LETTERS
Ordered silicon vacancies in the framework
structure of the zeolite catalyst SSZ-74
CHRISTIAN BAERLOCHER1 , DAN XIE1 , LYNNE B. MCCUSKER1 *, SON-JONG HWANG2 , IGNATIUS Y. CHAN3 ,
KENNETH ONG3 , ALLEN W. BURTON3 AND STACEY I. ZONES3
1
Laboratory of Crystallography, ETH Zurich, CH-8093 Zurich, Switzerland
2
The Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, California 91125, USA
3
Chevron Energy Technology Co., Richmond, California 94802, USA
* e-mail: mccusker@mat.ethz.ch
Published online: 11 July 2008; doi:10.1038/nmat2228
Physico-chemical characterization of the high-silica zeolite
catalyst SSZ-74 (ref. 1) suggested that it, like the related materials
TNU-9 (ref. 2) and IM-5 (ref. 3), has a multidimensional
10-ring channel system4 . Such pore systems are ideal for
many petrochemical applications, and indeed SSZ-74 has
been shown to be a good catalyst for a wide variety of
reactions1 . The elucidation of its framework structure, however,
proved to be difficult. Comparable problems were encountered
with TNU-9 and IM-5, which were synthesized with related
structure-directing agents. Their framework structures, which
are the two most complex ones known, both have 24 Si
atoms in the asymmetric unit, and were finally solved by
combining high-resolution powder diffraction data with
information derived from high-resolution electron microscopy
images5,6 . Therefore, a similar approach, using the powder
charge-flipping algorithm7 to combine the two types of data and
molecular modelling to help to locate the structure-directing
agent, was applied to SSZ-74. This procedure eventually
revealed a most unusual 23-Si-atom framework structure
(|(C16 H34 N2 )4 |[Si92 4 O184 (OH)8 ]) with ordered Si vacancies.
The three high-silica zeolites TNU-9, IM-5 and SSZ-74
constitute a synthesis series, in which the structure-directing agent
(SDA) differs only in the length of the methylene chain between
two N -methylpyrrolidinium rings. SSZ-74 is synthesized with
1,6-bis(N -methylpyrrolidinium)-hexane, whereas TNU-9 and
IM-5 are synthesized with the butane and pentane analogues,
respectively. TNU-9 (framework type code8 TUN) has a
three-dimensional 10-ring channel system reminiscent of that of
ZSM-5 (MFI), but its catalytic properties are quite distinct9 . IM-5
(IMF), on the other hand, has a two-dimensional 10-ring channel
system with a limited third dimension and has its own unique set
of catalytic properties10,11 . Adsorption experiments and catalytic
test reactions carried out on SSZ-74 indicated that it also had a
multidimensional 10-ring channel system, so the elucidation of its
structure was of considerable interest.
For structure analysis, high-resolution powder diffraction data
were collected on the as-synthesized material, because significant
line broadening had been observed on removal of the SDA. The
pattern could be indexed on a monoclinic unit cell (a = 20.507 Å,
b = 13.394 Å, c = 20.099 Å, β = 102.2◦ ), and the systematic
absences indicated that the most probable space groups were C 2/c
or Cc . Initial attempts to solve the structure from these data
nature materials ADVANCE ONLINE PUBLICATION www.nature.com/naturematerials
© 2008 Macmillan Publishers Limited. All rights reserved.
using the zeolite-specific programs FOCUS (ref. 12) and ZEFSA
II (ref. 13) failed, so the powder charge-flipping approach that
was used to solve the structure of IM-5 was tried. The original
charge-flipping algorithm introduced by Oszlányi and Sütő in
2004 for single-crystal data14,15 uses a deceptively simple electron
density modification procedure to reconstruct the phases of the
reflections (see the Methods section). The determination of these
phases, which are lost in a diffraction experiment, is the key to
structure solution.
For powder diffraction data, the relative intensities of reflections
with similar d -spacings are ambiguous, because they occur at
the same diffraction angle 2θ and only the total intensity can be
measured. For example, of the 3,258 reflections in the powder
diffraction pattern of SSZ-74 (dmin = 0.95 Å), 2,717 are within 0.3
times the full-width at half-maximum of a neighbouring one (more
than 80% overlap). These 2,717 reflections fall into 560 overlap
groups with 2–36 reflections per group (average 4.85). To address
this overlap problem, two independent research groups modified
the charge-flipping algorithm to include an intensity repartitioning
procedure for overlapping reflections7,16 . Baerlocher et al.7 coupled
the periodic repartitioning with a second type of electron density
modification based on histogram matching17 , and it is this
powder charge-flipping algorithm, implemented in the program
Superflip18 , that was used for SSZ-74. The reference histogram
used for this second perturbation simply reflects the chemical
composition of the material (the number and height of the peaks
that can be expected in the electron density map, but not their
location). One of the attractive features of the charge-flipping
approach is that it makes no assumptions about symmetry, so space
group ambiguities are irrelevant.
Because charge-flipping works in both real (model-building)
and reciprocal (diffraction) space, supplementary information
in either realm can be added easily. In the case of SSZ-74,
the high-resolution transmission electron microscopy (HRTEM)
image taken along the [110] direction showed the presence of
ordered pores (Fig. 1a). Fourier transformation of this image
using the CRISP software19 yielded a set of structure-factor
amplitudes and phases for 29 reflections. These were used to
construct a structure envelope20,21 that could be imposed in real
space to eliminate any electron density in these pores (Fig. 1b,c).
Although this envelope defines only a part of what was eventually
determined to be the full channel system, the information that
1
LETTERS
a b c
20 nm b

