Sei sulla pagina 1di 14

Some Mechanistic Insights into the Action of Facilitating

Agents on Gas Permeation through Glassy Polymeric


Membranes
Md Oayes Midda and Akkihebbal K. Suresh
Dept. of Chemical Engineering, Indian Institute of Technology, Bombay, Mumbai 400076, India

DOI 10.1002/aic.15873
Published online August 4, 2017 in Wiley Online Library (wileyonlinelibrary.com)

Incorporation of facilitating agents is one of the promising strategies being researched in recent years to cross the
Robeson bounds for gas separations using polymeric membranes. The ways in which such inclusions modify the perfor-
mance of membranes are not always clear. Here, we study the performance of two glassy membranes, Polyfurfuryl alco-
hol and Polysulfone, in O2/N2 and CO2/N2 separations, with Cobalt phthalocyanine in insoluble and solubilized forms
as the facilitating agent. The results show that in general, three effects are important: (1) a barrier effect, (2) a facilita-
tion effect, and (3) morphological effects on the polymer matrix due to an incompatibility between the particles and the
polymer. These results provide some insight into the action of facilitating agents in soluble and insoluble form, when
used as membrane additives. V C 2017 American Institute of Chemical Engineers AIChE J, 64: 186–199, 2018

Keywords: composite membrane, poly(furfuryl alcohol), polysulfone, cobalt phthalocyanine, facilitated transport, gas
separation

Introduction ammonia synthesis, Claus process, catalyst regeneration in


FCC, cement kilns, etc.) and medical11–13 (e.g., oxygen masks,
Membrane based separation technology has a significant
oxygen tents, incubators, and hyperbaric chambers) applica-
potential in the process industry to reduce equipment size,
tions, and membrane based technologies do have a potential
operating cost, energy utilization, and waste generation.1,2 The
here.
industrial application of membrane science and technology
Polymeric materials exhibit a trade-off between selectivity
has come after a long history in laboratory studies. Today, it is and permeability, which limits their application in gas sepa-
widely accepted as among the best available for water and rations. As their publication, Robeson’s “upper bounds”14–16
wastewater treatments, having shown significant improve- for the industrially important gas separations, have provided
ments in energy efficiency.3–5 With that history in the back- the benchmarks for membrane materials research.7,17,18
ground, membranes are being considered with ever increasing Efforts have been made to improve the performance of mem-
interest for gas separation, especially in the context of the cur- brane materials through the incorporation of various addi-
rent focus on the environmental impact of industrial processes tives. Facilitating agents, which selectively promote the
and climate change. While the work of last three decades has permeation of the desired component in the mixture, have
resulted in considerable improvements in membrane based gas been particularly attractive. However, in addition to facilita-
separations, challenges remain. Major breakthroughs have tion, other factors, such as polymer-additive compatibility,
been achieved for hydrogen separation.6 For certain other sep- often come into play and it can be difficult to predict the
arations such as CO2 separation, separations of hydrocarbon effect of such incorporation in the absence of insights into
gases and organic vapors from off-gas streams, and so forth, these various mechanisms. In this work, we have studied
membrane based technologies are often competitive with binary separations made up from oxygen, nitrogen and car-
existing technologies.7 Air separation however, remains a bon dioxide, using cobalt phthalocyanine in native and solu-
major area in which other technologies such as cryogenic dis- bilized forms, in an effort to understand the different ways in
tillation and pressure swing adsorption still remain the pre- which such additives influence membrane behavior. An
ferred options. Partial separations leading to oxygen enriched examination of the Robeson plots for the target separations
air (30–50 vol % O2) have a number of industrial1,2,8–10 (e.g., shows that in general, materials which perform at or near the
upper bound tend to be glassy polymers. We have, therefore,
chosen two glassy polymers, Polysulfone (PSf), and Poly(fur-
Additional Supporting Information may be found in the online version of this
article. furyl) alcohol (PFA) as the membrane materials for this
study. The former has been widely studied with much data
Correspondence concerning this article should be addressed to A. K. Suresh at available in the literature for comparison, while the latter has
aksuresh@iitb.ac.in and aksuresh@che.iitb.ac.in.
recently attracted attention45–47 as it can be synthesized from
C 2017 American Institute of Chemical Engineers
V renewable saccharidic sources.

186 January 2018 Vol. 64, No. 1 AIChE Journal


Table 1. Permeability and Selectivity of PSf Composite Membranes with Different Particulate Additives

Polymer Additives (%) P(O2) P (CO2) S(O2/N2) S (CO2/N2) Reference


PSf (Udel 3500) 0 5.97 24.8 35
1% GO-30% ZIF-301 23.8 60.7
2% GO-24% ZIF-301 21.8 60.3
3% GO-18% ZIF-301 17.7 65.8
4% GO-12% ZIF-301 15.05 68.6
5% GO-6% ZIF-301 11.8 70.8
PSf 0 5.55 29.23 36
10% F-MS 5.6 29.47
20% 6.23 29.67
30% 6.79 30.59
40% 8.46 33.31
PSf (Udel 3500) 0 5.9 24.3 37
8% MCM-41 12.6 36
PSf (Udel 3500) 0 1.6 5.9 4.7 24.3 38
8% HZS 2.3 7.2 6.9 41.7
PSf (Udel 3500) 0 1.6 5.9 4.7 24.3 39
16% ZIF-8 2.6 13 8.3 18
16% HKUST-S1C 1.7 8.4 8 38
PSf (Udel 3500) 0 0.84 3.9 4.94 22.94 40
5% SWNT 1.16 5.12 5.04 22.26
10% SWNT 1.23 5.19 5.34 22.56
15% SWNT 1.11 4.52 5.04 20.54
PSf (Udel P-1700) 0 1.4 5.5 5.6 22 41
5% fumed silica 1.75 6.9 5.6 22
10% fumed silica 1.96 8.25 5.3 22.3
15% fumed silica 3.36 11 5.1 16.7
20% fumed silica 5.18 17 4.5 14.8
PSf (Ultason S 6010) 0 1.5 6 5.9 20 42
7.5% MIL-101 (Cr) 2.6 14 5.1 24
14% MIL-101 (Cr) 4.5 23 5.2 23
19% MIL-101 (Cr) 5.8 32 5.8 26
24% MIL-101 (Cr) 7 36 5.4 27
PSf 1.4 5.6 5.6 22.4 43
PSf (Udel, BP-AMOCO) 0 1.3 5.9 44
15% zeolite A 1.5 6.4
25% zeolite A 1.8 7.7
PSf (Udel, AMOCO) 1.29 5.5 5.7 25 45
PSf (Udel P-1700) 0 2.6 6.5 1.86 4.64 46
10% 13X MS 1.4 6.1 5.6 23.46
20% 13X MS 1.6 6.1 5 19.1
[Unit of permeability is in Barrer].
1 Barrer 5 10210 cm3(STP).cm/cm2.s.cmHg.

