Local order of amorphous zeolite precursors from 29Si{ 1H} CPMAS and 27Al and 23Na MQMAS NMR and evidence for the nature of medium-range order from neutron diffraction

Article abstract of DOI:10.1021/jp037887g

Amorphous aluminum silicates, isolated from hydrothermal reactions used to form crystalline zeolite A, have been studied using a number of techniques. NMR studies using ²⁹Si{¹H} cross-polarization (CP) MAS and ²⁷Al and ²³Na multiple-quantum (MQ) MAS methods provide information about the local atomic structure of the solids. Materials isolated at the early stages of reaction are silicon rich, and some of the aluminum is found in six-coordinate sites as in the amorphous alumina starting material. ²⁹Si{¹H} CPMAS NMR spectra show that a range of silicon environments are present in the amorphous solids. The line widths in the ²⁷Al MQMAS NMR spectra of the amorphous materials suggest that a range of Al-O distances and/or Si-O-Al angles is present, rather than the well-defined units present in zeolite A. Although the average ²³Na isotropic chemical shift suggests a coordination number of close to six for all sodium throughout the crystallization, the sodium in the amorphous precursor phase can be distinguished from that in the crystalline zeolite. The neutron diffraction study of the amorphous aluminosilicates reveals information about atomic order over longer length scales. The first sharp diffraction in the neutron diffraction data indicates, for the first time, that the medium-range order of the amorphous precursors changes prior to crystallization. The shift in the T-O peak (T is a tetrahedral atom: Si or Al) in the radial distribution function is consistent with the incorporation of aluminum into the silicate network at the early stages of reaction, and this is accompanied by the growth of an Na-O peak, as more charge-balancing cations are required. The broadness of the T-T nonbonded correlation for the amorphous solid, compared with that seen for crystalline zeolite A, indicates that there are no well-defined secondary building units present before the formation of zeolite crystals.

Full text of DOI:10.1021/jp037887g

8208
J. Phys. Chem. B 2004, 108, 8208-8217
29
27
23
Local Order of Amorphous Zeolite Precursors from Si{1H} CPMAS and Al and Na
MQMAS NMR and Evidence for the Nature of Medium-Range Order from Neutron
Diffraction
Huaixin Yang, Richard I. Walton,* Sasa Antonijevic, and Stephen Wimperis
Department of Chemistry, UniVersity of Exeter, Stocker Road, Exeter EX4 4QD, U.K.
Alex C. Hannon
ISIS Facility, Rutherford Appleton Laboratory, Chilton, Didcot, OX11 0QX, U.K.
ReceiVed: December 16, 2003; In Final Form: March 29, 2004
Amorphous aluminum silicates, isolated from hydrothermal reactions used to form crystalline zeolite A, have
been studied using a number of techniques. NMR studies using 29Si{1H} cross-polarization (CP) MAS and
27Al and 23Na multiple-quantum (MQ) MAS methods provide information about the local atomic structure of
the solids. Materials isolated at the early stages of reaction are silicon rich, and some of the aluminum is
found in six-coordinate sites as in the amorphous alumina starting material. 29Si{1H} CPMAS NMR spectra
show that a range of silicon environments are present in the amorphous solids. The line widths in the 27Al
MQMAS NMR spectra of the amorphous materials suggest that a range of Al-O distances and/or Si-O-Al
angles is present, rather than the well-defined units present in zeolite A. Although the average 23Na isotropic
chemical shift suggests a coordination number of close to six for all sodium throughout the crystallization,
the sodium in the amorphous precursor phase can be distinguished from that in the crystalline zeolite. The
neutron diffraction study of the amorphous aluminosilicates reveals information about atomic order over
longer length scales. The first sharp diffraction in the neutron diffraction data indicates, for the first time, that
the medium-range order of the amorphous precursors changes prior to crystallization. The shift in the T-O
peak (T is a tetrahedral atom: Si or Al) in the radial distribution function is consistent with the incorporation
of aluminum into the silicate network at the early stages of reaction, and this is accompanied by the growth
of an Na-O peak, as more charge-balancing cations are required. The broadness of the T--T nonbonded
correlation for the amorphous solid, compared with that seen for crystalline zeolite A, indicates that there are
no well-defined secondary building units present before the formation of zeolite crystals.
Introduction
of literature has amassed that allows the steps in the formation
of crystalline silicate zeolites to be described. There is still the
need, however, to measure experimental data to verify the
proposed pathways involved in the assembly of a zeolite
structure; particularly data that concern the changes in atomic
arrangement in the transformation of chemical reagents into a
crystalline network structure.
The crystallization mechanism of microporous zeolites has
attracted a great deal of attention for many years now, and has
been widely reviewed; see, for example, refs 1-3. The diverse
industrial uses of microporous solids in heterogeneous catalysis,
ion exchange and gas separation give the materials huge
commercial value and drive continued research into zeolite
synthesis. More fundamental scientific interest in the materials
also remains owing to the structural complexity of zeolites and
the ability to include a huge variety of chemical elements in
zeolitic frameworks to confer novel properties in the solid state.4
Zeolite synthesis is often undertaken by a trial-and-error
approach, involving variation of a large number of synthetic
parameters until a new, crystalline material is produced in a
pure state. This approach has been, and continues to be, highly
productive, and it is believed that there remain many yet-
undiscovered zeolite structures that will be synthesized in time.5,6
Other research has addressed the question of zeolite crystal-
lization mechanism with the hope of developing rational
syntheses of new materials. A recent review by Cundy and Cox
describes the development of the understanding of zeolite
formation,3 and it is apparent that since the 1960s a large body
Zeolites are most commonly synthesized using hydrothermal
conditions, whereby chemical reagents, which might be solid
or liquid, are heated in a sealed reaction container in an aqueous
solution above 100 °C. For example, in the synthesis of an
aluminum silicate zeolite, silica and alumina are mixed in the
required ratio with a concentrated sodium hydroxide solution,
and the mixture heated in a sealed reaction container for periods
of hours to days. In recent years a number of elegant in situ
X-ray and neutron scattering techniques have been developed
that probe the changing long-range order of the solid state during
hydrothermal zeolite crystallization under real reaction condi-
tions,7-10 and these have allowed new information concerning
the transformation of starting materials into product to be
obtained, such as accurate crystallization curves and the
observation of crystalline intermediate phases. Wide-angle
scattering techniques allow the development of long-range order
to be determined by monitoring the appearance and growth of
* Corresponding author. E-mail: r.i.walton@exeter.ac.uk.
10.1021/jp037887g CCC: $27.50 © 2004 American Chemical Society
Published on Web 05/21/2004

NMR Studies of Amorphous Zeolite Precursors
J. Phys. Chem. B, Vol. 108, No. 24, 2004 8209
characteristic Bragg peaks (including those of any transient,
crystalline phases9), whereas small-angle diffraction studies
probe the growth (and decay) of small particles, prior to the
appearance of crystallographic order.11-13 The information
lacking in these experiments is a probe of local, atomic structure
during the transformation of starting materials to product.
Although both solution and solid-state NMR experiments have
been developed to study certain nuclei (such as 27Al, 19F, and
29Si) under hydrothermal conditions,14,15 the complexity of the
experiment (requiring heated sample vessels under pressure)
makes the method not generally applicable for in situ studies,
in particular when the application of advanced, high-resolution
NMR techniques would be desirable. To achieve some insight
into this problem, we have studied the structures of disordered
solids quenched from reacting mixtures of silica and alumina.
