Introduction
Fluorine is the ninth element in the periodic table. It has one stable and NMR-active
isotope (19 F), the highest electronegativity (3.98 on the Pauling scale), and the third-shortest
van der Waals radius of 1.47 Å, longer only than those of H (1.10 Å) and He (1.40
Å).[1 ] The C–F bonds are among the strongest in organic chemistry, with bond dissociation
energies as high as 130 kcal mol−1 , and an average length of 1.35 Å. Introduction of fluorine into the structures of
organic compounds, and especially the formal replacement of C–H with C–F bonds, has
profound effects on the structures and reactivities of said compounds. Since fluorine
is only a bit larger than hydrogen, such substitution changes the sterics minimally,
but the electronics profoundly. This feature led to an explosion of interest in the
incorporation of fluorine into pharmaceuticals,[2 ] and the creation of numerous modern methods for the functionalization of C–H bonds
with fluorine atoms or fluorine-containing functional groups.[3 ]
[4 ] Materials chemistry has also been exploring fluorination as a method of modulating
chemical and thermal stability, as well as physical properties, in polymers and other
classes of materials.[5 ]
[6 ]
Zhenglin Zhang (left) is from Ganzhou in China. He obtained his B.Sc. in Applied Chemistry (2011)
from Liaoning Shihua University and a M.Sc. in Chemical Engineering and Technology
(2014) from Beijing University of Technology. He has been working in Miljanic's group
at the University of Houston since October 2015. Currently, he is a Ph.D. candidate
with research interests in fluorescent and porous materials.
Ognjen Š. Miljanic (right) was born in Belgrade, then Yugoslavia, in 1978. He holds a Diploma from the
University of Belgrade (2000) and a Ph.D. from the University of California at Berkeley
(2005), where he worked with Prof. K. Peter C. Vollhardt. Following a postdoctoral
stay in Prof. J. Fraser Stoddart's group at UCLA (2005-2008), Ognjen joined the University
of Houston (UH) in 2008 as an Assistant Professor. He was promoted to Professor in
2019. He held visiting professorships at New York University Abu Dhabi and Ruprecht-Karls-Universität
in Heidelberg. Among other distinctions, he is the recipient of UH Teaching & Research
Excellence, NSF CAREER, and Cottrell Scholar awards, and a Fellow of the Royal Society
of Chemistry.
In this minireview, we will summarize the recent advances in the preparation and utilization
of fluorinated porous materials. Our text will be divided into four sections, discussing
the effects of fluorination on the structures and properties of (1) metal–organic
frameworks (MOFs), (2) covalent organic frameworks (COFs) and porous organic polymers
(POPs), (3) extrinsically porous molecular crystals (PMCs), and (4) intrinsically
porous molecular cages. These classes of materials are distinguished by the connections
between their building blocks, which are also responsible for the assembly of porous
solid-state structures. In MOFs, they are metal–ligand bonds; in crystalline COFs
and amorphous POPs, they are covalent bonds between organic fragments; in PMCs, weak
noncovalent interactions. Finally, porous molecular cages are constructed from covalently
connected elements that create a pore within a molecule. Cages then bring this pore
with them into the solid-state structure, in which noncovalent interactions between
the individual cages can vary in strength and directionality. We will not strive for
a comprehensive treatment of fluorinated MOFs in this short account. Instead, we will
focus on offering a personal perspective and—to the extent primary literature permits—the
comparison of properties of fluorinated porous materials with those of their nonfluorinated
counterparts. The reader is referred to two reviews on fluorinated MOFs,[7 ]
[8 ] as well as to numerous other general reviews of MOFs,[9 ]
[10 ]
[11 ]
[12 ] COFs,[13 ]
[14 ]
[15 ]
[16 ] and PMCs.[17 ]
[18 ]
[19 ]
[20 ]
[21 ]
[22 ]
Metal–Organic Frameworks
Fluorinated MOFs have been the subject of study since the mid-2000s. Early work in
this area focused on the hydrogen adsorption behavior in porous materials with exposed
fluorine atoms;[23 ]
[24 ] since then, other intriguing properties of this class of MOFs have been identified.