Literature 40 wt % CoPc content). Kuraoka et al.24 reported an increase


Among the methods which have been explored to improve in selectivity, but a decrease in permeability with CoPc sup-
the performance of native PSf membranes, a particularly dom- ported porous glass. Formation of a superoxide adduct
inant theme in recent years has been the incorporation of par- between O2 and CoPc has also been reported.25–27 Shoji
ticulate additives. Table 1 gives a summary of the literature on et al.28 showed a continuous increase in selectivity with Co(II)
such incorporation in PSf membranes for the two major gas porphyrin complex loading, in a Nafion membrane. Increased
separations of interest here, O2/N2 and CO2/N2. O2/N2 selec- selectivity and decreased permeability have been seen by
tivities of the order of 5–6 and CO2/N2 selectivities of 20–24 Chen and Lai29 for polycarbonate membrane and Co(SalPr)
are reported for native PSf membranes. With some exceptions complex; whereas Ruaan et al.30 reported increases in both
(e.g., MOFs and graphene oxide for CO2/N2 separations), the selectivity and permeability for the same membrane material.
improvements in selectivity have been modest, and this is Both increased selectivity and permeability have also been
especially true for O2/N2 separations because of the small dif- seen by Nishide et al.21 and Tsuchida et al.31 Nishide
ference in kinetic diameters (3.46 Å for O2 vs. 3.64 Å for N2). et al.20,22,32,33 report selectivities as high as 12 with the use of
One, therefore, looks for additives with particular affinity metal porphyrins in membranes. In general, the potential of
towards O2 (usually the more permeable component). Thus, CoPc has not been explored to the same extent as metal por-
for example, several Co(II) compounds have been investigated phyrins, although the two classes of compounds share struc-
as additives in various membranes, and Table 2 provides a tural similarities. Affinity-based additives have shown promise
summary. Transition metal macrocyclics such as pthalocya- in the case of CO2/N2 separations as well. Hosseinkhani
nines and porphyrins have been of particular interest. The et al.,34 used cobalt acetate tetrahydrate as a carrier complex
reversible O2 binding capacity of cobalt phthalocyanine incorporated in sulfonated ethylene-propylene-diene terpoly-
(CoPc), leads to facilitated transport by a hopping mecha- mer (EPDM) membranes, and reported a sevenfold increase in
nism.19–22 Preethi et al.23 studied the O2/N2 separation perfor- CO2 permeability and a threefold increase in CO2/N2 selectiv-
mance of CoPc-imidazole composite membranes and reported ity (also see Table 1 for results with particles functionalised
very high selectivity as well as permeability (as high as 56 for with 4-Aminophenazone39)

AIChE Journal January 2018 Vol. 64, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 187
Table 2. Reported Data on Improvement in O2/N2 Selectivity by the Addition of Co(II) Metal Complexes
in Different Glassy Polymers

Polymer/porous support; Co(II) Complex P S Pressure Temp


Complex (soluble/disperse form) (wt %) (O2) (O2/N2) (bar) (8C) Reference
OIm; CoPc (soluble) 10 20 4 – – 23
20 55 35
40 83.7 56
Porous glass; CoPc (encapsulated) without CoPc 49.7 GPU* 0.93 1.57 25 24
with CoPc 7.97 GPU* 2.9
Nafion membrane; CoTPP (soluble) 0 – 2.1 1 25 28
5 3.2
10 6.5
20 8.5
Nafion membrane; CoFPP (soluble) 0 – 2.1 28
5 5
10 10
20 14
Polycarbonate; Co(SalPr) (dispersed) 0 1 5 3 35 30
3 1.65 6.92
Polycarbonate; Co(SalPr) (dispersed) 0 0.37 8 3 5 29
3 0.33 15
MCAPE; cobalt(II) chelated membrane (dispersed) 0 2.4 2.5 2 30 47
5 5.1 4.9
7.2 6.6 6.4
12.3 8.5 8
14.5 9.1 8.4
Methylimidazole; CoP complex (dispersed) 0 6.4 3.2 5 mmHg 25 21
2.5 9.8 5.7
4.5 23 12
POMPy; CoS (dispersed) 0 1.95 2.3 10 mmHg 35 31
0.6 2.56 3.5
2.5 2.65 4.8
8 3.12 6.5
12 1.47 15
*Reported values are of permeance in GPU. The reference only reports these values without mentioning the thickness of the membrane.

It is seen from the above that while the incorporation of par- 0.10 g), Co(OAc)2 (0.154 mM, 0.027 g) and a catalytic
ticulate additives and facilitating agents often results in an amount of diazabicycloundecene (DBU) in n-pentanol (1 mL)
increase in selectivity, the effect on permeability is sometimes were heated to 1408C with stirring for 24 h under N2 atmo-
positive and sometimes negative. On the whole, the mecha- sphere. Then, the reaction mixture was cooled to room temper-
nisms which are brought into play by the incorporation of ature and allowed to precipitate by adding methanol.
additives are not clear; the variety of results seen suggest more Methanol was used several times for washing the precipitate
than one mechanism at work. and filtered off. The intense blue colour powder formed after
drying is tert-butyl cobalt phthalocyanine (tBu-CoPc).
Materials and Methods
Preparation of PFA and CoPc-PFA membranes
Material used
Since solution casting methods resulted in membranes with
Cylindrical small plate of polysulfone polymer (Udel P-
very poor fluxes, PFA membranes, both pure and CoPc
3500) was kindly gifted by Advanced Solvay Polymers (USA)
and was used to prepare the membranes. Reagent grade chlo- impregnated, were prepared by a spin coating technique in an
roform and dimethylformamide (DMF) supplied by Merck effort to reduce the membrane thickness. The PFA coating
(Germany) were used as solvents without any further purifica- was formed on a support of highly porous commercial ultrafil-
tion. Cobalt phthalocyanine (b-form, 97% dye content) was tration PSf membrane by the following procedure. A piece of
procured from Sigma-Aldrich. For the synthesis of PFA mem- dried PSf membrane (50 3 70 mm2) was fixed on to a glass
branes, reagent grade FA (98%, Sigma-Aldrich), sulfuric acid slide using a double-sided tape to prevent the membrane from
(98%, Sigma-Aldrich) and absolute ethanol (Merck) were rolling up and also to avoid sticking to the glass slide after
used without any further purification as polymeric precursor, polymerization. This was then coated with a PFA precursor
acid catalyst and solvent, respectively. O2, N2, CO2, and He solution (prepared by mixing FA, sulfuric acid and ethanol) in
gases with 99.99% purity were supplied by Med gas n a spin coater (1000 rpm for 60 s). The coated support was then
Equipment. heated at 808C overnight for FA polymerization and solvent
evaporation. Typically, in the fabrication of PFA-250 mem-
Synthesis of soluble tert-butyl CoPc compound brane, a mixture with a FA/H2SO4 molar ratio of 250 was pre-
A tert-butyl derivative of cobalt phthalocyanine was pre- pared by adding 10 g (0.1 mol) of FA into a solution
pared to solubilize the phthalocyanine in the solvents used for containing 0.0397 g (0.004 mol) of H2SO4 and 10 g (0.217
membrane synthesis. The procedure of making tert-butyl mol) of ethanol drop wise.49
cobalt phthalocyanine was as described in the work of Kai- CoPc-PFA composite membranes were prepared using the
pova et al.48 A mixture of 4-tert-butylphthalonitrile (0.46 mM, same procedure, except that the required amount of CoPc

188 DOI 10.1002/aic Published on behalf of the AIChE January 2018 Vol. 64, No. 1 AIChE Journal
Figure 1. Experimental gas permeation setup.
[Color figure can be viewed at wileyonlinelibrary.com]