We have studied the synthesis of zeolite A, because this is one
of the most well-characterized zeolites and therefore represents
a model system for investigation. After short reaction times it
is well-documented that amorphous aluminosilicate phases are
formed that on continued heating transform to the crystalline
zeolite.14,16-19 The local structure of these solids has been
investigated previously using NMR spectroscopy,16-18,20 but the
development of new, high-resolution NMR methods for qua-
drupolar nuclei such as 23Na and 27Al has allowed us to study
short-range atomic order in more detail than previously possible,
and to examine for the first time the local environment of sodium
using the 23Na MQMAS NMR method. We have also under-
taken what is to the best of our knowledge the first neutron
diffraction study of the amorphous aluminosilicates: this allows
us to determine the extent and the nature of medium-range order
in the solids.
ronmental SEM; this was used to provide further elemental
analysis by quantifying the relative intensities of characteristic
emissions of Si, Al, and Na with crystalline zeolite A as a
reference.
Solid-State NMR Studies. Spectra were recorded on a Bruker
Avance 400 spectrometer equipped with a widebore 9.4 T
magnet, operating at a Larmor frequency, ν0, of 105.8 MHz
3
5
for 23Na (I ) /2) and 104.3 MHz for 27Al (I ) /2), or on a
Bruker Avance 200 spectrometer equipped with a widebore 4.7
T magnet, operating at ν0 ) 39.7 MHz for 29Si (I ) 1/2). Samples
were packed inside 4 mm (9.4 T) or 7 mm (4.7 T) rotors, and
MAS rates were 10 and 5 kHz, respectively, were used. Cross-
polarized (CP) 29Si{1H} MAS NMR spectra were recorded using
a conventional cross-polarization sequence with 29Si and H
1
radio frequency field strengths, ν1Si ) ν1H, of ∼45 kHz and
with a contact time of 3 ms. Chemical shifts are reported in
ppm relative to an external standard of 0.15 M NaCl(aq) for
23Na, 1.0 M Al(NO3)3(aq) for 27Al, and TMS for 29Si. Two-
dimensional 23Na and 27Al triple-quantum MAS spectra were
recorded using the z-filtered sequence of Amoureux et al.21
Solid-state NMR spectra of quadrupolar nuclei, i.e., those with
spin quantum number I > 1/2, often have inherently low
resolution owing to the presence of significant anisotropic
quadrupolar broadening. This is a result of the interaction of
the nuclear quadrupole moment (eQ) with the electric field
gradient (eq). This quadrupolar coupling may be parametrized
by its magnitude
e2qQ
h
CQ )
(1)
and asymmetry, η, where 0 e η e 1. In addition to the
quadrupolar coupling, other mechanisms such as dipolar cou-
pling or chemical shift anisotropy further increase the observed
broadening, hindering the extraction of structural information.
The use of magic angle spinning (MAS),22 rotating the powdered
sample about an axis inclined at 54.7° to the external magnetic
field, removes the broadening to a first-order approximation,
thereby significantly increasing the spectral resolution.
Experimental Section
Sample Preparation and Characterization. Zeolite A and
its amorphous precursors were formed from a mixture of fumed
silica (Sigma) and an amorphous aluminum oxide, prepared by
thermal decomposition of aluminum nitrate hydrate (Aldrich)
at 500 °C for 17 h. The solids were shaken together to achieve
intimate mixing and stirred into a ∼1.95 M NaOH solution to
produce a paste. Batches of the mixture, having final composi-
tion Al2O3:2 SiO2:3.06 NaOH:128 H2O, were placed in polypro-
pylene bottles and heated at 100 °C for various periods of time
(see below). Solids were recovered by suction filtration, washed
with deionized water, and allowed to dry in air in a desiccator.
Samples are labeled LTAn, with n the reaction time (in hours)
used to prepare the sample. For the neutron diffraction study,
deuterated reagents were used: NaOD (Aldrich 40% in D2O),
D2O (Fluorochem Ltd), and Al2O3.nD2O (n ≈ 0.5) produced
by standing freshly prepared amorphous Al2O3 in a D2O
atmosphere. The solids were analyzed using powder X-ray
diffraction, recorded using a Bruker D8 diffractometer operating
with Cu KR radiation (average wavelength ) 1.5418 Å).
Powdered samples were pressed into aluminum sample holders
and data collected in Bragg-Brentano geometry from 5 to 70°
2θ with a step-size of 0.02° 2θ and a counting time of 1 s per
step. ICP analysis for aluminum and silicon was performed on
selected samples by Medac Ltd, U.K.. The water content of
the samples was determined using thermogravimetric analysis
performed using a Stanton Redcroft TG750 thermal analyzer,
with a ∼10 mg sample of each solid heated to 900 °C in air.
The densities of all samples were measured using a Quantach-
rome Micropycnometer with helium as the working fluid.
Energy-dispersive X-ray analysis (EDXA) was performed using
scanning electron microscopy with an Electroscan 2020 envi-
The quadrupolar interaction, however, may be very large
(often of the order of megahertz), and in this case, its second-
order effects result in anisotropic broadening that cannot be fully
removed by MAS alone. The multiple-quantum MAS NMR
technique (MQMAS),23-25 offers a method for obtaining truly
3
5
high-resolution spectra of half-integer spin (I ) /2, /2, etc.)
quadrupolar nuclei through the removal of all anisotropic
second-order quadrupolar broadening. The technique involves
the two-dimensional correlation of multiple- and single-quantum
coherences under MAS conditions. This results in a refocusing
of the anisotropic quadrupolar broadening, whereas the isotropic
chemical and quadrupolar shifts are retained. A two-dimensional
Fourier transform yields a spectrum containing ridge line shapes
that lie along a gradient equal to the MQMAS ratio, i.e., -7/9
for spin I ) 3/2 and +19/12 for spin I ) 5/2 for triple-quantum
MAS.23-25 A high-resolution spectrum may be obtained from
a projection orthogonal to this axis. It is possible to obtain values
of chemical shift and quadrupolar parameters from the two-
dimensional spectra: the δ1 and δ2 positions of the center-of-
gravity of a ridge line shape depend on both the isotropic
chemical shift, δiso, and the isotropic quadrupolar shift, δQ. For
3
spin I ) /2 triple-quantum MAS spectra,26
δ1 ) 3δiso + (6/5)δQ
δ2 ) δiso - (2/5)δQ
(2)
(3)

8210 J. Phys. Chem. B, Vol. 108, No. 24, 2004
Yang et al.
5
and for spin I ) /2 triple-quantum MAS spectra,
δ1 ) 3δiso - (4/5)δQ
δ2 ) δiso - (16/5)δQ
(4)
(5)
Thus both δiso and δQ may be determined. The composite
quadrupolar parameter,27 PQ ) CQ(1 + η2/3)1/2 may also be
extracted from δQ:
4I(2I - 1)ν
δ
0x
Q
PQ )
(6)
3000
The quadrupolar parameters, CQ and η, however, may not be
obtained individually by this method.
For amorphous or disordered solids, MAS NMR is a valuable
tool owing to its ability to probe local environment despite the
lack of long-range order. The range of local environments
encountered in such materials, however, leads to a distribution
in the isotropic chemical shift and/or quadrupolar parameters
of the nucleus studied.28 This further broadens the spectrum and
restricts the information available. The increased resolution
offered by MQMAS makes it a useful technique for the study
Figure 1. Laboratory powder X-ray diffraction patterns of aluminium
silicates prepared after (a) 12, (b) 7, (c) 5, (d) 2, and (e) 0.5 h. Weak
peaks at ∼38, 44, and 65° 2θ in some of the patterns are due to the
aluminium sample holder.