Particularly notable were the studies of Omary and co-workers, who have shown that
MOFs based on fluorinated triazolate ligands can be effective in hydrocarbon adsorption,
while adsorbing essentially no water even at 90% relative humidity.[25 ]
[26 ] This combination of properties was proposed as useful in the cleanup of oil spills
and hydrocarbon storage, and it suggested that the preparation and study of fluorinated
MOFs are very warranted. At the same time, true structure–property studies of fluorinated
MOFs were severely restricted by the paucity of the appropriately functionalized large
fluorinated ligands.
Our entry into this area was inspired by the desire to expand the range of organic
building blocks for use in MOF chemistry. To do so, we initiated a collaboration with
the group of Prof. Olafs Daugulis at the University of Houston. Initially, a series
of linear highly fluorinated carboxylic acids and tetrazoles was prepared using a
combination of Cu- and Pd-catalyzed chemistry. Our first joint publication[27 ] reported several MOFs derived from these ligands and two main conclusions: (1) MOFs
prepared from fluorinated linkers were superhydrophobic, but (2) were also thermally
and hydrolytically less stable than their nonfluorinated counterparts. This lowered
stability, especially in the case of carboxylate-based fluorinated MOFs, could tentatively
be explained by (a ) easier decarboxylation, which would result in the creation of aromatic anions stabilized
by the electron-withdrawing fluorine atoms, and (b ) lowered basicity of fluorinated carboxylates, which made them weaker ligands for
the metals.
By switching from linear to trigonal geometry and through the replacement of carboxylate
ligands with tetrazolates, more porous and more robust fluorinated MOFs could be prepared.
A material named MOFF-5 was prepared from trigonal linker 1 by reaction with CuCl2 ·H2 O ([Figure 1 ]).[28 ] Its stability was sufficient to allow full characterization and revealed a surface
area of ∼2400 m2 g−1 —the highest among the fluorinated MOFs at the time. This material has shown high
gravimetric adsorption capacities for a variety of fluorinated guests, including fluorocarbons,
hydrofluorocarbons (HFCs), and chlorofluorocarbons, reaching as much as 225 wt% for
perfluorohexane. These results have broader relevance in the improvement of efficiency
of air conditioners,[29 ] and the lowering of greenhouse emissions coming from HFCs, which are hundreds of
times more potent greenhouse gases than carbon dioxide.
Figure 1 Synthesis of a fluorinated metal–organic framework MOFF-5 . Crystal structure of MOFF-5 shown on the right, with element colors: C—gray, O—red, N—blue, F—lime, Cl—green,
Cu—cyan.
However, the stability of the tetrazolate-based framework was still not perfect: after
five or six sorption/desorption cycles it would start losing its crystallinity and
eventually completely decompose. Our attempt to solve this problem led us into the
field of PMCs, as will be described below.
Covalent Organic Frameworks and Porous Organic Polymers
Examples of fluorinated COFs in the literature are relatively rare, but several important
side-by-side comparisons have been made. Sun and coworkers prepared three isostructural
imine-based COFs from tetraphenylbenzene tetraaldehydes which were disubstituted with
hydrogen, methoxy, or fluoro groups.[30 ] Their structures were solved using continuous rotation electron diffraction and
were found to be five-fold interpenetrated and very similar to each other. The fluorinated
material had a much higher affinity for CO2 than the nonfluorinated framework, as well as a higher (50:1) IAST (ideal adsorbed
solution theory) selectivity for CO2 over N2 . In three separate studies, imine-[31 ]
[32 ] and azine-linked[33 ] COFs (example in [Figure 2 ], left) have been shown to organize better and lead to more crystalline materials
if fluorine atoms are included into the structures of their monomers. Evidence of
this improved crystallinity in fluorinated materials was found in the higher resolution
of their powder X-ray diffraction patterns and the clear observation of crystallites
by scanning electron microscopy. In the case of azine-linked COFs, an explanation
for the observed improvements in both crystallinity and porosity was that the fluorinated
layers aligned well with nonfluorinated ones through aromatic stacking, thus fixing
the interlayer arrangement in place and increasing order in the materials.[34 ] Nevertheless, single-crystalline fluorinated COFs have not been reported yet.