powder was added in the FA-ethanol solution and stirred mag- 12 h to completely remove the solvent. With this procedure,
netically for 5 hrs followed by 30 min of ultrasonication. The moisture trapping by DMF could be avoided, and defect-free
resulting CoPc-FA ethanol suspension was immediately mixed membranes produced.
with sulfuric acid diluted in ethanol under magnetic stirring
and then spin coated on the PSf membrane followed by heat- Membrane characterization methods
ing at 808C overnight as described earlier. While the concen- All the synthesized native and composite membranes were
tration of CoPc in the casting solution has been used in characterized by morphological analysis and thermal analysis.
designating the loading in the different composite membranes, Microscopic observations were performed by scanning elec-
the actual CoPc content of the membranes has been deter- tron microscopy to evaluate the membrane thickness and mor-
mined separately, as discussed in a subsequent section. phology. The (SEM) images were acquired on a JEOL JSM-
7600F microscope (Jeol Corp. Ltd., Tokyo, Japan). Both the
Preparation of PSf and CoPc-PSf membranes surface and cross-sectional samples were sputtered with plati-
PSf membranes were synthesized by solution casting num and the images were collected by E-SEM operated at
method described elsewhere.38–40 PSf was degassed under 5–20 kV in environmental mode. Cross-section sample of
vacuum at 808C for 24 h to remove adsorbed moisture. Then, PFA membranes were prepared by freeze fracture in liquid
a viscous solution was prepared by dissolving 1.2 g polymer nitrogen, while in the case of PSf membranes, such cross sec-
in 6 mL chloroform (13 wt % of polymer) followed by 24 h of tions were made by cutting the membrane with a very sharp
magnetic stirring. The solution was cast on a glass plate using and fine blade, since the freeze-fracture technique did not
Doctor’s blade (Sheen instruments, UK) and kept at room tem- work well with these composite membranes.
perature in a partially closed desiccator for 24 h to slow down The glass transition temperature (Tg) of pure and composite
the natural evaporation. After that the films were completely membranes were determined by DSC analysis (STA instru-
dried in vacuum at 808C for 24 h. Similarly, to synthesize dis- ments, NETZSCH PC-409, Germany). The temperature was
persed CoPc-PSf membranes, the required weight of CoPc increased from 100 to 3008C at a rate of 38C/min for two con-
was degassed as described above and magnetically stirred in secutive runs in N2 flow and the Tg was measured from middle
6 mL chloroform for 24 h to make a stable dispersion; and point of the slope transition in the DSC curve based on the sec-
then stirred again for 24 h after adding 1.2 g of polymer. This ond run. TG studies were performed in the same instrument
solution was cast on a glass plate and cured as before. from 30 to 10008C at a heating rate of 108C/min by placing 6–
In the case of PSf membranes with soluble CoPc (tBu- 8 mg sample in an 85 lL alumina pan.
CoPc), membranes synthesized as above with chloroform sol- The membranes were also characterized using Fourier trans-
vent showed defects. Defect free membranes could however form infrared spectroscopy (FTIR). FTIR spectra were
be prepared using DMF as the solvent instead of chloroform, recorded on a 3000 Hyperion Bruker Microscope with Vertex
with the following procedure. The required amount of tBu- 80 FTIR system (Germany) spectrometer.
CoPc was dissolved in 4mL DMF and then stirred for 24 h A schematic diagram and image of the gas permeation setup
after adding 0.8 g of polymer. Thereafter, the viscous solution is shown in Figure 1 which was used to measure membrane
was cast on a glass plate inside an oven (maintained at 458C parameters (permeability and selectivity). The setup was simi-
and kept for 15 mins). The glass plate was dried again consec- lar to the one described by David et al.50 The setup can be
utively at 608C and 708C for 15 mins each. Finally, the mem- used to study both pure as well as mixed gas permeation
brane was formed after heating the glass plate at 1508C for through membranes. Brooks-make calibrated mass flow

AIChE Journal January 2018 Vol. 64, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 189
Morphological studies using SEM
SEM images of CoPc powder are shown in Figures 3a, b.
The length of the long thin needle shaped particles varied
from 10 to 20 mm, with the thickness being in the range 1.7 to
2.4 mm. Figures 4a–d show surface images of CoPc-PFA com-
posite membranes for different loadings (wt %) of CoPc and
Figures 4e–f show the cross-sectional images of 3% CoPc-
PFA composite membranes. These images show that CoPc
Figure 2. Schematic diagram of the double O-ring crystals are essentially unmodified on incorporation in the
membrane test cell. membrane. The thickness of the skin layer on PSf support
[Color figure can be viewed at wileyonlinelibrary.com] made by spin coating was checked at different points along
the radius of the same sample, and the PFA layer thickness
varied from 1.7 to 2.07 mm. This is comparable to the thick-
controllers (MFC) were used to maintain the gas flow rate and
ness of CoPc needles, which are seen to by embed nearly par-
feed composition. The flow rate range of these MFCs was 0 to allel to the surface in the micrographs. For permeability
100 cm3 (STP) min21. A Tescom-make back pressure regula- calculation, a thickness of 2 mm (closer to the upper limit of
tor (BPR) was used on the retentate line to maintain the feed the range measured) has been assumed, because the actual
side pressure (pressure can be varied from 0 to 10 bar). Perme- thickness of the PFA layer might be higher owing to the pene-
ate side was at atmospheric pressure. A helium sweep was tration of PFA solution inside the porous PSf layer.
used to remove the permeated gas. The sweep gas flow rate Figures 5a–e represent the surface images of pure PSf mem-
was controlled using an MFC (range 0 to 50 cm3(STP) brane and CoPc-PSf composite membranes for different CoPc
min21). The permeate gas composition determined by an loadings. While the thickness of these membranes is much
online gas chromatograph fitted with a thermal conductivity more than that of the needles, a large number of CoPc particles
detector and two columns, (a) molecular sieve 5A, and (b) are seen on, or embedded fairly close to, the surface in these
Porapak-Q. The membrane test cell itself contains two com- images, with the density increasing with the particle loading
partments sealed by O-rings as shown in Figure 2. The thin as expected. The dispersion of CoPc particles on the surface is
flat sheet membrane was supported by a porous metallic plate, fairly homogeneous. Figures 6a–e show the cross-sectional
the exposed area of the circular flat sheet membrane being images of pure PSf and PSf-CoPc composite membranes. The
4.91 cm2. thickness of pure PSf membrane is seen to be about 20 mm. A
The gas permeability was calculated using Eq. 1 slightly higher gap had to be maintained in the Doctor’s blade
Qd for the casting of CoPc impregnated polymer so that the result-
P5 (1) ing membranes were defect-free. It was observed that the
ADp
thickness of composite membrane increased with increase in
Where Q is the permeate gas flow rate, d is the membrane CoPc content irrespective of maintaining the same gap. The
thickness, A is the exposed area of membrane and Dp is the thickness of the composite membranes (as measured from
(partial) pressure gradient across the membrane. The permeate cross sectional images) is reported in Table 9. The cross-
flow rate (Q) is the product of the gas concentration (mol/m3 sectional images do show the evidence for the presence of par-
or m3/m3) as measured by the GC and the helium sweep gas ticles, but probably to a lesser extent than the surface region.
flow rate.
FTIR studies
Results and Discussion Structural characterizations of CoPc, as well as the pure and
composite membranes using FTIR are shown in Figure 7. In
Characterization of membranes with and without
spectrum “a” for CoPc, sharp intense peaks associated with
dispersed CoPc different bond orientations may be distinguished: 731 cm21
Here, we describe the characterization of the additives stud- (C-H out of plane deformation), 780 cm21 (C–N in-plane
ied in this work and of the membranes with and without the stretching vibration), 912 cm21 (Metal ligand vibration),
additives. Table 3 shows the structure of the polymers as well 1087 cm21 (C–H in-plane deformation), 1120 cm21 (C–H
as the images of as-synthesized pure and composite mem- in-plane bending), 1166 cm21 (C–N in-plane bending),
branes (with CoPc) from the two different glassy polymers. 1288 cm21 (C–N stretching in isoindole), 1331 and 1423 cm21

Table 3. The Two Polymers Used in This Work and Images of Membranes Made from Them with and without CoPc [Color
table can be viewed at wileyonlinelibrary.com]

190 DOI 10.1002/aic Published on behalf of the AIChE January 2018 Vol. 64, No. 1 AIChE Journal
Figure 3. SEM images of CoPc particles (a) lower mag and (b) higher mag. (scale bar 5 1 mm).
[Color figure can be viewed at wileyonlinelibrary.com]

(C–C stretching in isoindole), 1521 cm21 (C 5 N stretching), from that in the solution used. The actual CoPc wt % was,
and 1605 cm21 (C 5 C macrocycle ring deformation).51 The therefore, determined by thermogravimetry on the delami-
spectra “b” and “d,” for pure PSf and PFA membranes, respec- nated PFA samples. Figure 8 shows the TGA mass loss curves
tively, clearly show the peaks expected from their structure. for the different membranes obtained at a heating rate of 108C/
Incorporation of CoPc does not change the signature of the min up to 6008C in N2 atmosphere. The reasons of choosing
polymer as seen from spectra “b-e.” In spectra “b” and “c,” the this condition are: a significant weight loss of the pure poly-
peak at 1160 and 1241cm21 corresponds to the vibration of dia- mer (TGA with pure PFA up to 8008C showed that about 95%
ryl sulfone (Ar-SO2-Ar) and diaryl ether (Ar-O-Ar) groups pre- of the mass loss occurs by 6008C) and no weight loss of pure
sent in PSf monomer unit. The peaks at 1494 and 1582 cm21 CoPc within the temperature range (CoPc starts to degrade
represent the vibration of aromatic C 5 C in PSf molecule.52,53 above 6008C), the difference in residual mass between differ-
The bands at 2970 and 3058 cm21 belong to the asymmetric and ent membranes is attributable to the amount of CoPc in the
symmetric stretching vibration of –CH3 group.54 In spectra “d” membrane. CoPc wt % in membrane was also cross-checked
and “e,” the broad band at 3237 cm21 is ascribed to the O-H with EDAX data, which is reported in Table 4 along with the
stretching. The peak at 1241 cm21 corresponds to the stretching result from TGA studies. TGA and EDAX data are in reason-
of C-O-C bond. The bands at 551, 841, 1100, 1241, 1486, and able agreement, and show that the actual concentration of
1582 cm21 in the PFA spectra are associated with furan rings. CoPc is less than what is expected on the basis of the concen-
tration in solution.
TGA analysis of CoPc-PFA composite membranes
During the preparation of CoPc-PFA composite mem- TGA analysis of CoPc-PSf composite membranes
branes, as the particle-polymer solution was coated by spin The TGA analysis of CoPc-PSf composite membranes was
coating, it is not safe to infer the CoPc wt % in membrane conducted in an air atmosphere up to 10008C. This change in

Figure 4. ESEM images of CoPc-PFA membranes (a) PFA surface, (b) 3%CoPc-PFA, (c) 7%CoPc-PFA, (d) 9%CoPc-
PFA; (e) x-section of 3%CoPc-PFA, (f) x-section at higher magnification.

