Enhancement in the gas permeabilities of novel polysulfones with pendant 4-trimethylsilyl-α-hydroxylbenzyl substituents

Article abstract of DOI:10.1021/ma0346411

A series of modified polymers with 4-trimethylsilyl-α-hydroxylbenzyl (HBTMS) substituents were made as new materials for membrane gas separation. HBTMS was introduced onto polysulfone (PSf), tetramethylpolysulfone (TMPSf), and hexafluoropolysulfone (6FPSf

Full text of DOI:10.1021/ma0346411

Macromolecules 2003, 36, 6807-6816
6807
Enhancement in the Gas Permeabilities of Novel Polysulfones with
Pendant 4-Trimethylsilyl-R-hydroxylbenzyl Substituents†
Yin g Da i,‡ Mich a el D. Gu iver ,*,‡ Gilles P . Rober tson ,‡ Yon g Soo Ka n g,§ a n d
Kw i J on g Lee|
National Research Council of Canada, Institute for Chemical Process and Environmental Technology,
Ottawa, Ontario, K1A 0R6, Canada, Korea Institute of Science and Technology, Center for Facilitated
Transport Membranes and Polymer Physics Lab, P.O. Box 131, Cheongryang, Seoul 130-650, Korea,
and Seoul National University, Hyperstructured Organic Materials Research Center and School of
Chemical Engineering, Seoul 151-744, Korea
Received May 15, 2003; Revised Manuscript Received J une 25, 2003
ABSTRACT: A series of modified polymers with 4-trimethylsilyl-R-hydroxylbenzyl (HBTMS) substituents
were made as new materials for membrane gas separation. HBTMS was introduced onto polysulfone
(PSf), tetramethylpolysulfone (TMPSf), and hexafluoropolysulfone (6FPSf) by forming lithiated polymer
intermediates followed by treatment with an electrophile, p-trimethylsilylbenzaldehyde. The resulting
side substituent contains the bulky trimethylsilyl (TMS) groups separated from the polymer chain by an
aromatic ring. The polymer structures were characterized by NMR spectroscopy, and thermal properties
were investigated by differential scanning calorimetry and thermogravimetry. The permeability coefficients
of the resulting polymers PSf-HBTMS, TMPSf-HBTMS, and 6FPSf-HBTMS showed considerable increases
over the unmodified polymers. TMPSf-HBTMS showed the greatest improvements in gas transport
properties with large increases in both CO2 and O2 permeabilities of ∼3-fold (72 and 18 Barrers,
respectively) compared with TMPSf. In the case of the CO2/N2 gas pair, there was an absence of the
typical accompanying tradeoff shown by decreases in permselectivity with increasing permeability.
Comparisons of this series of polymers containing the new HBTMS substituent were made with those
having other TMS pendant groups.
In tr od u ction
It is known that trimethylsilyl (TMS) groups in the
polymer backbone enhance polymer solubility without
significant loss of thermal stability. In addition, they
hinder chain motions and disrupt chain packing thereby
improving gas permeability with minimal loss in
permselectivity.11-17
The separation of gases by synthetic membranes is a
dynamic and rapidly growing field. Membrane separa-
tion processes offer a number of advantages over
traditional methods in terms of capital investments and
lower energy use.1 Polysulfone (PSf) has adequate gas
separation performance2-4 as well as good thermosta-
bility, mechanical properties, and chemical resistance.
It was the first commercial polymer used for fabricating
gas separation membranes in the form of hollow fibers
for the separation of H2 and the production of N2
because of its overall combination of adequate gas
permeabilities and relatively high permselectivities to
various gas pairs as well as its good mechanical proper-
ties for fiber spinning.5 The previous work most relevant
to the present work was by Koros and Paul whereby
parallel families of polysulfones with systematic struc-
tural modifications were correlated with their gas
separation properties.6-10
Both tetramethylpolysulfone (TMPSf) and hexafluo-
ropolysulfone (6FPSf) are more permeable than PSf but
maintain comparable permselectivity for many gas
pairs9,10 because tetramethyl and hexafluoroisopropyl-
idene groups sterically hinder both bond rotation and
intersegmental packing. The packing of 6FPSf may be
further inhibited by intermolecular repulsive forces
between fluorine atoms that have high electron density.9
The chemical structure and physical properties of the
membrane material influence the permeability and
permselectivity.18 Transport properties depend on the
free volume within the polymer and on the segmental
mobility of the polymer chains. The segmental mobility
of the polymer chains is affected by factors such as the
extent of unsaturation, degree of cross-linking, degree
of crystallinity, and nature of substituent. The glass
transition temperature (Tg) of polymers has a profound
influence on transport properties. For the majority of
polymers, those with lower Tg values possess greater
segmental mobilities, which result in higher diffusivi-
ties.19 Many studies have shown that an improvement
in gas transport properties could be obtained by modify-
ing or tailoring the polymer structure. Polarity and
steric characteristics of the polymeric material strongly
influence gas permeation. The size and shape of bulky
groups placed on the polymer chain affect certain
fundamental properties such as packing density and
rigidity. The substitution of bulky groups in the side
chains appears to have a greater influence on diffusivity
than substitution of these groups in the polymer back-
bone. An absence of such groups tends to increase the
structural regularity and increases density.1,19
* Corresponding author. Address: National Research Council,
Institute for Chemical Process and Environmental Technology,
1200 Montreal Road, Ottawa, Ontario, K1A 0R6, Canada. E-
mail: michael.guiver@nrc-cnrc.gc.ca.
Chemical modification of PSf’s was performed in order
to improve membrane separation properties. PSf’s were
activated by direct lithiation20 or by bromination-
lithiation21 to produce reactive lithiated intermediates
that were then converted by various electrophiles to
† NRCC No. 46443.
‡ National Research Council of Canada.
§ Korea Institute of Science and Technology.
| Seoul National University.