Figure 2 Examples of structural motifs found in fluorinated COFs (left), CTFs (top right),
and POPs (bottom right).
Wen and colleagues have shown that perfluorinated covalent triazine frameworks (CTFs,
example in [Figure 2 ], top right)[35 ] serve as excellent electrocatalysts for the conversion of CO2 into CH4 .[36 ] Postsynthetic introduction of sulfur functional groups into related CTFs could be
accomplished through an SN Ar reaction on the fluoro substituents; the resulting sulfur-containing CTFs were
of interest in lithium–sulfur batteries.[37 ]
[38 ]
[39 ] Fluorinated CTFs have also been utilized for selective and moisture-tolerant capture
of CO2 ,[40 ]
[41 ] again with high IAST selectivity over the adsorption of nitrogen.[42 ] A very recent study has identified a CTF based on fluorinated aryl ethers which
showed a CO2 uptake capacity of 6.58 mmol g−1 (29.0 wt%) at 273 K and 1 bar—among the highest values reported for POPs and related
materials to date.[43 ] This high sorption capacity was attributed to the to the presence of ultra-micropores
of 0.6–0.7 nm in size and strong electrostatic interactions with the residual fluorine
atoms within the framework. Molecular simulation suggested that CTF with F content
of ∼4.8 wt% and pore size distribution around ∼0.7 nm should have the highest CO2 uptake capacity, in agreement with the previous studies.[44 ] Studies of mechanical stiffness of CTFs substituted with trifluoromethyl groups
revealed behavior similar to glass polymers and zeolitic imidazolate frameworks.[45 ]
Qi and coworkers have prepared fluorinated POPs ([Figure 2 ], bottom right) by Pd-catalyzed C–H activation of 1,2,4,5-tetrafluorobenzene and
its subsequent coupling with trigonal and tetrahedral aryl bromides.[46 ] Using iron-porphyrin as a coupling partner, a series of fluorinated POPs was prepared
with Brunauer–Emmett–Teller (BET) surface areas ranging between 670 and 840 m2 g−1 .[47 ] These materials were shown to be competent heterogeneous catalysts for the Baeyer–Villiger
oxidation of ketones, and the fluorinated groups were speculated to help in increasing
the lifetime of the catalyst. Cooper and colleagues used a Sonogashira coupling of
1,3,5-triethynylbenzene with several extensively fluorinated aryl bromides to prepare
POPs.[48 ] In both Han's and Cooper's work, fluorination played a crucial role in the synthesis:
the C–H functionalization would not have proceeded on the nonfluorinated material,
nor would aryl bromides be active enough in the Sonogashira coupling without fluorine
substituents. In addition, fluorination influenced the properties of the prepared
materials: they were found to be more hydrophobic than their nonfluorinated counterparts,
and some showed high sorption capacities for electron-rich aromatics such as toluene.
