Modulator-Controlled Synthesis of Microporous STA-26, an Interpenetrated 8,3-Connected Zirconium MOF with the the-i Topology, and its Reversible Lattice Shift

Article abstract of DOI:10.1002/chem.201705136

A fully interpenetrated 8,3-connected zirconium MOF with the the-i topology type, STA-26 (St Andrews porous material-26), has been prepared using the 4,4′,4“-(2,4,6-trimethylbenzene-1,3,5-triyl)tribenzoate (TMTB) tritopic linker with formic acid as a modulating agent. In the as-prepared form STA-26 possesses Im (Formula presented.) m symmetry compared with the Pm (Formula presented.) m symmetry of the non-interpenetrated analogue, NU-1200, prepared using benzoic acid as a modulator. Upon removal of residual solvent there is a shift between the interpenetrating lattices and a resultant symmetry change to Cmcm which is fully reversible. This is observed by X-ray diffraction and ¹³C MAS NMR is also found to be remarkably sensitive to the structural transition. Furthermore, heating STA-26(Zr) in vacuum dehydroxylates the Zr₆ nodes leaving coordinatively unsaturated Zr⁴⁺ sites, as shown by IR spectroscopy using CO and CD₃CN as probe molecules. Nitrogen adsorption at 77 K together with grand canonical Monte Carlo simulations confirms a microporous, fully interpenetrated, structure with pore volume 0.53 cm³ g^?1 while CO₂ adsorption at 196 K reaches 300 cm³ STP g^?1 at 1 bar. While the pore volume is smaller than that of its non-interpenetrated mesoporous analogue, interpenetration makes the structure more stable to moisture adsorption and introduces shape selectivity in adsorption.

Full text of DOI:10.1002/chem.201705136

A Journal of
Accepted Article
Title: Modulator-Controlled Synthesis of Microporous STA-26, an
Interpenetrated 8,3-Connected Zirconium MOF with the the-i
Topology, and its Reversible Lattice Shift
Authors: Alice Bumstead, David Cordes, Daniel Dawson, Kristina
Chakarova, Mihail Mihaylov, Claire Hobday, Tina Dueren,
Konstantin Hadjiivanov, Alexandra Slawin, Sharon Elisabeth
Ashbrook, Ram Prasad, and Paul Anthony Wright
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201705136
Link to VoR: http://dx.doi.org/10.1002/chem.201705136
Supported by

10.1002/chem.201705136
Chemistry - A European Journal
Modulator-Controlled Synthesis of Microporous STA-26, an Interpenetrated 8,3-
Connected Zirconium MOF with the the-i Topology, and its Reversible Lattice Shift
Alice M. Bumstead,1 David B. Cordes,1 Daniel M. Dawson,1, Kristina K. Chakarova,2 Mihail
Y. Mihaylov,2 Claire L. Hobday,3 Tina Düren,3 Konstantin I. Hadjiivanov,2 Alexandra M. Z.
Slawin,1 Sharon. E. Ashbrook,1 Ram R. R. Prasad*,1 and Paul A. Wright*,1
1 EaStCHEM School of Chemistry, University of St Andrews, Purdie Building, North Haugh,
St Andrews KY16 9ST, United Kingdom.
2 Institute of General and Inorganic Chemistry, Bulgarian Academy of Sciences, Sofia 1113,
Bulgaria.
3Centre for Advanced Separations Engineering, Department of Chemical Engineering,
University of Bath, Bath BA2 7AY, United Kingdom.
*E-mail: rprr@st-andrews.ac.uk; paw2@st-andrews.ac.uk
Abstract A fully interpenetrated 8,3-connected zirconium MOF with the the-i topology
type, STA-26 (St Andrews porous material-26), has been prepared using the 4,4',4"-(2,4,6-
trimethylbenzene-1,3,5-triyl)tribenzoate (TMTB) tritopic linker with formic acid as a
modulating agent. In the as-prepared form STA-26 possesses Im-3m symmetry compared
with the Pm-3m symmetry of the non-interpenetrated analogue, NU-1200, prepared using
benzoic acid as a modulator. Upon removal of residual solvent there is a shift between the
interpenetrating lattices and a resultant symmetry change to Cmcm which is fully reversible.
13
This is observed by X-ray diffraction and C MAS NMR is also found to be remarkably
sensitive to the structural transition. Furthermore, heating STA-26(Zr) in vacuum
dehydroxylates the Zr6 nodes leaving coordinatively unsaturated Zr4+ sites, as shown by IR
spectroscopy using CO and CD3CN as probe molecules. Nitrogen adsorption at 77 K together
1
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
with grand canonical Monte Carlo simulations confirms a microporous, fully interpenetrated,
structure with pore volume 0.53 cm3 g−1 while CO2 adsorption at 196 K reaches 300 cm3 STP
g−1 at 1 bar. While the pore volume is smaller than that of its non-interpenetrated mesoporous
analogue, interpenetration makes the structure more stable to moisture adsorption and
introduces shape selectivity in adsorption.