10.1021/ma0346411 CCC: $25.00 © 2003 American Chemical Society
Published on Web 08/12/2003

6808 Dai et al.
Macromolecules, Vol. 36, No. 18, 2003
yield derivatives of PSf.11,12,22-26 In comparison with
previous work in which TMS, dimethylphenylsilyl
(DMPS), and methyldiphenylsilyl (MDPS) groups were
introduced on PSf and TMPSf,11,12,26 this work reports
the improvements in gas permeability for polymers PSf,
TMPSf, and 6FPSf modified to contain the new bulky
side substituent 4-trimethylsilyl-R-hydroxylbenzyl (HBT-
MS). HBTMS contains TMS, phenyl, and OH groups
and is distinct from TMS and DMPS since the TMS
group is spaced further from the polymer chain. The
phenyl spacer and hydroxyl group may contribute to
increase the rigidity because of changing structural
regularity and H-bonding. Another polymer, PSf-CH2-
TMS, was prepared containing a flexible -CH2- spacer
between polymer and TMS. In comparison with poly-
mers containing the rigid HBTMS groups, this polymer
had inferior gas transport properties.
was injected promptly into the reaction solution, which im-
mediately became cloudy white, and then the solution was
stirred with gradual warming to -15 °C. The polymer was
recovered by precipitation from ice water, washed with ice-
ethanol, and dried in an oven at 110 °C. The white product
(22.0 g) had a degree of substitution (DS) of 1.5.
Syn th esis of P Sf-o-HBTMS. A solution of PSf-Br2 (5.0 g,
8.33 mmol) in anhydrous THF (1.7 wt %) was cooled to -73
°C under argon, and 2.5 M n-butyllithium (7.2 mL, 17.9 mmol)
was injected dropwise by syringe pump over 15 min.21 Fol-
lowing addition, the lithiated polymer solution was stirred for
15 min and then warmed to -30 °C. TMSBA (4.45 g, 25 mmol)
was injected promptly into the reaction solution, which im-
mediately became cloudy white. The solution was stirred with
gradual warming to -15 °C over about 1 h. The polymer was
recovered by precipitation from ice water, washed with ice-
ethanol, and dried in an oven at 110 °C. The white product
(4.55 g) had a DS of 1.7.
Syn th esis of P Sf-o-CH2-TMS. A solution of PSf-Br2 (5.5
g, 9.16 mmol) in anhydrous THF (1.0 wt %) was lithiated with
10 M n-butyllithium (1.97 mL, 19.7 mmol) using the above
procedure. (Iodomethyl)trimethylsilane (9.8 g, 6.8 mL, 45.8
mmol) was injected promptly into the -40 °C reaction solution,
which immediately became cloudy white. The solution was
stirred with gradual warming to -5 °C over about 1 h. The
polymer was recovered by precipitation from 95% ethanol,
washed with ethanol, and dried in an oven at 110 °C. The
white product (4.59 g) had a DS of 1.1.
Syn th esis of 6F P Sf-Br 2. Excess bromine (45 mL, 139.5 g,
0.872 mol) was added to a magnetically stirred solution of
6FPSf8,9 (15 g, 27.2 mmol) in chloroform (300 mL) at room
temperature and under an argon atmosphere. The dark red
homogeneous mixture was left stirring at room temperature
for 16 h. The polymer was precipitated using methanol. The
recovered polymer was left standing in fresh methanol to leach
out residual free bromine and then filtered again and dried in
a vacuum oven for 24 h at 55 °C. White 6FPSf-Br2 polymer
(19 g; yield, 98%) was recovered. Elemental Analysis: calcd
for C27H14Br2O4S, 22.56% Br; found, 23.13% Br.
Exp er im en ta l Section
Mater ials. Monomers hexafluorobisphenol A (Aldrich Chemi-
cal Co.) and 4,4′-dichlorodiphenyl sulfone (Pfaltz & Bauer, Inc.)
were further purified by crystallization from ethanol 95%.
Before polymerization, all monomers were dried in a vacuum
oven at 55 °C for 24 h. Anhydrous potassium carbonate
(analytical reagent from BDH) was dried in an oven at 110 °C
for 24 h. N-Methylpyrrolidinone (NMP) was distilled over
barium oxide under a vacuum, and analytical reagent toluene
from BDH was used as received. PSf-Br2,21 TMPSf-Br2 and
12
6FPSf8,9 were made following previously reported synthetic
procedures. PSf Udel P-3500 (BP-Amoco), PSf-Br2, TMPSf-Br2,
and 6FPSf were dried at 110 °C for at least 24 h before
reaction. Reagent grade tetrahydrofuran (THF) was freshly
distilled over lithium aluminum hydride (LiAlH4). n-Butyl-
lithium (10 M in hexane), bromine, 1-bromo-4-(trimethylsilyl)-
benzene (BTMSB), (iodomethyl)trimethylsilane, and N,N-
dimethylformamide (DMF) were obtained from Aldrich Chem-
ical Co., and DMF was purified by azeotropic distillation with
benzene (10% v/v, previously dried over CaH2) before the
Syn th esis of 6F P Sf-o-HBTMS. A colorless solution of
6FPSf-Br2 (5.25 g, 7.42 mmol) in anhydrous THF (1.3 wt %)
was cooled to -76 °C under argon, and 10 M n-butyllithium
(1.6 mL, 16.0 mmol) was injected dropwise by syringe pump
over 15 min. Following addition, the clear orange solution was
stirred for 15 min and then warmed to -36 °C over about 1 h.
TMSBA (3.96 g, 22.3 mmol) was injected promptly into the
reaction solution, which immediately became cloudy yellow-
white. The solution was stirred with gradual warming to -15
°C. The polymer was recovered by precipitation from ice water,
washed with ice-ethanol, and dried in an oven at 110 °C. The
white product (3.98 g) had a DS of 1.7.
reaction. Modification reactions were conducted under
a
constant argon purge and with mechanical stirring. A mixture
of dry ice and ethanol was used for cooling reaction mixtures.
All modified polymers were recovered by precipitation from
ethanol using a Waring blender, washed thoroughly, and then
dried in a vacuum oven at 55 °C.
Syn th esis of p-Tr im eth ylsilylben za ld eh yd e (TMSBA).
A solution of colorless liquid BTMSB (38.50 g, 168 mmol) in
anhydrous THF (10 wt %) was cooled to -76 °C under argon,
and 10 M n-butyllithium (17.7 mL, 177 mmol) was injected
dropwise by syringe pump over 15 min. The reaction solution
remained colorless. Following addition, the reaction mixture
was stirred for 15 min to give a soluble lithiated intermediate;
then excess purified DMF (30.7 g, 32.4 mL, 420 mmol) was
added quickly into the reaction solution. When the solution
was warmed gradually to -30 °C over about 1 h, 300 mL of
saturated NH4Cl solution and 100 mL of ethyl ether were
added to the resulting colorless solution. The solution was
shaken well, and the organic phase was separated and then
washed with saturated NaCl solution, dried with anhydrous
Mg2SO4, and evaporated to remove solvents. The product was
purified by distillation under a vacuum to yield 26.76 g (89%)
of TMSBA as a colorless oil. The resulting product was
characterized by 1H NMR (CDCl3): CHO, 10.01 ppm, 1H, s;
Syn th esis of TMP Sf-HBTMS. A cloudy yellow solution of
12
TMPSf-Br2 (5.03 g, 7.66 mmol) in anhydrous THF (1.0 wt
%) was cooled to -77 °C under argon, and 10 M n-butyllithium
(1.7 mL, 16.5 mmol) was injected dropwise by syringe pump
over 15 min.12 Following addition, the lithiated polymer
solution was stirred for 15 min and then warmed to -30 °C.
