Mixed-matrix materials using metal-organic polyhedra with enhanced compatibility for membrane gas separation

Article abstract of DOI:10.1039/c8dt00082d

Discrete metal-organic polyhedra (MOPs) containing copper(ii), palladium(ii), and iron(ii) nodes were synthesized as fillers for mixed-matrix materials (MMMs) with a polyvinylidine fluoride (PVDF) polymer phase and contrasted against an MMM containing a metal-organic framework, MOF-5. When a given MOP was soluble in the precursor solutions, the resulting MMMs were thin, flexible, and homogeneous based on microscopy and SEM imaging. Analogous MMM formation using either insoluble MOPs or the inherent insoluble MOF-5 showed a higher degree of phase separation and inhomogeneity. Even when a MOP was not fully soluble, a significant particle size decrease was observed in contrast to the MOF-5 materials wherein the crystallites remained largely intact. This is a consequence of solubilizing the MOP fillers into the polymer solvent. The crystallinity and thermal stabilities of the MMMs were compared to pure PVDF using powder X-ray diffraction, and differential scanning calorimetry, indicating that the incorporation of MOPs both decreased overall crystallinity as well as increased thermal stability. In addition, MMMs containing PdMOP and FeMOP showed improved gas permeabilities relative to pure PVDF for H₂, N₂, CH₄, and CO₂, with the 10 wt% FeMOP membrane more selective for CO₂ over N₂ and H₂.

Full text of DOI:10.1039/c8dt00082d

Dalton
Transactions
View Article Online
View Journal
PAPER
Mixed-matrix materials using metal–organic
polyhedra with enhanced compatibility for
Cite this: DOI: 10.1039/c8dt00082d
membrane gas separation†
a
b
a
b
Cressa Ria P. Fulong,
Junyi Liu, Vincent J. Pastore,
Haiqing Lin * and
a
Timothy R. Cook
*
Discrete metal–organic polyhedra (MOPs) containing copper(II), palladium(II), and iron(II) nodes were syn-
thesized as fillers for mixed-matrix materials (MMMs) with a polyvinylidine fluoride (PVDF) polymer phase
and contrasted against an MMM containing a metal–organic framework, MOF-5. When a given MOP was
soluble in the precursor solutions, the resulting MMMs were thin, flexible, and homogeneous based on
microscopy and SEM imaging. Analogous MMM formation using either insoluble MOPs or the inherent in-
soluble MOF-5 showed a higher degree of phase separation and inhomogeneity. Even when a MOP was
not fully soluble, a significant particle size decrease was observed in contrast to the MOF-5 materials
wherein the crystallites remained largely intact. This is a consequence of solubilizing the MOP fillers into
the polymer solvent. The crystallinity and thermal stabilities of the MMMs were compared to pure PVDF
using powder X-ray diffraction, and differential scanning calorimetry, indicating that the incorporation of
MOPs both decreased overall crystallinity as well as increased thermal stability. In addition, MMMs con-
Received 8th January 2018,
Accepted 7th February 2018
DOI: 10.1039/c8dt00082d
taining PdMOP and FeMOP showed improved gas permeabilities relative to pure PVDF for H
2 2 4
, N , CH ,
rsc.li/dalton
and CO , with the 10 wt% FeMOP membrane more selective for CO over N and H
2
2
2
2
.
Currently, most commercialized synthetic membranes are
Introduction
organic polymer-based, due to their good processability and
th
4,5,8,10
Since the discovery of membrane technology in the 18
ease in scale-up.
However, traditional polymers exhibit a
century the field has grown to areas including controlled drug permeability/selectivity trade-off, i.e., when permeability is
1
2
18,19
delivery and industrial separation. Membranes are distin- increased, gas selectivity decreases.
In the last four
guished by their ability to control the permeation of chemical decades, several efforts have been made to overcome this
species, enabling moderation of the permeation rate of drugs inverse relationship by developing new inorganic materials
2
0
8
from reservoir to body or the transport of one component of a such as zeolites and metal–organic frameworks (MOFs) for
3
mixture while hindering transport of other components. This use in films and membranes with the intention of enhancing
property makes membranes an economical and energy both permeability and selectivity.
2
1
4
–6
efficient tool for industry. Membrane technology is popular
Since these inorganic materials are often difficult to fabri-
more specifically in cate into large-scale commercial membranes, mixed-matrix
hydrogen sep- materials (MMMs) present an attractive alternative to overcome
vapor recovery the permeability/selectivity trade-off. These membranes
systems, monomer recovery, organic solvent dehydration, combine the processability of organic polymers with the excel-
5
–9
in gas and small molecule separation,
1
0,11
12,13
natural gas sweetening,
helium recovery,
nitrogen enrichment from air,
1
4
15
aration,
1
6
17
8
3
and large molecule removal from organic solvents.
lent gas separation performance of porous filler materials.
Several variables need to be considered in tailoring the gas
9
separation performance of MMMs. The inclusion of filler
materials introduces a number of possible defects stemming
from phase separation, poor dispersion, and incomplete incor-
poration that can result in inhomogeneous membranes. When
not addressed, these defects can render a membrane impracti-
cal, regardless of its inherent permselectivity. Furthermore,
the batch-to-batch reproducibility of an MMM may suffer if
variables such as particle size, particle sedimentation agglom-
a
Department of Chemistry, University at Buffalo, The State University of New York,
Buffalo, New York 14260, USA. E-mail: trcook@buffalo.edu
b
Department of Chemical and Biological Engineering, University at Buffalo, The
State University of New York, Buffalo, New York 14260, USA.
E-mail: haiqingl@buffalo.edu
†
Electronic supplementary information (ESI) available. See DOI: 10.1039/
c8dt00082d
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
2
2
eration, and polymer–particle interface morphology are not
rigorously controlled.
Whereas early MMMs containing inorganic zeolites often
used micron-sized particles, it was quickly recognized that sub-
micron particles were essential in forming thinner membranes
that can provide more polymer–particle interfacial area for
9
high separation performance. Even when small particles are
used, precipitation and agglomeration of the filler material
results in inhomogeneous membranes with pinholes that are
2
3
considered non-selective membrane defects. A final concern
is the nanoscale morphology of MMM interface which is a
critical determinant of the overall transport property of mem-
2
2
branes. Interfacial defects may result from detachment of
polymer chains from the filler, formation of rigidified polymer
layer on the filler surface, or partial sealing of the filler surface
2
4
pores by rigidified polymer chains.
Fig. 1 Structures of (a) CuMOP (C = gray, O = red, S = yellow, H =
MOFs are highly porous coordination polymers and have white, Cu = gray sphere, CCDC no. 755923, ref. 50); (b) PdMOP (C =
4
gray, N = blue, H = white, Pd = blue sphere, CCDC no. 1208789, ref. 51);
c) FeMOP (C = gray, O = red, S = yellow, H = white, Fe = orange sphere,
emerged as promising filler materials for MMMs. Compared
(
to purely inorganic microporous zeolites, MOFs are highly
modular and possess a far greater structural diversity that can
facilitate better interactions with the organic polymer phase;
however, some of the challenges in using MOFs as fillers stem
from their propensity for hydrothermal instability and poor or
non-existent solubility that hinders homogeneous dispersion
in a membrane.