Keywords
Zirconium and hafnium MOFs; interpenetration; modulators; reversible lattice shift
Introduction
The diverse and growing family of porous metal-organic frameworks (MOFs),[1] constructed
from metal-based nodes and organic linkers, offers a range of potential applications in
adsorption,[2] sensing,[3] catalysis[4] and drug delivery.[5] Among these many materials, the
zirconium carboxylates offer some of the most porous and also the most chemically and
thermally stable examples.[6] Many of these are based on the Zr6O8 building block, which
may adopt 12-, 10-, 8- or 6-fold coordination with di-, tri- or tetracarboxylate groups of linker
molecules, to give structures with fcu, bct, csq and spn topologies.[7] A zirconium carboxylate
MOF with the 8,3-connected the topology type, formed with the 4,4',4"-(2,4,6-
trimethylbenzene-1,3,5-triyl)tribenzoate (TMTB) tritopic linker, was recently reported by two
groups and assigned the names NU-1200[8] and BUT-12.[9] The structure consists of a
primitive cubic arrangement of corner-sharing cages with octahedral geometry, each of which
possesses a sodalite-like (sod) cage structure (Figure 1). Each ‘superoctahedral’ cage
comprises six Zr6O8 clusters linked via tritopic TMTB linkers on the faces of the octahedron,
2
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
and has an internal free diameter of 1.4 nm. Each Zr6O8 cluster is linked to eight others in a
square prismatic arrangement. Charge balance in NU-1200 is reportedly achieved by the
addition of hydroxyl groups and water molecules to the zirconium clusters, giving an overall
framework formula Zr6(μ3O)4(μ3OH)4(OH)4(OH2)4(C30H24O6)8/3.[10,11] Between these
‘superoctahedral’ cages is a mesoporous pore space, accessible via 2 nm windows, which is
of great interest as a potential host for adsorbates or for species that are catalytically active
for the transformation of large molecules.
The reported syntheses of NU-1200 and BUT-12 are similar, but make use of different
carboxylic acids as modulators. Modulators have been observed to have a strong effect on
MOF crystallisation, controlling particle size and morphology, the density of ‘missing linker’
defects and even the phase that is formed.[12] Although the length and flexibility of linkers[13-
15]
and the influence of solvents[16,17] on regulating the degree of interpenetration has been
observed previously, the role of modulators in governing the synthesis of interpenetrated
MOFs has not, to the best of our knowledge.
Here we report the role of modulators in controlling the interpenetration in a zirconium
MOFs with the the framework topology, and its hafnium analogue. Interpenetration, in which
identical lattices have grown one within the other, is observed in many MOFs where their
geometry and dimensions allow, and where interactions between the interpenetrating lattices
are favourable.[18] The zirconium amino-terephthalate MOF UiO-66(NH2) and related
structural forms with longer ditopic linkers, for example, have been shown to form
interpenetrated frameworks, the crystallinity of which can be enhanced by the use of
modulators.[14]
We designate the interpenetrated Zr-MOF framework STA-26 (St Andrews porous material-
26) and have characterised its structure, porosity and properties fully. Complementary single
3
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
crystal X-ray diffraction and N2 and CO2 adsorption (together with molecular simulation of
the adsorption) show it is microporous rather than mesoporous; powder diffraction and solid
state NMR reveal details of a relative lattice shift upon post-synthetic treatment that modifies
the window size; IR spectroscopy is used to investigate its adsorption sites; and
interpenetration imparts enhanced stability to moisture. Furthermore, we have prepared the
hafnium-version of this material, in the knowledge that Hf has similar chemistry in MOF
frameworks but that Hf-MOFs can possess properties significantly different from those of
their Zr-analogues.[19]
Figure 1 (Left) Polyhedral and ball-stick representation of the structure of NU-1200, which has the
the topology type. The mesoporous cages of NU-1200 are shown by the blue sphere while
the space within the sodalite-like cages are shown by the green spheres. (Right) The 8-
connected Zr6 cluster of NU-1200. Hydrogen atoms are omitted for clarity.
4
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Results and Discussion
A modified NU-1200 synthesis, based on a literature procedure which used benzoic acid as a
modulator, gave a material with a PXRD pattern that matched that simulated from the
published, non-interpenetrated, structure of NU-1200 in Pm-3m, with unit cell parameter of
28.300 Å (Figures S1 and S2). [8] Initial synthesis attempts using formic acid as a modulator,
similar to those reported for the synthesis of BUT-12[9] and with a DMF:formic acid molar
ratio of 1:2 (8 ml: 8 ml) gave crystalline solids with a PXRD pattern which could be indexed
to a primitive cell with a very similar unit cell parameter to NU-1200 (28.169 Å) but with
1
strongly different diffraction intensities (Figure 2). Solution-state H NMR spectra of the
dissolved Zr-MOF showed peaks from the TMTB ligand and also from residual DMF. It was
possible to pick out single crystals from these preparations that were sufficiently large for
laboratory single crystal diffraction (Figure S3).
After screening different DMF:formic acid molar ratios (Figure S4), it was found that a 1:1.2
molar ratio of DMF:formic acid (10 ml: 6 ml) gave a microcrystalline material comprising
cubic crystals ca. 1-2 μm in dimension (Figure S5). In this case the PXRD pattern (Figure 2,
right) could be indexed as body-centred cubic, with a similar unit cell to those preparations
described above, and this Zr-MOF was designated STA-26(Zr) (STA = St Andrews porous
material). Notably, in the absence of formic acid modulator, no crystalline solid forms,
whereas adding smaller amounts of formic acid (DMF:formic acid molar ratio 5.8:1) or
higher amounts of formic acid (DMF:formic acid 1:2) gives solids for which the PXRD
shows considerable intensity in peaks that violate the I-centring that is clearly demonstrated
by pure STA-26(Zr) (Figure S4).
A hafnium version, STA-26(Hf), was also prepared with a PXRD pattern that could be
indexed to a body-centred cell, unit cell parameter 28.317 Å (Figure 2).
5
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Figure 2 (Left) (a) Simulated PXRD pattern of NU-1200(Zr) compared with (b) that of as-synthesised
STA-26(Zr) prepared with DMF:formic acid molar ratio 1:2. (Right) (a) Simulated PXRD of STA-
26(Zr) compared with those of (b) as-synthesised STA-26(Zr) and (c) as-synthesised STA-26(Hf)
(each with a 1:1.2 molar ratio of DMF:formic acid).