In this work we have reported the differential correlation
function, D(r):
of amorphous and disordered materials and applications of 11
B
Qmax
2
π
(I ) 3/2), 17O (I ) 5/2), 23Na (I ) 3/2), and 27Al (I ) 5/2) MQMAS
to such systems may be found in the literature.24,28-35 Ridge
line shapes are rarely observed, owing to the distribution of
shift parameters, but average values for δiso and PQ can still be
extracted by the center-of-gravity approach.
D(r) )
Q i(Q) M(Q) sin (rQ) dQ
(8)
∫
0
where r is the radial distance from an average origin atom. M(Q)
is a modification function, applied to reduce termination ripples
caused by the finite extent of data used. In this work we have
used the modification function due to Lorch:38
Neutron Diffraction Experiments. Neutron diffraction
experiments were performed using the GEM diffractometer at
ISIS, the UK spallation neutron source at the Rutherford
Appleton Laboratory. GEM is a time-of-flight diffractometer
that allows relatively rapid data collection owing to the large
area of the detector banks (10 m2, maximum azimuthal angle
∼45°).36 The time-of-flight method allows scattered neutrons
to be measured to high values of momentum transfer, permitting
high-resolution radial distribution functions to be obtained,
which was considered important in the current work because
we expected the materials to contain closely spaced atomic
correlations. Powdered samples were loaded into 8 mm diameter
vanadium cans and then dried gently in a vacuum oven at 70
°C to remove any residual protons due to surface water. The
cans were then sealed before mounting in the diffractometer.
The beam size was 40 × 15 mm, and data were collected from
each sample for a period of ∼5 h. A 6 mm vanadium rod was
used to measure the incident energy spectrum, to normalize the
time-of-flight data. The diffraction data were corrected using
standard procedures for container and instrument scattering,
attenuation, multiple scattering and inelasticity effects, using
the ATLAS suite of programs.37
sin (Q∆r)
M(Q) )
for Q < Qmax
for Q > Qmax
(9)
Q∆r
M(Q) ) 0
(10)
where ∆r ) π/Qmax. Qmax is the maximum useful momentum
transfer from the experimental data.
Results and Discussion
Laboratory Characterization and 29Si{1H} CPMAS NMR
Spectra. Figure 1 shows powder X-ray diffraction data mea-
sured from the samples isolated after varying reaction times.
The sole crystalline product after reaction times in excess of
12 h is zeolite A. All the peaks can be indexed with a refined
cubic unit cell of a ) 24.544(8) Å (space group Fm3hc); this
value agrees with previous structure determinations of the solid
(it should be noted that unit cell volume of hydrated zeolite A
is very similar to the dehydrated form).39-42 If reaction times
of shorter than 6 h are used, the solid isolated is amorphous.
At slightly longer times the material contains crystalline zeolite
A mixed with an amorphous phase. The Bragg peak widths due
to zeolite A decrease slightly after their first appearance,
suggesting that a period of crystal growth occurs at the
crystallites first formed. These powder X-ray diffraction data
are useful for determining, approximately at least, the extent of
reaction with time but give no information about atomic
arrangement in the solid state until the crystalline zeolite A
appears. Table 1 contains the Si:Al ratios determined by ICP
analysis for some of the samples, along with their water content
determined by TGA, and their experimentally measured densi-
ties. As expected, zeolite A has a strict 1:1 Si:Al ratio, but it
can be seen that at the early stages of reaction the amorphous
The quantity measured in the neutron diffraction experiment
is the differential cross section:
dσ
dΩ
) IS(Q) + i(Q)
(7)
where IS(Q) is the self-scattering, i(Q) is the distinct scattering,
and Q is the momentum transfer (the magnitude of the scattering
vector). The first, which can be calculated with knowledge of
the composition of the sample, is subtracted from the measured
differential cross section to yield the distinct scattering. Fourier
transformation then allows structural information to be obtained.

NMR Studies of Amorphous Zeolite Precursors
J. Phys. Chem. B, Vol. 108, No. 24, 2004 8211
Figure 2. 29Si{1H} CPMAS NMR spectra of (a) fumed silica, (b) LTA0.5, (c) LTA5, (d) LTA7, (e) LTA8, and (f) LTA12. The spectra are the result
of averaging 2048 transients with a recycle interval of 5 s. The MAS rate was 5 kHz. The contact pulse duration was 3 ms.
TABLE 1: Reaction Times Used To Prepare Aluminum
Silicates Studied and the Results of Analyses Performeda
TABLE 2: Average Chemical Shift Values of Features
Observed in the 29Si{1H} CPMAS NMR Spectra of the
Materials Studied, Extracted by Fitting of Lorentzian
Functions (See Figure 3)
preparation
time/h
water content after
air-drying (%)
density/
g cm-3
sample
Si/Alb
sample
δ[29Si] (ppm)
LTA0.5
LTA2
LTA5
LTA7
LTA12
0.5
2
5
7
12
2.7
1.8
1.6
1.4
1.0
15.5
17.4
17.8
17.3
22.3
1.56
2.11
2.11
1.99
2.01
fumed silica
LTA0
-110.7, -100.7, -91.0
-109.1, -99.5, -88.9
-98.5, -85.0
a
LTA0.5
LTA2
LTA5
LTA7
-94.9, -85.5
-93.2, -85.3
a Time of reaction at 100 °C from an initial reaction mixture of
composition Al2O3:2 SiO2:3.06 NaOH:128 H2O. Si/Al ratios are those
determined by ICP analysis. b Estimated error on the Si:Al ratio is (0.1,
based on repeated analyses.
-96.4, -89.2, -83.6
-94.5, -89.4, -84.4
-89.6
b
LTA8
LTA12 (zeolite A)
a Isolated after stirring the reagents at room temperature for 30 min.
b Analysis results are similar to sample LTA7.
solids formed are silicon rich. This is consistent with some
previous studies of related materials; for example, Engelhardt
et al. found that amorphous material with an Si:Al ratio of 2.4
formed at the early stages of the crystallization of zeolite A
from sodium silicate and sodium aluminate solutions.16,18 It
should be noted that a direct comparison of the evolution of
Si:Al ratio with reaction time between various previous studies
is not valid, because the zeolite crystallization rate depends on
many experimental factors, including temperature and specific
choice of starting materials (including the particle size of solids
used). We can make the general observation that at the early
stages of reaction only a small amount of aluminum is
incorporated into the silicate network with the remainder present
as a soluble form. As the reaction proceeds, more aluminum is
incorporated into the amorphous solid. The EDXA shows that
the Al:Si ratio is similar in all particles analyzed in each sample,
and also that the sodium content of the amorphous samples
increases as more aluminum is incorporated: this is expected
because the increasing negative charge of the aluminosilicate
network must be balanced by more sodium cations.