TMSBA (4.09 g, 23 mmol) was injected promptly into the
reaction solution, which immediately became cloudy yellow-
white. The solution was stirred with gradual warming to -15
°C over about 1 h. The polymer was recovered by precipitation
from ice-ethanol, washed with ice-ethanol, and dried in an oven
at 110 °C. The white product (3.98 g) had a DS of 1.5.
Ch a r a cter iza tion Meth od s. Nuclear magnetic resonance
spectra were recorded on a Varian Unity Inova spectrometer
at a resonance frequency of 399.961 MHz for 1H and 100.579
MHz for 13C. Samples were prepared using CDCl3 as the NMR
solvent, and the residual CHCl3 signal at 7.25 ppm was used
as the chemical shift reference. The inherent viscosities of
polymers were determined using an Ubbelohde viscometer at
a polymer concentration of 0.4 g/dL in NMP at 35 °C. The DS
4
H-2, 7.83 ppm, 2H, dd (3J ) 8 Hz, J ) 2 Hz); H-3, 7.68 ppm,
4
2H, dd (3J ) 8 Hz, J ) 2 Hz); CH3, 0.30 ppm, 9H, s.
Syn th esis of P Sf-s-HBTMS. A solution of PSf (15.0 g, 33.9
mmol) in anhydrous THF (1.7 wt %) was cooled to -73 °C
using a dry ice/alcohol bath under argon, and 10 M n-
butyllithium (7.46 mL, 74.6 mmol) was injected dropwise by
syringe pump over 15 min during which time color changes
were observed.20,22,23 Following addition, the lithiated polysul-
fone solution was stirred for 15 min and then warmed
gradually to -30 °C over about 1 h. TMSBA (16.3 g, 91.8 mmol)
1
of modified polymers was readily determined using H NMR
by comparison of the integration values for selected signals
(described later). Polymer thermal degradation curves were
obtained from thermogravimetric analysis (TGA) (TA Instru-

Macromolecules, Vol. 36, No. 18, 2003
Enhanced Gas Permeabilities of Polysulfones 6809
Sch em e 2. Syn th esis of P Sf-s-HBTMS
Sch em e 1. Syn th esis of TMSBA
ments model 2950), and glass transition temperatures (Tg)
were obtained from differential scanning calorimetry (DSC)
(TA Instruments model 2920). Polymer samples for TGA were
preheated to 60 °C for 2 h inside the TGA furnace for moisture
removal under nitrogen gas and then heated to 600 °C at 10
°C/min for degradation temperature measurement. Samples
for DSC were heated initially to at least 30 °C above Tg at 10
°C/min under a flow of 50 mL/min of nitrogen gas, quenched
with liquid nitrogen, and reheated at 10 °C/min for the Tg
measurement.
Sch em e 3. Syn th esis of 6F P Sf-o-HBTMS
X-ray diffraction was used to investigate d-spacing. A
Macscience model M18XHF22 was utilized with Cu KR radia-
tion of which the wavelength (λ) was 1.54 Å at 50 kV and 100
mA and the scanning speed was 5°/min. The value of the
d-spacing was calculated by means of Bragg’s law (d ) λ/2 sin
θ), using θ of the broad peak maximum.
Dense polymer films for permeability measurements were
made from 5 wt % polymer solutions in anhydrous THF that
were filtered through 1 µm poly(tetrafluoroethylene) filters and
then poured into flat-glass dishes and dried under a nitrogen
atmosphere at room temperature. The detached films were
further dried for 3 days in a vacuum oven at 40 °C to remove
the residual solvent. Optically clear films were obtained with
a thickness of about 40 µm in all cases. The absence of residual
solvent in the films was confirmed by observing Tg using DSC.
Permeability coefficients of carbon dioxide, oxygen, and nitro-
gen were measured by the constant volume method at 35 °C
with an upstream pressure of 1 atm.
Resu lts a n d Discu ssion
P r ep a r a tion of P olym er s. Aldehyde is strongly
reactive toward carbanions and other nucleophiles.
TMSBA was prepared as a reactive electrophile to
introduce HBTMS substituents onto PSf, TMPSf, and
6FPSf using a lithiation reaction. TMSBA was prepared
by lithium-bromine exchange followed by treatment
with DMF as shown in Scheme 1. The low reaction
temperature was necessary to obtain higher product
yield. This is an alternative and convenient method for
the synthesis of aromatic aldehydes that are often
prepared by oxidation of methylbenzenes or reduction
of acid chlorides.
tion in the reactive position because of four methyl
groups, whereas 6FPSf is deactivated because of the
electron-withdrawing effect of the two -CF3 groups.
The brominated polymers were first reacted with 2.1
mol equiv of n-butyllithium respectively in order to
exchange bromine for lithium, followed by reaction with
the electrophile TMSBA or (iodomethyl)trimethylsilane.
The synthetic routes are given in Schemes 2-6, and the
DSs of the polymers PSf-s-HBTMS, 6FPSf-o-HBTMS,
PSf-o-HBTMS, PSf-o-CH2-TMS, and TMPSf-HBTMS
were 1.5, 1.7, 1.7, 1.1, and 1.5, respectively, determined
by NMR. PSf-s-HBTMS had a lower DS than PSf-o-
HBTMS because of HBTMS substituent steric hin-
drance around the diphenyl sulfone segment. PSf-o-CH2-
TMS had a low DS compared with PSf-o-HBTMS
because (iodomethyl)trimethylsilane has lower reactiv-
ity than TMSBA. TMPSf-HBTMS had a lower DS
because of steric hindrance from -CH3, preventing
complete access of HBTMS substituents.