SCXRD of a selected crystal of the product from the preparation with a DMF:formic acid
molar ratio of 1:2 revealed a structure with cubic symmetry, space group Im−3m (Table 1 and
Figure 3, left). The PXRD pattern simulated from this structure solution is a close fit to that
observed for both Zr and Hf versions of STA-26 (Figure 2, right), with the small differences
in intensity in the low angle peaks attributable to the presence of disordered solvent
remaining in the pores. The pattern is very different from that of the non-interpenetrated NU-
1200, compared directly in Figure 3 (right). This indicates that the I-centred STA-26(Zr)
possesses an interpenetrated structure, with two identical lattices of the the topology, each
with a structure similar to that of the NU-1200 framework, displaced by a lattice vector
[½,½,½]. The TMTB linkers exhibit 90° rotation of the benzoate groups with respect to the
central mesitylene core that is characteristic of this linker, and which results in crystallisation
of the the 8,3-topology. In STA-26, the sodalite-like cages of one of the interpenetrating
frameworks now occupy the mesoporous cavities of the other.
6
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Figure 3. (Left) Structure of STA-26(Zr) with Im-3m symmetry. It is an interpenetrated analogue of
NU-1200, shown in Figure 1. (Right) PXRD patterns of as-synthesised (a) NU-1200(Zr) and (b) STA-
26(Zr)
The linkers of the interpenetrated networks at their closest point display a distance of 5.16 Å
between the centres of C atoms of the adjacent benzene rings. Pore openings with a width of
4.80 Å (once van der Waals radii are taken into account) lead to octahedral cages of internal
diameter 14 Å (similar to those of NU-1200). To the best of our knowledge, an
interpenetrated Zr-MOF with 8-connected Zr6O8 clusters has not been reported, although it
has been shown that by fine tuning the planarity of triangular tetrazolate-based linkers used in
the synthesis of copper benzene- or triazine-centered tetrazolates the the and the-i topology
can be achieved for 8,3-connected frameworks.[15]
Since both NU-1200 and STA-26 have the same overall composition (metal clusters and
linkers), the difference in structure must arise from the use of different modulators in the
solvothermal preparations. As the formic, benzoic and 2-fluorobenzoic acids have very
similar pKa values (3.8, 4.2 and 3.8)[20] this originates from the different steric effects when
attached to the Zr6 clusters during crystal growth. It should be noted that no formate species
7
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
were observed in the digested STA-26 by solution-phase NMR (Figure S6) although benzoic
acid was observed in digested NU-1200 (Figure S7).
Although the bulk sample from which the single crystal was taken showed reflections in the
powder pattern that violate body centring, prolonged immersion of the bulk ‘1:2’ sample in
DMF post-synthesis, followed by drying in air, gave a bulk sample with a body-centred
powder pattern similar to that observed for the small crystal ‘1:1.2’ preparation. This
suggested that both samples had the same interpenetrated structure, and that the additional
peaks derived from a reversible symmetry change related to residual DMF content (Figure
S8).
Table. 1 Crystallographic data for STA-26(Zr) and STA-26(Zr)-C
Label
STA-26(Zr)
C40H36O16 Zr3
1046.35
STA-26(Zr)-C
C40H36O16 Zr3
1046.35
Formula
Formula weight/g mol-1
Temperature/K
Crystal system
Space group
173
173
Cubic
Orthorhombic
Cmcm
Im m
28.16930 (14)
-
a/Å
40.3620 (8)
39.8384 (8)
27.9686 (5)
44972.3(15)
24
b/Å
c/Å
-
V/Å
22352.6(3)
12
Z
ρ(calcd) g cm-3
Radiation type
μ mm-1
0.933
0.927
Cu Kα
3.725
Cu Kα
3.703
F(000)
6288
12576
Rint
0.1007
1.069
0.1908
GOF
1.426
Final R1 values [I >2σ(I)]a
Final wR2 values (all data)b
0.0618
0.1977
0.1309
0.3662
8
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Structural changes upon activation for adsorption measurements
In order to optimise the accessible pore volume of STA-26(Zr), samples were repeatedly
washed with acetone and subsequently dried at 80 °C prior to measurement of N2 adsorption
at 77 K. Under these conditions, the PXRD pattern displays additional reflections that violate
the I-centring. Subsequent immersion in DMF and drying gave a sample with the original I-
centred pattern (Figure 4). STA-26(Hf) showed similar behaviour when treated under these
conditions (Figure S9). Further, VT-PXRD of STA-26(Zr) shows a loss of I-centring upon
heating, which was not reversed on cooling (Figure 5). Taken together, these observations
suggest that traces of residual DMF solvent keep the interpenetrated lattices apart, whereas
their removal results in a structural change that was reversible by re-introduction of DMF.
Figure 4 Comparison of (a) the simulated PXRD pattern for STA-26(Zr) with those of (b) as-
synthesised STA-26(Zr) (1:1.2 prep.), (c) STA-26(Zr) after 1 week in acetone with solvent
exchange every 24 h, followed by drying, (d) after immersion in DMF and drying.
9
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Figure 5 VT-PXRD of as-synthesised STA-26(Zr) heated from 25 °C to 200 °C and then cooled to
50 °C, compared with simulated pattern for as-synthesised STA-26(Zr), the black curve.
To investigate this further, SCXRD was performed on a single crystal of as-synthesised STA-
26(Zr) that had been kept in acetone for a week with the solvent replaced every 24 h and
subsequently dried at 80 °C. The structure was solved in the orthorhombic Cmcm space
group, and is here referred to as STA-26(Zr)-C (See Table 1 and the cif file in the ESI for
crystallographic details). There has been a relative movement of one lattice with respect to
the other, changing the distance between the TMTB linkers and the Zr6O8 clusters of the two
frameworks (Figure 6). Whereas in body-centred STA-26(Zr) the distance between the
central phenyl ring of the TMTB of adjacent linkers is the same in every case (4.74 Å), in
STA-26(Zr)-C there are three different distances, 3.571 Å, 4.74 Å and 7.41 Å. Consequently,
STA-26(Zr)-C possesses three different pore openings with free diameters 3.57 Å, 4.43 Å and
5.24 Å leading to cages of unchanged internal diameter (14 Å). This relative movement of the
interpenetrating frameworks changes the overall symmetry of the structure, so that instead of
10
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
a single crystallographic environment for the linkers in STA-26(Zr), there are now three
different sets of linker environments in STA-26(Zr)-C.
