Effect of isomerism on molecular packing and gas transport properties of poly(benzoxazole-co-imide)s

Article abstract of DOI:10.1021/ma501891m

A facile approach to synthesize poly(benzoxazole-co-imide)s without thermal rearrangement at high temperature is proposed. Poly(benzoxazole-co-imide)s with improved mechanical and solution-processable properties were prepared through polycondensation of 4

Full text of DOI:10.1021/ma501891m

Article
pubs.acs.org/Macromolecules
Effect of Isomerism on Molecular Packing and Gas Transport
Properties of Poly(benzoxazole-co-imide)s
Yongbing Zhuang,†,‡ Jong Geun Seong,† Yu Seong Do,† Hye Jin Jo,† Moon Joo Lee,§ Gang Wang,†
Michael D. Guiver,*,⊥,†,∥ and Young Moo Lee*,†
†Department of Energy Engineering, College of Engineering, Hanyang University, Seoul 133-791, Republic of Korea
‡College of Chemistry and Chemical Engineering, Hunan University of Arts and Science, Changde, Hunan 415000, P.R. China
§School of Chemical Engineering, College of Engineering, Hanyang University, Seoul 133-791, Republic of Korea
⊥State Key Laboratory of Engines, School of Mechanical Engineering, Tianjin University, Tianjin, 300072, P.R. China
S
* Supporting Information
ABSTRACT: A facile approach to synthesize poly-
(benzoxazole-co-imide)s without thermal rearrangement at
high temperature is proposed. Poly(benzoxazole-co-imide)s
with improved mechanical and solution-processable properties
were prepared through polycondensation of 4,4′-(hexafluor-
oisopropylidene) diphthalic anhydride (6FDA) with three
synthesized novel benzoxazole-containing diamines and a
commercial diamine. These poly(benzoxazole-co-imide)s had
high tensile strengths of 110.3−122.0 MPa and good
elongation at break of 11.9−26.3%, good thermal stability
and high glass transition temperatures (Tgs) of up 306 °C. The effect of chain isomerism on molecular packing and physical and
gas transport properties of the poly(benzoxazole-co-imide)s was investigated. The para-connecting isomers exhibited higher
molecular weights (Mws), better mechanical properties, higher Tgs, higher chain packing order and better overall performance for
CO2/CH4 and CO2/N2 separations as compared to the corresponding meta-connecting ones. This study guides molecular
architecture to improve particular membrane separation performance by introducing either para- or meta-connections into
polymeric main chains.
specific gases without significantly impairing their inherently
good selectivity.
INTRODUCTION
■
Polyimides (PIs) are one of the most attractive polymers for
membrane-based gas separations because they exhibit high gas
selectivity for major gas pairs (e.g., O2/N2 and CO2/CH4) and
desirable permeability.1−9 They also have good thermal stability
and chemical resistance, and excellent mechanical properties,
which make them suitable for use in harsh conditions, such as
high temperatures, high pressures, high stress, and aggressive
chemical environments. More importantly, the specific design
architecture within the diverse family of polyimides can be
tailored to improve their processability and other performance
parameters for particular applications by simply adjusting
diamine and dianhydride monomer structures. However, for
many polymeric membranes used for gas separation applications,
a gain in permeability is usually countered by a decline of
selectivity, which has been described by Robeson for several gas
pairs.10,11 In an effort to achieve optimum economic perform-
ance for membrane materials used as gas separation media, high
permeability along with high selectivity are desirable. Commer-
cial polyimides, such as Upilex and Kapton films usually exhibit
low permeability (e.g., the permeability to CO2 is less than 10
Barrer). Therefore, much of the research work is being addressed
to explore novel polyimides with improved permeability to
It is well-known that the gas permeability of a polymer can be
effectively improved by introducing bulky groups or contorted
moieties into backbones to prevent effective chain packing. For
example, polyimides incorporating large −C(CF3)2− linkages in
the backbones usually exhibited favorable permeability.12,13 Also,
the increase in polymer backbone rigidity improved gas
separation performance, especially in terms of enhancing size
sieving ability.14,15 Polybenzoxazole (PBO)16,17 is a representa-
tive example of highly rigid polymers that are thermally and
chemically resistant. Fluorinated PBO membranes have been
investigated as gas separation materials previously by Maruyama
et al.18 In recent years, high temperature conversion of α-
hydroxy-polyimide precursors into PBOs has been widely
utilized to construct high performance gas separation mem-
branes (so-called thermal rearrangement (TR) mem-
branes1,19−31). TR membranes exhibit extraordinary gas
separation properties as compared to other polymer membranes,
mainly due to the formation of microporosity and their particular
Received: September 11, 2014
Revised: October 31, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
Scheme 1. Synthesis of Benzoxazole-Containing Diamine Monomers (p-BFO, m-BFO, and m-BOA)
Scheme 2. Synthesis of the Poly(benzoxazole-co-imide)s Based on the Benzoxazole-Containing Diamines
pore size distributions within membranes derived from thermally
driven structural rearrangements in the solid state.20 However,
some of these TR membranes exhibit poor mechanical properties
compared to conventional polyimide membranes, and they are
also insoluble and cross-linked after α-hydroxypolyimide
precursors were treated at high temperatures (>400 °C).
Moreover, there is a possibility of chain decomposition during
high temperature treatment above 450 °C.
In view of the desirable properties of PIs and PBOs,
nonthermally rearranged poly(benzoxazole-co-imide)s are ex-
pected to integrate the advantages of both PIs and PBOs. A series
of nonthermally rearranged aromatic poly(benzoxazole-imide)s,
such as fluorinated aromatic poly(ether imide benzoxazole)s
(PEIBs), have been prepared previously for microelectronic
applications by using commercially available dianhydrides and
diamines containing preformed benzoxazole moieties,32−35 e.g.,
2,2′-o-phenylene-bis(6-aminobenzoxazole)32 and 2,2′-bis[2-(4-
aminophenoxy)benzoxazol-6-yl]hexafluoropropane.33,34 Fluori-
nated PEIBs showed excellent solubility and processability using
current commercial fabrication processes due to the introduction
of flexible ether bonds into chain backbones. However, they were
not suitable for membrane-based gas separation applications
because of their low gas permeability and poor thermal-resistance
properties, resulting from low chain rigidity.
flexibility affect the resulting properties of the corresponding
polymers, e.g., gas permeability.3,9,36−38 The effects of structural
symmetry and isomeric chain orientation on the physical and gas
separation properties of rigid polymers have been previously
reported for polymer membranes to a certain de-
gree.3,9,13,19,20,24,36,39,40 In general, changing the connecting
bond positions in polyimide backbones from para-connecting
linkages to meta-connecting ones appears to reduce their
permeability and diffusivity to gases, while correspondingly
increasing selectivity. For example, the para-connecting
polyimide PMDA-pp′ODA derived from pyromellitic dianhy-
dride (PMDA) and 4,4′-oxidianiline exhibited about 2.7 times
higher permeability to O2 and relatively lower selectivity of O2/
N2, in comparison with the one from the meta-connecting
PMDA-mp′ODA polymerized from PMDA and 3,4′-oxidiani-
line.13 To date, no related data was available on the gas transport
properties for these nonthermally rearranged poly(benzoxazole-
co-imide) membranes derived from the preformed benzoxazole-
containing diamines.
In this study, two fluorinated diamine monomers (p-BFO and
m-BFO) and one nonfluorinated diamine monomer (m-BOA)
containing benzoxazole moieties depicted in Scheme 1 were
synthesized. The corresponding poly(benzoxazole-co-imide)s
were prepared by using a dianhydride, 6FDA, having a bulky 6F
(−C(CF3)2−) group, with the three novel synthesized diamine
monomers along with a commercial diamine containing a
benzoxazole moiety, 5-amino-2-(4-aminobenzene)benzoxazole
The effects of structural symmetry and isomerism on the
polymer properties are of interest because they provide a means
to further understand how interchain packing and chain
B
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
an ethanol−water mixture, dried in vacuum at 80 °C, and further
purified by sublimation to give 6.94 g (yield: 76.9%) of yellow crystals.