Variants of Cooper et al.'s alkyne-based fluorinated POPs have been shown to be excellent
adsorbents for CO2 and CH4 (showing ∼100:1 and 8:1 selectivities over N2 , respectively),[49 ] as well as elemental iodine.[50 ] Fluorinated POPs incorporating alkyne functional groups and phenanthroline moieties
could coordinate Ag(I) and the resulting porous framework was capable of catalyzing
the reaction of CO2 with propargyl alcohols.[51 ] Very recently, Yavuz and coworkers have shown that alkyne-based perfluorinated POPs
can be synthesized in the absence of transition metal catalysts.[52 ]
Dichtel and coworkers have made a brilliant use of aromatic fluorine chemistry to
prepare cyclodextrin-based POPs.[53 ] Starting with the activated tetrafluoroterephthalonitrile 3 ([Figure 3 ]), an SN Ar reaction with ß -cyclodextrin (2 ) resulted in the replacement of 2.1–2.2 fluorines (on average) per molecule of 3 with the alkoxy groups of cyclodextrin, creating a porous polymer 4 (its structure in [Figure 3 ] is simplified, as other connectivities are certainly possible). In it, rigid and
hydrophobic fluoroarene and cyclodextrin moieties played a role in the creation of
materials with surface areas as high as 263 m2 g−1 ,[54 ] which could be used to remove nonpolar organic micropollutants from drinking water.
These results were built on the previous work of Deng and coworkers who used fluorinated
alkyne-based POPs to remove hydrophobic and heavy metal pollutants from water.[55 ] Inclusion of cotton fibers during the polymerization creates materials that can
capture volatile organic compounds from air.[56 ] By using a related material derived from the reaction of ß -cyclodextrin with decafluorobiphenyl, the same group showed efficient removal of
the notoriously environmentally persistent perfluorooctanoic acid (PFOA, used as a
surfactant and in the production of Teflon and fluorotelomers).[57 ] In very recent work, they demonstrated that the reduction of the –CN group into
–CH2 NH2 dramatically enhances the affinity of the material for PFOA.[58 ] Work on tailoring porous organic materials to capture water contaminants has recently
been reviewed.[59 ]
Figure 3 Polymerization of ß-cyclodextrin with tetrafluoroterephthalonitrile results in POPs
which are capable of removing organic micropollutants from water. Polymerization product’s
structure is simplified.
Porous Molecular Crystals
In 2015, our group made a serendipitous discovery. While attempting to prepare pyrazolate-based
fluorinated ligands for MOFs—in the hope of enhancing their stability relative to
their tetrazolate counterparts, mentioned earlier—we obtained single crystals of the
organic starting material 5 ([Figure 4 ], top left). This result would have normally been very disappointing. However, a
more detailed analysis of the crystal structure revealed that the precursor pyrazole
itself creates a porous structure ([Figure 4 ], top right), held together by a combination of [N–H···N] hydrogen bonds between
terminal pyrazoles ([Figure 4 ], bottom left) and aromatic [π···π] stacking interactions between electron-rich pyrazoles
and electron-poor 1,2,4,5-tetrafluorobenzenes ([Figure 4 ], bottom right). This framework proved to be remarkably robust in light of being
stabilized only by noncovalent interactions: it did not decompose until ∼380 °C and
did not lose its structural integrity until ∼280 °C. At that temperature, our material
transformed into another crystalline phase (as evidenced by variable temperature powder
X-ray diffraction) which was not porous. This observation was in line with the demonstrated
polymorphism of PMCs, which typically sees porous phases being kinetic products.[60 ]
[61 ]
Figure 4 Fluorinated aromatic pyrazole 5 self-assembled into a porous structure (top right), through a combination of hydrogen
bonds between terminal pyrazoles (bottom left) and [π···π] stacking between electron-rich
pyrazoles and electron-poor tetrafluorobenzenes (bottom right).