Solvent removal

Figure 6 Structural change of (left) STA-26(Zr) (Im−3m) to (right) STA-26(Zr)-C (Cmcm) upon
removal of residual solvent molecules. The relative direction of lattice movement of one of
the interpenetrated lattices (with clusters in red) is denoted by purple arrows.
Owing to its sensitivity to small changes in the local structure, solid-state 13C NMR
spectroscopy can be used to follow this structural transition. For body-centred STA-26(Zr),
each of the chemically-distinct carbon atoms of the TMTB linker structure gives a single
resonance, as expected from the symmetry (Figures 7 and S10, and S11 for the Hf-form).
Notably, the spectrum is similar to that observed for the non-interpenetrated NU-1200
(Figure S12), in which the ligands are all crystallographically equivalent, but with small
changes in chemical shift. In the 13C CP MAS NMR spectrum of STA-26(Zr)-C prepared by
solvent removal, each of the signals shows splittings that indicate the presence of distinct
crystallographic environments, due to the relative movement of the frameworks. As an
example, whereas there is only one crystallographically distinct CH3 in the I-centred
structure, there are eight in the C-centred structure, giving rise to the observed splitting in the
resonance at 16-18 ppm.
11
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
When the desolvated structure is reimmersed in DMF, the splittings of the spectral
resonances disappear and the spectrum is similar to that of the original STA-26(Zr) (Figure
7). This confirms that the I-centred structure is stabilised by the introduction of solvent
molecules residing in the pores. If these are removed either by continuous solvent exchange
and/or heating, one lattice moves with respect to each other to achieve a lower energy
position.
Figure 7 13C CP MAS NMR spectra of (a) as-synthesised STA-26(Zr), (b) STA-26(Zr)-C obtained
after prolonged solvent exchange with acetone and (c) after reimmersion in DMF.
It is notable that as-synthesised materials showing PXRD peaks of relatively low intensity
13
violating the I-centring also show splittings in the C CP MAS NMR spectrum, although
these are not as large as when washed and dried at 80 °C (Figure S13). It is likely that this is
12
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
a result of intermediate stages of framework movement where the structure contains different
levels of DMF.
Furthermore, the effect of immersion of the C-form of STA-26 in other solvent molecules
(Figure S14) showed that while DMF caused full return to I symmetry, tetrahydrofuran had
no effect while the bulkier N,N-dibutyl- and N,N-diethylformamide strongly reduced the
intensities of reflections attributed to the ‘off-centring’ of the second lattice. This indicates
solvent polarity plays a controlling role in this effect.
VT-PXRD of STA-26(Zr)-C showed that the lattice-shifted structure is stable and no further
symmetry changes occur up to 200 °C (Figure 8). The overall crystallinity of the structure is
retained after the experiment and when cooled down to room temperature, and there is now
very close agreement between observed solvent-free STA-26(Zr)-C and the pattern simulated
from the crystal structure.
Figure 8. VT-PXRD patterns of STA-26(Zr)-C from 25 °C to 200 °C and then cooled down to 50 °C
and 25 °C with simulated pattern of STA-26(Zr)-C.
13
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
TGA of as-prepared samples of STA-26(Zr) (Figure S15) indicates an initial weight loss of
8% corresponding to the removal of adsorbed volatile solvent molecules residing in the pores
and another loss of 8% around 150-300 °C, which can be attributed to the dehydroxylation of
the Zr6 clusters and removal of residual DMF. The structure decomposes above 400 °C.
Figure 9 shows the IR spectra of STA-26, as-prepared and evacuated at room temperature.
The two strong bands at 1600 and 1406 cm−1 are assigned to coordinated carboxylates.[21]
The as-prepared sample is characterized by a broad band at 3400 cm−1 due to adsorbed water.
Evacuation at RT removes this band completely, indicating weak water adsorption.
Therefore, the sample treated in this way has few strong adsorption sites such as co-
ordinately unsaturated Zr cations.
(OH)
756 -
(OH)
B
A
0.04
0.05
- 3674
RT
0.1
50oC
75oC
3720 3660 3600
3400
770
700
adsorbed H2O
as-prepared
100oC
125oC
DMF
acid
150oC
175oC
200oC
RT-outgassed
0.2
1658
1500
1736
2000
4000
3500
3000
2500
1000
500
3700 3650 3600
Wavenumber, cm-1
Wavenumber, cm-1
Figure 9. (A) IR spectra of STA-26(Zr) as-prepared (pink) and evacuated at room temperature
(violet) and (B) IR spectra showing the dehydroxylation of STA-26(Zr).
Water desorption leads to development of a sharp band in the OH stretch region at 3674 cm−1
due to isolated OH groups attached to zirconium oxide clusters.[21-23] For the related Zr MOF
14
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
NU-1000, this band was attributed to include vibrations of terminal and bridging OH
groups.[23] Surprisingly, three deformation OH bands at 745, 729 and 711 cm−1 also appeared,
but deuteration of the sample (see Figure S16) produced one OD deformation band. This
revealed that the three OH bands are not indicative of heterogeneity of OH groups but rather
are a consequence of a spectroscopic effect arising from mixing of OH deformation modes
with other MOF vibrations. These OH groups are of low acidity as the water H-bonded to
them can be easily removed. A weak and broad feature around 3645 cm−1 is probably due to a
small fraction of weakly H-bonded hydroxyls OH groups interacting with residual DMF
(1659 cm−1) or free acid (1736 cm−1).
The stability of intra-framework hydroxyls was studied upon stepwise heating in vacuum.