Mp: 253 °C. 1H NMR (300 MHz, DMSO-d6): δ 5.10 (H-9, s, 6H), 5.44
(H-1, s, 2H), 6.63−6.65 (H-3, d, J = 4.2 Hz, 2H), 6.73−6.75 (H-8, d, J =
3.9 Hz, 2H), 6.84 (H-6, s, 2H), 7.18−7.20 (H-4, t, J = 3.9 Hz, 2H), 7.25−
7.26 (H-7, d, J = 3.9 Hz, 2H), 7.36−7.38 (H-2, H-5, m, 4H). 13C NMR
(300 MHz, DMSO-d6): δ 64.3, 111.0, 112.2, 114.8, 117.6, 121.2, 123.1,
125.0, 126.4, 126.7, 129.1, 129.8, 141.6, 149.4, 164.6. FTIR (powder, ν,
cm−1): 3426, 3395, 3316, 3211 (N−H stretching), 1630, 1593 (cyclic
CN/CC stretching), 1480, 1464 (skeletal vibrations in conjugated
phenyl/benzoxazole ring), 1255 (Ar−C−O asymmetric stretching).
Anal. Calcd for C13H11N3O: C, 69.32; H, 4.92; N, 18.66. Found: C,
67.82; H, 4.96; N, 18.59.
(p-BOA). Obviously, the para-connecting p-BFO and p-BOA are
isomers of meta-connecting m-BFO and m-BOA, respectively.
The molecular packing and properties including single gas
permeability for the resulting four rigid poly(benzoxazole-co-
imide)s shown in Scheme 2 were investigated. The influence of
polymer structural isomerism on molecular packing and
properties were identified. Systematic investigations and analyses
provide a useful framework for designing modified polyimides to
meet the requirements in commercial membrane-based gas
separations such as desirable permeability, excellent mechanical
and thermal stability etc.
Synthesis of Poly(benzoxazole-co-imide)s. Poly(benzoxazole-
co-imide)s were prepared by using dianhydride with four benzoxazole-
containing diamines, p-BFO, m-BFO, p-BOA, and m-BOA as raw
materials via a two-step azeotropic imidization process, as shown in
Scheme 2. Using the preparation of 6FDA-p-BFO (PIa) as an example,
in a four-necked flask, p-BFO (1.7054 g, 30.0 mmol) was dissolved in
NMP (28 mL), and then 6FDA (1.3327 g, 30.0 mmol) was added. After
the mixture was stirred at room temperature for more than 12 h, 6 mL of
toluene as an azeotropic agent was added to the reaction solution, and
the solution mixture was then heated under reflux for at least 12 h under
a N2 atmosphere. While the toluene was refluxing, water was removed
using a Dean−Stark trap. The resulting brownish solution was poured
into a mixture of water and methanol (1 L, v/v = 1:1) under vigorous
stirring. The resulting fibrous precipitate was filtered off, washed with
cold water, methanol, and dried at 120 °C in a vacuum oven to yield the
6FDA-p-BFO (PIa) white fibrous powder.
EXPERIMENTAL SECTION
■
Materials. 2,4-Diaminophenol dihydrochloride (98.0%), 4-amino-
benzoic acid (>98.0%), 3-aminobenzoic acid (>98.0%), polyphosphoric
acid (PPA, 115%), tin(II) chloride dihydrate (98.0%), toluene (99.8%),
and N-methyl-2-pyrrolidinone (NMP, >99.0%) were purchased from
Sigma-Aldrich Chemical Co. (Milwaukee, WI) and used as received.
4,4′-(Hexafluoroisopropylidene) diphthalic anhydride (6FDA) was
obtained from Sigma-Aldrich Chemical Co. and purified by sublimation
before use. 2,2′-Bis(3-amine-4-hydroxyphenyl)hexafluoropropane
(APAF) was purchased from Central Glass Co. Ltd. (Tokyo, Japan)
and was purified by sublimation before use. Sodium carbonate (>99.0%)
was purchased from Tokyo Chemical Industry (Tokyo, Japan). 5-
Amino-2-(4-aminobenzene)benzoxazole (p-BOA) was obtained from
Changzhou Sunlight Medical Raw Material Co. Ltd. (Jiangsu, China)
and was used as received.
Synthesis of 2,2′-Bis[2-(4-aminophenyl)benzoxazol-6-yl]-
hexafluoropropane (p-BFO). Under a nitrogen atmosphere, PPA
(70.00 g) was added to a three-necked flask, and 4-aminobenzoic acid
(5.80 g, 42.3 mmol), APAF (7.33 g, 20.0 mmol), and tin(II) chloride
dihydrate (0.40 g, 1.8 mmol) were added. The mixture was heated to
110 °C with stirring overnight, and then the resulting solution was
heated to 195 °C and maintained at that temperature for more than 7 h.
After cooling to room temperature, the resulting reaction mixture was
neutralized with 10% sodium carbonate solution. The precipitate was
filtered off, rinsed with distilled water many times, recrystallized in an
ethanol−water mixture, and dried in vacuum at 80 °C to give 9.30 g
(yield: 81.9%) of white needle crystals. Melting point (Mp): 256 °C. 1H
NMR (300 MHz, DMSO-d6): δ 6.10 (H-1, s, 4H), 6.69−6.70 (H-2, d, J
= 4.5 Hz, 4H), 7.24−7.25 (H-5, d, J = 4.5 Hz, 2H), 7.62 (H-4, s, 2H),
7.74−7.75 (H-6, d, J = 4.5 Hz, 2H), 7.85−7.87 (H-3, d, J = 4.5, 4H). 13C
NMR (300 MHz, DMSO-d6): δ 64.1, 110.1, 112.1, 113.5, 120.1, 121.3,
123.2, 125.0, 125.5, 128.7, 129.3, 142.2, 149.9, 152.8, 165.1. FTIR
(powder, ν, cm−1): 3475, 3337, 3216 (N−H stretching), 1630, 1606
(cyclic CN/CC stretching), 1497, 1483 (skeletal vibrations in
conjugated phenyl/benzoxazole ring), 1251 (Ar−C−O asymmetric
stretching). Anal. Calcd for C29H18N4O2F6: C, 61.27; H, 3.19; N, 9.86.
Found: C, 60.98; H, 3.37; N, 10.07.
Synthesis of 2,2′-Bis[2-(3-aminophenyl)benzoxazol-6-yl]-
hexafluoropropane (m-BFO). The synthetic procedure was similar
to that used for diamine p-BFO except that 3-aminobenzoic acid was
used as raw material. White needle-like crystals with a yield of 86.5%
were obtained. Mp: 112 °C. 1H NMR (300 MHz, DMSO-d6): δ 5.54 (H-
1, s, 4H), 6.81−6.83 (H-3, d, J = 5.4 Hz, 2H), 7.23−7.26 (t, H-4, J = 3.9
Hz, 2H), 7.33−7.36 (H-7, H-8, t, J = 4.5 Hz, 4H), 7.43 (H-6, s, 2H), 7.77
(H-2, s, 2H), 7.86−7.88 (H-5, d, J = 4.5 Hz, 2H). 13C NMR (300 MHz,
DMSO-d6): δ 64.3, 111.0, 112.2, 114.8, 117.7, 121.2, 123.1, 125.0, 126.4,
126.7, 129.1, 129.8, 141.7, 149.5, 150.1, 164.5. FTIR (powder, ν, cm−1):
3452, 3354, 3226 (N−H stretching), 1623, 1593 (cyclic CN/CC
stretching), 1493, 1480 (skeletal vibrations in conjugated phenyl/
benzoxazole ring), 1252 (Ar−C−O asymmetric stretching). Anal. Calcd
for C29H18N4O2F6: C, 61.27; H, 3.19; N, 9.86. Found: C, 60.87; H, 3.36;
N, 10.02.