The synthetic route for the preparation of PSf-s-
HBTMS by direct lithiation is shown in Scheme 2. The
procedures for the synthesis of the other polymers
PSf-o-HBTMS, PSf-o-CH2-TMS, 6FPSf-o-HBTMS, and
TMPSf-HBTMS are all based on bromination-lithia-
21
12
tion. Brominated polymers PSf-Br2 and TMPSf-Br2
were prepared as previously reported. 6FPSf-Br2 was
prepared by the addition of excess bromine to a 6FPSf
polymer solution in CHCl3. The reactive substitution
position is situated ortho to the aryl ether linkage in
the bisphenol portion of the repeat units as shown in
Scheme 3. This is the most favorable site because it is
electrophilically activated by the oxygen atom. The
variation in reaction times for preparing TMPSf-Br2 (1
h), PSf-Br2 (6 h), and 6FPSf-Br2 (16 h) for DS ) 2 was
because of the different electrophilic reactivities. TMPSf
is the most electrophilically activated toward bromina-
Str u ctu r a l Ch a r a cter iza tion (NMR). The 1H NMR
spectra of PSf and 6FPSf derivatives with substituted
TMSBA are shown in Figures 1 and 2. The 1H NMR
chemical shift displacements (ppm) and multiplicity
data (s, singlet; d, doublet; m, multiplet) of aromatic and

6810 Dai et al.
Macromolecules, Vol. 36, No. 18, 2003
F igu r e 1. 1H NMR spectra of PSf-s-HBTMS and PSf-o-HBTMS.
aliphatic signals are listed in Tables 1 and 2. The
spectra show expansions of the aromatic regions where
the results of the substitution reaction can be observed
from the changing NMR signals. Evidence of the suc-
cessful incorporation of HBTMS groups on the main
polymer chain was also evident at high field where the
TMS signal could be observed (Tables 1 and 2).
the main chain aryl groups around the sulfone func-
tionality (H-8, 9, 12) each appear as two signals rather
than one signal; this is particularly obvious in the case
of H-9 and H-13 (Figure 1a). On the other hand, the
ortho-ether linkage substituted polymer (Figure 1b)
displays a much simpler spectrum where only the
hydrogen atoms of the polymer chain aryl groups
bearing the substituents appear as multiplets. Aryl
protons H-2, H-3 and H-8, H-12 are slightly split by
steric hindrance due to the substituents, while H-5 is
clearly more affected (two resolved signals) because of
its close proximity with HBTMS. The presence of an
asymmetric carbon (H-C*-OH) should not play a
significant role in this observed spitting of the signals
since the mixture of isomers is expected to be exactly
1:1 for this type of reaction on an aldehyde electrophile.
The chemical shift of H-13 on the asymmetric carbon
was observed to be dependent upon the site of substitu-
tion (ortho sulfone or ether). Differences in electron
shielding around H-13 due to its orientation in space
vary for each derivative. The distinct upfield signal for
H-13 was split into two for the ortho-sulfone HBTMS
substituted polymer, while the ortho-ether derivative
led to a singlet for the same H-13. This emphasizes what
was described earlier as restricted mobility of substit-
The NMR spectra of the aromatic regions of PSf
substituted at the ortho-sulfone and ortho-ether linkage
sites (Schemes 2 and 3) are displayed in parts a and b
of Figure 1, respectively. The high DSs obtained for each
of these derivatives (PSf-s-HBTMS DS ) 1.5, PSf-o-
HBTMS DS ) 1.7) resulted in simplified spectra due to
signals originating primarily from disubstituted repeat
units. On the other hand, the bulkiness of the substit-
uents had a significant effect on polymer chain protons
nearest to the substituents. This was especially appar-
ent in the ortho-sulfone derivative where both substit-
uents are localized around the sulfone linkage. This
hinders rotation about the aryl-sulfone-aryl bonds,
resulting in multiple signals for the same protons.
Moreover, possible hydrogen bonding between the -OH
groups could be another factor impacting rotational
mobility. As a result, most proton signals from HBTMS
substituents (H-13, 15, 16, 18, 19, -OH) as well as from

Macromolecules, Vol. 36, No. 18, 2003
Enhanced Gas Permeabilities of Polysulfones 6811
F igu r e 2. 1H NMR spectra of 6FPSf and selected derivatives.
uents near the sulfone linkage. Furthermore, the pro-
tons of the substituent aryl groups also display the same
behavior as seen for H-13: a multiple signal in the
ortho-sulfone polymer compared with two coupled dou-
blets for H-15, 19 and H-16, 18 for the ortho-ether
polymer. The -OH signals for both modified polymers
appear at high field around 3 ppm.
ether sites was easily confirmed by signals at similar
chemical shifts as seen before in Figure 1b. The two
spin-coupled doublets H-15, 19 and H-16, 18 as well as
the singlet H-13 at high field (6 ppm) and the -OH
signal (3 ppm) were the proof of HBTMS substitution.
The full 1H NMR spectrum and NMR data of the
HBTMS substituted polymer TMPSf-Br2 (Scheme 6) are
displayed in Table 3 and Figure 3. Although the
spectrum appears complicated, it was found that sub-
stitution occurred at both the ortho-sulfone and ortho-
ether sites as observed previously,12 due to nonregiose-
lectivity of the lithiation reaction for this particularly
hindered polymer. The complex NMR spectrum is a
result of a mixture of many different repeat units.
1
The H NMR chemical shift displacements (ppm) of
PSf-Br2 modified with (iodomethyl)trimethylsilane
(Scheme 4) are listed in Table 1. Once again, the high-
field trimethylsilyl signals (-0.19-0.21 ppm) as well as
the new CH2 singlet at 1.88 ppm are an indication of
the presence of the new substituent.
TMSBA was also substituted on 6FPSf-Br2 (Scheme
5). The 1H NMR spectrum of 6FPSf-Br2 (Figure 2b)
consists of the expected two spin systems: one involving
H-8, 12 with H-9, 11 and another system involving H-3
with H-2 and H-5. The spectrum of ortho-ether substi-
tuted 6FPSf-Br2 (Figure 2c) with TMSBA resembles
that of the PSf analogue (Figure 1b). The DS was 1.7,
hence the presence of mostly disubstituted repeat units.
The DS of all previously described modified polymers
was assessed by comparative integration of specific
NMR signals. One method consisted of comparing either
the intensity of the distinct H-13 signal or that of the
high-field TMS methyl groups with that of the six
isopropylidene protons of PSf. Another method derived
from ref 27 required the integration values of the signals
at low field exclusively. This method was utilized and
was particularly useful for measuring the DS of sub-
stituted 6FPSf.