contains the average chemical shift values of the main features
observed in the 29Si{1H} CPMAS spectra for all of the materials
studied. These were extracted by fitting with Lorentzian
functions, as shown in Figure 2. The use of Gaussian functions
would yield essentially identical results. 29Si MAS NMR has
been widely used to study zeolites and other aluminum
silicates,20,43 and the results of prior studies allow us to assign
the observed features. For the crystalline zeolite A, a single
resonance is observed, centered at δ ) -89.6 ppm, and this is
typical of the material: for example, Engelhardt et al. report
the resonance to be centered at δ ) -89.4 ppm18 and
Kulshreshtha et al. reported the same chemical shift value.44
These values are typical of Q4(4Al) silicons.43 The SiO2 starting
material contains three broad features centered at δ ) -91.0,
-100.7, and -110.7 ppm. These spectral features are well-
established for samples of fumed silica:45,46 the last two features
correspond to Q3 and Q4 silicon environments ([Si-O]3Si[OH]
and Si[O-Si]4) whereas the peak at -91 ppm is due to Q2
silicons, i.e., those that have two surface hydroxyl groups
attached ([Si-O]2Si[OH]2). For the amorphous material isolated
at room temperature (after stirring the chemical reagents together
for 30 min) the Si{1H} CPMAS spectra (not shown) show peaks
centered at δ ) -88.9, -99.5, and -109.1 ppm, consistent with
a network structure consisting largely of corner-shared silicate
units with some surface silanol groups, similar to that seen in
Figure 2 shows 29Si{1H} CPMAS spectra measured from
some of the amorphous materials, the fumed silica starting
material and the crystalline zeolite A. The non-cross-polarized
spectrum was also measured for sample LTA8 to confirm the
agreement in line shapes between the two methods. The CP
method was then selected for its greater sensitivity. Table 2

8212 J. Phys. Chem. B, Vol. 108, No. 24, 2004
Yang et al.
the SiO2 starting material. This indicates that only a trace of
aluminum is incorporated into the solid after this short time
and prior to heating the reagents.
After 0.5 h of reaction time there are at least two broad
features: one at -98.5 ppm and the other centered at -85.0
ppm. The former is consistent with the presence of silicon-rich
regions of sample, containing Q2-4(0Al) type silicons, as found
in the starting material. The feature centered at -85 ppm may
be assigned as arising from a broad range of Qn(mAl) silicons:
the chemical shift range for Q2(mAl), Q3(mAl), and Q4(mAl)
type silicons spans the range -75 to -100 ppm. Shi et al.
studied a precursor to zeolite A and assigned peaks in the range
-75 to -95 ppm to the same species,14 Interestingly, Engelhardt
et al. observed a single broad peak centered at -85.0 ppm for
samples with an Si:Al ratio of 1:1, whereas for silicon-rich
samples (Si:Al 2:1), the resonance is centered at -92.5 ppm.16,18
Thus we propose that for our silicon-rich material prepared using
the shortest of reaction times, a range of tetrahedral silicon sites
are present: some as Q2-4(0Al) sites, where the network
structure is siliceous, and some as Q2-4(mAl) where aluminum
incorporation has taken place. The sample isolated after 5 h of
reaction time is rather similar.
Figure 3. 27Al MAS NMR spectra of (a) LTA0, (b) LTA0.5, (c) LTA3,
(d) LTA7, (e) LTA8, and LTA12. The spectra are the result of averaging
320 transients with a recycle interval of 0.5 s. The MAS rate was 10
kHz.
The samples isolated after 7 and 8 h of reaction time show
the first sign of the sharp resonance at -89.6 ppm that is due
to crystalline zeolite A [Q4(4Al)], and weaker features at around
-84 ppm and around -95 ppm still remain that are due to the
amorphous component. This is complete agreement with the
powder XRD data that show a mixture of zeolite A and
amorphous material. Interestingly, the feature at close to -84
ppm, at higher ppm than the crystalline resonance, and which
we attribute to Q3(3Al) type silicons, indicates that a significant
number of Si-OH or Si-O- moieties remain in the structure.
Although 29Si MAS NMR has been widely applied to the
study of zeolites and amorphous silicates, and our new 29Si-
{1H} CPMAS data are similar to those previously reported for
amorphous aluminosilicates, we can use the data to pinpoint
changes in the local structure with increasing preparation time
of the solids. In addition, our data are consistent with the
observations from powder X-ray diffraction and provide a
starting point for the interpretation of the remaining NMR data,
and neutron diffraction data.
27Al and 23Na MQMAS NMR Spectra. Figures 3 and 4
show the 27Al MAS and triple-quantum MAS NMR spectra
recorded from the samples studied. The conventional MAS
spectrum of the amorphous alumina starting material (not
shown) indicates the presence of at least three distinct sites for
aluminum; these have chemical shifts in the ranges expected
for four-, five- and six-coordinated aluminum, which is con-
sistent with the well-known observation that in samples of
amorphous alumina a variety of coordination geometries are
seen.47 After the alumina and silica starting materials are stirred
in sodium hydroxide solution at room temperature and the solid
formed is recovered by filtration, the NMR data show evidence
that much of the aluminum remains in octahedrally coordinated
sites. The conventional MAS spectrum of this material indicates
the presence of at least two distinct sites for aluminum in the
solids, and the two-dimensional triple-quantum MAS NMR
spectrum confirms the presence of two signals, albeit broadened
as expected for an amorphous solid. One broad signal has an
average isotropic chemical shift of 16.5 ppm, whereas the other
is a complex signal made up of two components with average
chemical shift values of 74.0 and 62.6 ppm. The first signal
lies in the range expected for octahedrally coordinated alumi-
num, and the second, complex signal indicates tetrahedrally
coordinated aluminum.47
The aluminum silicate sample isolated after 0.5 h of reaction
time retains only a trace of octahedrally coordinated aluminum;
one explanation for this is the presence of some unreacted
amorphous alumina. The strongest signal, however, is that
characteristic of tetrahedrally coordinated aluminum. All further
samples contain solely tetrahedrally coordinated aluminum, but
as progressively longer periods of heating are used, the signal
becomes less broadened. For the sample isolated after 7 h of
reaction time, the two-dimensional triple-quantum MAS NMR
spectrum allows two distinct signals to be resolved: the
conventional spectrum provides little evidence for the presence
of these two signals, whereas in the two-dimensional spectrum
a distinct “lobe” can be seen on the broad feature. One of these
signals is broad and corresponds approximately to the signal of
the amorphous material isolated at the beginning of the reaction,
whereas the other is considerably sharper. Isotropic chemical
shift values extracted from the two-dimensional spectra are given
in Table 3. By examining the 27Al triple-quantum MAS NMR
spectrum of crystalline zeolite A and the PQ values, we can see
that the sharper signal in the sample isolated after 7 h of reaction
time is due to the presence of the crystalline zeolite in the
product mixture. The broadness of the signal of the amorphous
materials indicates a range of Al-O bond distances or
Al-O-Si angles are present, although the value of isotropic
chemical shift is always in the region for tetrahedrally coordi-
nated aluminum.
Figures 5 and 6 show the 23Na MAS and triple-quantum MAS
NMR spectra, respectively, of the amorphous materials. At a
first glance the spectra show a single feature that is broadened
to varying degrees, but a closer inspection reveals that the
crystalline zeolite A exhibits a signal with an isotropic chemical
shift value of 1.9 ppm, whereas for the first materials isolated,
after stirring the reagents at room temperature, the signal is
centered at 0.4 ppm. The fact that we only observe one
resonance in the 23Na NMR data, when there are three
crystallographically unique sodium ions, is consistent with
previous 23Na studies of the zeolite: for example, Veeman and

NMR Studies of Amorphous Zeolite Precursors
J. Phys. Chem. B, Vol. 108, No. 24, 2004 8213
Figure 5. 23Na MAS NMR spectra of (a) LTA0, (b) LTA0.5, (c) LTA3,
(d) LTA7, (e) LTA8, and (f) LTA12. The spectra are the result of
averaging 320 transients with a recycle interval of 0.5 s. The MAS
rate was 10 kHz.