6FDA-p-BFO (PIa). ATR-FTIR (membrane, ν, cm−1): 1787 (imide
CO symmetric stretching), 1724 (imide CO asymmetric
stretching), 1623, 1608 (cyclic CN/CC stretching), 1503, 1479
(skeletal vibrations in conjugated phenyl/benzoxazole ring), 1362
(imide −C-N), 1249 (Ar−C−O asymmetric stretching). Molecular
weight: Mw = 64 × 103 with a polydispersity of 2.2. Anal. Calcd for
C48H20F12N4O6: C, 59.03; H, 2.06; N, 5.74. Found: C, 58.19; H, 2.18; N,
5.70.
6FDA-m-BFO (PIb). ATR-FTIR (membrane, ν, cm−1): 1786 (imide
CO symmetric stretching), 1723 (imide CO asymmetric
stretching), 1625, 1608 (cyclic CN/CC stretching), 1492, 1477
(skeletal vibrations in conjugated phenyl/benzoxazole ring), 1367
(imide −C−N), 1249 cm−1 (Ar−C−O asymmetric stretching).
Molecular weight: Mw = 114 × 103 with a polydispersity of 2.5. Anal.
Calcd for C48H20F12N4O6: C, 59.03; H, 2.06; N, 5.74. Found: C, 58.54;
H, 2.04; N, 5.71.
6FDA-p-BOA (PIc). 1H NMR (300 MHz, DMSO-d6): δ 7.54 (br, s,
1H), 7.74 (br, s, 2H), 7.78 (br, s, 2H), 7.944 (s, 1H), 7.99 (m, 3H), 8.23
(br, s, 2H), 8.389 (br, s, 2H). ATR-FTIR (membrane, ν, cm−1): 1785
(imide CO symmetric stretching), 1719 (imide CO asymmetric
stretching), 1622, 1608 (cyclic CN/CC stretching), 1502, 1478
(skeletal vibrations in conjugated phenyl/benzoxazole ring), 1361
(imide −C−N), 1252 cm−1 (Ar−C−O asymmetric stretching).
Molecular weight: Mw = 115 × 103 with a polydispersity of 3.3. Anal.
Calcd for C32H13F6N3O5: C, 60.67; H, 2.07; N, 6.63. Found: C, 59.63;
H, 1.90; N, 6.45.
6FDA-m-BOA (PId). 1H NMR (300 MHz, DMSO-d6): δ 7.53 (br, s,
1H), 7.74−7.80 (m, 4H), 7.93−8.00 (m, 4H), 8.22−8.40 (m, 4H).
ATR-FTIR (membrane, ν, cm−1): 1785 (imide CO symmetric
stretching), 1720 (imide CO asymmetric stretching), 1625, 1610
(cyclic CN/CC stretching), 1492, 1477 (skeletal vibrations in
conjugated phenyl/benzoxazole ring), 1367 (imide −C−N), 1252 cm−1
(Ar−C−O asymmetric stretching). Molecular weight: Mw = 188 × 103
with a polydispersity of 2.9. Anal. Calcd for C32H13F6N3O5: C, 60.67; H,
2.07; N, 6.63. Found: C, 59.07; H, 1.72; N, 6.56.
Membrane Preparation. The poly(benzoxazole-co-imide) fibrous
powders were dissolved in NMP to form a 20 wt % polymer solution and
were cast onto clean glass plates after filtering with 1.0 μm Nylon (NY)
filter cartridges. The glass plates were placed in a vacuum oven and
slowly heated from 60 to 250 °C to evaporate solvent by successive
heating at 60, 100, 150, 200, and 250 °C for 1 h at each temperature. The
Synthesis of 5-Amino-2-(3-aminobenzene)benzoxazole (m-
BOA). The synthetic procedure was similar to that used for diamine p-
BFO except that 3-aminobenzoic acid and 2,4-diaminophenol
dihydrochloride were used as raw materials, respectively. The precipitate
was filtered off, rinsed with distilled water many times, recrystallized in
C
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
resulting membranes were removed from the glass plates by immersion
in hot water, washed with deionized water, and dried overnight in a
vacuum oven at 120 °C. The dry membranes were further heated up to
300 °C in a muffle furnace at a rate of 5 °C/min with a Eurotherm
controller, and were maintained at 300 °C for 1 h to obtain the fully
imidized poly(benzoxazole-co-imide) membranes. The thickness of all
films after treatment was approximately 42−62 μm.
where θ is “time-lag” and l is the membrane thickness. Solubility
coefficient (S) was obtained indirectly via the equation:
(5)
S = P/D
RESULTS AND DISCUSSION
■
Monomer Syntheses and Electronic Effect. Three new
benzoxazole-containing diamine monomers, p-BFO, m-BFO,
and m-BOA, were synthesized in PPA at elevated temperatures
using 4-aminobenzoic acid/3-aminobenzoic acid with 2,2′-bis(3-
amine-4-hydroxyphenyl)hexafluoropropane (APAF)/2,4-diami-
nophenol dihydrochloride following literature procedures,33,34 as
shown in Scheme 1. The structures of diamine monomers were
confirmed by FTIR spectroscopy, elemental analyses, 1H NMR
(Figure 1) and 13C NMR recorded in DMSO-d6. In 1H NMR, the
Characterization. An infrared microspectrometer (IlluminatIR,
SensIR Technologies, Danbury, CT) was used to characterize the
diamine monomers and polymer membranes. Nuclear magnetic
resonance (NMR) spectra were obtained with a Mercury Plus 300
MHz spectrometer (Varian, Inc., Palo Alto, CA). Molecular weight was
measured by gel permeation chromatography (Waters GPC Systems,
Milford, MA) with polystyrene as an external standard and NMP as the
eluent. Elemental analyses for the monomers and polymers were
performed with a Thermofinnigan EA1108 (Fisions Instrument Co.,
Milan, Italy) elemental analyzer. Mechanical properties were obtained
using a Universal Testing Machine, UTM (AGS-J, Shimadzu, Kyoto,
Japan) with specimens prepared according to ASTM D638-Type 5
recommendations, and the result of each sample was averaged with at
least five specimens. Differential scanning calorimetry (DSC) analysis
and thermogravimetric analysis (TGA) were recorded on a TA
Instruments Q-20 calorimeter and TGAQ50 instrument (TA Instru-
ment, New Castle, DE), respectively, at a heating rate of 10 °C/min
under a nitrogen atmosphere. The glass transition temperatures (Tgs) of
the polyimide membranes were determined by second scan. Wide angle
X-ray diffractometry (WAXD) spectra were recorded in the reflection
mode at room temperature by using a Rigaku Denki D/MAX-2500
(Rigaku, Tokyo, Japan) diffractometer with Cu Kα (wavelength λ = 1.54
Å) radiation. The net charges by the Hirshfeld method were calculated
by using the GGA-BLYP/DND method within Dmol3 module
embedded in Materials Studio 5.0 program (Accelrys, San Diego,
CA).41 The geometry of a repeat unit was optimized by using the
COMPASS force-field. The conformation rotation of single bonds was
conducted using the conformer modulus.
Densities of the membranes were measured using a density
measurement kit (Sartorius LA 120S, Sartorius AG, Goettingen,
Germany) by the buoyancy method. The liquid used for measurement
was 2,2,4-trimethylpentane (Sigma-Aldrich). Fractional free volume
(FFV, Vf) was calculated as follows:
M0
ρ
V =
sp
(1)
V − 1.3 × Vw
sp
V =
f
V
sp
(2)
where Vsp is the molar volume of polymers determined by the measured
density and Vw is the van der Waals molar volume based on Bondi’s
group contribution theory.
The constant volume method, so-called time-lag method, was used to
measure the gas permeability of the membranes as described in our
previous study.42 The permeability of all gases was measured at 1 atm at
35 °C and was calculated by using the following equation:
⎛
⎜
⎝
⎞
⎟
dp
273.15Vl
P =
1
Figure 1. H NMR spectra (300 MHz, in DMSO-d6) of synthesized
76TΔpA dt
⎠
(3)
benzoxazole-containing diamines (p-BFO, m-BFO, and m-BOA) and
the commercial p-BOA for comparison.
where P (Barrer) is the gas permeability, l (cm) is membrane thickness,
V (cm3) is the downstream chamber volume, T (K) is the measurement
temperature, Δp (cmHg) is the pressure difference between upstream
and downstream, A (cm2) is the effective membrane area, and dp/dt is
the rate of the pressure rise in the downstream chamber at steady state.