CCDC no. 688687, ref. 52); and (d) MOF-5 (C = gray, H = white, O = red
spheres, Zn = purple sphere, CCDC no. 256965, ref. 53).
fillers were solubilized in aprotic polar solvents wherein the
PVDF was also soluble. The complementary solubilities were
hypothesized to facilitate polymer-filler integration to ulti-
mately form homogeneous membranes with good interface
morphologies. The fabricated membranes were characterized
using various microscopic, gravimetric, spectroscopic, and
diffraction methods and compared to a MOF-containing MMM
that served as a control representing an insoluble filler system.
Gas permeability and selectivity of the pure polymer and
MMMs were compared. In addition, due to the innate semi-
crystallinity of PVDF, the change in crystallinity upon mixing
the filler material was used as a good indication of polymer-
Recently, there has been interest in exploring metal–
organic polyhedra (MOPs) as fillers for gas separation
4
,25–30
membranes.
These materials are similar to MOFs in that
they are formed via self-assembly reactions driven by metal–
ligand bond formation. In stark contrast to MOFs, however,
they are discrete molecular species rather than infinite frame-
works. As a result, they have well-defined uniform sizes, typi-
cally between 1–5 nm and exhibit solubility in a range of sol-
3
1
vents. The motivation behind their use is the hypothesis that
soluble filler materials will form composite materials with
good chemical compatibility and homogeneous dispersion
within the continuous phase of MMMs in contrast to insoluble
fillers that are fabricated from suspensions.
36–38
filler integration.
Herein, we explore the use of a suite of MOPs for the for-
mation of highly dispersed and homogeneous MMMs using
PVDF as a polymer phase. Although PVDF has been used to
Results and discussion
Fabrication of MMMs
3
2–34
form MMMs with MOFs,
our work is the first example of Four novel MMMs were made using one of two methods (see
using this binder with discrete metallacages. For membranes Experimental section). The first method occasionally resulted
that formed free of defects, gas permeabilities and selectivities in nanoscopic defects that prevented permeability measure-
were measured, revealing enhancements over pure PVDF. To ments. The second method was used to fabricate defect-free
date, an Cu24
L24 cuboctahedral cage previously reported by membranes. Both methods were used with three MOP fillers:
3
5
Yaghi and co-workers is the only MOP filler used in MMMs, CuMOP, PdMOP, and FeMOP (Fig. 1) and one coordination
specifically with poly(1-trimethylsilyl-1)propyne, Boltorn polymer, MOF-5. CuMOP is a M24
3
9
L24 cuboctahedral cage
W3000, Matrimid, polysulfone, poly(2-hydroxyethyl methacry- formed from Cu(II) paddlewheel nodes and sulfonated-1,3-ben-
4
0
late), poly(methylmethacrylate) as organic phases. This family zenedicarboxylate ligands (Fig. 1a). PdMOP is a M L trun-
6
4
of Cu24
good gas permeability and selectivity.
suitability of three MOP fillers differing in both architecture (TPT) ligand (Fig. 1b). FeMOP is a M L tetrahedral cage that
L24 assemblies has been fabricated into MMMs with cated tetrahedral cage formed from 2,2′-bipyridine (2,2′-bpy)-
2
5–30
In this work, the capped Pd(II) nodes and 2,4,6-Tris(4-pyridyl)-1,3,5-triazine
4
1
4
6
24 6 4 4 6
(M L24, M L , M L ) and metal ions (Cu(II), Fe(II), and Pd(II)), was formed from Fe(II) nodes and using in situ subcomponent
were evaluated as fillers for PVDF-based MMMs (Fig. 1). To self-assembly with 4,4′-diaminobiphenyl-2,2′-disulfonate and
minimize particle sedimentation and agglomeration, the MOP 2-formylpyridine (Fig. 1c). MOF-5 is a metal–organic frame-
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
4
work formed from Zn O nodes and 1,4-benzenedicarboxylate
4
2
ligands (Fig. 1d).
PVDF was chosen as the organic polymer to fabricate all
MMMs. Though it is semi-crystalline fluoropolymer with at
4
3,44
least four phases, α, β, γ, and δ,
it has very good film
forming capability, and an ideal matrix to explore the inter-
actions between the MOPs and polymer, and the effect of
MOPs loading on gas transport properties.
Solubility of MOPs
MMMs fabricated by Method I involved dissolving PVDF and
the filler material in the same solvent, where both polymer
and filler are soluble. Solubilizing both polymer and filler in
the same solvent should improve the filler–solvent–polymer
interaction, which is one of the major factors that can affect
filler integration into the polymer and the resulting polymer–
2
2
particle interface morphology. PVDF is soluble in polar
aprotic solvents such as N,N-dimethylacetamide (DMA), N,N-
dimethyformamide (DMF), N,N-dimethylsulfoxide (DMSO),
and N-methylpyrrolidone (NMP). It was observed that the MOP
fillers are soluble in DMF and DMSO at varying degrees.
MOF-5, on the other hand, is effectively insoluble in those sol-
vents. Table 1 summarizes the solubility of each MOP in their
respective solvents. CuMOP has a 0.68 wt% solubility in DMF,
PdMOP has 5.8 wt% solubility in DMSO, while FeMOP has
Fig. 2 Optical microscope images at 5.0 magnification of polymer and
filler powders and membranes: (a) PVDF in DMF; (b) PVDF in DMSO; (c)
CuMOP; (d) PdMOP; (e) FeMOP; and (f) MOF-5.
0
.281 wt% solubility in DMSO.
Fixed loadings of 5, 20, and 30 wt% filler into PVDF were
prepared for each filler material. The concentration of MOP in
the MMM ink solution is summarized in Table 2. MMMs at 5,
Microscopy
2
0
0
0, and 30 wt% final loading have MOP concentrations of
.13, 0.53, and 1.05 wt% in precursor DMF solutions and 0.12,
.48, and 0.95 wt% in precursor DMSO solutions. Based on
Microscopy was used to image and characterize the fabricated
membranes. Fig. 2 illustrates the optical microscope images
(
5.0 magnification) of the powdered starting materials, i.e.
these solubilities at 30 wt% of CuMOP and 20 wt% FeMOP the
cages will not be completely solubilized in the MMM ink solu-
tion prior to casting.