a
b c

a
c

Figure 1 Pore information on SSZ-74 obtained from an HRTEM image. a, The HRTEM image taken along the [110] direction with an enlarged inset showing a
translationally averaged image and the unit cell. b, Four unit cells of the [110] projection of the structure envelope calculated from the 29 structure-factor amplitudes and
phases derived from the image in a. c, A different orientation of a single unit cell showing the three-dimensional character of the structure envelope.

these parts of the unit cell should not contain framework atoms
served to facilitate structure solution. In contrast to TNU-9 and
IM-5, where HRTEM images along three different directions
were obtained, only one was available for SSZ-74. This meant
that the phase information that could be extracted was very
limited, so it was used to construct a structure envelope (see the
Supplementary Information) rather than to modify the (random)
starting phase sets.
In general, the major differences between the diffraction pattern
of an as-synthesized zeolite and that of a calcined one (that is,
without organic species in the pores) occur at low angles. To
simulate the calcined state while retaining the sharper lines of the
as-synthesized material, the intensities of the first 11 reflections
were artificially increased by a factor of 1.5 to approximate those
in the pattern of the calcined material. With this modified set
of 3,258 intensities and the structure envelope described above,
a series of powder charge-flipping runs was started. The electron
density maps with the best R-values were examined by eye,
and those with reasonable-looking pores and the semblance of
a framework structure were averaged and used as a seed within
Superflip to generate 100 new starting phase sets. That is, the
phases were calculated from the map and then the phase for each
reflection was allowed to vary by up to 25% in a random fashion.
In this way, 100 different but related phase sets were generated
from the seed map. This is similar to the procedure used for
IM-5, where an approximate (but incorrect) framework model
was used to generate starting phase sets. The ten best maps from
these 100 new runs were averaged, and a framework structure
emerged. The symmetry of the final map was ambiguous, so
the centrosymmetric space group C 2/c was assumed. Eleven Si
positions were taken directly from the powder charge-flipping map,
and a twelfth one was added by hand to yield a fully 4-connected
framework with a powder diffraction pattern very similar to the
measured one.
Because the peak shapes in the original data were extremely
asymmetric and difficult to model, new data were collected for
Rietveld refinement. Difference Fourier maps showed a cloud of
electron density in the pores, but the individual atoms of the
SDA could not be resolved, so molecular modelling was used
to estimate their positions. For this, energy-optimization docking
calculations were carried out according to the procedure described
by Burton et al.22 , which is similar to the one initially described
by Lewis et al.23 . Rietveld refinement of this model, with geometric
restraints imposed on the bond distances and angles of both the
framework and the SDA, proceeded smoothly, except that the
geometry of the Si atom that had been added was severely distorted.
This Si was bonded to an oxygen atom lying on an inversion
centre, so it was initially assumed that this was the cause of the
problem, and the oxygen was moved off the special position. This
did not relieve the strain, however, so the symmetry of the entire
structure was reduced to Cc with 24 Si atoms in the asymmetric
unit. This led to some improvement in the profile fit, but the
coordination environment of the two Si atoms connected by the
problematic oxygen atom remained distorted. The cause of the
problem became apparent only when the occupancy parameters
of these two Si atoms were refined. One disappeared entirely while
the other remained intact, and the profile fit improved significantly
(Rwp decreased from 0.159 to 0.123). Furthermore, the geometric