7 and 8 h of reaction time, the spectra contain signals
characteristic of both the crystalline zeolite A and its amorphous
precursor. To extract isotropic chemical shift values from spectra
containing a complex signal made of more than one component,
we simply estimated, by inspection, the position of each center-
of-gravity; see Table 3. The validity of this approach was tested
by extracting δiso and PQ values from spectra that contained a
single, symmetrical signal and comparing these with those
extracted by a more rigorous approach based on numerical
determination of the true center-of-gravity; excellent agreement
was seen in these cases. The observation of two signals for
samples LTA7 and LTA8 is consistent with the powder X-ray
diffraction results, and the 27Al triple-quantum MAS NMR
spectra, but it is noteworthy that sodium in the crystalline and
amorphous phases can be distinguished by NMR, even if the
chemical environment of sodium in each phase is very similar.
We have previously investigated whether any correlation exists
between the 23Na isotropic chemical shift and the local
coordination environment of sodium in oxides by studying a
range of solids that contain sodium in different environments.35
We concluded that there is a general trend of increasing chemical
shift with decreasing coordination number and average Na-O
interatomic distance, which allows information about the
coordination number and interatomic distance to be inferred,
although the chemical shift is not an unambiguous parameter
as it is for other nuclides, for example, 27Al. The values of
isotropic chemical shifts we see here for both the crystalline
zeolite and its amorphous precursors are consistent with the
presence of sodium bound to six oxygen atoms at 2.45-2.55
Å; sodium with fewer oxygen near-neighbors at shorter distances
typically exhibits higher values of isotropic chemical shifts (up
to 27 ppm).35
Figure 4. Two-dimensional 27Al triple-quantum MAS NMR spectra
of (a) LTA0, (b) LTA0.5, (c) LTA3, (d) LTA7, (e) LTA8, and (f) LTA12,
recorded using a z-filtered pulse sequence. In (a) 576, (b)-(e) 144,
and (f) 72 transients were acquired for each of (a)-(d) 128, (e) 256,
and (f) 1024 t1 increments of 8.33 µs. In each case, the recycle interval
was 0.5 s and the MAS rate was 10 kHz. Contour levels are drawn at
4, 10, 20, 30, 50, 70, and 90% of the maximum value. Spinning
sidebands are indicated by *.
TABLE 3: Average Isotropic Chemical Shifts (δiso) for 23Na
and 27Al, Relative to 0.15 M NaCl(aq) and to 1.0 M
Al(NO3)3(aq), Respectively, and Average Quadrupolar
Products (PQ) for Each Sample Studied, Extracted from
Centers-of-Gravity of Two-Dimensional Triple-Quantum
MAS NMR Line Shapesa
27
23
δ
[ Al]
PQ[27Al]
(MHz)
δ
[ Na]
iso
PQ[23Na]
(MHz)
iso
sample
LTA0
(ppm)
(ppm)
74.0
62.6
16.5
64.3
63.0
63.0b
4.4
2.7
4.6
3.0
2.7
2.5b
0.4
1.6
LTA0.5
LTA3
LTA7
-0.1
-0.7
2.0
1.9
2.0
1.3
2.0
1.3
2.0
1.3
-0.2
LTA8
63.0b
2.5b
1.5
-0.6b
1.9
LTA12
61.2
1.6
a Typical measurement errors in the values quoted are (1.0 ppm
(δiso) and (0.3 MHz (PQ). b Multiple sites were impossible to separate
because of their close proximity (see Figure 4).
The local environment of sodium in dehydrated, crystalline
zeolites is rather different than typically seen in many condensed
solids, and often low coordination numbers and irregular
geometries are found.49 On hydration, however, the coordination
number of sodium is increased by the presence of directly
co-workers reported the same result in their 23Na nutation NMR
study of zeolite A hydration and suggested that the mobility of
the sodium ions or water molecules, averages the three crys-
tallographically unique sites.48 For the materials isolated after

8214 J. Phys. Chem. B, Vol. 108, No. 24, 2004
Yang et al.
Figure 7. Correction of the raw neutron scattering data (GEM 90°
detector bank) from sample LTA5 for the incoherent scattering of
residual protons for data from one GEM bank. Note that for this detector
bank, data in the range 8-30 Å-1 were used to produce the complete
i(Q).
distances in hydrated zeolite A observed by X-ray diffraction
are a consequence of framework vibration, and the “real” Na-O
distances are lower than the values taken between centroids of
electron density.49
Our new 23Na NMR data are thus consistent with the presence
of 5- or 6-coordinate sodium ions in the hydrated, crystalline
zeolite A, and also in the amorphous aluminum silicates, because
the value of δiso is always between +2.1 and -1.6 ppm. Our
results are in agreement with the widely accepted crystallization
model for the formation of many zeolites, in which hydrated
sodium ions, Na(H2O)n+ (n ≈ 6 for the first coordination shell)
act as templates for the building up of the aluminosilicate
network.3 Sodium remains coordinated to around 6 oxygen
atoms throughout the crystallization process.
Neutron Diffraction Data. The samples for the neutron
diffraction experiment were prepared using deuterated reagents
to avoid the large incoherent scattering background due to the
presence of protons. Although our samples were dried carefully,
they apparently exchange water with the atmosphere readily
because all samples exhibit the characteristic sloping background
of the incoherent scattering due to the presence of some protons.
For most materials it is possible to perform a successful
calculation of the Q-dependence of the self-scattering, Is(Q),
by use of a Placzek correction50 that takes into account the
effects of inelasticity within an approximation.51,52 However,
for samples that contain very light nuclei, especially hydrogen
and deuterium, the effects of inelasticity are too severe and the
approximation breaks down. Instead, the self-scattering can be
subtracted from the differential cross section by an empirical
approach whereby a smooth, nonoscillatory function is fitted
to the data.53,54 In this work we have used a cubic spline with
a variable knot spacing to perform this correction. Figure 7
shows such a correction for the data from one of the detector
banks on GEM (note that each detector bank does not measure
the full extent of scattering data, and overlapping regions of
data from each bank are used, so that a nonoscillatory
background is always fitted).
Figure 6. Two-dimensional 23Na triple-quantum MAS NMR spectra
of (a) LTA0, (b) LTA0.5, (c) LTA3, (d) LTA7, (e) LTA8, and (f) LTA12,
recorded using a z-filtered pulse sequence. In (a)-(d), (f) 480 and (e)
384 transients were acquired for each of (a)-(d) 64 and (e), (f) 128 t1
increments of 16.67 µs. In each case, the recycle interval was 0.5 s
and the MAS rate was 10 kHz. Contour levels are drawn at 10, 20, 30,
50, 70, and 90% of the maximum value.
coordinating water molecules. Water in zeolite cages is often
disordered, and the electron density due to its presence is usually
modeled as a number of sites having fractional occupancy; this
makes the identification of the precise local environment of
sodium ions difficult. A number of structure determinations of
zeolite A have been undertaken over the years,39-41,49 but the
most useful for us to consider are those by Ikeda et al. who
studied both dehydrated and hydrated zeolite A using a
combination of powder diffraction and maximum-entropy
methods.42 For dehydrated zeolite A, three crystallographically
unique sodium ions are coordinated to either three or four
framework oxygen atoms (at distances between 1.9 and 2.9 Å)
and reside in positions close to the center of the four-, six-, and
eight-membered rings. On hydration, the sodium ions migrate
closer to the center of the R and â cages, i.e., away from the
windows, and water molecules lie very close to the cations.42
The coordination of each of the sodium ions then increases to,
on average, between 5 and 6 (this includes both framework
oxygens and water molecules), and these oxygen atoms lie at
distances between 1.9 and 2.9 Å. The upper value is perhaps
too large to be considered a chemical bond, but these oxygens
are nevertheless rather close to the sodium ions and must be
considered as part of the chemical environment of the ions.