PVDF polymer and filler materials, and the fabricated MMMs
at 20 wt% loading. The pure PVDF that were made using both
DMF and DMSO as solvents serve as blank membranes (Fig. 2a
and b). Fig. 2c–f shows the starting powdered materials and
MMMs fabricated from CuMOP, PdMOP, FeMOP, and MOF-5,
respectively. In all cases, the MMM films were very thin, flex-
Table 1 Solubility of MOPs in DMF and DMSO
MOP
Solvent
Solubility, wt% ible, and free of macroscopic defects. At 5.0 magnification, all
MMMs with MOP fillers are homogeneous with MOP particles
CuMOP
PdMOP
FeMOP
DMF
DMSO
DMSO
0.68 ± 0.02
5.8 ± 0.2
0.281 ± 0.001
size significantly reduced compared to the starting material.
Considerable color changes of the MMMs were also noted
upon addition of CuMOP, PdMOP, or FeMOP fillers into PVDF.
The even dispersion of color suggests good integration of MOP
fillers into the PVDF. In contrast, phase separation was
observed in the MOF-5 MMM.
Table 2 Concentration of MOP in the MMM ink solution
SEM imaging was employed to examine the surface of the
bulk starting materials and the membranes at a microscopic
DMSO scale. Fig. 3 illustrates the SEM images with EDS analysis of
b
MOP ink, wt% MOP
a
MMM, wt% MOP
DMF
the powdered polymer and filler materials. The SEM image of
PVDF powder showed semi-crystalline spheres of <1 µm size
5
2
3
0.13
0.53
1.05
0.12
0.48
0.95
0
0
(
Fig. 3a). Fig. 3b–e shows the SEM images of CuMOP, PdMOP,
a
b
FeMOP, and MOF-5, respectively. The CuMOP powder is com-
posed of ∼2.5 µm semi-cubic block crystals. The PdMOP
wt% = (mass MOP)/(mass MOP + mass PVDF) × 100%. wt% = (mass
MOP)/(mass MOP + mass solvent) × 100%.
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Fig. 4 (a) SEM images of PVDF powder and membranes in DMF and
DMSO; and SEM and YAGBSE images of filler powder and 20 wt% mem-
branes: (b) CuMOP; (c) PdMOP; (d) FeMOP; and (e) MOF-5.
Fig. 3 SEM images and EDS spectra of (a) PVDF; (b) CuMOP; (c) <100 µm (Fig. 4c). The DMSO solubility of the PdMOP was the
PdMOP; (d) FeMOP; and (e) MOF-5 powder.
highest among all fillers, which is the main factor in creating
such well-dispersed MMM. In contrast, FeMOP is the least
soluble. At 20 wt%, FeMOP is not completely solubilized in the
powder has prismatic crystals of about 100 µm. FeMOP
MMM ink solution. The MMM formed from FeMOP shows
powder is composed of irregularly-shaped crystals of varying
how the particles, though well-dispersed, are inconsistent in
sizes from 100–200 µm. The MOF-5 powder shows its charac-
their size distribution (Fig. 4d). Overall, the MOP-based
teristic cubic block crystals of <100 µm size. All EDS spectra
MMMs are more homogeneous with well-dispersed particles
confirm the identity and purity of each material (Fig. 3).
than the MOF-based MMM (Fig. 4e). In the MOF-5 MMM, the
Fig. 4 summarizes the SEM images of the fabricated mem-
particles are not solubilized in the MMM solution which
branes at 100 µm scale for pure PVDF and MOP-based MMMs
resulted in the same particle size of filler materials that are
and 200 µm scale for MOF-5 MMM. The representative MMM
randomly agglomerated across the film.
shown are all at 20 wt% filler loadings. Yttrium Aluminium
Garnet Backscattered Electron (YAGBSE) scintillation counting
was used to show features up to a depth of 2 µm below the
Thermogravimetric analysis
surface. YAGBSE images show regions of high electron density Thermal stability of membranes is very important especially in
as bright spots. In the case of heterogeneous MMMs, crystal- industrial gas separation which sometimes requires elevated
7
line domains of the MOP fillers can be readily identified. This temperatures, e.g. CO2 separation at greater than 100 °C.
was a helpful tool in confirming the presence of integrated Thermogravimetric analysis (TGA) was used to evaluate
filler materials below the surface of the MMM. In all cases, improvements in thermal stability of the MMMs by comparing
very thin (<100 µm thick) membranes that are free of micro- changes in the decomposition temperatures. The TGA curves
scopic defects were made. The pure PVDF is homogeneous for the starting materials and the MMMs with 20 and 30 wt%
and formed a defect-free film (Fig. 4a). Although the CuMOP loadings are summarized in Fig. 5.
was fully soluble in DMF during casting, as the solvent evapor-
TGA curves of pure PVDF powder and membranes show the
ated, crystallites reformed (Fig. 4b). These crystallites are well characteristic decomposition temperature at 450 °C (Fig. 5a).
dispersed through the MMM, leading to a high quality film. In Fig. 5b–e illustrates the TGA curves for the powders and
the case of the PdMOP MMM, the filler particles were not only MMMs made with CuMOP, PdMOP, FeMOP, and MOF-5,
well-dispersed but their particle sizes were also reduced to respectively. The TGA curve of CuMOP shows two mass-loss
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
In addition, it was observed in all MMMs TGA curves that
the PVDF decomposition process shifted to a higher tempera-
ture, i.e. 10 °C higher for PdMOP MMM and 30 °C higher for
CuMOP, FeMOP, and MOF-5 MMM. This indicates an improve-
ment in the thermal stability of the PVDF upon addition of the
filler materials.
X-ray methods
As previously demonstrated, all starting materials including
PVDF and all filler materials are crystalline in nature. To deter-
mine the crystallinity and crystal structures of these starting
materials and the fabricated membranes, powder X-ray diffrac-
tion (PXRD) was used. Fig. 6 summarizes the collected diffrac-
tion patterns for the powdered starting materials, including
PVDF, CuMOP, PdMOP, FeMOP, and MOF-5, and membranes,
including pure PVDF using DMF and DMSO as solvents and
MMMs fabricated at 20 and 30 wt% loadings.
The change in crystal structure and crystallinity of PVDF
will dictate some of its physical and mechanical properties. As
previously mentioned, pure PVDF crystallizes in at least four
different polar and nonpolar phases that easily transforms
Fig. 5 TGA curves for PVDF and filler powders and membranes: (a)
PVDF; (b) CuMOP; (c) PdMOP; (d) FeMOP; and (e) MOF-5.
steps at 70 °C and 160 °C for the removal of CH OH and DMA,
3
respectively, followed by a major cage decomposition step at
3
00 °C and two minor decomposition steps at 500 and 700 °C
3
9
(Fig. 5b). In the case of PdMOP, three mass-loss events were
observed at 100, 250, and 350 °C, which are attributed to H O-
2
4
5
46
removal, followed by decomposition of 2,2′-bpy and TPT,
respectively (Fig. 5c). The FeMOP TGA curve also has three
notable mass-loss steps at 100, 350, and 450 °C (Fig. 5d).