restraints on both the framework and the SDA could be relaxed
and all bond distances and angles remained reasonable. Refinement
of this model with just 23 Si atoms (and a vacancy) in the
asymmetric unit (Fig. 2) converged with RF = 0.035 and the profile
fit shown in Fig. 3. Further crystallographic details are provided in
the Supplementary Information.
The 29 Si and 1 H magic-angle-spinning (MAS) NMR spectra
of SSZ-74 support this final model (see the Supplementary
Information). The former indicates the presence of an unusually
high percentage (Q3 /(Q3 + Q4 ) ∼ 15%) of Q3 species (3- rather
than 4-connected Si atoms)4,24 , and the latter has a strong signal at
10 p.p.m., which is consistent with a silanol/siloxy nest25 . From the
model, the fraction of Q3 sites should be 4Q3 /23Si or 17.4%, but
the 29 Si spectrum is quite broad and there is no clear distinction
between the Q3 and Q4 sites, so the number of Q3 sites is
slightly underestimated.
This ordered Si vacancy is perhaps the most striking feature
of the SSZ-74 framework structure. Such an ordered defect,
with four framework oxygen atoms forming an approximate
tetrahedron around a vacancy, has never been observed
before in a high-silica zeolite. Two of these oxygen atoms
make the closest contacts to the two nitrogen atoms of the
doubly charged 1,6-bis(N -methylpyrrolidinium)-hexane species
(O33 ··· N5 = 3.62 Å; O3 ··· N13 = 3.58 Å), so it is assumed
from charge-balance arguments that they are siloxy rather than
silanol oxygens. The other two, O16 and O32, are presumably

2 nature materials ADVANCE ONLINE PUBLICATION www.nature.com/naturematerials

© 2008 Macmillan Publishers Limited. All rights reserved.

 LETTERS
a
N5
N13

c O3
O33
a O32 O16
b O3
O33
N13
N5

a a
a

c b b

Figure 2 The crystal structure of SSZ-74. a, A portion of the structure showing the Si vacancy and the interaction between O3 and O33 of the framework with N13 and N5
of the SDA. b, Projections of the framework structure down the [010], [110] and [001] directions showing the arrangement of the vacancies and of the 10-ring pore openings.
Bridging oxygen atoms in the framework have been omitted for clarity. The [110] projection corresponds to that of the HRTEM image in Fig. 1.

latter corresponds to approximately 5% of the total hydrogen

30,000 concentration, which is consistent with two OH groups per SDA
molecule (34 hydrogen atoms). The peak disappears on calcination,
indicating the loss of siloxy–silanol pairs.
The fact that the diffraction pattern of calcined SSZ-74 has
20,000
broader peaks is also consistent with this structural model.
4,000
The vacancies in the framework are stabilized by electrostatic
interactions between the siloxy groups and the positive charge of
10,000 the SDA cations. Analogies can be drawn with layered silicates, in
2,000 which the structures are stabilized by the electrostatic interactions
between the organocations and the siloxy groups. Usually such
structures become amorphous on calcination unless siloxy/silanol
0 0 groups in adjacent layers are able to condense. Defects in all-silica
5 10 15 20 25 zeolites, on the other hand, generally occur as a result of broken
2θ 2θ Si–O–Si links that form silanol/siloxy nests. On calcination, these
defects are able to anneal by condensation with neighbouring
silanol groups to form an intact framework structure. This is not
Figure 3 Observed (top), calculated (middle) and difference (bottom) profiles possible for SSZ-74, because not only are there no Si atoms in the
for the Rietveld refinement of SSZ-74. The higher angle data have been scaled up vacancies, but the geometry at the vacancy is incompatible with that
by a factor of four to show more detail. l = 0.51981 Å. of a SiO4 tetrahedron.
Attempts to refine the structure in the higher space group
C 2/c assuming the vacancy to be disordered over the two
silanols, forming H-bonds to the siloxy oxygens and giving rise symmetry-equivalent positions (that is, with a Si occupancy of
to the 10 p.p.m. signal in the 1 H MAS NMR spectrum25 . The 1/2) led to significantly higher R-values and a poorer profile fit.