In a subsequent exhaustive study,[62 ] we have prepared a series of derivatives of 5 , with the intention of identifying the molecular features necessary for the formation
of the porous structure. The tetrafluorinated ring proved to be critical, as its replacement
with a nonsubstituted benzene ring or a triple bond led to materials that could either
not be crystallized or that crystallized in nonporous structures. This study identified
three novel precursors to PMCs. They all shared certain structural features: trigonal
structures, perfluorinated rings capable of aromatic stacking, and a hydrogen-bonding
terminal group. Linear and bent precursors resulted in close-packed structures. In
the solid state, 5 and 6 were isostructural ([Figure 5 ], top); 7 , as would be expected, had an expanded unit cell ([Figure 5 ], bottom). Curiously, tetrazole 1 —which is a precursor to the highly porous MOFF-5 ([Figure 1 ])—also formed a porous solid-state structure on its own. The structure is different
from those of 5 –7 , in which the coplanar triplets of hydrogen bonds were replaced with infinite one-dimensional
(1D) chains of hydrogen-bonded tetrazoles ([Figure 6 ], left), resulting in a more compact organization ([Figure 6 ], right).
Figure 5 Trigonally shaped, fluorinated, and rigid trispyrazoles 6 and 7 self-assemble into porous structures isoreticular to that of 5 in the solid state.
Figure 6 Crystal structure of 1 shows infinite 1D chains of hydrogen-bonded tetrazoles (left) and is porous, but
more compact (right) than the structures of the closely related 5 and 6 .
The most extensively investigated PMC in our group was based on compound 5 . This material was found to be an excellent adsorbent for a variety of fluorinated
species—with capacities slightly lower than those observed for MOFF-5 , but with vastly improved stability. Compound 5 was also a competent adsorbent for a number of fluorinated anesthetics.[63 ] In addition, PMCs based on 5 proved to be piezochromic: upon guest entry into the pores, the pores would shrink
in the hydrogen-bonding plane, but expand along the [π···π] stacking direction.[64 ] This change, although small, was measurable and reproducible and constitutes a sensing
mechanism for detecting guest entry into the pores.
PMCs can be dissolved in solvents such as DMF, which disrupts the hydrogen bonding
between individual molecules. Their solubility allowed direct comparisons of optical
properties in the solution and in the solid state. In the case of isostructural 5 and 6 , solid-state absorption and fluorescence emissions were found to be virtually identical.
However, upon dissolution, the two compounds behaved differently: 5 was fluorescent in dilute DMF solution, while 6 was not.[65 ] With addition of water (which is a poor, aggregation-inducing solvent for both 5 and 6 ), the triazine-centered 6 exhibited aggregation-induced turning ON of emission (AIE effect), while the emission
of 5 stayed more or less the same. This difference in emission properties was explained
through the computational analysis of rotational barriers in 5 and 6 . Because there are no hydrogen atoms on the central triazine ring of 6 , rotation of the tetrafluorinated rings is relatively unrestricted; to become emissive,
the intramolecular rotations of 6 have to be limited through water-induced aggregation. On the other hand, the fluorinated
rings of 5 are not as flexible as those of 6 and thus remain in a close-to-emissive conformation regardless of the aggregation
stage. We have further shown that this switch of 6 from the freely rotating to the emissive conformation can be induced not only by
aggregation, but also by supramolecular assembly.[66 ] Namely, upon exposure to dicarboxylic acids with convergently positioned –COOH groups,
such as 1,2-benzendicarboxylic acids or cis -alkenediacids, two molecules of 6 would come together into a dimer which restricts nonradiative decay (intramolecular
rotations) and elicits emission. In a very recent work, we have examined a fluorinated
tetragonal precursor to PMCs based on the central tetraphenylene ethylene unit, which
is a well-documented AIEgen.[67 ] This material shows switchable emission in the solid state, which can be toggled
between green and blue by the removal and re-addition of the DMF solvent.
Hisaki and coworkers have recently incorporated ortho -fluorine substituents into the structures of their porous hydrogen-bonded networks
built on carboxylic acids.[68 ] Relative to the nonfluorinated system, fluorination resulted in the loss of coplanarity
between the benzene ring and the –COOH group (common to the behavior of fluorinated
MOFs) and significant rotational disordering of the –COOH groups.