Interestingly, Pluth and Smith suggested that the long Na-O
Figure 8 shows the distinct scattering, i(Q), for each of the
amorphous samples studied, along with that of crystalline zeolite
A, plotted to 20 Å-1 to allow comparison. The most striking
feature of the functions is their overall similarity (ignoring the
sharp Bragg peaks of zeolite A), suggesting that the network

NMR Studies of Amorphous Zeolite Precursors
J. Phys. Chem. B, Vol. 108, No. 24, 2004 8215
Figure 8. Distinct scattering, i(Q), for samples (a) LTA12, (b) LTA7,
(c) LTA5, (d) LTA2, and (e) LTA0.5. The dotted line indicates the
variation of the position of the first sharp diffraction peak between
samples.
Figure 10. Calculated D(r) for crystalline, hydrated zeolite A using
the structural model of Ikeda et al. and comparison with the experi-
mental function (see Table 4 for assignment of peaks).
each of the solids has a similar atomic arrangement, because
peaks in the radial distribution occur at similar positions and
have similar relative intensities for each material. A detailed
examination of the data, however, reveals significant changes
taking place with increasing sample preparation time: for
example, it can be seen that the first peak in D(r) (at around
1.6 Å) shifts with increasing sample preparation time, the
relative intensity of a peak at ∼3.2 Å changes with time, and
in the region 3.5-5 Å significant changes are observed. To
assign these peaks to specific atomic pairs and understand their
variation from sample to sample, we consider previous crystal
structures of zeolites and relate this to our experimental
correlation function for the crystalline material.
The shortest interatomic separation expected in zeolite A will
be that of the O-H pair at ∼1 Å due to the presence of occluded
water molecules. It should be noted that the relative height of
the O-H/D peak in the D(r) function will be affected severely
by the relative amount of protons and deuterons in the sample:
because H and D neutron scattering lengths have opposite signs,
-3.74 and +6.67 fm, respectively,57 the O-D and O-H
contributions will tend to cancel each other. (Note that all
isotopes of the other elements in the samples have positive
scattering lengths.) Despite the many structural investigations
of zeolite A, there is no full description of its structure that
includes all proton positions for the fully hydrated material, but
we can examine the crystal structures of other hydrated zeolites
to provide reference distances for O-H/D bonds. The peak we
see in D(r) of zeolite A lies at 1.14 Å, and this compares well
with O-D distances determined by neutron diffraction in the
zeolites sodalite58 (1.06 Å), hydroxosodalite59 (0.90-1.15 Å),
zeolite Y60 (0.98 and 1.14 Å), and laumonite61 (0.95-1.21 Å).
It should be noted that the first three zeolites here are sodium
aluminum silicates whose structure is made up from the same
sodalite cage that is seen in zeolite A, so these provide very
good systems for direct comparison with zeolite A.
Figure 10 shows the calculated D(r) from the crystal structure
of hydrated zeolite A determined by Ikeda et al. from powder
X-ray diffraction data.42 The D(r) calculation was performed
using the program XTAL62 using a constant peak broadening
of 0.04 Å. The structure solution of Ikeda et al. for Na95Si97-
Al95O384(H2O)248 is the most complete available for any hydrated
sample of zeolite A, although their structural model does not
include any proton positions and only the positions of oxygen
atoms of water molecules (as well as sodium cations and
framework atoms). Our material has a similar composition:
Figure 9. The differential correlation function, D(r), for samples (a)
LTA12, (b) LTA7, (c) LTA5, (d) LTA2, and (e) LTA0.5
.
structure of the amorphous materials resembles to some extent
that of the final crystalline product. A closer inspection of the
data, however, does reveal some subtle differences, even
between the amorphous samples prepared using the shortest
reaction times. The “first sharp diffraction peak”, indicated in
Figure 8, is observed to shift depending on the preparation time
of the samples, and this is particularly noticeable between the
amorphous materials LTA2 and LTA5. This diffraction feature
is widely supposed to arise from the presence of medium-range
order (i.e., beyond the first few coordination shells of each atom,
extending between 5 and 20 Å), although it is extremely difficult
to extract quantitative information regarding order over such
length scales.55,56 Our data thus provide the first indication that
although the local atomic structure probed by NMR is rather
similar for the amorphous zeolite precursors, particularly in
terms of coordination number, their structure does change prior
to zeolite crystallization.
To extract quantitative information from the neutron diffrac-
tion regarding the atomic arrangement in the amorphous
materials, radial distribution functions were produced by Fourier
transformation of the interference function according to eq 8
and using Qmax ) 30 Å-1. Figure 9 shows the differential
correlation function, D(r), for each of the materials studied. Note
that any peaks below 1 Å are not physically meaningful and
are probably due to some residual background that has not been
corrected fully. Once again, an initial comparison indicates that
between Na97Si97Al97O384(H2O)221 and Na97Si97Al97O384(D2O)199
,

8216 J. Phys. Chem. B, Vol. 108, No. 24, 2004
Yang et al.
units.55 This peak overlaps with peaks due to O--O pairs between
water molecules and framework oxygens (∼2.8-3.0 Å). The
similar positions of these peaks in all samples shows the
similarity of the short-range order in all samples, but it should
be noted that this peak broadens to the higher radial distance
side as crystallization proceeds. This can be simply interpreted
as arising from a greater amount of Na(H2O)6+ within the solid,
consistent with the decreasing Si:Al ratio, and the water analysis
results (see Table 1); thus a larger number of O--O pairs between
water and framework oxygens are present.
TABLE 4: Selected Radial Distances in Hydrated Zeolite A
(from the Crystal Structure Determination of Ikeda et al.)42
That Have Been Used To Interpret the Correlation
Functions of the Amorphous Solidsa
radial distance
(Å)
atomic correlation
1.59
1.73
Si-O
Al-O
Na-O
2.10-2.90
2.20-2.90
2.60-2.75
2.80-3.00
3.20-3.25
3.20-3.80
O--O (hydrogen-bonded water molecules)
O--O (within tetrahedral units)
O--O (water-framework atoms)
T--T
A very important atomic correlation to consider is that due
to neighboring tetrahedral centers (Si--Al in a 1:1 aluminum
silicate), because this gives information about the average
T-O-T angle, and thus the presence of different secondary
building units. In zeolite A, which is constructed solely of four-
and six-membered rings, the shortest T--T distances are found
at ∼3.2 Å, and this peak is indeed observed in the radial
distribution function. (It should be noted that Na--T distances
are seen at similar distances in zeolite A, so these will also
contribute to the radial distribution in this region.) It is
noteworthy that for the amorphous materials the T--T peak at
∼3.2 Å is of much lower intensity than the O--O peak at 2.75
Å in D(r). This suggests that the amorphous material is made
of a variety of structural units; rings of a variety of sizes, and
perhaps aluminosilicate chains, which possess a variety of
T-O-T angles and hence T--T distances. This is not true for
all zeolite crystallizations: for example, Twu et al. in their
vibrational spectroscopy study of the formation of mordenite
observed four- and five-membered rings in the amorphous
precursor gel.63 For zeolite A, there is no evidence for a
preponderance of units with specific T--T correlations until the
long-range order, characteristic of the zeolite, is present.