1
However, the H NMR spectrum was more complicated
because it had more overlapping signals than previous
spectra; H-2, 3, and 5 could not be assigned. In contrast,
the presence of the HBTMS substituents at the ortho-

6812 Dai et al.
Macromolecules, Vol. 36, No. 18, 2003
Ta ble 1. 1H NMR Da ta for P Sf a n d Selected Der iva tives
proton number
H2
PSf
PSf-s-HBTMS
6.78-6.93 (m)
PSf-o-HBTMS
PSf-o-CH2-TMS
6.80 (d, J 8.4)
6.93 (d, J 8.8)
6.73 (d, J 8.4)
6.74 (d, J 8.4)
H3
H5
7.24 (d, J 8.8)
7.24 (d, J 8.8)
6.93 (d, J 8.8)
7.13-7.26 (m)
7.13-7.26 (m)
6.78-6.93 (m)
7.11 (dd, J 8.4, 2.3)
7.09 (dd, J 8.4, 2.3)
7.62 (d, J 2.4)
6.88 (dd, J 8.4, 2.2)
6.91-7.40 (m)
7.58 (d, J 2.4)
H6
H-3′,5′
H-2′,6′
H8
7.24 (d, J 8.8)
7.01 (d, J 8.8)
6.91-6.97 (m)
7.80-7.88 (m)
7.00 (d, J 8.8)
7.84 (d, J 8.8)
6.78-6.93 (m)
8.07 (d, J 8.8)
7.95 (d, J 8.8)
6.78-6.86 (m)
7.68-7.77 (m)
H9
H11
H12
H13
7.84 (d, J 8.8)
7.00 (d, J 8.8)
6.78-6.86 (m)
6.78-6.86 (m)
5.88 (s)
7.80-7.88 (m)
6.91-6.97 (m)
1.88 (s)
6.98-7.09 (m)
6.53 (d, J 4.0)
6.44 (d, J 3.0)
7.42 (d, J 8.0)
H-15, 16, 18, 19
7.35 (d, J 7.5)
7.19 (d, J 7.5)
7.38 (d, J 8.0)
7.13-7.26 (m)
6.98-7.09 (m)
3.44 (s)
OH
2.3-2.7 (broad)
3.01 (s)
H-20 (SiCH3)
CH3-C-CH3
0.24 (s)
1.67 (s)
0.19 (s)
1.71 (s)
-0.19-0.21 (m)
1.60-1.75 (m)
1.69 (s)
Ta ble 2. 1H NMR Da ta 6F P Sf a n d Selected Der iva tive
Sch em e 5. Syn th esis of P Sf-o-CH2-TMS
proton number
6FPSf
6FPSf-Br2
6FPSf-HBTMS
H2
H3
H5
H6
7.02 (d, J 8.8) 7.02 (d, J 8.8)
7.41 (d, J 8.8) 7.33 (dd, J 8.8, 2.4) 6.7-8.0
7.41 (d, J 8.8) 7.69 (d, J 2.4)
7.02 (d, J 8.8)
6.7-8.0
6.7-8.0
H8-12
H9-11
H13
7.09 (d, J 8.8) 7.06 (d, J 8.8)
7.92 (d, J 8.8) 7.93 (d, J 8.8)
6.75-6.90 (m)
7.70-7.90 (m)
5.93 (s)
5.97 (s)
H15, 16, 18, 19
7.13-7.20 (m)
7.30-7.40 (m)
Sch em e 4. Syn th esis of P Sf-o-HBTMS
Sch em e 6. Syn th esis of TMP Sf-HBTMS
H13
Ha
n
)
Solu bility Test. Solubility was observed during the
day of sample preparation and then 14 days later to
determine if the solutions were stable. All the polymers
dissolved in CHCl3.
16 - n + 4n
hence
16H13
(Ha - 3)H13
Th er m a l Mea su r em en ts. The thermal stabilities of
PSf derivatives were measured by TGA, and the weight-
loss data are shown in Table 4. All derivative polymers
were thermally less stable than the respective parent
polymers. The curves of all the polymers exhibited a
two-step degradation. This suggests that during the
n )
for PSf and 6FPSf
where H13 is the intensity of the H-13 signal, Ha is the
total intensity of aromatic signals, and n is the degree
of substitution (0-2).

Macromolecules, Vol. 36, No. 18, 2003
Enhanced Gas Permeabilities of Polysulfones 6813
Ta ble 3. 1H NMR Da ta for TMP Sf a n d
Selected Der iva tives
be a combination of steric hindrance about the diphenyl
sulfone site and a bulky side group as well as the
strong electron-withdrawing ability of the sulfone
group, resulting in a weakening of the carbon-silicon
bond.11,17
proton number
TMPSf
6.94 (s)
TMPSf-Br2
7.44 (s)
TMPSf-HBTMS
H3
H3′
6.6-8.3
6.6-8.3
All PSf polymer derivatives had lower Tg values
than PSf. PSf-s-HBTMS had a Tg ∼13 °C higher than
that of PSf-o-HBTMS. PSf-o-CH2-TMS had a very low
Tg of 142.7 °C most likely because of the relatively
greater mobility of the alkyl spacer compared with
the HBTMS substituent. 6FPSf-o-HBTMS showed a
Tg reduction of 10 °C compared with 6FPSf. This
was possibly due to a loss in symmetry of the phenyl-
ene ring by the introduction of HBTMS. The Tg of
TMPSf-HBTMS was approximately the same as
TMPSf. This suggests the HBTMS increased the chain
stiffness while reducing the intermolecular interac-
tions.