AIChE Journal January 2018 Vol. 64, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 191
Figure 5. Surface images of (a) pure PSf, (b) 5%CoPc-PSf, (c) 10%CoPc-PSf, (d) 15%CoPc-PSf,
and (e) 20%CoPc-PSf.

procedure and conditions (from what were used with PFA apparent weight losses: one at around 425–5308C and the other
membranes) was necessitated since the decomposition of PSf at 540–7408C. The first weight loss (around 50 wt %) might be
membranes was incomplete under the previously used condi- attributed to pyrolysis processes, releasing SO2, benzene, phe-
tions. Therefore, the corresponding TG curves (shown in Fig- nol, toluene, styrene, xylene, and so forth.37 The second
ure 9) were taken in an oxidizing atmosphere up to 10008C to weight loss (around 47 wt %) is the result of complete degra-
ensure the stability of TGA curves and the conversion of CoPc dation of the polymer chain. The residual mass when CoPc
into cobalt oxide. In general, the membranes showed two was subjected to thermogravimetry under the same conditions,

Figure 6. X-sectional images of (a) pure PSf, (b) 5%CoPc, (c) 10%CoPc, (d) 15%CoPc, and (e) 20%CoPc composite
membranes.
[Color figure can be viewed at wileyonlinelibrary.com]

192 DOI 10.1002/aic Published on behalf of the AIChE January 2018 Vol. 64, No. 1 AIChE Journal
Table 4. Actual CoPc Content in PFA Composite
Membranes
%CoPc content %CoPc content
Membrane (based on TGA) (based on EDAX)
PFA-5%CoPc 3.11 3.03 6 0.5
PFA-10%CoPc 6.95 5.21 6 0.6
PFA-15%CoPc 9.06 11.48 6 0.9

higher than the pure membrane, which implies CoPc impreg-


nation does not significantly affect the polymer properties.
In similar manner, the Glass transition temperature (Tg) of
pure PFA was determined to be 1458C by DSC analysis, which
is close to the value reported in the literature.57
Volume fraction of CoPc in the composite membranes
For explaining the experimental results with composite
Figure 7. FTIR spectra of powder CoPc, pure and com- membranes (in which CoPc particles are dispersed in the poly-
posite membranes. mer matrix) on the basis of available theories, we need to
[Color figure can be viewed at wileyonlinelibrary.com] know the volume fraction of the particles in the membrane.
This was determined by independently determining the densi-
corresponds to Co2O3. Therefore, from the residual mass of ties of the CoPc particles and the native polymer membrane,
TGA curves, exact CoPc content of the membranes were back using Eq. 2
calculated. CoPc wt % in membrane was cross checked with  
particle wt %
EDAX data. All results are reported in Table 5. Once again, particle density
u5   (2)
the TGA and EDAX results are comparable, but the values for particle wt % ð1002particle wt %Þ
PSf are closer to what is expected on the basis of the concen- particle density 1 polymer density
trations in the solutions used than in the case of PFA.
where the particle wt % is determined from TGA studies as
described in previous section. The polymer density was deter-
DSC Analysis of CoPc-PSf Composite Membranes mined as follows. A known weight of the polymer was placed
Phase transition of polymer from glassy to rubbery state in the pycnometer, and soaked in diesel (chosen because both
occurs at the glass transition temperature, much earlier than the polymers show negligible swelling in diesel) by filling the
polymer decomposition temperature. The glass transition tem- pycnometer. The weight of the filled pycnometer gives the
perature (Tg) was measured from DSC analysis. The DSC weight of diesel. Using the density of diesel (measured in the
analyses of all the membranes were conducted by heating the same pycnometer) this weight was converted to volume. The
sample up to 1008C and then to 3008C at 38C/min for two con- polymer density was then calculated as
secutive runs. The polymer glass transition temperature (Tg) mpol mpol
qpol 5 5  (3)
was taken from middle point of the slope transition in the DSC Vpol Vpycnometer 2 mdiesel
qdiesel
curve based on the second run. It is a good idea to heat the
sample a second time because the first heating freezes any pos- The particle density was also similarly determined. The densi-
sible metastable states.55,56 Figure 10 shows the variation in ties so measured are shown in Table 6. Based on TG and den-
glass transition temperature (Tg varies from 186 to 188.58C) sity studies, the physical characteristics of the membranes
with CoPc wt %. Tg of composite membranes are only slightly were determined as shown in Table 7.

Figure 8. Thermograms of CoPc-PFA composite Figure 9. Thermograms of CoPc-PSf composite


membranes. membranes.

AIChE Journal January 2018 Vol. 64, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 193
Table 5. CoPc Content in PSf Composite Membranes Table 6. Density of Polymers and Powder CoPc
Verified by Different Techniques
Density (g/cc)
CoPc in Solution (wt %CoPc Content %CoPc Content Diesel 0.91
CoPc/wt Polymer) Based on TGA Based on EDAX PFA Membrane 1.50
5%CoPc 4.48 4.49 6 0.5 PSf Membrane 1.31
10%CoPc 9.36 10.45 6 0.7 CoPc Particle 1.69
15%CoPc 13.44 14.8 6 0.9
20%CoPc 19.51 19.36 6 1
“hop”). The facilitation, however, is not very effective in the
insoluble state of the CoPc, and, in any case, not significant
Performance of PFA and dispersed CoPc-PFA enough to counteract the barrier effect. The trends observed
composite membranes can thus be qualitatively explained as being due primarily to a
barrier effect of the particles, with some facilitation coming
The performance of PFA and CoPc-PFA composite mem- into play as the loading is increased. Cussler et al.59 have mod-
branes with pure gases (O2 and N2) is shown in Table 8 and eled the barrier effect due to impermeable flakes and lamellae
Figure 11. The permeabilities are quite high in general, in several idealized configurations. For the case of randomly
although the selectivities are not, being in the range of 2–3. placed flakes, the model gives
Our figures for the permeabilities of oxygen and nitrogen in
the pure PFA membrane may be compared with those reported Pm lb2 /2
511 (4)
by He et al.,58 of 3.33 Barrer for oxygen and 2.67 Barrer for PC 12/
nitrogen. Our permeabities (as well as the selectivity) are
higher; differences in the preparation method may account for where m is a geometric factor which may be calculated for our
this (He et al.49 found the FA to acid ratio to influence the per- case by fitting our data to Eq. 4. Such a procedure gives a
meation property of the membrane in H2/N2 separation; fur- value of 1.869 for l, and the fit of the model is shown by
ther, our membranes were made by a spin-coating technique dashed lines in Figure 11. The fit is better for oxygen than for
nitrogen. As this model does not take any facilitation into
while theirs were made by a hand-casting technique). Both
account, the selectivity improvement seen in the experiments
oxygen and nitrogen permeabilities show a decreasing trend
would not be explained by the model. While the fit and selec-
with CoPc loading (the permeability for oxygen is almost con-
tivity predictions could be improved by fitting different values
stant up to a loading of 3 wt % CoPc). However, the decrease
of l to oxygen and nitrogen, the precepts of the model are not
in the permeability of nitrogen is more pronounced, the net
consistent with such an assumption; such a sophistication was
result being a continuous increase in membrane selectivity
also not considered warranted by the limited amount of data
from 1.6 for pure PFA membrane to 2.64 for the highest CoPc
available here.
loading studied, a 65% increase. These results may be com-
Figure 12 represents the performance of these membranes
pared with the literature studies reported in Table 2. While in relation to the Robeson’s bounds. It is seen that the perfor-
most papers report an increase in selectivity with particle load- mance of the membrane in general moves closer to the bound
ing, a decrease in O2 permeability such as observed for PFA in as the CoPc loading is increased.
this study has only been reported by Chen and Lai 29 for poly-
carbonate membrane. As the CoPc is insoluble and the par- Performance of PSf and dispersed CoPc-PSf composite
ticles are nonporous, the observed trends are suggestive of a membranes
barrier effect as the loading is increased. The barrier effect is Table 9 gives a summary of the experimental results on
less pronounced for oxygen probably because of facilitation PSf-CoPc composite membranes. Gas permeability and selec-
by the surface molecules on the particles, as the loading tivity values are tabulated with the percent decrease or
increases and the average distance between CoPc sites increase (with respect to the native membrane) shown in
decreases (as this is the distance the oxygen molecules have to parenthesis. In addition to air separation data, the table also
shows some data on CO2/N2 separation, which will be dis-
cussed in the succeeding paragraphs. All the permeability and
selectivity values for the pure PSf (without CoPc) membranes
compare well with the literature values quoted in Table 1.
With the O2/N2 system, it is seen that the PSf membranes in
general have a better selectivity but a lower permeability as
compared to the PFA membranes. The selectivity shows a
marginal increase with loading. The permeabilities for both