The ideal selectivity (αx/y) for components x and y was defined as the
ratio of gas permeability of the two components. The diffusion
coefficient (D), could be calculated from the time-lag apparatus by using
the following equation:
proton signals of the amine groups appeared obviously at 5.10−
6.10 ppm depending on the diamine structures, as shown in
Figure 1. As expected, m-BOA has two obvious amine proton
signals located at 5.10 (H-9) and 5.44 ppm (H-1) corresponding
to the amine protons linked to benzoxazole ring and phenyl ring,
respectively, indicating the asymmetric nature of the m-BOA.
Only one amine proton signal was observed for both the p-BFO
and m-BFO monomers near 6.10 (H-1) and 5.54 ppm (H-1),
l2
D =
(4)
6θ
D
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
Figure 2. Net charges of nitrogen atoms in amine groups of benzoxazole-containing diamines by the Hirshfeld method, calculated using the Dmol3
program embedded in the Materials Studio package.41
Table 1. Molecular Weights and Packing Parameters of the Poly(benzoxazole-co-imide)s
a
b
b
polymer
code
inherent viscosity (dL/g)
Mn (×10−3
)
Mw (×10−3
)
PDI (Mw/Mn)
ρ (g/cm3)
FFV
6FDA-p-BFO
6FDA-m-BFO
6FDA-p-BOA
6FDA-m-BOA
PIa
PIb
PIc
PId
0.31
0.35
0.72
1.14
28.6
45.6
34.7
64.0
64
114
115
188
2.2
2.5
3.3
2.9
1.4388
1.4484
1.4528
1.4349
0.170
0.165
0.153
0.164
a
b
Inherent viscosity measured at a concentration of 0.5 g/dL in NMP at 25 °C. Relative to polystyrene standards.
a
Table 2. Solubility of Poly(benzoxazole-co-imide) Fibrous Powders
solvent
DMAc
code
acetone
THF
DMSO
NMP
methanol
ethanol
chloroform
PIa
PIb
PIc
PId
−
−
−
−
−
−
+
−
−
+
+
+
+
+
+
+
+
+
−
−
−
−
−
−
−
−
+
+
+
+
+
+
a
Key: +, soluble at room temperature; −, insoluble at room temperature.
respectively, indicative of their symmetric structures. In addition,
the chemical shifts of the amine protons are significantly affected
by the amine substitution positions in the phenyl rings and are
also affected by the introduction of -C(CF3)2- in the backbones.
As shown in Figure 1, a downfield shift of 0.36 ppm occurs for the
H-1 amine protons in para-BOA (5.80 ppm) versus meta-BOA
(5.44 ppm), and a downfield shift of 0.56 ppm occurs for the
para-BFO (6.10 ppm) versus the meta-BFO (5.54 ppm),
highlighting the strong resonance induced electron-withdrawing
effect on the substituted aniline.43 This effect is greater in the case
of the para-substituted amine groups than that of the meta-
substituted ones, resulting in a downfield chemical shift of the
para-substituted amine proton. Also, the electron-withdrawing
effect of -C(CF3)2- on amine H-1 induced a small 0.30 ppm
downfield shift to 6.10 ppm in p-BFO in comparison with a 5.80
ppm in p-BOA.
The net charges in the amine N atoms were also calculated by
the Hirshfeld method using the Dmol3 program41 embedded in
the Materials Studio 5.0 package as shown in Figure 2. Obviously,
the net charges of the amine N atoms in para-substituted anilines
are significantly lower than those from the corresponding meta-
substituted ones, and the net charges of the amine N atoms in 6F-
containing diamines are significantly lower than the correspond-
ing ones in non-6F diamines, because of the electron-
withdrawing effect of the −C(CF3)2− groups. Since the
nucleophilicity of diamines decreases by introducing strong
electron-withdrawing −C(CF3)2− and oxazole substituents into
the polymer backbones, it can be inferred that for polymerization
with the same dianhydride, the para-substituted diamines and the
6F-containing diamines would possess lower nucleophilicity than
the corresponding meta-substituted and non-6F ones, respec-
tively. Therefore, the nucleophilicity of the diamines would
increase in the order p-BFO (least nucleophilic) < m-BFO/p-
BOA < m-BOA (most nucleophilic).
Polymerization. The novel poly(benzoxazole-co-imide)s
were synthesized directly from 6FDA and benzoxazole-
containing diamines by a two-step procedure via poly-
(benzoxazole-co-amic acid) intermediates, as shown in Scheme
2. The results of the polymerization are summarized in Table 1.
The inherent viscosities and Mws of poly(benzoxazole-co-imide)s
were in the range of 0.31−1.14 dL/g and 64−188 × 103
depending on the diamine residues. The Mws of these polyimides
follows the sequence: PIa < PIc/PIb < PId, which is exactly the
same as the nucleophilicity order of the diamines from the results
of 1H NMR and molecular modeling. The higher nucleophilicity
of the N atom will induce higher reactivity for the
polycondensation reaction, implying that the Mws of the resulting
poly(benzoxazole-co-imide)s depend significantly on the reac-
tivity of the corresponding diamine monomers when the
polymerizations were performed with the same dianhydride,
6FDA. The solubility of these poly(benzoxazole-co-imide)
fibrous powders was tested in various common solvents, as
shown in Table 2. All the polyimides were soluble in NMP, N,N-
dimethylacetamide (DMAc) and chloroform, indicating that
they have good solution processability.
The anticipated chemical structures of the poly(benzoxazole-
co-imide) membranes resulting from the fabrication process are
supported by ATR-FTIR (Figure 3) and elemental analyses. All
E
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
Figure 4. WAXD curves of poly(benzoxazole-co-imide) membranes.
Figure 3. ATR-FTIR spectra of the poly(benzoxazole-co-imide)
membranes. Gray band refers to the characteristic CO stretching
domain from −CONH of the unimidized intermediate poly(amic acid)s
(PAAs).
backbones.35,44 In addition, the XRD pattern of the PIa
membrane shows an additional low angle peak at about 2θ =
9.0° (peak C), which may correspond to a (00l) diffraction peak
derived from the order along the molecular chain axis.5,7
membranes exhibited typical characteristic imide absorption
bands at around 1785 cm−1 (imide carbonyl symmetric
stretching), 1720 cm−1 (imide carbonyl asymmetric stretching)
and 1367 cm−1 (−C−N stretching), respectively, but bands at
about 1660 cm−1 (CO stretching from the intermediate
−CONH, gray band domain in Figure 3) were not detected,
indicating that imidization was complete. Additionally, character-
istic absorptions at about 1252 cm−1 (Ar−C−O asymmetric
stretching) and 1053 cm−1 (Ar−C−O symmetric stretching),
were indicative of the incorporation of benzoxazole. Moreover,
the experimental elemental analyses of the poly(benzoxazole-co-
imide)s agree well with the calculated values of carbon, hydrogen
and nitrogen.
Molecular Packing and FFV. The chain packing density of
poly(benzoxazole-co-imide)s was evaluated by their FFV, which
was determined by density measurements of membranes. The
FFV values (Table 1) were in the range of 0.153−0.170, which
are close to those of conventional polyimide membranes, such as
6FDA-pPD (FFV = 0.16112) and 6FDA-mPD (FFV = 0.16012)
derived from 6FDA and phenylenediamine, and also the
TRPBO-co-PI (5:5) (FFV = 0.1561). However, they are
significantly less than those of typical TR membranes, such as
TR-cPBO (FFV = 0.3525) and TR-aPBO (FFV = 0.2225), which
exhibited better microporosity and particular pore size
distributions due to the thermally driven structural rearrange-
ments.20 This indicated that the microporosity within the
membranes played a key role in improving the FFV of the
corresponding membranes. Additionally, the para-connecting
PIa exhibited slightly higher FFV than the corresponding meta-
connecting PIb, which is similar to previous findings.12,13
However, a comparison of para-connecting PIc and meta-
connecting PId shows opposite behavior, i.e. the FFV of PIc is
less than that of PId.