H5
H8
H9
H11
H12
H-13
OH
H15, 16, 18, 19
H-20 (SiCH3)
ArCH3
6.94 (s)
6.6-8.3
6.6-8.3
6.6-8.3
6.6-8.3
6.83 (d, J 8.8) 6.83 (d, J 9.2)
7.81 (d, J 8.8) 7.84 (d, J 8.8)
7.81 (d, J 8.8) 7.84 (d, J 8.8)
6.83 (d, J 8.8) 6.83 (d, J 9.2)
6.6-8.3
5.7-6.5 (m)
2.5-3.4 (m)
6.6-8.3
0.18-0.26 (m)
2.03 (s)
C2-CH3 2.09 (s) 1.4-2.2 (m)
C6-CH3 2.11 (s)
CH3-C-CH3 1.69 (s)
1.67 (s)
1.4-2.2 (m)
Ta ble 4. Gla ss Tr a n sition Tem p er a tu r es (Tg’s) a n d
Ga s Tr a n sp or t P r op er ties. A tradeoff relationship
is usually observed between permeability (P) and perm-
selectivity (R) for common gases in glassy or rubbery
polymers. That is, higher permeability is gained at the
cost of lower permselectivity and vice versa. Upper
bound performance lines for the relationship between
gas permeability and permselectivity have been pro-
posed by Robeson et al.3 A new calculation method to
show the distance (δ) from the permeability point (Pi,
vs Rij, where permselectivity Rij ) Pi/Pj) to the upper
bound line3 was developed in this work. This gives a
quantitative and convenient measure of the overall
improvement in permeability/permselectivity that oc-
curs from a polymer modification. The following equa-
tion was used to calculate the upper bound line accord-
ing to Robeson’s work:3
Th er m a l Degr a d a tion of P olym er s
extrapolated
onset, °C
polymer
DS Tg, °C actual onset, °C
PSf
188.1
2.0 191.0
1.5 179.7
1.7 166.2
1.1 142.7
199.4
455.0
287.6
304.4
319.8
339.4
486.8
465.2
343.4
396.0
310.0
312.2
521.1
430.4
303.1
335.7
393.7
514.0
505.2
358.9
432.1
425.2
320.7
PSf-Br2
PSf-s-HBTMS
PSf-o-HBTMS
PSf-o-CH2-TMS
6FPSf
6FPSf-Br2
2.0 196.4
6FPSf-o-HBTMS 1.7 186.5
TMPSf
231.3
2.0 275.4
1.5 230.3
TMPSf-Br2
TMPSf-HBTMS
heating process, HBTMS was lost from polymer first.
For example, the extrapolated onsets of weight loss for
PSf-o-HBTMS occurred at 353.7 °C and the actual
onsets of degradation occurred at 321.2 °C. The ini-
tial-step degradation of PSf-o-HBTMS is believed to be
the loss of -Si(CH3)3 since the weight loss is close to
the theoretical weight loss of 17% from PSf-o-HBTMS
with DS ) 1.7. The second degradation for PSf-o-
HBTMS is believed to be the loss of the entire substi-
tution group since the weight loss is close to the
theoretical weight loss of 40% from PSf-o-HBTMS with
DS ) 1.7. These data are in agreement with the DS
results from NMR. A possible explanation for the lower
thermal stability of the ortho-sulfone derivatives could
Pi ) k1Rnij1
(1)
(2)
log Rij ) -n log Pi + log k
where n ) dj - di where dj is the kinetic diameter of
the lower permeability gas and di is the kinetic diameter
of the higher permeability gas. The kinetic diameters
as reported by Breck28 are listed in Table 5. The exact
position of the upper bound line was chosen from the
best data of Park et al.2 For the O2/N2 gas pair, poly-
(tetrabromofluorene tetrabutyl isophthalate) gives the
F igu r e 3. 1H NMR spectra of TMPSf-HBTMS.

6814 Dai et al.
Macromolecules, Vol. 36, No. 18, 2003
Ta ble 5. Len n a r d -J on es Kin etic Dia m eter s of
Va r iou s Ga sesa
gas
He
H2
CO2
3.3
O2
3.46
N2
CH4
3.8
kinetic diameter (Å) 2.6
2.89
3.64
a
Reference 28.
F igu r e 5. Gas permeabilities of modified PSf’s for CO2 and
N2.
dashed lines represent the Robeson upper bound lines
whereas the solid lines represent the generated upper
bound lines from the data of Park et al.3 Each point on
the upper bound lines in Figures 4 and 5 refers to poly-
(tetrabromofluorene tetrabutyl isophthalate), the refer-
ence polymer giving the best data from Park et al. for
both O2/N2 and CO2/N2 gas pairs.
F igu r e 4. Gas permeabilities of modified PSf’s for O2 and N2
in relation to the upper bound line of Robeson (ref 3) (dashed
line) and one generated from data in Park et al. (ref 2) (solid
line).
best data (P(O2) ) 16.8, R(O2/N2) ) 5.9) and was selected
as the point from which to draw the upper bound line
according to eq 2. If instead Robeson’s data3 were to be
selected, the distance to the upper bound (δ) would be
shorter by ∼15 × 10-3 for the O2/N2 gas pair and ∼5 ×
10-3 for the CO2/N2 gas pair. Using the data of Park et
All the polymers containing the HBTMS substituent
were significantly more permeable to all gases than
their parent polymers PSf, 6FPSf, and TMPSf. What is
notable is the small or absent decrease in permselec-
tivity, particularly for the CO2/N2 gas pair. The HBTMS
substituent has a TMS group spaced from the polymer
chain by a rigid phenyl and H-bonded hydroxyl groups.
The open structure leads to high gas permeability, while
the rigid substituent provides a selective diffusion
environment.
al.2 and n ) dN - dO ) 3.64 - 3.46 ) 0.18, a value of
2
k in eq 2 can be calcu2lated. Therefore the upper bound
line equation can be obtained. Using a mathematical
method, the distance from a permeability/permselectiv-
ity point to the upper bound line can be calculated
according to eq 3.