Table 7. Particle Volume Fraction of Different CoPc-PSf


Composite Membranes
CoPc wt % CoPc wt % Volume
in Solution (Measured by TGA) Fraction (/)
1 1 0.008
2 2 0.016
5 4.5 0.039
10 9.4 0.082
Figure 10. Variation in Tg of PSf membranes with CoPc 15 13.4 0.119
loading. 20 19.5 0.176

194 DOI 10.1002/aic Published on behalf of the AIChE January 2018 Vol. 64, No. 1 AIChE Journal
Table 8. O2/N2 Separation Performance of PFA Composite
Membranes with CoPc Loading at 6 Bar and 308C (percent
change from the value for native membrane is shown in
parenthesis)
CoPc P (O2) P (N2) S
wt % (Barrer) (Barrer) (O2/N2)
0 14.80 9.2 1.61
3.11 15.33 (3.6%) 8.66 (–5.9%) 1.77 (10%)
6.95 13.88 (–6.2%) 6.39 (–30.5%) 2.17 (34.9%)
9.06 11.93 (–19.4%) 4.53 (–50.8%) 2.64 (63.8%)

oxygen and nitrogen are lower with CoPc incorporation, but


the trends are different from what was seen for PFA. The per-
meabilities first decrease and then increase as the loading is
increased, with a clear minimum at a loading of 5%. The vari- Figure 12. O2/N2 separation performance of PFA com-
ation in permeability is less pronounced for N2; the values for posite membranes with respect to Robeson
oxygen increase almost to the level of the native membrane as upper bound.
the loading is increased. Figure 13 shows the variation in per-
meability and selectivity with CoPc wt %. In Figure 15, these
The comparison with Robeson’s limits (Figure 17) also under-
results are juxtaposed against the Robeson upper bound, and it
lines this observation.
is seen that 10% loading represents an optimum in terms of
Studies in the literature attribute any decrease in permeabil-
approach to the bound.
ity on particle incorporation in membranes to a simple barrier
Metal pthalocyanins and porphyrins are molecules inspired
effect (as seen above in this work also, for PFA membranes)
by hemoglobin. Hemoglobin shows a reversible binding abil-
or to polymer rigidification around the particle (which in turn
ity not only with oxygen but also with CO2. Since CO2/N2 sep-
contributes to a barrier effect). Thus, Li et al.60,61 observed a
aration is also an industrially important problem which we are
decreased permeability and an increased selectivity on incor-
addressing in our lab, we thought it would be interesting to see poration of zeolite 5A in their polyether sulfone (PES) mem-
how the CoPc-PSf membranes perform on this separation as branes. Similar results have also been reported by Suer et al.62
well. The work of Hosseinkhani et al.,34 which argues a facili- for 13X-zeolite-PES membrane, and by Mahajan et al.63 for
tated transport of CO2 transport through incorporation of a 4A zeolite-polyimide membranes. While CoPc does have a
cobalt complex (cobalt acetate tetrahydrate) has been dis- reversible binding ability with both oxygen and carbon diox-
cussed earlier. The reversible binding ability is explained by ide, with the CoPc being in insoluble form, a “hopping” of
four- or five-coordinated square complex formation. This bound molecules from one site to another only becomes signif-
square complex formation is reported by Tsuchida et al.31 and icant when the sites are not too far apart, that is, at higher load-
Nishide at al.,21 mainly for facilitated O2 transport through ings. Therefore, the permeability data for low loadings (5% or
CoPc or CoP impregnated membranes, but has been seen in less) was fitted with the models for barrier membrane as
CO2 transport as well. explained in the previous section (Eq. 4) for randomly placed
Results of permeation studies on the CO2/N2 system are impermeable flakes. The fitted value of the geometric factor
summarized in Table 9 and Figure 16. The changes in perme- (l) for randomly placed particles is 11.78. The fitted curves
ability and selectivity follow nearly the same trend as for are represented by dashed lines in Figures 13,14,16. Cussler
O2-N2 system, but overall, the results are even more encourag- et al.59 have also modeled the case of regularly placed flakes,
ing than for O2/N2 separations, with permeability values for which predicts the decrease in permeability in terms of two
CO2 exceeding that for native membrane at high loadings, and parameters, an aspect ratio for the flakes and an aspect ratio
a maximum improvement in selectivity of the order of 33%. for the pores. Using the value for the former from the SEM
micrographs and fitting the value of the latter gave a slightly
better fit to the data than Eq. 4. These results are not shown
here as a random placement of flakes was considered more
realistic.
As the CoPc loading is increased, it is seen that the perme-
abilities increase. The increase in permeability between load-
ings of 5 and 20% is in the order CO2 > N2 > O2. This
ordering suggests that facilitation is unlikely to be the sole
mechanism, as nitrogen certainly is not influenced by such a
mechanism. A combination of facilitation (if any) and mor-
phological effects arising out of an incompatibility between
the particulate phase and the polymer matrix (the SEM studies
of Figure 6 do suggest such possibilities) are likely to be the
cause here. A simple way to model this increase is to use the
Maxwell equation for the effect of a “permeable” particle
incorporation in the membrane. This assumes that having a
Figure 11. Effect of CoPc loading on PFA membrane particle such as CoPc (with any associated void spaces created
performance. in the matrix due to incompatibility between the polymer and

AIChE Journal January 2018 Vol. 64, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 195
Table 9. Membrane Performances of CoPc-PSf Composite Membranes at 6 Bar and 308C

Membrane
Membrane Tg (8C) thickness (mm) P (O2) P (N2) P (CO2) S (O2/N2) S (CO2/N2)
PSf 186.1 20 1.53 0.327 7.04 4.68 21.5
1% CoPc 187.2 28.1 1.37 (–10.5%) 0.281 (–14.2%) 6.11 (–13.2%) 4.88 (4.3%) 21.7 (1.2%)
2% CoPc 187.3 28.4 1.32 (–13.7%) 0.266 (–18.8%) 5.75 (–18.3%) 4.97 (6.2%) 21.6 (0.6%)
5% CoPc 187.4 30.02 1.26 (–17.5%) 0.225 (–31.4%) 5.07 (–27.9%) 5.62 (20.3%) 22.6 (5.1%)
10% CoPc 187.4 31.84 1.41 (–7.9%) 0.232 (–29%) 6.07 (–13.8%) 6.06 (29.6%) 26.1 (21.4%)
15% CoPc 187.8 33.89 1.43 (–6.5%) 0.256 (–21.9%) 7.35 (4.4%) 5.58 (19.2%) 28.7 (33.5%)
20% CoPc 188.4 34.6 1.45 (–3.9%) 0.287 (–12.3%) 7.7 (9.9%) 5.12 (9.5%) 26.9 (25.3%)
Parenthesis values represent percent increase or decrease in permeability or selectivity value.