Na--T
a T represents a tetrahedral center (aluminum or silicon). A non-
bonded atomic correlation is indicated by --.
based on all chemical analysis results. The agreement between
the experimental D(r) for our sample of zeolite and the
calculated curve is most informative. It must be noted that the
relative intensities of peaks in the measured D(r) is affected by
the finite Q range of the data,62 but the positions of major peaks
show agreement. The O-D/H peak at 1.14 Å that we observe
experimentally is clearly not predicted by the model of Ikeda
et al. (which contains no H/D positions), but all other significant
peaks in the experimental D(r) are accounted for by this model.
The contribution of atomic correlations that include H/D must
be responsible to some extent for discrepancies between the
model, where they are not included, and the measured data
(particularly given the large amount of H2O/D2O in the samples)
and it is also likely that there is a large degree of disorder for
the proton positions in the samples. Nevertheless, the compari-
son of the functions for crystalline zeolite A is valuable because
it allows us to assign the peaks seen in D(r) plots, and then to
infer information about order in the amorphous solids. Table 4
lists the peak assignments made by direct comparison of the
calculated and experimental functions.
The second distinct peak in the D(r) functions can be assigned
to Si-O and Al-O pairs (T-O, in general): these are expected
to occur at 1.59 and 1.73 Å, respectively, in hydrated zeolite
A.42 The position of this first peak in D(r) shifts as increasing
reaction time is used to prepare the samples: from 1.63 Å for
LTA0.5 to 1.65 Å for LTA12. This is entirely consistent with
the elemental analysis showing the materials prepared after the
shortest period of time to be silicon rich; thus the average T-O
distance is initially somewhat shorter. For a material with an
Si:Al ratio of 3:1 (close to that of the amorphous sample
prepared using the shortest reaction time) the average T-O
distance is expected to be 1.625 Å, and for a material with a
Si:Al ratio of 1:1, 1.66 Å. The next sharp peak in the D(r)
functions lies at ∼2.4 Å and by using the above NMR results
and the zeolite A crystal structure, this can be assigned as arising
from the closest Na-O contacts. This peak becomes more
intense (relative to the T-O peak) as increasing reaction time
is used, which is entirely consistent with the fact that as more
aluminum is incorporated into the solid, more sodium must be
present to balance the net negative charge of the network. This
result indicates that the amorphous network is made up of
{AlO4} and {SiO4} units. It should also be noted that inter-
molecular O--O distances between hydrogen-bonded water
molecules are expected to occur at 2.4 Å in zeolite A42 and so
must also contribute toward the intensity of this peak.
As we examine the D(r) to higher radial distance, interpreta-
tion becomes more difficult, because peaks become increasingly
overlapped with their neighbors. It is important to note, however,
that the region 3-5 Å in D(r) begins to resemble that of the
crystalline zeolite for the amorphous materials prepared with
the longer reaction times (LTA5, for example), whereas for the
silicon-rich solids isolated at the early stages of reaction, the
medium-range order is rather different. This appears to show
that considerable structural rearrangement of the amorphous
phase takes place even prior to the formation of zeolite crystals.
Conclusions
The amorphous materials isolated from reactions used to
prepare zeolite A have local atomic order that resembles that
of the crystalline zeolite. The use of sensitive solid-state NMR
methods has provided new information about the solids. For
example, by studying the 23Na MQMAS spectra, we can
ascertain that hydrated sodium ions are present at all stages of
crystallization, and that despite having a similar chemical shift
throughout the crystallization process, sodium in crystalline and
amorphous phases can be distinguished in mixtures of the two.
To the best of our knowledge our neutron diffraction data are
the first measured from amorphous zeolite precursors. Despite
the complexity of the system (five chemical elements and ready
exchange with atmospheric water), we are able to detect the
presence of medium-range order in the amorphous materials
and relate this to the structure of the crystalline zeolite finally
produced. Our work has highlighted the absence of complete
structural models for hydrated zeolites, and this must be the
subject of future work: it is possible that total diffraction
measurements might have a role to play in solving this problem.
At the present time, even without knowledge of H/D positions,
The complex, intense peak in D(r) at ∼2.75 Å contains a
large component due to O--O pairs within {TO4} tetrahedral
units (T ) Al or Si), expected at 2.6 Å; and this is well-known
for materials constructed from tetrahedral primary building

NMR Studies of Amorphous Zeolite Precursors
J. Phys. Chem. B, Vol. 108, No. 24, 2004 8217
(15) Gerardin, C.; Haouas, M.; Lorentz, F.; Taulelle, F. Magn. Reson.
Chem. 2000, 38, 429.
(16) Engelhardt, G.; Fahlke, B.; Magi, M.; Lippmaa, E. Zeolites 1983,
3, 292.
we can obtain information about the structure of the amorphous
aluminosilicate network and its interaction with the charge-
balancing hydrated sodium cations.
We have studied solid phases isolated from zeolite crystal-
lizations, but it is also important to bear in mind the role of the
solution in crystallization. It has been established that zeolite
crystallization may occur via homogeneous routes (for example,
from clear solutions by dissolution of starting materials and/or
intermediate phases) or by heterogeneous processes (directly
from the amorphous precursor gel by interaction with the
solution). Thus methods that probe the structure of the solution
will always be important in providing complementary informa-
tion about zeolite crystallization.19,64 In situ scattering studies
of the heterogeneous mixtures present during crystallization
would obviously be extremely desirable, but the data from a
mixture of solution and solid, both of changing composition,
would be extremely difficult to interpret
(17) Mu¨ller, D.; Starke, P.; Jank, M.; Wendlandt, K.-P.; Bremer, H.;
Scheler, G. Z. Anorg. Allg. Chem. 1984, 517, 167.
(18) Engelhardt, G.; Fahlke, B.; Ma¨gi, M.; Lippmaa, E. Zeolites 1985,
5, 49.
(19) Serrano, D. P.; Grieken, R. V. J. Mater. Chem. 2001, 11, 2391.
(20) Bell, A. T. Colloids Surf. A: Phys. Eng. Asp. 1999, 158, 221.
(21) Amoureux, J. P.; Fernandez, C.; Steuernagel, S. J. Magn. Reson.
A 1996, 123, 116.
(22) Andrew, E. R.; Bradbury, A.; Eades, R. G. Nature 1958, 182, 1659.
(23) Frydman, L.; Harwood, J. S. J. Am. Chem. Soc. 1995, 117, 5367.
(24) Fernandez, C.; Amoureux, J. P. Solid State Nucl. Magn. Reson.
1996, 5, 315.
(25) Brown, S. P.; Wimperis, S. J. Magn. Reson. 1997, 128, 42.
(26) Bodart, P. R. J. Magn. Reson. 1998, 133, 207.
(27) Mueller, K. T.; Baltisberger, J. H.; Wooten, E. W.; Pines, A. J.
Phys. Chem. 1992, 96, 7001.
(28) Kentgens, A. P. M. Geoderma 1997, 80, 271.
(29) Hunger, M.; Sarv, P.; Samoson, A. Solid State Nucl. Magn. Reson.
1997, 9, 115.
Further interpretation of the neutron scattering data is
extremely complex, and even for binary systems, such as SiO2,
only with advanced computer modeling methods and models
containing hundreds of atoms can the subtle features in the radial
distribution functions be accounted for. Nevertheless, for our
system (which contains five chemical elements) the assignment
of peaks by direct comparison with an appropriate crystalline
material allows us to interpret the data from the amorphous
solids, providing valuable, new information about the presence
and nature of medium-range order in these materials. A
noteworthy result is that there is no evidence for the presence
of distinct secondary building units in the amorphous materials
prior to the formation of the zeolite. This result is consistent
with some previous solution NMR studies of aluminosilicate
solutions taken from zeolite crystallization that show the
presence of simple monomeric or dimeric units rather than the
building units found in the crystalline solid.65,66 The amorphous
solids we have studied may be thought of as reservoirs for the
reactive solution species that allow zeolite crystal growth but
are not actually built up of only the secondary building units
found in the zeolite itself.