Barbour and coworkers reported a porous halogen-bonded framework, which was obtained
by evaporating an acetone solution of 8 and 9 ([Figure 7 ]).[69 ] The evaporation produced a solvated yellow porous molecular crystal, wherein I···N
interactions connect 8 and 9 in a sinusoidal infinite (8 ···9 ···)
n
chain. Adjacent chains interact with each other by weak [C–H···F] contacts and [π···π]
stacking between tetrafluorodiiodobenzenes and thiophenes, finally forming a porous
structure with 1D guest-accessible channels. The weak intermolecular interactions
allow the pores to change their shape and volume in response to guest molecules. For
example, the voids are discrete when CO2 is introduced as the guest into the activated crystals of 8 ·9 but become 1D open channels when ethane is the guest. The volume per guest binding
site changes from 55 to 143 Å3 . The I···N interactions almost do not change, but the [C–H···F] interactions break
during such structure expansion.
Figure 7 Co-crystallization of compounds 8 and 9 (top) results in a porous framework consisting of discrete 1D halogen-bonded chains
(bottom, compound 9 shown in red). Flexible interactions between these chains allow their adjustment
to various guests.
As we have seen in our work, the flexibility of fluorinated PMCs is also reflected
in their fluorescence properties. Naka and coworkers prepared platinum dihalide complexes
which crystallize in low-porosity structures with BET surface areas of 20 and 42 m2 g−1 , respectively.[70 ] Such PMCs are characterized by [C···F] short contacts between two fluorinated benzenes
in edge-to-side mode and [π···π] stacking between the fluorinated and nonfluorinated
aromatic rings. On account of their flexible structures, these PMCs can be transformed
to nonporous crystals by exposure to CH2 Cl2 , which also results in the turning ON of red luminescence. Such change can be reversed
by exposing the nonporous crystals to hexafluorobenzene.
Zhang and coworkers synthesized the single crystals of benzo-trisimidazole 10a ([Figure 8 ], top) by the solvothermal reaction of 1,3,5-triamino-2,4,6-trinitrobenzene,[71 ] formic acid, and sodium formate at 160 °C. In the solid state, each molecule of
10a is connected to three adjacent molecules by six [N–H···N] bonds ([Figure 8 ], center), resulting in a two-dimensional extended framework without significant
porosity.[72 ] After the removal of trapped H2 O, the activated crystals of 10a kept their original crystal structure, where the isolated cavities were blocked by
two adjacent layers. Molecule 10b , functionalized with a CF3 group, formed a porous crystal with three-dimensional (3D) intersecting channels that
had increased volume (36%) and BET surface area (131 m2 g−1 ) compared to the parent molecule. The introduction of CF3 groups resulted in steric hindrance that twisted the adjacent molecules of 10b with respect to each other ([Figure 8 ], bottom), opening up pores. The fluorinated material was capable of separating O2 from Ar and N2 ; the selectivity was attributed to its aperture size that resulted in only O2 being adsorbed effectively.
Figure 8 Benzo-trisimidazoles 10a and 10b (top) crystallize in very different ways. While the parent 10a can establish six [N–H···N] hydrogen bonds to its neighbors (center), the larger
10b cannot because of its bulky CF3 group. Instead, it twists out of plane (bottom) and crystallizes in a porous structure.
Porous Fluorinated Cages
In the construction of 3D molecular cages, fluorinated moieties can find themselves
either on the inside (endofluorinated) or the outside (exofluorinated) of the cage.
In 2006, Fujita and coworkers reported the synthesis of four endofluorinated molecular
cages 12 by the reactions of bent-shaped dipyridines 11 and Pd(NO3 )2 in DMSO-d
6 ([Figure 9 ]).[73 ] The crystal structure of 12a reveals an ellipsoid-shaped cage, with dimension of 4.9 × 4.2 nm and 24 perfluoroalkyl
groups disordered as in a liquid state. Calculations indicated that there is a void
at the very center of the cage, surrounded by the disordered and partially aggregated
fluorinated groups that could not be crystallographically located. The solutions of
12a can selectively dissolve 5.8 molecules of perfluorooctane per capsule, and that number
can be modulated by adjusting the length of the perfluoroalkyl groups in cages 12a –d .