the particles) is equivalent to having a dispersed phase of The O2 and N2 separation performance of all these mem-
higher permeability in the polymer matrix. According to the branes (prepared using DMF) was studied at a feed pressure of
Maxwell model 6 bar, and the results are shown in Table 10 and Figure 18.
 21 Figure 18 compares the effect of CoPc loading in dispersed vs.
PC a1Ap
511ð11AÞ/ 2/ (5) in soluble form, on O2 permeability and O2/N2 selectivity. It is
Pm a21 seen that, while the O2 permeability decreases with increasing
CoPc wt % at low loadings of dispersed CoPc, it increases
where the parameter Ap depends on the shape of particle
continuously within the studied range for soluble CoPc. Table
(Ap 5 1 for transverse cylinder and 2 for spherical particles)
10 shows that the presence of solube CoPc has no significant
and a is the permeability ratio of dispersed phase to continu-
effect on the permeability of nitrogen. As a result, the O2/N2
ous phase. As, the CoPc particles are needle shaped as seen
from SEM image, the A value has been taken as 1 (for trans- selectivity increases with tBu-CoPc loading, and the increase
verse cylinder) and a estimated so as to provide a best fit to is larger than in the case of CoPc particles.
the experimental data of CO2, O2, and N2. The value of a for These results, therefore, indicate that the presence of molec-
CO2, O2, and N2 is 22.2, 2.75, and 3.99 respectively. The fit of ularly dispersed tBu-CoPc molecules facilitates the transport
Eq. 5 is shown in Figures 13,14,16 for the rising part of the of oxygen as expected, leaving that of nitrogen essentially
permeability curve. unchanged. Nishide et al.64 have studied the facilitated O2
transport through soluble cobalt porphyrin dispersed in polyvi-
Performance of soluble CoPc-PSf composite membranes nylidene dichloride membrane using an extended dual-mode
The results reported above show that in general, any facili- transport model represented by Eq. 6.
tation due to CoPc incorporation is not much in evidence = =
CH bDH CC DC K
when the CoPc is in the form of insoluble needles, any facilita- P5kD DD 1 1 (6)
11bp 11Kp
tion being limited by the fact that it is only the surface mole-
cules of the CoPc particles that can contribute in this fashion. Here, the first term is due to the diffusion of dissolved gas
It was therefore felt that, if the CoPc could be molecularly dis- molecules through polymer matrix, the concentration of the
persed by solubilizing it, better results could be expected. dissolved gas being governed by Henry’s law. The second
Here, we present the characterization and performance results term is due to the surface diffusion of adsorbed gas, the
with tBu-CoPc, a soluble derivative of CoPc, synthesized with adsorption being governed by the Langmuir isotherm. These
this objective. PSf having shown better performance in the two terms govern the transport in the native polymer. In the
earlier described experiments, it was chosen as the membrane presence of a facilitating agent, the third term is included, and
material for these studies. Details of characterization of the represents the effect of additional sorption-type sites contrib-
synthesized tBu-CoPc and the soluble tBu-CoPc-PSf compos- uted by the agent for the particular species whose transport it
ite membranes are available in the Supporting Information. facilitates (oxygen in the present case). We have used this

Figure 13. O2 separation performance of PSf mem- Figure 14. N2 separation performance of PSf mem-
branes with CoPc wt %. branes with CoPc wt %.

196 DOI 10.1002/aic Published on behalf of the AIChE January 2018 Vol. 64, No. 1 AIChE Journal
Figure 15. O2/N2 separation performance of CoPc-PSf Figure 17. CO2/N2 separation performance of CoPc-
composite membranes with respect to PSf composite membranes with respect to
Robeson upper bound. Robeson upper bound.

model for the case of tBu-CoPc dispersed in PSf, in the fol- the parameters for nitrogen transport through the native PSf
lowing manner. First, we include the third term only for oxy- membrane are also compared with the data in Table 10 and
gen and not for nitrogen whose transport is not influenced by Figure 18.
the presence of the facilitating agent. Assuming the additional
sites for oxygen contributed by tBu-CoPc to be proportional to Conclusions
its volume fraction, we write C0C 5a/L in Eq. 6. Denoting aC 5
In recent years, there has been much interest in the use of
aDC we finally get
particulate additives and facilitation agents to improve the per-
C0H bDH aC K/L formance of polymeric membranes in gas separation. This
P5kD DD 1 1 (7)
11bp 11Kp work is focussed on the mechanisms that come into play with
such incorporations, and the complex ways in which they
The values of the parameters were determined as follows. The interact to modify the performance of the membrane. In gen-
value of kD, b and C0H were calculated for each of oxygen and eral, three effects have been recognized—a barrier effect and a
nitrogen by fitting the experimental sorption isotherm data structural modification of the polymer matrix when the
published by Kim and Marand;65 then the values of DD and
DH were calculated by fitting our own experimental data (not
Table 10. Experimental and Calculated Data of Solubilized
reported here) on pure gas permeabilities as a function of par-
CoPc-PSf Membranes
tial pressure. For oxygen, the value of equilibrium constant
(K) was calculated from the DH and DS value of the reversible P (O2) P (N2) S (O2/N2)
reaction reported by Preethi et al.23 and the estimated value of Exp Cal Exp Cal Exp Cal
K is 11.94. The value of aC was then calculated for oxygen by PSf 1.62 1.51 0.321 0.325 5.03 4.63
fitting the data of permeability at different loading from this PSf-1% tBu-CoPc 1.58 1.61 0.288 0.325 5.49 4.96
work, reported below. The fit of Eq. 7 is represented by the PSf-2% tBu-CoPc 1.74 1.72 0.29 0.325 5.99 5.29
solid line in Figure 18. The fitted value of aC (for O2) is 4.93e- PSf-3% tBu-CoPc 1.82 1.83 0.294 0.325 6.2 5.61
09. The fit of Eq. 7 to the data is seen to be satisfactory. The
selectivity predictions based on the fitted value of aC and

Figure 18. Comparison of membrane performance


Figure 16. CO2 separation performance of PSf mem- between dispersed and soluble CoPc-PSf
branes with CoPc wt %. membranes