The dehydroxylation of STA-26(Zr) sample starts at 125 oC, proceeds gradually with heating,
o
and is completed at 200 C (Figure 9). This process was found to be reversible. The
dehydroxylation route can be described as condensation of hydroxyl groups: 2OH  O2- +
H2O. This is in accordance with the TGA of STA-26(Zr). The picture is similar to that
observed for NU-1200 and is attributed to the loss of water and hydroxyl groups from the
Zr6O8 clusters.[10,11]
To investigate changes in the internal surface chemistry of STA-26(Zr) upon heating, CO,
CD3CN and CO2 were used as probe molecules in an extended IR spectroscopic study. The
adsorption of probe molecules was examined for both hydroxylated and dehydroxylated
forms of STA-26(Zr).
Surface acidity was assessed by adsorption of CO and CD3CN at cryogenic and ambient
temperature, respectively (Figure 10). With the hydroxylated sample both probes reveal the
existence of weak Brønsted acidity which is typical of MOF materials.[24,25] Adsorption of
CO leads to development of an OH···CO band at 2149 cm-1 (Figure 10A). Simultaneously, a
15
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
red shift of the OH stretching band of 68 cm−1 is detected (note that the largest such shift
reported with MOF materials is 90 cm−1 [26]). The OH···NCCD3 complexes were detected at
2273 cm-1.
The bands associated with OH···CO and OH···NCCD3 species appeared with strongly
reduced intensities on the dehydroxylated sample. However, in this case, bands indicative of
the presence of open Zr4+ sites were detected (Zr4+···CO at 2170 cm−1 and Zr4+···NCCD3 at
2300 cm−1, see Figs. 10B and D, respectively). Note that, as a strong base, CD3CN can
displace water adsorbed on Lewis acid sites and can thus detect some open Zr4+ sites that are
not visible by CO as a probe. However, it appears that the concentration of such sites is
negligible, as indicated by the very weak feature around 2300 cm-1 in Figure 10C. These
results rule out the existence in our sample of a large fraction of clusters terminated by water
ligands only.
16
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
2200
0.1
2150
2100
2320
0.05
2280
2240
A
C
- 2262
...
OH CO
OH···NCCD3
Phys.
ads. CO
Phys. ads.
CD CN
3
2133
2130
B
D
0.1
- 2262
0.05
Zr4+···NCCD3
...
OH CO
Phys.
ads. CO
Phys. ads.
4+...
Zr
CO
CD CN
3
2200
2150
2100
2320
2280
2240
Wavenumber, cm-1
Wavenumber, cm-1
Figure 10. (A, B) IR spectra of CO adsorbed at 100 K at an initial pressure of 5 mbar, followed by
gradual evacuation (blue to green spectra) and (C, D) CD3CN adsorbed at room temperature at an
initial pressure of 10 mbar followed by evacuation (blue to green), on STA-26(Zr) evacuated at room
temperature (upper panels) and at 175C (bottom panels).
Spectra of adsorbed CO2 were acquired at room temperature with CO2 equilibrium pressure
varied in the 1 – 50 mbar range. The as(CO2) bands obtained with the dеhydroxylated STA-
26(Zr) sample (Figure 11B) are less than one-half the intensity of the hydroxylated material
(Figure 11A).This indicates that the OH groups can play an important role as adsorption sites
for CO2. The second derivatives of the spectra (shown in the left insets) reveal that the main
band consists of two components. The high frequency component can be related to CO2
adsorbed on structural OH groups (OH-OCO) because (i) it is hardly observed with the
17
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
highly dehydroxylated sample and (ii) it correlates with the shift of the OH stretching band
(see the right inset in Figure 11A). The low frequency component can be attributed to weakly
and non-specifically adsorbed CO2.
- 3645.3
 = 28.1
0.005
2339.5 -
2336.5 -
0.01
0.02
3720 3660 3600
0.02
3673.4 -
- 2335.2
2340 2330
2340 2330
3720 3660 3600
50 mbar CO2
20 mbar CO2
10 mbar CO2
5 mbar CO2
1 mbar CO2
A
B
0.04
0.04
2380 2360 2340 2320 2300
Wavenumber, cm-1
2380 2360 2340 2320 2300
Wavenumber, cm-1
Figure 11. (A) IR spectra of CO2 adsorption of STA-26(Zr) evacuated at room temperature and (B)
at 150 C.
The pore structure of STA-26 was probed by N2 and CO2 adsorption measurements, after an
activation protocol that, according to our IR and PXRD studies, dehydroxylates the clusters
and results in conversion to the STA-26(Zr)-C form. N2 adsorption at 77 K gives a Type I
isotherm typical of a microporous solid (Figure 12, left). This contrasts strongly with that
reported for NU-1200[8] and the isotherm measured for the NU-1200 prepared in this work
(Figure S17), both of which show a Type IV isotherm typical of a mesoporous structure.
18
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Figure 12. (Left) N2 adsorption isotherm at 77 K and (right) CO2 adsorption isotherm at 196 K
measured for STA-26(Zr) activated at 150 °C for 16 h (closed symbols), with desorption (open
symbols).
Simulated nitrogen adsorption isotherms in STA-26(Zr)-C and NU-1200 (Figure 13) show
reasonable agreement with the experimentally measured isotherms, although the total pore
volume is higher than that measured, presumably because not all the pore space is accessible,
or there is some loss of structure. For STA-26, around 75% of the calculated value is
observed. Note that the lattice shift (I to C transition) has only a very minor effect on the
simulated adsorption isotherm.
Figure 13. Simulated N2 adsorption isotherms for STA-26(Zr)-I, STA-26(Zr)-C and NU-
1200 at 77 K.