Figure 9 Synthesis of the internally fluorinated cages 12 . (Reproduced with permission from Ref. 73.)
The fluorination of starting materials can facilitate the porous materials’ synthesis.
Mastalerz and coworkers synthesized three shape-persistent tetrahedral boronic ester
cages 15a –c ([Figure 10 ]) by the [4+6] condensation between nonfluorinated or fluorinated diboronic acids
13a –c and a brominated triptycene-based hexaol 14 .[74 ] The fluorinated diboronic acids 13b and 13c have higher Lewis acidity than the nonfluorinated 13a , which reduced the reaction time from 16 h to 3–4 h. At the same time, the fluorinated
cages are more susceptible to solvolysis in MeOH, on account of the increased electrophilicity
of the boron centers. Single crystals of 15a and 15b were obtained by slow vapor diffusion of n -pentane or n -hexane into CDCl3 solutions. The crystal structure of the nonfluorinated 15a features [π···π] stacking and short B···Br and Br···π contacts. In the case of fluorinated
15b , [π···π] stacking is observed, along with apparent [C–F···π] and [C–F···B] interactions.
Despite similar covalent radii of hydrogen and fluorine, fluorinated cages had dramatically
lower BET surface areas of 69 m2 g−1 (15b ) and 33 m2 g−1 (15c ) compared with the nonfluorinated 15a (577 m2 g−1 ).
Figure 10 Synthesis of porous fluorinated and nonfluorinated cages 15a–c through [4+6] condensation of diboronic acids 13a–c with the triptycene-based hexaol 14 .
Cooper and coworkers performed a side-by-side comparison of the fluorinated and nonfluorinated
imine-based cages.[75 ] Using a [4+6] cycloimination of 1,3,5-triformylbenzene (16 , [Figure 11 ]) with (R ,R )-1,2-diphenylethylenediamine (17a ) and (R ,R )-1,2-bis(4-fluorophenyl)ethane-1,2-diamine (17b ), they synthesized two novel imine cages 18a (R = H) and 18b (R = F). The intention behind this study was to test whether the bulky aryl groups
at the cages’ vertices would frustrate the efficient packing and lead to increased
porosity (as well as to probe the influence of fluorination). Cage 18a was found to crystallize in two major forms, which belonged to space groups P 3 (prior to desolvation) and R 3 (after desolvation), while 18b quickly desolvated to a structure in the R 32 space group. A face-to-window stacking along the crystallographic c -axis was observed for both polymorphs of 18a (P 3 and R 3). The major difference between the two crystal forms is that the intrinsic pores
connect to adjacent extrinsic pores in 18a (P 3), but not in 18a (R 3). Neither of the crystal forms shows significant short contacts (< 2.5 Å, electrostatic
interactions or hydrogen bonding) among the cages. For a desolvated crystal of fluorinated
18b (R 32), each molecule of 18b packs alternately in window-to-window and face-to-face with the two adjacent molecules
of 18b along the crystallographic c -axis. Such different packing modes between 18a and 18b are due to the introduction of fluorinated aromatic rings, which engage in [C–H···F]
hydrogen bonding and quadrupole-complementary [π···π] stacking. The overall result
is a more rigid structure, in which extrinsic channels are sealed. This anisotropic
ordering on account of fluorination may be the reason why other polymorphs of 18b have not been identified. The tested BET surface area of the fluorinated 18b (533 m2 g−1 ) is comparable to that of 18a -R 3 (575 m2 g−1 ), but almost twice lower than that of 18a in the P 3 space group (952 m2 g−1 ).
Figure 11 Synthesis of porous fluorinated and nonfluorinated imine cages 18a–b through [4+6] condensation of trialdehyde 16 with diamines 17a and 17b .