Similar to that of PdMOP, FeMOP decomposition has a H
2
O-
removal step followed by two ligand decomposition steps that
4
6
are attributed to 4-cyanopyridine and 4,4′-diaminobiphenyl-
4
7
2
,2′-disulfonate degradation, respectively. Lastly, the TGA
curve of MOF-5 shows two mass-loss events at 160 and 500 °C,
4
8
which are due to DMF-loss and framework decomposition,
respectively (Fig. 5e).
The TGA curves for all MMMs shows all mass-loss events
that occurred for the decomposition of both filler and PVDF,
which suggests some level of filler integration into the
polymer membrane. In the case of CuMOP and PdMOP MMM,
increasing the filler loading by 10 wt% resulted in ∼10%
increase in mass loss due to the filler material decomposition.
This is not the case for both FeMOP and MOF-5 MMM, which
can be attributed to FeMOP saturation at about 20 wt%
loading and inefficient filler integration of MOF-5 into the
MMM.
Fig. 6 Powder X-ray diffraction patterns for PVDF and filler powders
and membranes.
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Table 3 Values of 2θ and respective d spacing observed in the PVDF
Fourier transform infrared spectroscopy
The FT-IR spectra of PVDF, filler materials, and the MMMs
Material
2θ
d, Å
4.833
PVDF phase
fabricated were shown in Fig S5.† PVDF has strong character-
PVDF powder
18.34
α
−1
−1
istic peaks at 3000 cm for C–H stretching, 1400 cm for C–F
1
2
9.90
6.80
4.459
3.324
4.763
4.437
3.363
4.311
4.304
4.306
4.331
10.014
4.410
stretching, and 1350 and 970 cm− for C–H bending.
1
43,49
All
PVDF film (DMF)
PVDF film (DMSO)
CuMOP film
PdMOP film
18.61
α
MMMs have peaks that are similar to the fingerprint region of
the PVDF spectrum. This further supports the enhanced inte-
gration of the filler materials into the PVDF polymer
membrane.
1
2
9.99
6.48
20.59
20.62
20.61
20.49
8.82
β
β
β
β
β
a
a
a
FeMOP film
MOF-5 film
a
Differential scanning calorimetry
20.12
The DSC curves of the MMMs are shown in Fig. 7. Melting
a
Films are MMMs at 20 wt% filler loading.
temperatures (T ), heats of fusion (ΔH) and crystallinities (X)
m
of the mixed-matrix materials are given in Table 4. Crystallinity
is calculated by the equation below:
into other phases upon exposure to different conditions such
ΔH
as mechanical stretching, high temperature, and electric
x ¼
ꢀ 100%
WΔH°
4
4
field. The most common crystal form of pure PVDF and
membranes that are cast and evaporated at temperatures
where ΔH° is the heat of fusion of the PVDF with 100% crystal-
linity, which is taken as 90.4 J g for α phase and 104.5 J g
for β phase. W is the polymer content of the MMMs.
Interesting results were observed in the MOP-containing
MMMs. Unlike other PVDF-based MMMs with non-MOP
−
1
−1
4
3
above 110 °C is the nonpolar α phase. It was also previously
reported that a mixture of α and β phases will be present if the
membrane was evaporated at temperatures between 70 and
5
0
1
10 °C, while β phase will predominate if evaporation occurs
4
4,51,52
fillers,
the crystallinity of the MMMs here decreased sig-
nificantly as compared to pure PVDF, in agreement with our
below 70 °C. Table 3 summarizes the 2θ and d-spacing of all
diffraction peaks observed for pure PVDF powder, pure PVDF,
and 20 wt% loadings of CuMOP, PdMOP, FeMOP, and MOF-5.
The powdered PVDF shows 2θ peaks at 18.34°, 19.90°, and
2
6.80°, which corresponds to the (020), (110), and (021) planes
of the α phase, respectively. This is similar to what was
observed with the PVDF made using DMF as the solvent. PVDF
crystalline phase change from α to β occurred when DMSO was
used as the solvent and upon addition of all filler materials. In
all of these cases, the characteristic diffraction peak at 20.6°
for the sum of the diffractions at (110) and (200) planes of the
β phase was observed. It is also noteworthy that a peak at 8.82°
was observed for MOF-5 MMM which is attributed to the
crystal structure of MOF-5.
Characteristic diffraction peaks of each starting material
were present in the collected diffraction patterns. The sharp
peaks observed in PVDF and MOF-5 suggest high crystallinity
of the starting powdered materials. Although the broad peaks
on the CuMOP, PdMOP, and FeMOP suggests a large degree of
amorphous phases. In all of the membranes fabricated, the
predominant diffraction peaks represent the crystal structure
Fig. 7 DSC curves of pure PVDF and MMMs.
of PVDF with considerable decrease in peak intensities. This Table 4 Values of melting temperature, fusion heat and respective
crystallinity calculated in the PVDF
implies disruption of the crystalline features of the MOP fillers
and decrease in overall crystallinity of the PVDF, which indi-
cates good integration of the fillers into PVDF. It has been pre-
ΔH (J g−
1
)
X (%)
Material
T
m
(°C)
viously suggested that decrease in crystallinity or increase in PVDF
159
152
161
160
163
161
165
164
160
43.3
25.3
14.1
23.5
23.3
25.1
18.6
22.1
11.2
48
30
23
28
32
30
25
27
16
2
3
2
0 wt% Cu-PVDF
amorphous phase of the membrane results in more free
0 wt% Cu-PVDF
0 wt% Fe-PVDF
3
6,38
volume cavities for gas transport.
In the case of MOF-5
MMM, the crystalline features of MOF-5 becomes more appar- 30 wt% Fe-PVDF
2
3
2
0 wt% Pd-PVDF
0 wt% Pd-PVDF
0 wt% MOF 5-PVDF
ent at higher loading resulting in a more crystalline MMM.
This is a consequence of poor solubilisation of MOF-5 into the
MMM ink solution.
30 wt% MOF 5-PVDF
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
PXRD results and indicative of strong interactions between the selectivities for MMMs. The ideal gas separation properties
polymer and filler. For the CuMOP, PdMOP, and MOF-5 remain the same, except for the FeMOP MMM. Addition of
MMMs, the crystallinity further decreased with higher load- MOP fillers alters the selectivity due to introduction of gas
ings, which suggests an increasingly stronger interaction diffusion channels and transition metal sites that can facilitate
between the filler and polymer matrix. However for FeMOP, transport of high affinity gases. A remarkably high CO /N
2
2
there is a relative increase in crystallinity due to the poor solu- selectivity was observed as compared with pure PVDF. We attri-
bility of the cage.
bute this improved CO
2
selectivity over N
2
to iron’s capacity for
This selectivity suggests that FeMOP
has potential for superior CO /N separations, especially if an
inherently selective polymer binder is used rather than
5
7,58
CO chemisorption.
2
2
2
Gas permeability and selectivity
5
9
Fig. 8 shows gas permeability as a function of critical volume PVDF.