19
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
These results emphasise that, while non-interpenetrated NU-1200 is mesoporous, with pore
volume deriving both from the space within the sod cages and the mesoporous space between
them, in STA-26 the mesoporous cavities of one the framework are occupied by sod cages of
the other, giving a microporous material. This confirms our interpretation of the single crystal
XRD, which might without the N2 adsorption data also be interpreted as a partially
interpenetrated structure, with each of the lattices having the same fractional occupancy. Such
a structure would possess some mesoporosity and so show an inflexion in the N2 adsorption
isotherm.
CO2 adsorption measured at 196 K approaches 300 cm3 STP g−1 , 1 bar (Figure 12 right), and
at 298 K reaches 160 cm3 STP g−1 at 10 bar (Figure S18), in each case showing an inflection
at around 80 cm3 STP g−1. GCMC simulations of the adsorption isotherms of CO2 on STA-
26(Zr) and NU-1200 at 196 K are shown in Figure S19. Comparison with the observed
isotherm confirms that STA-26 is fully interpenetrated and microporous. The simulations also
yield detailed information about the guest-framework interaction and the location of the guest
molecules. Figure 14 shows histograms of the simulated CO2 / MOF interaction energies at 1
bar. While NU-1200 has some low energy sites (i.e. strong interaction sites) mainly around
the Zr-nodes of the framework (Figure 14d), the adsorption behaviour is dominated by weak
adsorption sites (interaction energies ~ −2 kJ mol−1) in the centre of the octahedral pores. In
STA-26(Zr)-C the Zr-clusters act as strong adsorption sites, and furthermore interpenetration
in STA-26(Zr)-C results in smaller pore sizes and stronger adsorption sites (Figure 14a and
c). Simulated adsorption surfaces of CO2 at 0.01 bar on STA-26(Zr)-I and C show that the
lattice shift gives rise to an additional adsorption site in the void between phenyl groups of
neighbouring interpenetrating nets (Figure S20).
20
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Figure 14. Energy density distributions of CO2/MOF interactions in (a) STA-26(Zr)-C and (b) NU-
1200 at 196 K and 1 bar. (c) and (d) Surface plots of the CO2/MOF interaction energy, where blue = 0
kJ mol-1 and red = -30 kJ mol-1.
To demonstrate the difference in window size between microporous STA-26(Zr) and
mesoporous NU-1200, dye uptake and catalytic behaviour of the MOF were also studied.
Following the previous dye uptake studies performed by Lin et al.,[27] NU-1200 and STA-26
were kept in ethanolic solutions of Fuchsin basic dye for 10 h, followed by subsequent
filtration and multiple washes with ethanol. Mesoporous pore openings enable NU-1200 to
achieve higher levels of dye uptake in comparison with STA-26(Zr), as clearly indicated by
the difference in intensity of colour (Figure 15).
Figure 15. NU-1200 and STA-26(Zr) after exposure to a solution of Fuchsin basic dye (right).
21
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Furthermore, in the Lewis acid-catalysed condensation of 4’-fluoroacetophenone and
benzylamine (Scheme S1), STA-26(Zr) gave only 28% conversion towards the imine, in
comparison with 51% for the NU-1200 (Figure S21). As both MOFs possess similar Zr6O8
clusters, the higher conversion rates of NU-1200 are attributed to the free movement of the
large reactant and product molecules within its mesopores, while the lower observed
conversion over microporous STA-26 could result from surface catalysis.
Thermal and water stability of MOFs are two major concerns that limit their application.
While the thermal stability has been steadily improved[7, 28] the water instability of MOFs
remains a significant limitation. One cause of instability, giving rise to extremely low surface
areas after activation from water, is the high capillary forces acting on the walls of pores
containing water.[29] Consequently, MOFs are best activated after being kept in low boiling
point solvents or in supercritical CO2.[30] As interpenetration has been reported to increase the
overall stability of MOFs,[13] we tested the water stability of STA-26(Zr). Following the
protocol by Farha et al., as-prepared STA-26(Zr) was kept in water for 48 h and subsequently
activated at 100 C.[31] PXRD showed peaks corresponding to change in symmetry and can
be indexed to STA-26(Zr) showing the I to C symmetry change. The overall crystallinity of
the MOF was well-preserved under these conditions (Figure 16). By contrast, when NU-1200
was kept in water and activated under similar conditions, some broadening of reflections
occurred. Furthermore, at 77 K, the N2 adsorption isotherm of NU-1200 activated directly
from water showed significantly lower porosity, while STA-26(Zr) retained its porous nature.
This indicates that the interpenetrated geometry of STA-26 enhances its stability to water.
22
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Figure 16. Left, (a) Simulated PXRD pattern for STA-26(Zr)-C, (b) as-prepared STA-26(Zr), (c)
STA-26(Zr) after 48 h water stability test and direct activation from water. Inset: N2 isotherm of STA-
26(Zr) activated at 150 °C after solvent exchange with acetone (black) and directly activated from
water (red). Right, (a) simulated PXRD pattern for NU-1200, (b) as-prepared NU-1200, (c) NU-1200
after water stability test and direct activation from water. Inset: N2 isotherm of NU-1200 activated at
150 °C after solvent exchange with acetone (black) and directly activated from water (red).
Conclusions
A route has been devised to a fully interpenetrated version of the Zr-based MOF, NU-1200,
which we denote STA-26(Zr). The Hf-analogue of STA-26 has also been prepared. As in
NU-1200, the TMTB linker in STA-26 shows the characteristic geometry, where the presence
of the methyl groups on the central benzene ring of the linker results in rotation of the
benzoate groups and consequently favours the the topology type with eight-connected Zr6O8
clusters. In the STA-26 structure, two lattices with the the topology type interpenetrate.
Whether the interpenetrated or non-interpenetrated form crystallises depends on the
modulating agent used in our syntheses. Use of benzoic acid gives NU-1200 whereas in the
presence of formic acid the interpenetrated structure results. It is likely that during
23
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
crystallisation the observed presence of benzoate groups bound to the clusters sterically
inhibits the growth of a second, interpenetrating, lattice.