Since the calculated d-spacing values of peak A at about
2θ=15° (∼5.9 Å) and B at 24.5° (∼3.6 Å) were close to the
interchain distance and typical π−π stacking distances of
aromatic polyimides,45 respectively, peak A (near 2θ = 15°)
likely corresponds to the average interchain packing in the
ordered domains, whereas peak B (2θ = 24.5°) could be assigned
to the π−π stacking order. The presence of the ordered π−π
stacking implies that some planar and rigid aromatic heterocyclic
rings in the polymer backbones are arranged parallel to each
other. From peak A, the order of the interchain distances (d-
spacing values) in the ordered domains is PIa (5.98 Å) > PIc
(5.86 Å) > PIb (5.81 Å) > PId (5.80 Å). This indicates that the
meta-substituted isomers showed closer chain packing as
compared to the corresponding para-substituted ones in the
ordered domains. It also implies that the introduction of para-
connections in the polymer chain is more effective than meta-
connections for increasing interchain distance.3,9,36 However, the
overall mean interchain distances should be estimated by
considering the contributions of chain packing d-spacing from
both the ordered and the amorphous domains. The overall mean
d-spacing value in the whole membrane can be estimated by42
Aa
Aa + Ab
Ab
Aa + Ab
dmean
=
da +
db
(6)
where Aa and Ab are the diffraction peak areas from the
amorphous diffraction peak and ordered diffraction peak,
respectively, and da and db are the d-spacing values of
macromolecules from the amorphous domains and the ordered
domains, respectively. When Hb = (Ab/(Aa + Ab)), eq 6 can be
rewritten in the following form:
dmean = da − (da − db)Hb
(7)
In order to probe the cause for the abnormal FFV behavior of
PIc and PId, out-of-plane wide-angle X-ray diffraction (WAXD)
measurements were conducted to examine the polymer chain
packing, as shown in Figure 4. All poly(benzoxazole-co-imide)
membranes exhibited at least two diffraction peaks, A and B,
located at about 2θ = 15.0° and 24.5°, respectively, although they
are relatively broad in comparison with those from typical
crystalline polymers. This indicates the existence of some
ordered packing domains combined with primarily amorphous
morphology domains in the membranes, which is similar to other
reported polyimides containing benzoxazole rings in the
Obviously, the mean interchain distance in the whole
membrane is affected by Hb in addition to da and db according
to eq 7. It is well-known that the chain packing in completely
amorphous domains is looser than that in ordered domains,46
implying that da > db (that is da − db > 0). If Δk = da − db, eq 7 can
be further converted to
dmean = (db + Δk) − ΔkHb(Δk > 0)
(8)
From eq 8, a smaller overall mean interchain distance (dmean
may be obtained by increasing the Hb or decreasing the db (the d-
)
F
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
spacing value of macromolecules from the ordered domains
within the membranes). To estimate the difference of the overall
mean interchain distances within the combined membrane
domains between the para-connecting poly(benzoxazole-co-
imide) membranes and the corresponding meta-connecting
ones, eq 9 can be used as
corresponds to a very small positive “dbp − dbm” value in eq 9.
However, the degree of order of PIc (X = 9.91 U0/I0) is about 1.5
times higher than that in PId (X = 6.60 U0/I0), implying Hbm
≪
Hbp. Since Hbm is much smaller than Hbp, as the Δkp is similar to
Δkm, it will result in a large negative “ΔkmHbm − ΔkpHbp” value in
eq 9. Therefore, the combined result due to the large negative
contribution from “ΔkmHbm − ΔkpHbp” value and small positive
contribution from “dbp - dbm” value resulted in a negative Δd value
in eq 9, dmean‑para (PIc) < dmean‑meta (PId), and then a lower FFV
value of the para-connecting PIc relative to that of meta-
connecting PId.
Δd = dmean−para − dmean−meta = (dbp − dbm)
+ (Δkp − Δkm) + (ΔkmHbm − ΔkpHbp)
(9)
where subscripts p and m correspond to the related parameters
from the para-connecting poly(benzoxazole-co-imide) mem-
branes and the corresponding meta-connecting ones, respec-
tively. The relative value of the overall mean interchain distances
between two membranes is either positive or negative, which is
determined by a combined contribution from various chain
packing parameters, including the shift of the interchain distances
in the ordered packing domains, “dbp − dbm” value (represented
by the first term of eq 9), but also the other two terms in eq 9,
“Δkp − Δkm” and “ΔkmHbm − ΔkpHbp”.
Mechanical and Thermal Properties. Good thermal and
mechanical properties are key factors for gas separation
membrane applications, especially under harsh environments
such as elevated temperature and high stress. Thermal analyses
were performed by TGA (Figure 5) and DSC (Figure 6) and the
As reported previously, the average interchain distance in
completely amorphous domains may be obtained by deconvolut-
ing and fitting the WAXD patterns with Gaussian broadening
functions on a single baseline and estimating the central position
of broad amorphous diffraction halo.42,45,47 However, this not
easy to do in this study due to the monotonicity of these WAXD
patterns (Figure 4). Then, it is difficult to obtain the exact values
of “Δkp”, “Δkm” and also “Δkp − Δkm” in eq 9 in this study.
Similar to the d-spacing value in amorphous domains, the Hbp
and Hbm are also very difficult to estimate directly in this study
because contributions due to the ordered packing domains and
completely amorphous domains are not easily identified in the
WAXD patterns. However, in order to estimate the differences in
the ordered degrees in these poly(benzoxazole-co-imide)
membranes, we calculated the ordered degree of macro-
molecules (X) according to eq 1048 based on the characteristic
peak A at around 2θ=15° as follows:
Figure 5. TGA curves of poly(benzoxazole-co-imide) membranes.
IX
UX
U0
I0
X =
×
× 100%
(10)
where U0 and Ux represent the backgrounds of the reference and
experimental samples, whereas I0 and Ix are the integral
intensities of the diffraction lines of the reference and
experimental samples, respectively. The ordered degree values
of these poly(benzoxazole-co-imide) membranes gradually
decreased in the order of PIa (11.22 U0/I0) > PIc (9.91 U0/I0)
> PIb (6.73 U0/I0) > PId (6.60 U0/I0), indicating the para-
connecting linkages significantly improve the packing ordered
degrees of the macromolecules. This also implies that changing
the para-connecting linkages into meta-connecting ones induce a
significant decrease in Hb values for the corresponding
membranes, that is Hbp ≫ Hbm.
According to the results from wide-angle X-ray diffraction
spectra (Figure 4), the interchain distance (dbp = 5.96 Å) of the
PIa membrane in the ordered domains is much larger than that
(dbm = 5.81 Å) of the PIb membrane. From eq 9, a very large
positive “dbp − dbm” value can significantly benefit a positive Δd
value, dmean‑para (PIa) > dmean‑meta (PIb), which was confirmed by
the order of their FFV values: PIa (0.170) > PIb (0.165). In the
case of the comparison of PIc with PId, the interchain distance
(dbp = 5.86 Å) in the ordered domains of the PIc membrane is
very similar to that (dbm = 5.80 Å) of the PId membrane. This
Figure 6. DSC curves of poly(benzoxazole-co-imide) membranes.