(30) Stebbins, J. F.; Oglesby, J. V.; Xu, Z. Am. Mineral. 1997, 82, 1116.
(31) Hwang, S. J.; Fernandez, C.; Amoureux, J. P.; Han, J. W.; Cho, J.;
Martin, S. W.; Pruski, M. J. Am. Chem. Soc. 1998, 120, 7337.
(32) Angeli, F.; Charpentier, T.; Faucon, P.; Petit, J. C. J. Phys. Chem.
B 1999, 103, 10356.
(33) McManus, J.; Ashbrook, S. E.; MacKenzie, K. J. D.; Wimperis, S.
J. Non-Cryst. Solids 2001, 282, 278.
(34) Ashbrook, S. E.; McManus, J.; MacKenzie, K. J. D.; Wimperis, S.
J. Phys. Chem. B 2000, 104, 6408.
(35) Antonijevic, S.; Ashbrook, S. E.; Walton, R. I.; Wimperis, S. J.
Mater. Chem. 2002, 12, 1469.
(36) Hannon, A. C. In Encyclopedia of Spectrsocopy and Spectrometry;
Lindon, J., Tranter, G., Holmes, J., Eds.; Academic Press: London, 2000;
Vol. 2, p 1479.
(37) Hannon, A. C.; Howells, W. S.; Soper, A. K. IOP Conf. Ser. 1990,
107, 193.
(38) Lorch, E. J. Phys. C 1969, 2, 229.
(39) Gramlich, V.; Meier, W. M. Z. Kristallogr. 1971, 133, 134.
(40) Cheetham, A. K.; Eddy, M. M.; Jefferson, D. A.; Thomas, J. M.
Nature 1982, 299, 24.
(41) Adams, J. M.; Haselden, D. A.; Hewit, A. W. J. Solid State Chem.
1982, 44, 245.
(42) Ikeda, T.; Izumi, F.; Kodairi, T.; Kamiyama, T. Chem. Mater. 1998,
10, 3996.
(43) Handbook of Porous Solids; Schu¨th, F., Sing, K. S. W., Weitkamp,
J., Eds.; Viley-VCH: Weinheim, 2002; Vol. 1, p 478.
(44) Kulshreshtha, S. K.; Vijayalakshami, R.; Sudarsan, V. Stud. Surf.
Sci. Catal. 1998, 113, 699.
(45) Liu, C. C.; Maciel, G. E. J. Am. Chem. Soc. 1996, 118, 5103.
(46) Maciel, G. E.; Sindorf, D. W. J. Am. Chem. Soc. 1980, 102, 7606.
(47) Kunath-Fandrei, G.; Bastow, T. J.; Hall, J. S.; Ja¨ger, C.; Smith, M.
E. J. Phys. Chem. 1995, 99, 15138.
(48) Tijink, G. A. H.; Janssen, R.; Veeman, W. S. J. Am. Chem. Soc.
1987, 109, 7301.
(49) Pluth, J. J.; Smith, J. V. J. Am. Chem. Soc. 1980, 102, 4704.
(50) Placzek, G. Phys. ReV. 1985, 86, 377.
(51) Howe, M. A.; McGreevy, R. L.; Howells, W. S. J. Phys. Condens.
Matter. 1989, 1, 3433.
Acknowledgment. We thank the Royal Society for the
award of a K. C. Wong Fellowship to H.X.Y., the EPSRC for
funding (GR/N07622), and the CCLRC for provision of
beamtime at ISIS. We are grateful to Dr. S. E. Ashbrook of the
Department of Earth Sciences, University of Cambridge, for
advice on the interpretation of the 29Si NMR data and to Mr. J.
Hawkins of the Interface Analysis Centre, University of Bristol,
for the EDXA.
References and Notes
(1) Barrer, R. M. Hydrothermal Chemistry of Zeolites; Academic
Press: London, 1982.
(52) Wright, A. C. J. Non-Cryst. Solids 1985, 76, 187.
(53) Farman, H.; Blakey, D.; Dore, J. C.; Bellissent-Funel, M.-C.; Elliott,
S. R. Phys. Scripta 1994, 49, 634.
(2) Francis, R. J.; O’Hare, D. J. Chem. Soc., Dalton Trans 1998, 3133.
(3) Cundy, C. S.; Cox, P. A. Chem. Soc. ReV. 2003, 103, 663.
(4) Cheetham, A. K.; Ferey, G.; Loiseau, T. Angew. Chem., Int. Ed.
Engl. 1999, 38, 3268.
(54) Yanwei, Z.; Fagherazzi, G.; Polizzi, S. J. Mater. Sci. 1995, 30, 2153.
(55) Elliott, S. R.; Rao, C. N. R.; Thomas, J. M. Angew. Chem., Int.
Ed. Engl. 1986, 25, 31.
(56) Elliott, S. R. Nature 1991, 354, 445.
(5) Klinowski, J. Curr. Opin. Solid State Mater. Sci. 1998, 3, 79.
(6) Newsam, J. M.; Bein, T.; Klein, J.; Maier, W. F.; W, W. S.
Microporous Mesoporous Mater. 2001, 48, 355.
(7) Norby, P. J. Am. Chem. Soc. 1997, 119, 5215.
(8) Cheetham, A. K.; Mellot, C. F. Chem. Mater. 1997, 9, 2269.
(9) Walton, R. I.; O’Hare, D. Chem. Commun. 2000, 2283.
(10) Walton, R. I.; Smith, R. I.; O’Hare, D. Microporous Mesporous
Mater. 2001, 48, 79.
(57) International Tables for Crystallography; Kluwer Academic Pub-
lishers: Dordrecht, 1995; Volume C.
(58) Felsche, J.; Luger, S.; Fischer, P. Acta Crystallogr. C 1987, 43,
809.
(59) Wiebcke, M.; Engelhardt, G.; Felsche, J.; Kempa, P. B.; Sieger,
P.; Schefer, J.; Fischer, P. J. Phys. Chem. 1992, 96, 392.
(60) Czjek, M.; Jobic, H.; Fitch, A. N.; Vogt, T. J. Phys. Chem. 1992,
96, 1535.
(11) Dougherty, J.; Iton, L. E.; White, J. W. Zeolites 1995, 15, 640.
(12) Kirschhock, C. E. A.; Ravishankar, R.; Jacobs, P. A.; Martens, J.
A. J. Phys. Chem. B 1996, 103, 11021.
(13) Moor, P. P. E. A. d.; Beelen, T. P. M.; Santen, R. A. v.; Beck, L.
W.; Davis, M. E. J. Phys. Chem. B 2000, 104, 7600.
(14) Shi, J. M.; Anderson, M. W.; Carr, S. W. Chem. Mater. 1996, 8,
369.
(61) Stahl, K.; Artioli, G. Eur. J. Mineral. 1993, 5, 851.
(62) Hannon, A. C. Rutherford Appleton Laboratory Report RAL-93-
063, 1993.
(63) Twu, J.; Dutta, P. K.; Kesge, C. T. J. Phys. Chem. 1991, 95, 7.
(64) Davies, M. E.; Lobo, R. F. Chem. Mater. 1992, 4, 756.
(65) Knight, C. T. G. Zeolites 1990, 10, 140.
(66) Swaddle, T. W. Coord. Chem. ReV. 2001, 219-221, 665.