In comparison with PSf, PSf-o-HBTMS showed large
increases in CO2 and O2 permeability (∼13-fold and 10-
fold) with a simultaneously lesser reduction in perm-
selectivity as measured by a lessening of distances δ to
the upper bound line (from 0.387 to 0.054 and from
0.213 to 0.133, respectively) and the d-spacing increased
from 5.0 to 5.3 Å. As shown in Table 6 and Figures 4
and 5, the gas transport data for PSf-s-HBTMS and PSf-
o-HBTMS were better than those of PSf. The substitu-
tion site for the HBTMS group is important especially
in terms of the extent of permeability increase since the
data for PSf-o-HBTMS were better than those of PSf-
δ ) (log k - log Rij - n log Pi) cos(arc tan(n)) (3)
Single gas permeability coefficients (P) were measured
on polymer dense films for CO2, O2, and N2 using the
constant downstream volume method. A summary of
d-spacing, permeability coefficients, and permselectivi-
ties for these gases and the δ values are summarized
in Table 6, and the gas permeation improvement trends
are shown in Figures 4 and 5. The distances in Table 6
are to the upper bound lines that were generated from
the data of Park et al.2 In both Figures 4 and 5, the
Ta ble 6. Ga s P er m ea bilities of Mod ified P Sf’s
P, Barrera
Rb
δ × 103
CO2/N2
polymer
d-spacing
CO2
O2
N2
CO2/N2
O2/N2
O2/N2
PSf
5.0
5.3
5.3
5.1
5.3
5.5
5.1
5.1
5.2
5.3
5.3
5.6
21
70
18
6.1
29
12.0
105
21.0
32
1.4
5.10
14.6
4.0
1.4
7.1
3.4
23.9
5.6
6.95
17.6
0.25
0.96
3.29
0.95
0.27
1.3
0.67
5.63
1.06
1.51
3.36
22.4
22.2
21.3
18.9
22.6
22.3
17.9
18.6
19.8
21.3
21.4
5.6
387
204
54
293
372
159
373
54
213
137
133
255
246
99
187
114
132
174
74
PSf-s-HBTMS
PSf-o-HBTMS
PSf-o-CH2-TMS
PSf-o-DMPSc
PSf-o-TMSd
6FPSf
5.31
4.44
4.2
5.18
5.3
5.1
4.25
5.3
6FPSf-o-HBTMS
TMPSf
252
164
48
TMPSf-TMSe
TMPSf-HBTMS
4.60
4.97
72
a
b
Permeability coefficients measured at 35 °C and 1 atm pressure. 1 Barrer ) 10-10 [cm3 (STP)‚cm]/(cm2‚s‚cm Hg). Ideal permselectivity
R ) (Pa)/(Pb). c Reference 26. Reference 11. e Reference 12.
d

Macromolecules, Vol. 36, No. 18, 2003
Enhanced Gas Permeabilities of Polysulfones 6815
s-HBTMS. This shows a trend similar to our previous
observations,11,17,26 which is that substitution at the
ortho ether site is more effective in increasing perme-
ability than at the ortho sulfone sites. This is due to
the fact that the ether linkage is more flexible and is
hindered more readily by bulky substituents than the
already hindered sulfone linkage. Since this effect was
observed for PSf-HBTMS, the ortho sulfone site poly-
mers TMPSf-s-HBTMS and 6FPSf-s-HBTMS were not
prepared.
as well as TMPSf and PSf. 6FPSf-o-HBTMS was the
most permeable polymer among the HBTMS polymers
because of effects of both the HBTMS and the hexaflu-
oroisopropylidene (6F) groups. Bulky HBTMS groups
are responsible for forcing chain segments apart, thereby
increasing permeability. From other literature studies,
especially polyimides, the 6F group is known to signifi-
cantly improve permeability and permselectivity by
increasing chain stiffness, reducing intersegmental
chain packing, and reducing interchain interactions
such as charge-transfer complexes (CTCs), since CF3
groups are electron-withdrawing and reduce CTC for-
mation.
The highest overall increase in P versus R was
achieved with TMPSf-HBTMS, which had a distance δ
of only 0.048 for the CO2/N2 and 0.074 for the O2/N2 gas
pair. Structurally, TMPSf-HBTMS contains both the
tetramethylbisphenol segment as well as the HBTMS
substituent, both of which increase the interchain
distance. However, the HBTMS substituent is consider-
ably larger than a methyl group such that the methyl
groups may fill some space between the polymer chains
and further increase chain stiffening. This may explain
why TMPSf-HBTMS has both increased permeability
and permselectivity for the CO2/N2 gas pair as shown
in Figure 4. In the case of the O2/N2 gas pair, the
TMPSf-HBTMS polymer also had the highest P versus
R of all the derivatives, though there was a decline in
permselectivity.
Apart from the HBTMS substituent, PSf’s with other
silyl substituents were compared for gas permeability
data as listed in Table 6. PSf-o-DMPS26 and PSf-o-
TMS11 were prepared previously, whereas PSf-o-CH2-
TMS containing an alkyl spacer was prepared for this
study. Compared with other structural silyl substitu-
ents, PSf-o-HBTMS had significantly higher P(CO2) and
P(O2) and maintained good permselectivity against N2.
Although PSf-o-CH2-TMS showed increases in P(CO2)
and P(O2) compared with PSf as evidenced by a slightly
larger d-spacing of 5.1 Å compared with 5.0 Å for the
PSf, it had inferior gas transport properties because the
concomitant permselectivities were rather lower. This
was somewhat expected because of the flexible -CH2-
spacer between the polymer chain and the TMS group.
On the other hand, the silicon atom is bonded directly
with the aromatic polymer chain in the PSf-o-DMPS
polymer and the DMPS substituent contains a rigid
phenyl ring on the Si(CH3)2. The overall effect of
introducing this substituent was negligible since both
P and R remained approximately the same or slightly
increased. It seems likely that two opposing effects lead
to this result. The steric effect of the substituent causes
an increase in interchain spacing from 5.0 to 5.3 Å26
that would be expected to show in an increase in P.
However, the planarity of the phenyl rings on the DMPS
substituent is capable of allowing more interchain
packing thereby reducing the fractional free volume
while maintaining permselectivity through structural
rigidity. The third comparative polymer PSf-o-TMS had
a good combination of P and R. PSf-o-HBTMS had
higher permeabilities and slightly lower permselectivi-
ties than PSf-o-TMS, which had less space between the
TMS group and the polymer chain. This is in contrast
with the work of Nagai et al.15 In this work, it was found
that when a phenyl group was inserted between the
TMS group and the main chain of poly[1-(trimethylsi-
lyl)-1-propyne], interchain interactions increased result-
ing in a large permeability decrease.
The HBTMS derivatives of all the polymers PSf,
6FPSf, and TMPSf all resulted in high P versus R for
the CO2/N2 gas pair, all giving approximately the same
distance δ ∼ 0.05 to the upper bound line as shown in
Table 6 and Figure 4. This is remarkable in the case of
PSf and TMPSf, which both had almost the same values
for P and R. For the O2/N2 gas pair, the HBTMS
derivatives of PSf and 6FPSf gave an almost identical
P versus R trend, whereas the TMPSf derivative showed
less reduction in permselectivity.