AIChE Journal January 2018 Vol. 64, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 197
additives are insoluble or partially soluble, and a facilitation b = sorption affinity toward gas molecules
effect, which occurs whether the particles are insoluble or sol- C0H = maximum sorption uptake
uble. CoPc-PFA and CoPc-PSf were the candidate systems DD, DH = diffusion coefficient for Henry and Langmuir mode
respectively
used in the study. CoPc particles are insoluble in the formula- DC = diffusion coefficient of gas molecules by facilitation through
tion and remain dispersed in the polymer matrix at all loadings fixed carrier complex
tested, down to 1%. Incorporation of CoPc into the polymeric K = equilibrium constant
matrix improves the selectivity for CO2 and O2 over N2, and uL = particle loading (wt %)
in general moves the membrane performance closer to the
Robeson upper bounds for these mixtures. However, in dis-
persed form, facilitation is unlikely to play a significant role, Literature Cited
as seen by the permeability data as a function of CoPc loading. 1. Bernardo P, Drioli E, Golemme G. Membrane gas separation: a
The barrier effect is evident because of which the permeability review/state of the art. Ind Eng Chem Res. 2009;48(10):4638–4663.
of all gases decreases in general. While permeabilities increase 2. Baker RW. Membrane Technology and Applications, 2nd ed. New
York: Wiley, 2004.
at higher loadings with PSf, this is likely to be due more to 3. Neoh CH, Noor ZZ, Mutamim NSA, Lim CK. Green technology in
morphological changes wrought in the polymer matrix by the wastewater treatment technologies: integration of membrane bioreac-
addition of CoPc particles than to any facilitation effect per se. tor with various wastewater treatment systems. Chem Eng J. 2016;
Facilitation, however, comes into its own when the CoPc is 283:582–594.
derivatized so as to make it soluble in the polymer matrix, and 4. Hussain A, Al-Rawajfeh AE, Alsaraierh H. Membrane bio reactors
(MBR) in waste water treatment: a review of the recent patents.
much better results are obtained in terms of performance for Recent Pat Biotechnol. 2010;4(1):65–80.
both gas mixtures studied here. 5. Tan JM, Qiu G, Ting YP. Osmotic membrane bioreactor for munici-
The observed trends with dispersed CoPc have been ratio- pal wastewater treatment and the effects of siler nanoparticles on
nalized on the basis of known concepts of barrier effect and system performance. J Cleaner Prod. 2015;88:287–297.
permeation in composite membranes. The results with the sol- 6. Stolten D, Emonts B. Hydrogen Science and Engineering. Wiley-
VCH Verlag GmbH & Co. KGaA, 2016.
ubilized CoPc has been explained on the basis of an extended 7. Baker RW, Low BT. Gas separation membrane materials: a perspec-
dual mode sorption mechanism, in which the facilitation effect tive. Macromolecules. 2014;47(20):6999–7013.
of the tBu-CoPc molecules is modeled in terms of additional 8. Bhasin D, Liebelson M, Chapman G. Oxygen increases FCC thruput.
sorption sites becoming available. Hydrocarbon Process. 1983;62(9):85–88.
9. Tong J, Yang W, Zhu B, Cai R. Investigation of ideal zirconium-
doped perovskite-type ceramic membrane materials for oxygen sepa-
Acknowledgments ration. J Membr Sci. 2002;203(1–2):175–189.
The authors are acknowledged to the Australia India Stra- 10. Wang H, Werth S, Schiestel T, Caro J. Perovskite hollow-fiber mem-
branes for the production of oxygen-enriched air. Angew Chem Int
tegic Research Fund (AISRF) and Department of Science Ed. 2005;44(42):6906–6909.
and Technology (DST, India) for financial support. We also 11. Holley MT, Schmidt JL. Hyperbaric exercise facility, hyperbaric
thank Mr. Sunit Kumar (student of Prof. Mangalampalli dome, catastrophe or civil defense shelter. US Patent, US8739792
Ravikanth, Chemistry Dept., IIT Bombay) for the prepara- B2, 2014.
tion and characterization of soluble tert-butyl cobalt phthalo- 12. Slaker BF, Gettys A, Seward JA, Valley BO, Jaffer AN, Maida JR,
Hoover F-AE. Respiratory treatment delivery system. US Patent,
cyanine compound. US8733355 B2, 2013.
13. Turiello AJ. Breathable air safety system and method. US Patent,
Notation US8733355 B2, 2014.
14. Robeson LM. Correlation of separation factor versus permeability
Co(SalPr) = di(salicylal)-3,3’-diimino-di-n-propylamin for polymeric membranes. J Mater Chem. 1991;62:165–185.
CoP = cobalt(II) porphyrin 15. Robeson LM. The upper bound revisited. J Membr Sci. 2008;320(1–
CoPc = cobalt(II) phthalocyanine 2):390–400.
CoS = N,N’-Disalicylideneethylenediamine cobalt(II) 16. Robeson LM, Freeman BD, Paul DR, Rowe BW. An empirical cor-
DMF = dimethylformamide relation of gas permeability and permselectivity in polymers and its
F-MS = functionalized mesoporous silica
theoretical basis. J Membr Sci. 2009;341(1–2):178–185.
GO = graphene oxide
17. Sanders DF, Smith ZP, Guo R, Robeson LM, McGrath JE, Paul DR,
HZS = hollow zeolite spheres
Freeman BD. Energy-efficient polymeric gas separation membranes
MCAPE = 2-methylacrylic acid 3-(bis-carboxymethyl-amino)-2-
for a sustainable future: a review. Polymer. 2013;54(18):4729–4761.
hydroxy-propyl ester
18. Yampolskii Y. Polymeric gas separation membranes. Macromole-
MS = mesoporous silica
OIm = poly(octyl methacrylate-co-vinylimidazole) cules. 2012;45(8):3298–3311.
PFA = polyfurfuryl alcohol 19. Sakamoto K, Ohno-Okumura E. Syntheses and functional properties
POMPy = poly [(octyl methacrylate)-co-(4-vinylpyridine)] of phthalocyanines. Materials. 2009;2(3):1127–1179.
PSf = polysulfone 20. Nishide H, Kawakami H, Suzuki T, Azechi Y, Tsuchida E.
SWNT = single walled nano tubes Enhanced stability and facilitation in the oxygen transport through
tBu-CoPc = tert butyl cobalt phthalocyanine cobalt porphyrin polymer membranes. Macromolecules. 1990;23(15):
3714–3716.
21. Nishide H, Ohyanagi M, Okada O, Tsuchida E. Dual-mode transport
Abbreviations of molecular oxygen in a membrane containing a cobalt porphyrin
J0, JC = permeation flux of bare membrane and composite membrane complex as a fixed carrier. Macromolecules. 1987;20:417–422.
respectively 22. Nishide H, Suzuki T, Kawakami H, Tsuchida E. Cobalt porphyrin-
Pm, PC = permeability of bare membrane and composite membrane mediated oxygen transport in a polymer membrane: effect of the
respectively cobalt porphyrin structure on the oxygen-binding reaction, oxygen-
a = selectivity diffusion constants, and oxygen-transport efficiency. J Phys Chem.
b = particle aspect ratio 1994;98(19):5084–5088.
u = volume fraction 23. Preethi N, Shinohara H, Nishide H. Reversible oxygen-binding and
m = geometric factor for randomly placed particles facilitated oxygen transport in membranes of polyvinylimidazole
Ap = particle shape factor complexed with cobalt-phthalocyanine. React Funct Polym. 2006;
kD = Henry’s law constant 66(8):851–855.