5
3
of gas molecules for pure PVDF, and MMMs with 10 wt%
FeMOP, and 20 wt% PdMOP, the two MOPs that resulted in
defect-free films. Although the SEM images show that all
MMMs were free of microscopic pores, nanoscopic pore
defects in the CuMOP MMM resulted in leaky films, prevent-
Experimental
Materials and methods
ing permeability measurements. The presence of these nano- All reagents and solvents were reagent grade and used as
scopic defects resulted in increasing permeability measure- received without further purification. Terephthalic acid
ment as the gas pressure was increased. For the MMMs, there (H BDC), 2-formylpyridine, monosodium 5-sulfoisophthalate
2
is a significant increase in gas permeability due to the (5-SO -1,3-BDC), and copper(II) acetate monohydrate
3
decrease in crystallinity as well as the fast gas diffusion chan- (Cu (OAc) ·2H O) were purchased from TCI Chemicals. N,N-di-
2
4
2
nels created by the MOPs. This increase in permeability due to methylacetamide (DMA) and tetramethylammonium hydrox-
decrease in crystallinity follows the two-phase model proposed ide pentahydrate were purchased from Acros Organics.
by Michaels and Bixler, which describes the inverse depen- 4-Cyanopyridine, 4,4′-diaminobiphenyl-2,2′-disulfonic acid
dence of diffusivity and crystallinity of pure polymeric hydrate (up to 30% water), and polyvinylidene fluoride (PVDF)
5
4–56
membranes.
Since the permeability of each gas increases were purchased from Alfa Aesar. Methanol (CH OH), aceto-
3
to the same extent upon a loss of crystallinity, there is no net nitrile (CH CN), N,N-dimethylsulfoxide (DMSO), acetone
3
effect on the overall selectivity, which takes into account the ((CH ) CO), hydrochloric acid (HCl), potassium hydroxide,
3
2
ratio of these permeabilities. Table 5 summarizes the pure gas silver nitrate (AgNO3), and iron(II) sulfate heptahydrate
FeSO ·7H O) were purchased from Fisher Scientific. N,N-
Dimethylformamide (DMF) and nitric acid (HNO , 65%) were
purchased from EMD Millipore. Palladium(II) chloride and
,2′-bipyridine (bpy) were purchased from Strem Chemicals.
(
4
2
3
2
Ethanol (EtOH, 190 proof) was purchased from Decon Labs,
Inc. Zinc nitrate hexahydrate (Zn(NO ) ·6H O) was purchased
3
2
2
from Aqua Solutions Inc. 1,4,7,10,13,16-hexaoxacycloocta-
decane (18-crown-6) was purchased from Aldrich Chemical
Company, Inc. Pyridine was purchased from Beantown
Chemical, Corp. Aqueous ammonia (NH ) was purchased from
3
JT Baker Chemicals. Deionized water was used in all aqueous
synthesis.
All UV-Vis spectra were acquired from an Agilent Cary 8454
UV-Vis Diode Array system with a 10 mm rectangular quartz
cuvette. Blank spectra with pure solvents were acquired in
between runs. All measurements are done in triplicate.
1
H NMR spectra were recorded on either Varian 300 or
Fig. 8 Permeability curve of pure PVDF and MMMs.
Table 5 Pure gas selectivity of MMMs
400 MHz spectrometer. Chemical shifts (δ) are reported in
parts per million (ppm) referenced using the residual protio-
solvent peaks as internal standards. Coupling constants (J) are
quoted in Hertz (Hz), and the following abbreviations are used
to describe the signal multiplicities: s (singlet); d (doublet); t
Selectivity
Material
(
triplet); q (quartet); m (multiplet).
CO /N
2 2
CO
2
/H
2
CO
2
/CH
4
Powder X-ray diffraction (PXRD) patterns were collected
from a Rigaku Ultima IV system with inert atmosphere attach-
ment and heated sample chamber equipped with Cu source
(0.15418 nm Kα) and operated at 1.76 kW power (40 kV,
PVDF
13
13
23
1.3
1.1
1.5
24
19
5.6
2
1
0 wt% Pd-PVDF
0 wt% Fe-PVDF
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
3
α
3
4
5
4 mA). Diffraction patterns were measured within 2θ range of = 9.34 (d, 24H, J = 5.7 Hz, pyridine H ), 8.79 (d, 24H, J =
−
1
β
3
to 35° at a rate of 1° min
.
5.7 Hz, pyridine H ), 8.38 (d, 12H, J = 8.9 Hz, 6,6′-bipyridine),
Thermogravimetric analysis (TGA) curves were measured 8.26 (t, 12H, J = 7.8 Hz, 5,5′-bpy), 7.58 (d, 12H, J = 5.5 Hz,
using a TA Instruments DSC SDT Q600 instrument at a 3,3′-bpy), 7.48 (t, 12H, J = 6.6 Hz, 4,4′-bpy). FT-IR (ATR, cm ):
heating rate of 5° min from 30–600 °C under constant flow 3415 (w), 3076 (w), 1608 (w), 1576 (w), 1520 (w), 1472 (w), 1451
3
3
3
−1
−
1
−
1
of nitrogen (100 mL min ).
(w), 1363 (w), 1332 (m), 1309 (m), 1161 (w), 1108 (w), 1060 (w),
Fourier transform-infrared (FT-IR) spectra were acquired 1042 (w), 1025 (w), 877 (w), 862 (w), 814 (w), 766 (w), 720 (w),
from a PerkinElmer 1760 FT-IR spectrometer equipped with 675 (w).
horizontal attenuated total reflectance (HATR) from 4000 to
FeMOP was self-assembled using the following procedure.
00 cm− wavenumbers (ν). The following abbreviations are 1.00 g (2.03 mmol) 4,4′-diaminobiphenyl-2,2′-disulfonic acid,
used to describe signal intensities: w (weak); m (medium); s 0.435 g (4.06 mmol) 2-formylpyridine, 0.737 g (4.06 mmol)
strong). tetramethylammonium hydroxide pentahydrate, and 0.376 g
Field Emission Scanning Microscope (FESEM) images were (1.35 mmol) iron(II) sulfate heptahydrate were added to a
captured using a Hitachi SU70 FESEM with Oxford Energy-dis- 100 mL Schlenk flask and evacuated for 15 min. Then, 25 mL
persive X-ray Spectrometer (EDS) set to 20 keV accelerating deionized H O was added to the flask. The mixture was stirred
voltage. under N for 20 h at 50 °C. The product (1.13 g) was then iso-
1
5
(
2
2
Differential Scanning Calorimetry (DSC) Melting tempera- lated at quantitative yield as dark purple crystals after acetone
ture OF MMMs were studied by differential scanning calorime- layering into the aqueous solution. The TGA curve, PXRD
try (DSC, Q2000, TA Instruments, New Castle, DE). Samples pattern, and FT-IR spectrum for the synthesized powder are
1
were scanned under a dry N purge at a flow rate of 50 ml shown in Fig. 5, 6, and 7, respectively. The H NMR spectrum
2
−
1
−1
1
min . A typical scanning rate was 10 °C min , and the scan- is shown in Fig. S1b.† H NMR (400 MHz, D
2
O, 25 °C):
3
ning temperature range is 60 °C to 200 °C cycles for each δ (ppm) = 9.34 (s, 12H, imine), 8.71 (d, 12H, J = 7.9 Hz, 3-pyri-
3
sample.