nature materials ADVANCE ONLINE PUBLICATION www.nature.com/naturematerials 3

© 2008 Macmillan Publishers Limited. All rights reserved.
LETTERS
b METHODS
a
c inverse Fourier transform (equation (1)).
Figure 4 The framework structure and channel system of SSZ-74. Si: yellow;
O: red. The details of the shape and size of the periodic nodal surface used to depict
the channel system are only approximate and should not be over-interpreted. The
surface is meant to indicate only the approximate arrangement of the channels
within the framework structure. Comparison of the channel system with the
structure envelope used for the structure solution (Fig. 1c) shows that although the
latter has some of the features of the former, it is incomplete. The undulating form of
the channels along the [110] direction is approximated by a straight channel, and
the connecting channels along the b axis are missing.
The vacancies seem to be truly ordered. The whole framework has
adjusted to the lower symmetry, but perhaps the most obvious
deviations from ideal C 2/c symmetry, besides the vacancy itself,
can be seen in the organic molecule, which has lost its two-fold
axis and adopted a conformation that allows interaction with the
terminal O3 and O33 oxygen atoms (Fig. 2a and Supplementary
Information). The geometric restraints placed on the bond
distances and angles of the SDA were strong enough to enforce
a chemically sensible geometry during the refinement, but soft
enough to allow the atoms to adjust their positions. No restraints
on the N ··· O distances were imposed, so the fact that these refined
to chemically reasonable values is an indication that the refinement
can be trusted.
As the physico-chemical characterization indicated, SSZ-74 has
a three-dimensional 10-ring channel system. Undulating channels
along the [110] direction are connected to one another by relatively
straight 10-ring channels along the [001] direction (Fig. 4). This
in itself makes SSZ-74 a promising catalyst, but the presence of
structured vacancies with terminal oxygen species poses intriguing
possibilities for incorporating extra functionality. A competitive
cracking reaction test (n-hexane versus 3-methylpentane) showed
the Al form of SSZ-74 to be more open than ZSM-5 (ref. 1). More of
the branched isomer cracks on SSZ-74, as might be expected, given
the larger pores. This positions SSZ-74 rather uniquely as a 10-ring
zeolite catalyst that can admit larger reactant molecules than can
ZSM-5, and yet provide more selectivity than a large-pore zeolite.
Once again, the key to the structure determination of a
complex zeolite framework structure has proved to be the
combination of powder diffraction and electron microscopy data
within the charge-flipping algorithm. The power and sensitivity
of this algorithm are remarkable. Not only could the 89 atom
(23Si + 48O + 16C + 2N = 89 atoms/asymmetric unit) structure
of this polycrystalline zeolite be solved, but in retrospect it is also
clear that the powder charge-flipping map that was used to generate
the initial model actually showed the totally unexpected Si vacancy.
This approach to structure determination is generally applicable,
and holds much promise for all those polycrystalline materials with
structures that have hitherto resisted solution.
4 nature materials ADVANCE ONLINE PUBLICATION www.nature.com/naturematerials
© 2008 Macmillan Publishers Limited. All rights reserved.
The high-resolution powder diffraction data used for the structure
determination were collected on beamline X16C at the National Synchrotron
Light Source in Brookhaven, New York, with a wavelength of 0.70035 Å, and
those for the structure refinement on the Swiss Norwegian Beamlines at the
European Synchrotron Radiation Facility with a wavelength of 0.51981 Å.
Further details are given in the Supplementary Information.
The charge-flipping approach to structure determination can be described
quite simply. The structure-factor amplitudes (|F hkl |), which are derived from
the experimentally measured reflection intensities, are first assigned random
phases (φ hkl ) to generate a (random) electron density map (ρ xyz ) through an
X
ρ xyz = |F hkl |e−φ hkl . (1)
The signs of all electron densities in this map that are below a user-defined
threshold δ (a small positive number) are then reversed to produce a perturbed
map. Essentially, all negative electron densities, which have no physical meaning,
are made positive. Fourier transformation of this map yields a modified set of