198 DOI 10.1002/aic Published on behalf of the AIChE January 2018 Vol. 64, No. 1 AIChE Journal
24. Kuraoka K, Ueda T, Fujiwara M, Sato M. Ship-in-a-bottle synthesis 47. Wang CC, Cheng MH, Chen CY, Chen CY. Facilitated transport of
of a cobalt phthalocyanine/porous glass composite membrane for molecular oxygen in cobalt-chelated copolymer membranes prepared
oxygen separation. J Membr Sci. 2006;286(1–2):12–14. by soap-free emulsion polymerization. J Membr Sci. 2002;208(1–2):
25. Bohrer FI, Colesniuc CN, Park J, Ruidiaz ME, Schuller IK, Kummel 133–145.
AC, Trogler WC. Comparative gas sensing in cobalt, nickel, copper, 48. Kaipova S, Dinçer H, Altindal A. Synthesis, characterization, con-
zinc, and metal-free phthalocyanine chemiresistors. J Am Chem Soc. duction, and dielectric properties of tetra tert-butylsulfanyl
2009;131(21):478–485. substituted phthalocyanines. J Coordination Chem. 2015;68(4):717–
26. Cardenas-Jiron GI, Leon-Plata P, Cortes-Arriagada D, Seminario JM, 731.
Electrical characteristics of cobalt phthalocyanine complexes 49. He L, Li D, Dong D, Yao J, Huang Y, Wang H. Effects of polymer-
adsorbed on graphene. J Phys Chem C. 2011;115(32):16052–16062. ization conditions on the properties of poly (furfuryl alcohol) com-
27. Cardenas-Jiron GI, Leon-Plata P, Cortes-Arriagada D, Seminario JM, posite membranes. J Appl Polymer Sci. 2012;124:3383–3391.
Electron transport properties through graphene oxide-cobalt phthalo- 50. David OC, Gorri D, Urtiaga A, Ortiz I. Mixed gas separation study
cyanine complexes. J Phys Chem C. 2013;117:23664–23675. for the hydrogen recovery from H2/CO/N2/CO2 post combustion
28. Shoji M, Oyaizu K, Nishide H. Facilitated oxygen transport through mixtures using a Matrimid membrane. J Membr Sci. 2011;378(1–2):
a Nafion membrane containing cobaltporphyrin as a fixed oxygen 359–368.
carrier. Polymer. 2008;49(26):5659–5664. 51. Verma D, Dash R, Katti KS, Schulz DL, Caruso AN. Role of coor-
29. Chen SH, Lai JY. Polycarbonate/(N,N’-dialicylidene ethylene dinated metal ions on the orientation of phthalocyanine based coat-
diamine) cobalt(II) complex membrane for gas separation. J Appl ings. Spectrochim Acta A Mol Biomol Spectrosc. 2008;70(5):1180–
Polymer Sci. 1996;59:1129–1135. 1186.
30. Ruaan R, Chen S, Lai J. Oxygen/nitrogen separation by polycarbon- 52. Ismail A, Ng B, Abdul Rahman WA. Effects of shear rate and
ate/Co(SalPr) complex membranes. J Membr Sci. 1997;135:9–18. forced convection residence time on asymmetric polysulfone mem-
31. Tsuchida E, Nishide H, Ohyanagi M, Kawakami H. Facilitated trans- branes structure and gas separation performance. Sep Purif Technol.
port of molecular oxygen in the membranes of polymer-coordinated 2003;33(3):255–272.
cobalt schiff base complexes. Macromolecules. 1987;20:1907–1912. 53. Rafiq S, Man Z, Maulud A, Muhammad N, Maitra S. Separation of
32. Nishide H, Ohyanagi M, Okada O, Tsuchida E. Highly selective CO2 from CH4 using polysulfone/polyimide silica nanocomposite
transport of molecular oxygen in a polymer containing a cobalt por- membranes. Sep Purif Technol. 2012;90:162–172.
phyrin complex as a fixed carrier. Macromolecules. 1986;19:494– 54. Zhang Y, Shan L, Tu Z, Zhang Y. Preparation and characterization
496. of novel Ce-doped nonstoichiometric nanosilica/polysulfone compos-
33. Nishide H, Kawakami H, Suzuki T, Azechi Y, Soejima Y, Tsuchida E. ite membranes. Sep Purif Technol. 2008;63(1):207–212.
Effect of polymer matrix on the oxygen diffusion via a cobalt porphyrin 55. Sakurai K, Maegawa T, Takahashi T. Glass transition temperature of
fixed in a membrane. Macromolecules. 1991;24(23):6306–6309. chitosan and miscibility of chitosan/poly(N-vinyl pyrrolidone)
34. Hosseinkhani O, Kargari A, Sanaeepur H. Facilitated transport of blends. Polymer. 2000;41:7051–7056.
CO2 through Co(II)-S-EPDM ionomer membrane. J Membr Sci. 56. Sakurai K. Strain-induced crystallization in polynorbornene. J Appl
2014;469:151–161. (November):
Polymer Sci. 1989;38(6):1191–1194.
35. Sarfraz M, Ba-Shammakh M. Synergistic effect of incorporating
57. Li X, Nicollin A, Pizzi A, Zhou X, Sauget A, Delmotte L. Natural
ZIF-302 and graphene oxide to polysulfone to develop highly selec-
tannin–furanic thermosetting moulding plastics. RSC Adv. 2013;
tive mixed-matrix membranes for carbon dioxide separation from
3(39):17732–17740.
wet post-combustion flue gases. J Membr Sci. 2016;514:35–43.
58. He L, Li D, Wang K, Suresh AK, Bellare J, Sridhar T, Wang H.
36. Waheed N, Mushtaq A, Tabassum S, Gilani MA, Ilyas A, Ashraf F,
Synthesis of silicalite-poly(furfuryl alcohol) composite membranes
Jamal Y, Bilad MR, Khan AU, Khan AL. Mixed matrix membranes
for oxygen enrichment from air. Nanoscale Res Lett. 2011;6(1):637
based on polysulfone and rice husk extracted silica for CO2 separa-
59. Cussler EL, Hughes SE, Ward WJ, Aris R. Barrier Membranes.
tion. Sep Purif Technol. 2016;170:122–129.
J Membr Sci. 1988;38:161–174.
37. Zornoza B, Irusta S, Tellez C, Coronas J. Mesoporous silica sphere-
60. Li Y, Chung TS, Cao C, Kulprathipanja S. The effects of polymer
polysulfone mixed matrix membranes for gas separation. Langmuir.
2009;25(10):5903–5909. chain rigidification, zeolite pore size and pore blockage on polye-
38. Zornoza B, Esekhile O, Koros WJ, Tellez C, Coronas J. Hollow thersulfone (PES)-zeolite A mixed matrix membranes. J Membr Sci.
silicalite-1 sphere-polymer mixed matrix membranes for gas separa- 2005;260(1–2):45–55.
tion. Sep Purif Technol. 2011;77(1):137–145. 61. Li Y, Guan HM, Chung TS, Kulprathipanja S. Effects of novel
39. Zornoza B, Seoane B, Zamaro JM, Tellez C, Coronas J. Combina- silane modification of zeolite surface on polymer chain rigidification
tion of MOFs and zeolites for mixed-matrix membranes. Chemphy- and partial pore blockage in polyethersulfone (PES)-zeolite A mixed
schem. 2011;12(15):2781–2785. matrix membranes. J Membr Sci. 2006;275(1–2):17–28.
40. Kim S, Chen L, Johnson JK, Marand E. Polysulfone and functional- 62. S€uer MG, Baç N, Yilmaz L. Gas permeation characteristics of
ized carbon nanotube mixed matrix membranes for gas separation: polymer-zeolite mixed matrix membranes. J Membr Sci. 1994;91(1–
theory and experiment. J Membr Sci. 2007;294(1–2):147–158. 2):77–86.
41. Ahn J, Chung W-J, Pinnau I, Guiver MD. Polysulfone/silica nano- 63. Mahajan R, Burns R, Schaeffer M, Koros WJ. Challenges in forming
particle mixed-matrix membranes for gas separation. J Membr Sci. successful mixed matrix membranes with rigid polymeric materials.
2008;314(1–2):123–133. J Appl Polymer Sci. 2002;86(4):881–890.
42. Jeazet HBT, Koschine T, Staudt C, Raetzke K, Janiak C. Correlation 64. Nishide H, Tsukahara Y, Tsuchida E. Highly selective oxygen per-
of gas permeability in a metal-organic framework MIL-101(Cr)-poly- meation through a poly (vinylidene dichloride)-cobalt porphyrin
sulfone mixed-matrix membrane with free volume measurements by membrane: hopping transport of oxygen via the fixed cobalt porphy-
positron annihilation lifetime spectroscopy (PALS). Membranes. rin carrier. J Phys Chem B. 1998;102:8766–8770.
2013;3(4):331–353. 65. Kim S, Marand E, Ida J, Guliants VV. Polysulfone and mesoporous
43. Aitken CL, Koros WJ, Paul DR. Effect of structural symmetry on molecular sieve MCM-48 mixed matrix membranes for gas separa-
gas transport properties of polysulfones. Macromolecules. 1992; tion. Chem Mater. 2006;18:(4):1149–1155.
25(13):3424–3434. 66. Sakamoto K, Ohno E. Synthesis of soluble metal phthalocyanine
44. Wang H, Holmberg B. A, Yan Y. Homogeneous polymer zeolite derivatives and their electron transfer properties. JSDC. 1996;112:
nanocomposite membranes by incorporating dispersible template- 368–374.
removed zeolite nanocrystals. J Mater Chem. 2002;12(12):3640– 67. Tolbin AY, Pushkarev VE, Balashova IO, Dzuban AV, Tarakanov
3643. PA, Trashin SA, Tomilova LG, Zefirov NS. A highly stable double-
45. Ghosal K, Chern RT, Freeman BD, Carolina N, Daly WH, coordinated 2-hydroxy-tri(tert-butyl)-substituted zinc phthalocyanine
Negulescu II. Effect of basic substituents on gas sorption and perme- dimer: synthesis, spectral study, thermal stability and electrochemical
ation in polysulfone. Macromolecules. 1996;29:4360–4369. properties. New J Chem. 2014;38(12):5825–5831.
46. Gur TM. Permselectivity of zeolite filled polysulfone gas separation
membranes. J Membr Sci. 1994;93(94):283–289. Manuscript received Apr. 13, 2017, and revision received June 30, 2017.

AIChE Journal January 2018 Vol. 64, No. 1 Published on behalf of the AIChE DOI 10.1002/aic 199

Potrebbero piacerti anche