data are summarized in Table 3. The observed 5% and 10%
weight losses of the poly(benzoxazole-co-imide)s were in the
ranges of 528−559 °C and 548−586 °C, which confirms their
high thermal stability characteristics. The Tgs were in the range of
306 to 366 °C depending on the polymer structure and
isomerism. The para-connecting polyimide isomers exhibited
relatively higher Tgs than those from the meta-connecting ones. It
is well-known that the Tg values of polymers are closely related to
the chain rigidity and interchain interactions. In order to
understand the reason for the differences in Tgs between the
para-connecting and meta-connecting isomers, geometric
optimization conformations were performed by using Materials
Studio software. There is some conformational flexibility about
the single bonds in each repeat unit shown in Figure 7, in which
G
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
Table 3. Mechanical and Thermal Properties of the Poly(benzoxazole-co-imide) Membranes
a
b
c
d
polymer code
tensile strength (MPa)
tensile modulus (GPa)
elongation at break (%)
Tg (°C)
T5 (°C)
T10 (°C)
char yield (%)
PIa
PIb
PIc
PId
110.3 4.5
122.0 4.9
120.0 7.4
115.1 7.8
1.90 0.06
1.80 0.10
2.10 0.04
1.90 0.08
22.3 2.4
11.9 0.6
26.3 6.5
14.9 1.7
366
306
357
317
528
553
531
559
548
576
556
586
47.5
46.6
50.7
54.1
a
b
Midpoint temperature of baseline shift on the second DSC heating trace of the sample after quenching. Temperature at which 5% weight loss was
c
d
recorded by TGA. Temperature at which 10% weight loss was recorded by TGA. Residual weight retention when heated to 800 °C in nitrogen.
typical dihedral angles, δ and Φ are illustrated in the chain
segment. The energy variation of the poly(benzoxazole-co-
imide)s as a function of the dihedral angles, δ and Φ, is presented
in Figure 7. Obviously, both of the dihedral angles, δ and Φ, with
the lowest energy conformation in the para-connecting
polyimides are closer to 0°, indicative of more coplanar chain
conformations, in comparison with the corresponding meta-
connecting ones with conformations significantly deviating from
coplanarity. It is well-known that strong interchain interactions in
polyimides originate mainly from interchain charge transfer
complex (CTC) formation. The large amount of coplanar/
conjugated segments in the para-connecting polyimide back-
bones significantly limit the rotational freedom around single
bonds and also favor interchain interactions, resulting in stronger
CTC interactions and higher rigidity of macromolecules, which
are responsible for the higher Tgs of para-connecting isomers
(see Table 3 and Figure 6) as compared to the corresponding
ones from meta-substituted isomers. In addition, the Tg values of
the PIc (357 °C) and PId (317 °C) are slightly higher than the
reported 6FDA-pPD derived from 4,4′-phenylenediamine and
6FDA (Tg = 351 °C12) and the 6FDA-mPD derived from 3,3′-
phenylenediamine and 6FDA, (Tg = 298 °C12), respectively,
showing that benzoxazole units increase Tg.
The poly(benzoxazole-co-imide) membranes exhibited ex-
cellent mechanical properties with tensile strengths of 110.3−
122.0 MPa, elongation at break of 11.9−26.3% and initial
modulus values of 1.8−2.1 GPa. The para-connecting isomers
revealed slightly higher modulus and significantly larger
elongation at break than the corresponding meta-connecting
ones. Higher chain rigidity usually induces higher molecular in-
plane orientation, as found in the para-connecting polyimide
Figure 7. Energy variation of the poly(benzoxazole-co-imide)s with the
rotation angles Φ and δ in backbones (Scheme 2), calculated using the
conformers package in Materials Studio.
isomer, resulting in higher in-plane modulus.49 A much higher
level of molecular ordering and more limited chain flexibility in
the para-connecting isomers results in significantly improved
Table 4. Single Gas Permeabilities (P) and Ideal Selectivities (α) for the Poly(benzoxazole-co-imide) Membranes
b
ideal selectivity (α)
code
variable
He
H2
N2
O2
CH4
CO2
H2/N2
O2/N2
CO2/CH4
CO2/N2
H2/CO2
H2/CH4
a
PIa
P
D
S
99
1939
0.39
89
95
1692
0.43
68
1.98
16
10
54
1.06
1.54
5.21
0.41
0.74
4.26
0.54
0.78
5.26
0.43
0.78
4.22
47
19
19
18
8.8
16
25
5.10
37
19
8
48
108
0.44
86
5.11
3.47
1.47
6.02
5.14
1.17
5.71
3.33
1.72
6.31
3.69
1.71
44
12
24
1.21
20
2.04
90
90
1098
0.08
166
0.96
0.79
6.39
0.94
0.98
4.83
1.54
0.89
9.01
0.75
1.42
4.75
33
3.60
44
0.02
3.82
109
0.04
2.76
107
0.03
4.31
118
0.04
a
PIb
P
D
S
23
4055
0.17
76
956
0.54
69
150
0.58
70
12
1.37
17
1294
0.13
127
1.10
5.61
16
3.65
46
a
PIc
P
D
S
25
673
0.86
98
551
0.95
80
114
0.61
90
6.57
7.04
44
1.06
24
706
2.65
5.59
33
0.18
186
a
PId
P
D
S
21
3733
0.20
945
0.64
105
0.86
10
0.89
24
1215
0.15
1.28
18
4.18
a
b
Units: P (Barrer), 10−10 cm3(STP)/cm s cmHg; D, 10−9cm2/s, and S, cm3(STP)/cm3 atm, measured at 1 atm, 35 °C. Ideal selectivity α = P1/P2.
H
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
■
●
▲
Figure 8. Robeson plots relevant to PIa (red ), PIb (red ), PIc (blue ), PId (blue ★) for (a) H2/CH4, (b) H2/N2, (c) H2/CO2, (d) CO2/CH4, (e)
CO2/N2, and (f) O2/N2 gas pairs (solid lines represent the 2008 upper bound). Data points are from reported 6FDA-pPD12 (1), 6FDA-mPD12 (2),
6FDA-pp′ODA13 (3), 6FDA-mp′ODA13 (4), and Matrimid51 (5) along with TR-αPBO25 (6), TR-PBO-co-PI (5:5)1 (7), and AD-cPBO−PI (5:5)30 (8)
membranes for comparison.
elongation at break in comparison with the corresponding meta-
connecting ones.49 Additionally, significantly greater tensile and
much higher elongation at break are observed for these 6FDA-
based poly(benzoxazole-co-imide) membranes relative to the
conventional TR membranes such as TR-1−45020 (tensile
strength: 98 MPa, elongation at break: 3.9%) and even TR
poly(ether-imide)28 (sample 450−1, tensile strength: 54 MPa,
elongation at break: 4.0%). Moreover, the incorporation of
benzoxazole rings also significantly improved the tensile
properties as compared with the conventional polyimides, as
shown by comparing PIc with the reference 6FDA-pPD
membrane reported previously.49 The pPD is believed to assume
a conjugated planar conformation and thus the corresponding
6FDA-pPD polymer necessarily possesses high chain rigidity.50
Both the tensile strength (120 MPa) and elongation at break
(26.3%) of the PIc are higher than the corresponding values of
6FDA-pPD (tensile strength of 108 MPa and elongation at break
of 6%49). Especially, the elongation of the PIc was about 4.4
times higher than that of 6FDA-pPD. This indicated that
incorporation of benzoxazole rings in polyimide significantly
improved the tensile properties of the resulting copolymer
membrane. Furthermore, all para-connecting poly(benzoxazole-
co-imide) membranes exhibited much higher tensile strength,
modulus, and also slightly higher elongation at break in
comparison with Matrimid,51 which has a tensile strength of
72.3 MPa, a modulus of 1.4 GPa and an elongation at break of
19.4%. The excellent tensile strength and thermal properties of
the poly(benzoxazole-co-imide) class of membranes is of interest
for commercial gas separation applications.