Upon removal of solvent from STA-26, there is a shift in the position of one interpenetrating
lattice with respect to the other, resulting in a symmetry change from Im−3m to Cmcm, which
is reversible upon reimmersion in DMF (the synthesis solvent). The post-synthetic structural
13
changes are observed not only by X-ray diffraction, but also, remarkably, by solid-state C
NMR, which shows peak splittings as the symmetry changes. Heating at 125 °C and above
results in dehydroxylation of the cluster and IR spectroscopy using Lewis basic probe
molecules reveals the presence of Lewis acid sites, likely to be coordinatively unsaturated
Zr4+ cations.
Activated samples of STA-26 give N2 adsorption 77 K isotherms of Type I, confirming the
microporous nature of the STA-26 structure, compared to the Type IV isotherm of the
mesoporous non-interpenetrated analogue NU-1200. CO2 adsorption shows gradual pore
filling at 196 K up to ca. 0.2 bar, which is well described by grand canonical Monte Carlo
simulations of adsorption on the microporous interpenetrated structure. While the
microporosity and especially the reduced window size of ca. 5 Å will make STA-26
unfavourable for the adsorption and catalytic transformations of large molecules, it possesses
good porosity for N2 and CO2 even after direct activation from water, demonstrating that it is
a stable and size-selective adsorbent.
24
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
Experimental Section
Materials and Methods The synthesis of TMTB, 4,4’,4’’-(2,4,6-trimethylbenzene-1,3,5-
triyl)tribenzoic acid, was performed by modifying previously reported methods.[8,9,32]
A
detailed experimental procedure is provided in the Supporting Information. NU-1200(Zr)
was synthesised by modifying the published work by Liu et al.[8] ZrCl4 (48 mg, 0.20 mmol),
TMTB (40 mg, 0.08 mmol) and benzoic acid/2-fluorobenzoic acid (1.40/1.61 g, 11.5 mmol)
were ultrasonically dissolved in DMF (8 mL) in a pyrex vial. The solution was heated at 120
°C for 48 h in an oven before being cooled to room temperature. The supernatant solution
was extracted with a syringe and the resulting white solid was washed multiple times with
DMF followed by acetone and dried in an oven at 80 °C (26.0 mg). Single crystals of STA-
26(Zr) were prepared using a reaction procedure similar to that of Wang et al for the
synthesis of BUT-12.[9] ZrCl4 (48 mg, 0.20 mmol) and TMTB (40 mg, 0.08 mmol) were
ultrasonically dissolved in a solvent mixture of DMF (8 mL) and formic acid (8 mL) in a 20
mL Pyrex vial. The resulting solution was gradually heated to 120 °C with a ramp rate of 3
°C min−1 and kept there for 48 h before being cooled to room temperature. The supernatant
solution was removed and the colourless crystals obtained were washed multiple times with
DMF followed by acetone, then dried in an oven at 80 °C (35.9 mg).
For the synthesis of single phase STA-26(Zr), ZrCl4 (48 mg, 0.2 mmol) and TMTB (40 mg,
0.08 mmol) were ultrasonically dissolved in a solvent mixture of DMF (10 mL) and formic
acid (6 mL) in a 20 mL Pyrex vial. The resulting solution was gradually heated to 120 °C
with a ramp rate of 3 °C min−1 and kept there for 48 h before being cooled to room
temperature. The supernatant solution was removed and the resulting white solid was washed
multiple times DMF followed by acetone, then dried in an oven at 80 °C (53.0 mg). The
related STA-26(Hf) was synthesised using HfCl4 (64.1 mg, 0.20 mmol) and TMTB (40 mg,
0.08 mmol), ultrasonically dissolved in a solvent mixture of DMF (10 mL) and formic acid (6
25
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
mL) in a Pyrex vial. The resulting solution was gradually heated to 120 °C with a ramp rate
of 3 °C min−1 and kept there for 48 h before being cooled to room temperature. The
supernatant solution was removed and the white solid product was washed multiple times
with DMF followed by acetone and dried in an oven at 80 °C (55.3 mg).
Materials and methods Powder X-ray diffraction (PXRD) patterns of finely ground powder
were collected in Debye-Scherrer geometry from Stoe STAD i/p diffractometers with
primary monochromation (Cu Kα1, λ = 1.54056 Å), using 0.7 mm glass capillaries. Variable
temperature PXRD (VT-PXRD) measurements were performed in air, with a ramp rate of 5
°C min−1 and held at the desired temperature for 10 min prior to measurements. Data were
collected from 25 to 200 °C and after cooling down to 25 °C for STA-26(Zr) and STA-
26(Zr)-C, using a Cobra Plus non-liquid-nitrogen cryostream (Oxford Cryosystems).
For solution-state nuclear magnetic resonance (NMR) spectra of the MOF samples, the
protocol developed by Lin et. al[33] was used. Prior to analysis, 10 mg of MOF in DMSO-d6
(0.5 mL) was added to a saturated solution of K3PO4 in D2O (0.5 mL) and stirred at room
temperature until the MOF was completely dissolved.
Solid-state NMR spectra were recorded using a Bruker Avance III spectrometer equipped
1
with a 9.4 T superconducting magnet (Larmor frequencies of 400.1 MHz for H and 100.6
MHz for 13C). For the 13C NMR spectra, samples were packed into standard ZrO2 rotors with
an outer diameter of 4 mm and rotated at the magic angle at a rate of 12.5 kHz. Spectra were
1
1
recorded with cross polarisation (CP) from H with a contact pulse (ramped for H) of 2 ms.
Signal averaging was carried out for between 1024 and 19968 transients with a recycle
1
interval of 3 s. Two-pulse phase modulation (TPPM) decoupling of H (ν1  100 kHz) was
1
carried out during acquisition. For the H NMR spectra, samples were packed into standard
26
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
ZrO2 rotors with an outer diameter of 1.9 mm and rotated at the magic angle at a rate of 40
kHz. Spectra were recorded using a rotor-synchronised spin-echo pulse sequence with an
echo delay of 25 μs. Signal averaging was carried out for 64 or 128 transients with a recycle
interval of 3 s. Chemical shifts are reported in ppm relative to TMS, using L-alanine as a
secondary solid reference (CH3 δ = 20.5 ppm and NH3 δ = 8.5 ppm).