dine), 8.40 (t, 12H, J = 8.1 Hz, 4-pyridine), 7.77 (t, 12H, 5-pyri-
dine), 7.54 (d, 12H, J = 5.5 Hz, 6-pyridine), 7.14 (d, 12H, J =
.4 Hz, 6,6′-benzidine), 6.44 (s, 12H, 3,3′-benzidine), 5.84 (d,
3
3
Syntheses
8
3
9
40
41
3
+
All filler materials (CuMOP,
PdMOP,
FeMOP,
and 12H, J = 3.6 Hz, 5,5′-benzidine), 7.14 (s, [NMe ] ). FT-IR (ATR,
) were synthesized as published in the literature. cm ): 3407 (w), 1608 (w), 1553 (w), 1491 (w), 1471 (w), 1299
4
4
2,48
−1
MOF-5
The synthesized materials were dried in a vacuum oven at (w), 1182 (w), 1097 (w), 1080 (w), 1038 (w), 1021 (w), 997 (w),
1
20 °C for 24 h and stored in a desiccator (Drierite) until use.
CuMOP was self-assembled by dissolving 0.269
949 (w), 832 (w), 774 (w), 750 (w), 729 (w), 698 (w), 670 (w), 620
(w), 578 (w), 551 (w).
g
(1.00 mmol) of 5-SO
3
-1,3-BDC in 8 mL CH
3
OH and sonicating
MOF-5 self-assembly was adapted from previously pub-
for 30 min. This was added to a solution of 0.200 g lished procedures. 1.00 g (6.02 mmol) H BDC, 5.37 g
2
(1.10 mmol) Cu
2
(OAc)
4
·2H
2
O
3 2 2
in 15 mL 1 : 1 (v/v) (18 mmol) Zn(NO ) ·6H O, and 169 mL of DMF were added to
CH OH : DMA. The mixture was stirred for 10 min at room a 250 mL round bottom flask. The mixture was sonicated for
3
temperature. Then 5 mL of DMA was layered over the mixture. 15 minutes until all the solid materials were dissolved. Then
Light-blue powder of CuMOP (0.388 g, 54% yield) were formed the solution was heated to 100 °C for 24 hours to obtain color-
after 10 days. The product was collected and washed with DMA less cubic crystals of MOF-5. The DMF solution was decanted
and EtOH. The TGA curve, PXRD pattern, and FT-IR spectrum and MOF-5 crystals (1.4 g, 85% yield) were washed with DMF.
for the synthesized powder are shown in Fig. 5, 6, and 7, The TGA curve, PXRD pattern, and FT-IR spectrum for the syn-
−
1
respectively. FT-IR (ATR, cm ): 3375 (w), 1611 (m), 1564 (w), thesized powder are shown in Fig. 5, 6, and 7, respectively.
1
1
(
(
528 (w), 1435 (m), 1356 (s), 1310 (w), 1220 (m), 1202 (m), 1181 FT-IR (ATR, cm− ): 1648 (w), 1599 (m), 1503 (w), 1384 (m),
s), 1120 (w), 1045 (s), 991 (w), 923 (w), 876 (w), 770 (m), 730 1297 (w), 1251 (w), 1150 (w), 1110 (w), 1087 (w), 1060 (w), 1014
s), 666 (m), 619 (s), 565 (s). (w), 884 (w), 820 (w), 743 (m), 672 (w), 653 (w), 583 (w),
PdMOP was self-assembled by mixing 0.16 g (0.40 mmol) 522 (m).
6
0
palladium(II) bipyridine nitrate (Pdbpy(ONO
2
)
2
)
and 0.090 g
4
6,61
Solubility of MOPs
(0.30 mmol) 2,4,6-Tris(4-pyridyl)-1,3,5-triazine (TPT)
in
1
0 mL solvent mixture of 1 : 1 (v/v) CH OH : H O. The suspen- The solubilities of CuMOP, PdMOP, and FeMOP in their
3
2
sion was refluxed at 80 °C overnight. Then the mixture was respective solvents were determined using their measured
centrifuged and the solution decanted and filtered. The yellow absorption coefficients in wt%− cm (CuMOP: 3.20; PdMOP:
1
−1
powder product obtained after solvent removal was washed 302; and FeMOP: 12.1) at λmax (725 nm for CuMOP; 300 nm for
three times with 100 mL portions of CH
3
CN, then dissolved in PdMOP; and 580 nm for FeMOP). Standard solutions of each
1
00 mL H O. Pure yellow powder product (0.24 g, 96% yield) MOP were prepared by serial dilution of 0.10 wt% solution of
2
was obtained after solvent removal. The TGA curve, PXRD CuMOP in DMF (0.0005–0.01 wt%), 0.50 wt% solution of
pattern, and FT-IR spectrum for the synthesized powder are PdMOP in DMSO (0.00005–0.05 wt%), and 0.50 wt% solution
1
shown in Fig. 5, 6, and 7, respectively. The H NMR spectrum of FeMOP in DMSO (0.00005–0.05 wt%). The UV-Vis absor-
1
is shown in Fig. S1a.† H NMR (300 MHz, D O, 25 °C): δ (ppm) bance of a saturated solution of each MOP were measured to
2
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
calculate the solubility in wt%. The saturated solutions were polymer phase. The use of the soluble MOPs as fillers led to
diluted as needed to keep the total absorbance below 1.00: homogeneous films when conditions were identified such that
1
00-fold for CuMOP and 10 000-fold for PdMOP and FeMOP.
both the polymer and metallacages were mutually soluble. The
homogeneities of the resulting MMMs were greatly improved
relative to that of a MOF control MMM, indicating superior
Mixed-matrix membrane preparation
Method I: Membranes were prepared using an adapted pro- polymer-filler integration. This improved homogeneity was
6
2
cedure from a previously published strategy. A solution of established using microscopy that revealed that the mem-
pure PVDF was made by mixing 0.125 g PVDF in 3.5 mL DMF branes containing MOPs were well-dispersed with no observa-
or DMSO in a vortex mixer to prepare 3.6 wt% solution in DMF ble microscopic defects. In contrast, the MOF-5 analogue had
and 3.1 wt% solution in DMSO. Filler solutions at 0.5, 2.0, and significant phase separated domains. PdMOP was the most
4
0
.0 wt% were prepared by mixing 0.00625 g, 0.0250 g, and soluble filler in the membrane solvent, resulting in the most
.050 g filler in 1.25 mL solvent (DMF for CuMOP and MOF-5 homogeneous membrane. The microcrystalline precursor fully
or DMSO for PdMOP and FeMOP), then sonicating for 30 min. dissolved such that no discrete domains were observable by
Mixed-matrix membranes (MMMs) of MOP and PVDF at 5, 20, SEM. In addition, the semi-crystalline binder, PVDF, showed a
and 30 wt% were made by mixing pure PVDF and filler solu- significant decrease in crystallinity upon addition of MOP
tions at 0.5, 2.0, and 4.0 wt%, respectively. The mixtures were fillers. TGA of the fabricated films show an improvement in
sonicated for 30 min, then cast on a 10 × 10 cm rectangular the thermal stability of PVDF upon addition of filler materials.