Single Gas Permeability. Six gases (He, H2, CO2, O2, N2
and CH4) were used to determine the gas permeabilities, which
are presented in Table 4. The sequence in permeability
coefficients from highest to the lowest for all the poly-
(benzoxazole-co-imide) membranes was He > H2 > CO2 > O2
> N2 > CH4, following the gas kinetic diameter order, i.e., He
(2.60 Å) < H2 (2.89 Å) < CO2 (3.30 Å) < O2 (3.46 Å) < N2 (3.64
Å) < CH4 (3.80 Å). The permeability coefficients of the
membranes strongly depended on the chemical structures of the
corresponding polymers. The para-connecting PIa exhibited
higher permeability coefficients than meta-connecting PIb for all
I
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
gases (e.g., CO2 permeability was about 2.6 times higher). For
larger gases (N2, O2, CH4, and CO2) the para-connecting isomer
membrane revealed much higher diffusivity coefficients than the
corresponding meta-connecting ones. These results correlate
well with the FFV values of the para- and meta-linked PIa
isomers. A previous investigation of cavity size distributions and
diffusion in para- and meta-linked isomers by molecular
simulation studies confirmed that the average cavity size in
para-linked isomers is larger than the corresponding ones in
meta-linked isomers.36 The solubility coefficients of the para-
isomers had slightly higher values for the larger N2, O2, CH4 and
CO2 gases than the corresponding meta-isomers, which is
consistent with an interpretation previously reported by Neyertz
et al.52 In contrast, the PIc para-isomer had lower permeability
coefficients for small size He and H2 gases, due to a lower FFV
than the meta-isomer. However, for large gases (N2, O2 CH4 and
CO2) the PIc para-isomer had higher permeabilities than the
meta-isomer, even though the former had relatively smaller FFVs.
This is due to higher solubility coefficients for these larger gases
(see Table 4), which depend on both the polymeric chain
structures and the interfaces formed between the polymer chains
and permeating gases.52
The incorporation of highly rigid benzoxazole rings into the
polyimide backbones can significantly improve the permeability,
as can be confirmed by comparing PIc and PId with reference
membranes 6FDA-pPD and 6FDA-mPD, respectively, reported
previously.12 Obviously, the permeabilities to CO2 of the PIc
membrane (25 Barrer) and the PId membrane (19 Barrer)
increased 63% and 107%, respectively, as compared to the
corresponding reference membranes 6FDA-pPD (15.3 Barrer)
and 6FDA-mPD (9.2 Barrer). Also, in comparison with
membrane made from commercial Matrimid, the poly-
(benzoxazole-co-imide) membranes exhibited significantly high-
er permeability, e.g., the CO2 permeability of PIa was about 8.7
times higher than that of Matrimid (5.39 Barrer51). However,
these nonthermally rearranged poly(benzoxazole-co-imide)
membranes are much less permeable than the TR-PBO-co-PI
membranes. Note that thermal rearrangement is conducted in
the solid state and therefore creates change in chain angles, which
would force open free volume spaces.20 However, if the
benzoxazole group is already formed, and then cast as a polymer
solution, the same process would not occur. In this case,
interchain packing would be closer, because it had more degrees
of freedom to move and thus had more packing density.
The influence of structural isomerism of poly(benzoxazole-co-
imide)s on gas transport properties is readily visualized in the
Robeson permeability/selectivity trade-off plots, as shown in
Figure 8. The permeability/selectivity trade-off performance of
the present membranes is clearly well below that of thermally
rearranged (TR) membranes such as TR-αPBO, especially for
the CO2/CH4, CO2/N2 and O2/N2 gas pairs.25 The poly-
(benzoxazole-co-imide) meta-isomers had better overall trade-off
performance relative to the para-isomers for separating smaller
gases (e.g., H2) from larger gases (e.g., N2 and CO2), as shown in
Figure 8, parts b and c. In contrast, the para- isomers showed
better overall performance for larger gas pairs such as CO2/CH4
and CO2/N2 (Figure 8, parts d and e). It is noteworthy that
among these poly(benzoxazole-co-imide) membranes, the PId
meta-isomer membrane had the best performance for the H2/
CH4, H2/N2 and H2/CO2 gas pairs. In particular, the separation
performance for H2/CH4 is very close to the Robeson 2008
upper bound, exceeding conventional/commercial polyimide
membranes, e.g., 6FDA-pPD,12 6FDA-mPD,12 6FDA-
pp′ODA,13 6FDA-mp′ODA,13 Matrimid,51 and even TR-PBO-
co-PI (5:5),1 AD-cPBO−PI (5:5),30 and TR-αPBO25 for the H2/
CH4 and H2/CO2 gas pairs. However, the para-connecting PIa
had the best performance for the CO2/N2 and CO2/CH4 gas
pairs among the poly(benzoxazole-co-imide) membranes,
exceeding the above conventional/commercial polyimide
membranes, but below the TR-PBO-co-PI (5:5),1 AD-cPBO−
PI (5:5),30 and TR-αPBO25 membranes.
The present nonthermally rearranged poly(benzoxazole-co-
imide) membranes also exhibited improved mixed gas transport
and plasticization resistance properties as compared to conven-
tional polyimides. As shown in the Supporting Information
(Table S1), the nonthermally rearranged PIa exhibited retained
CO2/CH4 selectivity (42.2) as compared to the commercial
Matrimid membrane (structure shown in Scheme S1 in the
Supporting Information) with a CO2/CH4 selectivity of 36.4, for
the 50/50 CO2/CH4 mixed gas at 600 psig and 50 °C. Also, the
un-cross-linked PDMC membrane53 (6FDA-DAM3:DABA2), a
representative 6FDA-based polyimide membrane, exhibited
increased CO2 permeability as the increase of the CO2 feed
pressure (Figure S1 in the Supporting Information), indicative of
CO2−induced plasticization phenomenon. However, similar to
that of the cross-linked PDMC, the CO2 permeability of the PIa
membrane decreased from 5 to 25 bar and then remained almost
constant in the range of 30−40 bar as increasing CO2 feed
pressure (Figure S1), implying it has good plasticization
resistance. The enhanced CO2 plasticization resistance of the
PIa membrane is mainly attributed to an introduction of the
highly rigid benzoxazole units into the polymer chain backbone.
Further study will be conducted to demonstrate the effect of the
introduction of rigid moieties on gas transport properties
including mixed gas permeation as well as plasticization
resistance.
CONCLUSIONS
■
Three novel benzoxazole-containing diamine monomers, p-
BFO, m-BFO, and m-BOA, were synthesized via a high-yield
one-step route from commercially available starting materials.
The resulting poly(benzoxazole-co-imide)s were prepared by
using 6FDA with three synthetic diamine monomers and also a
commercial diamine, p-BOA. Because of introduction of
benzoxazole moieties into the polyimides, the resulting poly-
(benzoxazole-co-imide)s exhibited an excellent combination of
properties, such as good solubility, excellent thermal and
mechanical properties, and moderate permeability. Structural
isomerism in the linkages had a significant impact on molecular
packing and properties of the resulting poly(benzoxazole-co-
imide) membranes. The para-connecting poly(benzoxazole-co-
imide) isomers exhibited higher molecular weights (Mws), higher
modulus and significantly larger elongation at break, higher Tgs,
higher ordering of macromolecules, along with better overall
performance for CO2/CH4 and CO2/N2 separations as
compared to the corresponding meta-connecting ones. However,
the changes in FFV values from para-connecting polymer
isomers to meta-connecting ones greatly depended on a
combined contribution from both the chain packing and
interchain distances in ordered domains. Systematic inves-
tigations and analyses of the correlation between molecular
packing and properties of these poly(benzoxazole-co-imide)s in
this study provide useful insight for adjusting specific perform-
ance parameters to meet the requirements of membrane-based
gas separations.
J
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules
Article
(22) Soo, C. Y.; Jo, H. J.; Lee, Y. M.; Quay, J. R.; Murphy, M. K. J.
Membr. Sci. 2013, 444, 365−377.
ASSOCIATED CONTENT
■
S
* Supporting Information
(23) Calle, M.; Chan, Y.; Jo, H. J.; Lee, Y. M. Polymer 2012, 53, 2783−
2791.
Mixed gas permeation test, a scheme showing the chemical
structure of commercial Matrimid 5218, a table of pure gas and
mixed gas transport properties for PIa and Matrimid, and CO2
plasticization resistance of membranes, including a figure
showing pure CO2 permeation isotherms. This material is
available free of charge via the Internet at http://pubs.acs.org.