Thermogravimetric analysis (TGA) of all samples were carried out on a Netzsch TGA 760
for a temperature range of 20 - 900 °C at a heating rate of 5 °C min−1 in a continuous air
flow. N2 adsorption isotherms for STA-26(Zr) and NU-1200 samples were measured
volumetrically on a Micrometrics Tristar. The CO2 adsorptions isotherms were measured
gravimetrically on Hiden IGA porosimeter. All samples were activated at 150 °C for 16 h
prior to measurements unless otherwise stated.
The IR spectra of STA-26(Zr) on KBr pellet were recorded using a Nicolet Avatar 360
spectrometer (DTGS detector) and the spectra of a self-supporting STA-26(Zr) pellet were
recorded using a Nicolet 6700 FTIR spectrometer (MCT detector). During the experiments,
up to 128 scans were accumulated at a spectral resolution of 2 cm−1 with an accuracy of 0.01
cm−1. Specially designed IR cells were used for the experiments. One cell, equipped with
CaF2 windows, permitted the measurement of spectra at low (100 K) and ambient
temperatures. Another, equipped with KBr windows, allowed the measurement of spectra at a
pre-set temperature in the range between 100 and 300 K. The cells were directly connected to
a vacuum-adsorption apparatus with a residual pressure lower than 10−3 Pa.
For catalytic studies, both NU-1200 and STA-26(Zr) were activated at 150 °C for 16 h under
vacuum. In the test imine formation, 4’-fluoroacetophenone (1 mmol) and benzylamine (1
mmol) in dry, degassed toluene (5 mL) were added to the activated MOF (2 mol%, 42 mg)
27
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
and heated with stirring at 90 °C for 24 h under argon. After completion of the reaction, the
conversion rates were determined by 19 F NMR.
Crystal Structure determination
Single crystal X-ray diffraction (SCXRD) data for body centred STA-26(Zr), here STA-
26(Zr)-I, and STA-26(Zr)-C were collected at 173 K using a Rigaku MM-007HF High
brilliance RA generator/confocal optics and Rigaku XtaLAB P100 system, with Cu Kα
radiation (λ = 1.54187 Å). Intensity data for all compounds were collected using both ω and
φ steps, accumulating area detector images spanning at least a hemisphere of reciprocal
space. All data were corrected for Lorentz polarization effects. A multiscan absorption
correction was applied by using CrysAlisPro.[34] Structures were solved by dual-space
methods (SHELXT)[35] and refined by full-matrix least-squares against F2 (SHELXL-
2016/6).[36] Non-hydrogen atoms were refined anisotropically, and alkyl and aryl hydrogen
atoms were refined using a riding model. Water hydrogens in STA-26(Zr) were located from
the difference map and refined subject to distance and angle restraints. Similar water
hydrogens could not be located for STA-26(Zr)-C. All calculations were performed using the
CrystalStructure interface.[37] The data for STA-26(Zr)-C showed non-merohedral twinning
(twin law - 0.503 - 0.5 1.014 - 0.487 - 0.5 - 1.014 0.487 - 0.5 0.014, twin fraction 15.5 %),
which was accounted for in the refinement, and also showed signs that several further minor
twins might also be present; although attempts to refine the data taking these into account did
not improve structure quality. Both structures showed large voids, 12119 Å3 for STA-26(Zr)-
I (54.2 % of unit cell volume) and 25893 Å3 for STA-26(Zr)-C (57.6 % of unit cell volume)
and the SQUEEZE[38] routine implemented in PLATON[39] was used to remove the
contribution of the disordered electron density in the void spaces. The crystallographic data
(excluding structure factors) of the structures reported in this paper have been deposited into
28
This article is protected by copyright. All rights reserved.

10.1002/chem.201705136
Chemistry - A European Journal
the Cambridge Crystallographic Data Centre with CCDC Numbers 1571655 for STA-26(Zr)-
I and 1571656 for STA-26(Zr)-C.
Simulation details
Gas adsorption was simulated using grand canonical Monte Carlo (GCMC) simulations,
implemented in the multipurpose code MuSiC.[40] Framework structures were taken from
measured or reported crystallographic cif files. The structures were assumed to have
dehydrated Zr6O8 cores. In addition, disordered H-atoms on the CH3 groups were ordered for
the initial optimisation of the structures. This lowered the symmetry of STA-26(Zr) from Im-
3m to I432 and STA-26(Zr)-C from Cmcm to C222. Due to the μVT ensemble imposed in
GCMC the framework atoms were fixed in the optimised positions. For each gas (CO2 and
N2) at each pressure, 1x109 Monte Carlo steps were performed where each step consists of
either a random translation, insertion or deletion, and random rotation – all equally weighted.
The first 40 % of the steps were used for equilibration and the remaining used to calculate the
ensemble averages. Standard Lennard-Jones (LJ) potentials were used to model the dispersive
interactions between the framework and gases. The Trappe force field was used to model
CO2, and N2.[41] Coulombic interactions were included and calculated using Wolf Coulombic
summations which are much more efficient than Ewald summations.[42] Interactions beyond
18 Å were neglected. To calculate the gas-phase fugacity the Peng–Robinson equation of
state was used.[43]
Acknowledgements We acknowledge the support of the EPSRC/St Andrews Criticat
CDT (RRRP, PAW) and the European Community Seventh Framework Program (FP7/2007-
2013 number 608490] (project M4CO2) (KKC, MYM, KIH, PAW). SEA would like to thank
29
This article is protected by copyright. All rights reserved.