glass plates (GE Health Bio-Sciences Corp.) using a casting A significant increase in gas permeability was observed for the
knife (∼50 mm width Adjustable Film Applicator, MTI Corp.) MMMs. More importantly, this increased permeability did not
set to 0.20 mm height. Each membrane was dried in a vacuum come at the cost of selectivity. Instead, interesting gas separ-
oven at 80 °C for at least 3 h. After cooling to room tempera- ation properties for the MMMs were observed, particularly for
ture, the membranes were delaminated by immersing in CO
2
/N
approach to forming high quality, homogeneous MMMs from
Method II: Membranes were made with some modifications MOPs will facilitate a matching of separation performances
2
selectivity by the FeMOP. The generality of this
3 2
(CH ) CO solution.
4
4,52
to the basic membrane preparation.
MMM solutions of between filler and polymer, providing a route to superior gas
PVDF and filler at 20 wt% in DMF and DMSO were prepared by separation materials using self-assembled fillers and highly
mixing 0.025 g filler material with 4.75 mL solvent (DMF for processable polymers.
CuMOP and MOF-5 or DMSO for PdMOP and FeMOP). The
solution was sonicated for 30 min. Then 0.002 g PVDF (2% of
total PVDF needed) was added and the mixture was stirred for
Conflicts of interest
1
3
2 h at room temperature. The mixture was then sonicated for
0 min before the remaining 0.098 g PVDF was added. The There are no conflicts of interest to declare.
resulting mixture was stirred for 48 h at room temperature.
After stirring, the mixture was sonicated for 30 min then cast
the same way as the basic membranes. The membrane was
dried in a vacuum oven at 80 °C for at least 3 h after which,
Acknowledgements
the temperature was increased to 120 °C and the membrane TRC and HL thank the University at Buffalo Innovative Micro-
was left to dry for 24 h. After cooling to room temperature, the Programs Accelerating Collaboration in Themes (IMPACT)
membrane was delaminated by immersing in (CH ) CO program for support.
3
2
solution.
Gas permeability and selectivity
Notes and references
A constant-volume and variable-pressure system was used to
6
3
1 D. F. Stamatialis, B. J. Papenburg, M. Gironés, S. Saiful,
S. N. M. Bettahalli, S. Schmitmeier and M. Wessling,
J. Membr. Sci., 2008, 308, 1–34.
determine pure gas permeability at 35 °C. A low pressure
transducer made by Honeywell is used to monitor pressure
change in the downstream volume. In a typical case, the
MMMs films were masked using a aluminium foil with an
2
H. Strathmann, in Ullmann’s Encyclopedia of Industrial
Chemistry, Wiley-VCH Verlag GmbH & Co. KGaA, 2000,
DOI: 10.1002/14356007.a16_187.
2
open area of 0.32 cm . The films were mounted in permeation
cells for the measurement. Ideal gas selectivity is the ratio of
64
3 R. W. Baker and I. Books24x, Membrane technology and
applications, third edition, John Wiley & Sons, Chichester,
West Sussex, U.K, 3rd edn, 2012.
the gas permeability.
4
J. Dechnik, J. Gascon, C. J. Doonan, C. Janiak and
C. J. Sumby, Agnew. Chem., Int. Ed., 2017, 56, 9292–9310.
Conclusions
Soluble metal–organic polyhedra, such as CuMOP, PdMOP,
and FeMOP, were used as fillers for MMMs using PVDF as
5 D. Bastani, N. Esmaeili and M. Asadollahi, J. Ind. Eng.
Chem., 2013, 19, 375–393.
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.

View Article Online
Paper
Dalton Transactions
6
7
8
9
M. Wang, J. Zhao, X. Wang, A. Liu and K. K. Gleason, 33 Y. Han, P. Qi, J. Zhou, X. Feng, S. Li, X. Fu, J. Zhao, D. Yu
J. Mater. Chem. A, 2017, 5, 8860–8886.
Y. Zhang, J. Sunarso, S. Liu and R. Wang, Int. J. Greenhouse
Gas Control, 2013, 12, 84–107.
and B. Wang, ACS Appl. Mater. Interfaces, 2015, 7, 26608–
26613.
34 M. S. Denny Jr., J. C. Moreton, L. Benz and S. M. Cohen,
Nat. Rev. Mater., 2016, 1, 16078.
H. B. Tanh Jeazet, C. Staudt and C. Janiak, Dalton Trans.,
2
012, 41, 14003–14027.
35 M. Eddaoudi, J. Kim, J. B. Wachter, H. K. Chae,
M. O’Keeffe and O. M. Yaghi, J. Am. Chem. Soc., 2001, 123,
4368–4369.
T.-S. Chung, L. Y. Jiang, Y. Li and S. Kulprathipanja, Prog.
Polym. Sci., 2007, 32, 483–507.
1
0 S. Wang, X. Li, H. Wu, Z. Tian, Q. Xin, G. He, D. Peng, 36 Q. Xin, Y. Zhang, T. Huo, H. Ye, X. Ding, L. Lin, Y. Zhang,
S. Chen, Y. Yin, Z. Jiang and M. D. Guiver, Energy Environ.
Sci., 2016, 9, 1863–1890.
1 G. George, N. Bhoria, S. AlHallaq, A. Abdala and V. Mittal,
Sep. Purif. Technol., 2016, 158, 333–356.
2 J. Sunarso, S. S. Hashim, Y. S. Lin and S. M. Liu, Sep. Purif.
Technol., 2017, 176, 335–383.
3 C. A. Scholes and U. K. Ghosh, Membranes, 2017, 7, 1–13.
4 H. Li, Z. Song, X. Zhang, Y. Huang, S. Li,
H. Wu and Z. Jiang, J. Membr. Sci., 2016, 508, 84–93.
37 X. Guo, H. Huang, Y. Ban, Q. Yang, Y. Xiao, Y. Li, W. Yang
and C. Zhong, J. Membr. Sci., 2015, 478, 130–139.
38 S. Zhao, X. Cao, Z. Ma, Z. Wang, Z. Qiao, J. Wang and
S. Wang, Ind. Eng. Chem. Res., 2015, 54, 5139–5148.
39 J.-R. Li and H.-C. Zhou, Nat. Chem., 2010, 2, 893–898.
40 T. Kusukawa and M. Fujita, J. Am. Chem. Soc., 2002, 124,
13576–13582.
