. Parsegain and R. P. Rand, Biochemistry,
1981, 20, 1771–1777.
26 A. Abragam, Principles of nuclear magnetism, Oxford Science
Press, 1986, p. 239.
27 P.-O. Eriksson, G. Lindblom and G. Arvidson, J. Phys. Chem.,
1987, 91(4), 846–853.
31 M. Carvell, D. G. Hall, I. G. Lyle and G. J. T. Tiddy, Faraday
Discuss., Chem. Soc., 1986, 81, 223–237.
28 S. A. Buckingham, C. J. Garvey and G. G. Warr, J. Phys. Chem.,
1993, 97, 10236–10244.
32 J. N. Israelachvili and H. Wennerstrom, J. Phys. Chem., 1992, 96,
520–532.
¨
29 D. M. LeNeveu, R. P. Rand and V. A. Parsegian, Nature, 1976,
259, 601–603.
33 M. W. Anderson, C. C. Egger, G. J. T. Tiddy and J. L. Casci,
Studies in Surface Science and Catalysis, 2002, 142, 1149.
T h i s j o u r n a l i s & T h e O w n e r S o c i e t i e s 2 0 0 5
P h y s . C h e m . C h e m . P h y s . , 2 0 0 5 , 7 , 1 8 4 5 – 1 8 5 5
1855