structure-factor amplitudes and phases. The new phases are then combined
with the experimentally determined amplitudes to calculate a new map. This
cycle is repeated until the calculated amplitudes agree with the experimental
ones (convergence) or until a predefined number of cycles has been reached.
After convergence, most of the phases should be correct and the electron
density map can usually be interpreted to yield a structural model suitable for
least-squares refinement. For more information on charge flipping, see the
recent review by Oszlányi and Sütő26 .
All solid-state MAS NMR spectra were recorded using a Bruker DSX-500
with operating frequencies of 500.23 and 99.4 MHz for 1 H and 29 Si nuclei,
respectively, and using a Bruker 4 mm cross-polarization MAS probe. The
typical spinning rates were 14 kHz for 1 H and 8 kHz for 29 Si, and the chemical
shifts were referred to tetramethylsilane. For Q3 quantitation purposes, 29 Si
direct polarization (Bloch decay experiment) spectra were obtained using
repetition delay times (>1,000 s) that allowed full relaxation (T1 ) recovery.
Received 16 March 2008; accepted 18 June 2008; published 11 July 2008.
References
1. Zones, S. I., Burton, A. W. & Ong, K. Molecular sieve SSZ-74 composition of matter and synthesis
thereof. International Patent WO/2007/079038 (2007).
2. Hong, S. B. et al. Synthesis, structure solution, characterization, and catalytic properties of TNU-10:
A high-silica zeolite with the STI topology. J. Am. Chem. Soc. 126, 5817–5826 (2004).
3. Benazzi, E., Guth, J. L. & Rouleau, L. IM-5 zeolite, method of preparation and catalytic applications
thereof. Patent Cooperation Treaty WO 98/17581 (1998).
4. Burton, A. W. et al. A comparative study of three closely related unsolved zeolite structures. Studies
Surf. Sci. Catal. 170, 690–697 (2007).
5. Gramm, F. et al. Complex zeolite structure solved by combining powder diffraction and electron
microscopy. Nature 444, 79–81 (2006).
6. Baerlocher, Ch. et al. Structure of the polycrystalline zeolite catalyst IM-5 solved by enhanced charge
flipping. Science 315, 1113–1116 (2007).
7. Baerlocher, Ch., McCusker, L. B. & Palatinus, L. Charge flipping combined with histogram matching
to solve complex crystal structures from powder diffraction data. Z. Kristallogr. 222, 47–53 (2007).
8. Baerlocher, Ch., McCusker, L. B. & Olson, D. H. Atlas of Zeolite Framework Types 6th edn (Elsevier,
Amsterdam, 2007).
9. Hong, S. B. et al. Synthesis, crystal structure, characterization, and catalytic properties of TNU-9.
J. Am. Chem. Soc. 129, 10870–10885 (2007).
10. Serra, J. M., Guillon, E. & Corma, A. A rational design of alkyl-aromatics dealkylation-transalkylation
catalysts using C8 and C9 alkyl-aromatics as reactants. J. Catal. 227, 459–469 (2004).
11. Zones, S. I. et al. Indirect assessment of unknown zeolite structures through inference from zeolite
synthesis comparisons coupled with adsorption and catalytic selectivity studies. J. Catal. 250,
41–54 (2007).
12. Grosse-Kunstleve, R. W., McCusker, L. B. & Baerlocher, Ch. Powder diffraction data and crystal
chemical information combined in an automated structure determination procedure for zeolites.
J. Appl. Crystallogr. 30, 985–995 (1997).
13. Falcioni, M. & Deem, M. W. A biased Monte Carlo scheme for zeolite structure solution. J. Chem.
Phys. 110, 1754–1766 (1999).
14. Oszlányi, G. & Sütő, A. Ab initio structure solution by charge flipping. Acta Crystallogr. A 60,
134–141 (2004).
15. Oszlányi, G. & Sütő, A. Ab initio structure solution by charge flipping. II. Use of weak reflections.
Acta Crystallogr. A 61, 147–152 (2005).
16. Wu, J. S., Leinenweber, K., Spence, J. C. H. & O’Keeffe, M. Ab initio phasing of X-ray powder
diffraction patterns by charge flipping. Nature Mater. 5, 647–652 (2006).
17. Zhang, K. Y. J. & Main, P. Histogram matching as a new density modification technique for phase
refinement and extension of protein molecules. Acta Crystallogr. A 46, 41–46 (1990).
18. Palatinus, L. & Chapuis, G. Superflip—a computer program for the solution of crystal structures by
charge flipping in arbitrary dimensions. J. Appl. Crystallogr. 40, 786–790 (2007).
19. Hovmöller, S. CRISP: Crystallographic image processing on a personal computer. Ultramicroscopy
41, 121–135 (1992).
20. Brenner, S., McCusker, L. B. & Baerlocher, Ch. Using a structure envelope to facilitate structure
solution from powder diffraction data. J. Appl. Crystallogr. 30, 1167–1172 (1997).