Con clu sion
Modified PSf, 6FPSf, and TMPSf containing the novel
substituent HBTMS were successfully synthesized by
lithiating polymers and then treating with TMSBA. The
resulting side substituent contains the bulky TMS
groups separated from the polymer chain by an aromatic
ring. The open structure leads to high permeability,
while the substituent rigidity provides a selective dif-
fusion environment. PSf-o-HBTMS, TMPSf-HBTMS,
and 6FPSf-o-HBTMS for CO2 and O2 were significantly
more permeable than PSf, TMPSf, and 6FPSf with little
or no reduction in permselectivity (R) particularly for
the CO2/N2 gas pair. The O2/N2 gas pair showed overall
improvements in P versus R, especially for the TMPSf
derivative, whose high Tg is the same as that of the
starting polymer. The distances δ to the Robeson upper
bound line3 for all the HBTMS derivatives were much
shorter in comparison with PSf, TMPSf, and 6FPSf,
especially for the CO2/N2 gas pair, which remarkably
all had values of δ ∼ 0.05. The typical P versus R
tradeoff observed for polymers is notably absent for the
CO2/N2 gas pair in the HBTMS derivatives. The large
increases in P observed for gases (up to 13-fold) are
ascribed to the steric restrictions of the HBTMS sub-
stituent that decrease the packing efficiency, while the
retention of good permselectivity is imparted by the
rigidity of the substituent.
Both 6FPSf and TMPSf have higher P than PSf
because of the nature of the 6F group and the steric
bulkiness of the tetramethyl segment. Thus, the incre-
mental increases in P observed with the introduction
of the HBTMS substituent were less than those for PSf.
The observed increases in P(CO2) and P(O2) for 6FPSf-
o-HBTMS were ∼9-fold and ∼7-fold in comparison with
6FPSf, respectively. The polymer maintained excellent
R(CO2/N2), with no reduction evident. The overall result
for this gas pair was a reduction in the δ distance from
0.373 to 0.054, which is close to the upper bound line.
However, there was an observed decrease for R(O2/N2),
with the δ distance being reduced from 0.187 to 0.114.
It is clearly illustrated in Figure 4 that the decline in
R(O2/N2) for 6FPSf and PSf is greater than that of the
TMPSf derivative. This is in contrast to Figure 5 which
shows that little or no decreases in R(CO2/N2) are
observed for the polymer HBTMS derivatives of 6FPSf

6816 Dai et al.
Macromolecules, Vol. 36, No. 18, 2003
(14) Sisto, R.; Bonfanti, C.; Valenti, C. J . Membr. Sci. 1994, 95,
135.
(15) Nagai, K.; Masuda, T.; Nakagawa, T.; Freeman, B. D.;
Pinnau, I. Prog. Polym. Sci. 2001, 26, 721.
Ack n ow led gm en t. The authors gratefully acknowl-
edge partial financial support from the international
collaboration project between the National Research
Council of Canada and the Korea Institute of Science
and Technology.
(16) Madkour, M. Polymer 2000, 41, 7489.
(17) Lee, K. J .; J ho, J . Y.; Kang, Y. S.; Dai, Y.; Robertson, G. P.;
Guiver, M. D.; Won, J . J . Membr. Sci. 2003, 212, 147.
(18) Aoki, T. Prog. Polym. Sci. 1999, 24, 951.
Refer en ces a n d Notes
(19) George, S. C.; Thomas, S. Prog. Polym. Sci. 2001, 26, 985.
(20) Guiver, M. D.; ApSimon, J . W.; Kutowy, O. J . Polym. Sci.,
Polym. Lett. Ed. 1988, 26, 123.
(21) Guiver, M. D.; Kutowy, O.; ApSimon, J . W. Polymer 1989,
30, 1137.
(22) Guiver, M. D.; Robertson, G. P. Macromolecules 1995, 28, 294.
(23) Guiver, M. D.; Robertson, G. P.; Foley, S. Macromolecules
1995, 28, 7612.
(24) Guiver, M. D.; Zhang, H.; Robertson, G. P.; Dai, Y. J . Polym.
Sci., Part A: Polym. Chem. 2001, 39, 675.
(25) Guiver, M. D.; Robertson, G. P.; Yoshikawa, M.; Tam, C. M.
In Membrane Formation and Modification; ACS Symposium
Series, No. 744; Pinnau, I., Freeman, B., Eds.; American
Chemical Society: Washington, DC, 1999; Chapter 10, p 137.
(26) Kim, I.-W.; Lee, K. J .; J ho, J . Y.; Park, H. C.; Won, J .; Kang,
Y. S.; Guiver, M. D.; Robertson, G. P.; Dai, Y. Macromolecules
2001, 34, 2908.
(1) Pandey, P.; Chauhan, R. S. Prog. Polym. Sci. 2001, 26, 853.
(2) Park, J . Y.; Paul, D. R. J . Membr. Sci. 1997, 125, 23.
(3) Robeson, L. M. J . Membr. Sci. 1991, 62, 165.
(4) Robeson, L. M.; Smith, C. D.; Langsam, M. J . Membr. Sci.
1997, 132, 33.
(5) Henis, J . M. S.; Tripodi, M. K. Sep. Sci. Technol. 1980, 15,
1059.
(6) Aitken, C. L.; Koros, W. J .; Paul, D. R. Macromolecules 1992,
25, 3424.
(7) Aitken, C. L.; McHattie, J . S.; Paul, D. R. Macromolecules
1992, 25, 2910.
(8) McHattie, J . S.; Koros, W. J .; Paul, D. R. Polymer 1992, 33,
1701.
(9) McHattie, J . S.; Koros, W. J .; Paul, D. R. Polymer 1991, 32,
2618.
(10) McHattie, J . S.; Koros, W. J .; Paul, D. R. Polymer 1991, 32,
840.
(11) Guiver, M. D.; Robertson, G. P.; Rowe, S.; Foley, S.; Kang,
Y. S.; Park, H. C.; Won, J .; Le Thi, H. N. J . Polym. Sci., Part
A: Polym. Chem. 2001, 39, 2103.
(12) Dai, Y.; Guiver, M. D.; Robertson, G. P.; Bilodeau, F.; Kang,
Y.; Lee, K.; J ho, J .; Won, J . Polymer 2002, 43, 5369.
(13) Zhang, J .; Hou, X. J . Membr. Sci. 1994, 97, 275.
(27) Zaidi, S. M. J .; Mikhailenko, S. D.; Robertson, G. P.; Guiver,
M. D.; Kaliaguine, S. J . Membr. Sci. 2000, 173, 17.
(28) Breck, D. W. In Zeolite Molecular Sieves; J ohn Wiley &
Sons: New York, 1974; Chapter 8, p 636.
MA0346411