1
1
1
1
Y. Mao, H. J. Ploehn, Y. Bao and M. Yu, Science, 2013, 342, 41 P. Mal, D. Schultz, K. Beyeh, K. Rissanen and
5–98. J. R. Nitschke, Angew. Chem., Int. Ed., 2008, 47, 8297–8301.
5 W. J. Koros and R. Mahajan, J. Membr. Sci., 2000, 175, 181– 42 H. Li, M. Eddaoudi, M. O’Keeffe and O. M. Yaghi, Nat.
9
1
1
96.
Chem., 1999, 402, 276–279.
1
1
1
1
2
6 M. Ulbricht, Chem. Ing. Tech., 2001, 73, 1608–1608.
7 R. W. Baker, Ind. Eng. Chem. Res., 2002, 41, 1393–1411.
8 L. M. Robeson, J. Membr. Sci., 1991, 62, 165–185.
9 L. M. Robeson, J. Membr. Sci., 2008, 320, 390–400.
0 N. Kosinov, J. Gascon, F. Kapteijn and E. J. M. Hensen,
J. Membr. Sci., 2016, 499, 65–79.
43 R. Gregorio, J. Appl. Polym. Sci., 2006, 100, 3272–3279.
44 E. Ahmadi Feijani, A. Tavasoli and H. Mahdavi, Ind. Eng.
Chem. Res., 2015, 54, 12124–12134.
45 L. Tian, N. Ren, J.-J. Zhang, S.-J. Sun, H.-M. Ye, J.-H. Bai
and R.-F. Wang, J. Chem. Eng. Data, 2009, 54, 69–74.
46 M.-X. Li, Z.-X. Miao, M. Shao, S.-W. Liang and S.-R. Zhu,
Inorg. Chem., 2008, 47, 4481–4489.
47 J. Fang, X. Guo, S. Harada, T. Watari, K. Tanaka, H. Kita
and K.-I. Okamoto, Macromolecules, 2002, 35, 9022–9028.
48 N.-D. H. Gamage, K. A. McDonald and A. J. Matzger, Angew.
Chem., Int. Ed., 2016, 55, 12099–12103.
2
2
2
1 G. Liu, W. Jin and N. Xu, Angew. Chem., Int. Ed., 2016, 55,
1
3384–13397.
2 R. Mahajan and W. J. Koros, Ind. Eng. Chem. Res., 2000, 39,
692–2696.
3 M. G. Buonomenna and G. Golemme, Advanced Materials
2
for Membrane Preparation, SAIF Zone, Bentham Science 49 H.
Publishers, 1st edn, 2012.
Farrokhzad,
T.
Kikhavani,
F.
Monnaie,
S. N. Ashrafizadeh, G. Koeckelberghs, T. Van Gerven and
2
2
4 C. M. Zimmerman, A. Singh and W. J. Koros, J. Membr. Sci.,
B. Van der Bruggen, J. Membr. Sci., 2015, 474, 167–174.
50 K. Nakagawa and Y. Ishida, J. Polym. Sci., Part B: Polym.
Phys., 1973, 11, 1503–1533.
1
997, 137, 145–154.
5 M.-S. Kim, J. Perry Iv, T. C. M. Julien, E. Marangon,
C. Manouat, J. F. Eubank and J. P. Harmon, J. Mater. Chem. 51 P. Costa, J. Silva, V. Sencadas, C. M. Costa, F. Van Hattum,
A, 2015, 3, 13215–13225.
6 M. Kitchin, J. Teo, K. Konstas, C. H. Lau, C. J. Sumby,
J. G. Rocha and S. Lanceros-Méndez, Carbon, 2009, 47,
2590–2599.
2
2
A. W. Thornton, C. J. Doonan and M. R. Hill, J. Mater. 52 Y. Shen and A. C. Lua, Polym. Eng. Sci., 2012, 52, 2106–
Chem. A, 2015, 3, 15241–15247. 2113.
7 J. Ma, Y. Ying, Q. Yang, Y. Ban, H. Huang, X. Guo, Y. Xiao, 53 M. El-Hibri and D. Paul, J. Appl. Polym. Sci., 1986, 31, 2533–
D. Liu, Y. Li, W. Yang and C. Zhong, Chem. Commun., 2015,
1, 4249–4251.
2560.
5
54 H. Lin and B. D. Freeman, J. Membr. Sci., 2004, 239, 105–
2
8 E. V. Perez, K. J. Balkus Jr., J. P. Ferraris and
I. H. Musselman, J. Membr. Sci., 2014, 463, 82–93.
117.
55 A. S. Michaels and H. J. Bixler, J. Polym. Sci., 1961, 50, 413–
2
9 C. Zhao, N. Wang, L. Wang, H. Huang, R. Zhang, F. Yang,
439.
Y. Xie, S. Ji and J.-R. Li, Chem. Commun., 2014, 50, 13921– 56 Z. Mogri and D. R. Paul, Polymer, 2001, 42, 7765–7780.
1
3923.
57 J. Clarke, M. E. Dry, J. J. Montano and W. J. van Zyl, Trans.
Faraday Soc., 1961, 57, 2239–2246.
3
3
3
0 C. Zhao, N. Wang, L. Wang, S. Sheng, H. Fan, F. Yang, S. Ji,
J.-R. Li and J. Yu, AIChE J., 2016, 62, 3706–3716.
1 T. R. Cook, Y.-R. Zheng and P. J. Stang, Chem. Rev., 2013,
58 A. Hakim, T. S. Marliza, N. M. Abu Tahari, R. W. N. Wan
Isahak, R. M. Yusop, W. M. Mohamed Hisham and
A. M. Yarmo, Ind. Eng. Chem. Res., 2016, 55, 7888–7897.
1
13, 734–777.
2 J. B. DeCoste, J. M. S. Denny, G. W. Peterson, J. J. Mahle 59 J. Liu, X. Hou, H. B. Park and H. Lin, Chem. – Eur. J., 2016,
and S. M. Cohen, Chem. Sci., 2016, 7, 2711–2716. 22, 15980–15990.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Dalton Transactions
Paper
6
0 S. Wimmer, P. Castan, F. L. Wimmer and N. P. Johnson, 62 M. S. Denny and S. M. Cohen, Angew. Chem., Int. Ed., 2015,
J. Chem. Soc., Dalton Trans., 1989, 403–412, DOI: 10.1039/
DT9890000403.
1 H. L. Anderson, S. Anderson and J. K. M. Sanders, J. Chem.
54, 9029–9032.
63 H. Lin, T. Kai, B. D. Freeman, S. Kalakkunnath and
D. S. Kalika, Macromolecules, 2005, 38, 8381–8393.
6
Soc., Perkin Trans. 1, 1995, 2231–2245, DOI: 10.1039/ 64 S. Shah, J. Liu, S. Ng, S. Luo, R. Guo, C. Cheng and H. Lin,
P19950002231. J. Polym. Sci., Part B: Polym. Phys., 2016, 54, 1924–1934.
This journal is © The Royal Society of Chemistry 2018
Dalton Trans.