21. Brenner, S., McCusker, L. B. & Baerlocher, Ch. The application of structure envelopes in structure
determination from powder diffraction data. J. Appl. Crystallogr. 35, 243–252 (2002).
22. Burton, A. W., Lee, G. S. & Zones, S. I. Phase selectivity in the syntheses of cage-based zeolite
structures: An investigation of thermodynamic interactions between zeolite hosts and structure
directing agents by molecular modeling. Microporous Mesoporous Mater. 90, 129–144 (2006).
23. Lewis, D. W., Freeman, C. M. & Catlow, C. R. A. Predicting the templating ability of organic additives
for the synthesis of microporous materials. J. Phys. Chem. 99, 11194–11202 (1995).
24. Engelhardt, G. & Michel, D. High-Resolution Solid-State NMR of Silicates and Zeolites (Wiley,
Chichester, 1987).
25. Koller, H., Lobo, R. F., Burkett, S. L. & Davis, M. E. SiO− ···HOSi hydrogen bonds in as-synthesized
high-silica zeolites. J. Phys. Chem. 99, 12588–12596 (1995).
26. Oszlányi, G. & Sütő, A. The charge flipping algorithm. Acta Crystallogr. A 64, 123–134 (2008).
Supplementary Information accompanies this paper on www.nature.com/naturematerials.
Acknowledgements
We thank the beamline scientists at Beamline X16C at the National Synchrotron Light Source,
Brookhaven, and at the Swiss Norwegian Beamlines at the European Synchrotron Radiation Facility,
nature materials ADVANCE ONLINE PUBLICATION www.nature.com/naturematerials
© 2008 Macmillan Publishers Limited. All rights reserved.
LETTERS
Grenoble, for their assistance with the powder diffraction measurements. Research was carried out in
part at the National Synchrotron Light Source, Brookhaven National Laboratory, which is supported
by the US Department of Energy, Division of Materials Sciences and Division of Chemical Sciences.
The NMR facility at Caltech was supported by the National Science Foundation (NSF) under Grant
Number 9724240 and partially supported by the MRSEC Program of the NSF under Award Number
DMR-0520565. Funding from the Swiss National Science Foundation (C.B., D.X. and L.B.M.) is
gratefully acknowledged.
Author contributions
C.B., D.X. and L.B.M. carried out the structure analysis, S.-J.H. the MAS NMR analysis, I.Y.C. the
electron microscopy experiments, K.O. the initial powder diffraction data collection and analysis and
S.I.Z. the synthesis of the material. A.W.B. carried out the molecular modelling and coordinated
the project.
Author information
Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions.
Correspondence and requests for materials should be addressed to L.B.M.
5