(24) Han, S. H.; Kwon, H. J.; Kim, K. Y.; Seong, J. G.; Park, C. H.; Kim,
S.; Doherty, C. M.; Thornton, A. W.; Hill, A. J.; Lozano, A. E.;
Berchtoldf, K. A.; Lee, Y. M. Phys. Chem. Chem. Phys. 2012, 14, 4365−
4373.
(25) Han, S. H.; Misdan, N.; Kim, S.; Doherty, C. M.; Hill, A. J.; Lee, Y.
M. Macromolecules 2010, 43, 7657−7667.
(26) Choi, J. I.; Jung, C. H.; Han, S. H.; Park, H. B.; Lee, Y. M. J. Membr.
Sci. 2010, 349, 358−368.
AUTHOR INFORMATION
■
Corresponding Authors
(27) Park, H. B.; Han, S. H.; Jung, C. H.; Lee, Y. M.; Hill, A. J. J. Membr.
Sci. 2010, 359, 11−24.
*(M.D.G.) E-mail: michael.guiver@outlook.com.
*(Y.M.L.) Telephone: +82-2-2220-0525. Fax: +82-2-2291-5982.
E-mail: ymlee@hanyang.ac.kr.
(28) Calle, M.; Lee, Y. M. Macromolecules 2011, 44, 1156−1165.
(29) Calle, M.; Doherty, C. M.; Hill, A. J.; Lee, Y. M. Macromolecules
2013, 46, 8179−8189.
Notes
(30) Kim, S.; Woo, K. T.; Lee, J. M.; Quay, J. R.; Murphy, M. K.; Lee, Y.
M. J. Membr. Sci. 2014, 453, 556−565.
The authors declare no competing financial interest.
∥M.D.G. is a visiting professor at Hanyang University.
(31) Li, S.; Jo, H. J.; Han, S. H.; Park, C. H.; Kim, S.; Budd, P. M.; Lee,
Y. M. J. Membr. Sci. 2013, 434, 137−147.
(32) Preston, J.; Carson, J. W. Polymer 1993, 34, 830−834.
(33) Mercer, F. W.; McKenzie, M. T. High Perform. Polym. 1993, 5,
97−106.
ACKNOWLEDGMENTS
■
This research was supported by the Korea Carbon Capture &
Sequestration R&D Center (KCRC) through the National
Research Foundation of Korea (NRF) funded by the Ministry of
Science, ICT & Future Planning (NRF-2014M1A8A1049305).
(34) Mercer, F. W.; McKenzie, M. T.; Bruma, M.; Schulz, B. High
Perform. Polym. 1996, 8, 395−406.
(35) Zhuang, Y.-B.; Liu, X.-Y.; Gu, Y. Polym. Chem. 2012, 3, 1517−
1525.
(36) Wang, X.-Y.; in ’t Veld, P. J.; Lu, Y.; Freeman, B. D.; Sanchez, I. C.
Polymer 2005, 46, 9155−9161.
REFERENCES
■
(1) Jung, C. H.; Lee, J. E.; Han, S. H.; Park, H. B.; Lee, Y. M. J. Membr.
Sci. 2010, 350, 301−309.
(37) Chen, C.; Yokota, R.; Hasegawa, M.; Kochi, M.; Horie, K.;
Hergenrother, P. High Perform. Polym. 2005, 17, 317−333.
(38) Kawakami, H.; Nakajima, K.; Nagaoka, S. J. Membr. Sci. 2003, 211,
291−298.
(2) Kim, T. H.; Koros, W. J.; Husk, G. R.; O’Brien, K. C. J. Membr. Sci.
1988, 37, 45−62.
(3) Coleman, M. R.; Koros, W. J. J. Membr. Sci. 1990, 50, 285−297.
(4) White, L. S.; Blinka, T. A.; Kloczewski, H. A.; Wang, I. F. J. Membr.
Sci. 1995, 103, 73−82.
(39) Chang, K.-S.; Wu, Z.-C.; Kim, S.; Tung, K.-L.; Lee, Y. M.; Lin, Y.-
F.; Lai, J.-Y. J. Membr. Sci. 2014, 454, 1−11.
́ ́
(40) Comesana-Gandara, B.; Calle, M.; Jo, H. J.; Hernandez, A.; de la
̃
(5) Calle, M.; Lozano, A. E.; de Abajo, J.; de la Campa, J. G.; Alvarez, C.
J. Membr. Sci. 2010, 365, 145−153.
Campa, J. G.; de Abajo, J.; Lozano, A. E.; Lee, Y. M. J. Membr. Sci. 2014,
450, 369−379.
(6) Qiu, W.; Chen, C.-C.; Xu, L.; Cui, L.; Paul, D. R.; Koros, W. J.
Macromolecules 2011, 44, 6046−6056.
(41) Wang, X.; Chen, F.; Gu, Y. J. Polym. Sci., Part A: Polym. Chem.
2011, 49, 1443−1452.
(7) Calle, M.; Garcia, C.; Lozano, A. E.; de la Campa, J. G.; de Abajo, J.;
Alvarez, C. J. Membr. Sci. 2013, 434, 121−129.
(42) Zhuang, Y.-B.; Seong, J. G.; Do, Y. S.; Jo, H. J.; Cui, Z.-L.; Lee, J.;
Lee, Y. M.; Guiver, M. D. Macromolecules 2014, 47, 3254−3262.
(43) Hilborn, J. G.; Labadie, J. W.; Hedrick, J. L. Macromolecules 1990,
23, 2854−2861.
(8) Kratochvil, A. M.; Koros, W. J. Macromolecules 2010, 43, 4679−
4687.
(9) Aitken, C. L.; Koros, W. J.; Paul, D. R. Macromolecules 1992, 25,
3424−3434.
(44) Kolhe, N. B.; Asha, S. K.; Senanayak, S. P.; Narayan, K. S. J. Phys.
Chem. B 2010, 114, 16694−16704.
(10) Robeson, L. M. J. Membr. Sci. 1991, 62, 165−185.
(11) Robeson, L. M. J. Membr. Sci. 2008, 320, 390−400.
(12) Tanaka, K.; Okano, M.; Toshino, H.; Kita, H.; Okamoto, K.-I. J.
Polym. Sci., Part B: Polym. Phys. 1992, 30, 907−914.
(13) Tanaka, K.; Kita, H.; Okano, M.; Okamoto, K.-I. Polymer 1992,
33, 585−592.
(45) Wakita, J.; Jin, S.; Shin, T. J.; Ree, M.; Ando, S. Macromolecules
2010, 43, 1930−1941.
(46) Baseri, S.; Karimi, M.; Morshed, M. Polym. Bull. 2013, 70, 953−
969.
(47) Zhuang, Y.-B.; Gu, Y. J. Polym. Res. 2012, 19, 14.
(48) Godocikova, E.; Balaz, P.; Boldizarova, E. Hydrometallurgy 2002,
65, 83−93.
(49) Ree, M.; Kim, K.; Woo, S. H.; Chang, H. J. Appl. Phys. 1997, 81,
698−708.
(50) Chen, H. L.; Ho, S. H.; Wang, T. H.; Chen, K. M.; Pan, J. P.; Liang,
S. M.; Hung, A. J. Appl. Polym. Sci. 1994, 51, 1647−1652.
(51) Zhao, H.-Y.; Cao, Y.-M.; Ding, X.-L.; Zhou, M.-Q.; Liu, J.-H.;
Yuan, Q. J. Membr. Sci. 2008, 320, 179−184.
(52) Neyertz, S.; Brown, D. J. Membr. Sci. 2014, 460, 213−228.
(53) Hillock, A. M. W.; Koros, W. J. Macromolecules 2007, 40, 583−
587.
(14) Carta, M.; Malpass-Evans, R.; Croad, M.; Rogan, Y.; Jansen, J. C.;
Bernardo, P.; Bazzarelli, F.; McKeown, N. B. Science 2013, 339, 303−
307.
(15) Guiver, M. D.; Lee, Y. M. Science 2013, 339, 284−285.
(16) Hsiao, S. H.; Huang, Y. H. Eur. Polym. J. 2004, 40, 1127−1135.
(17) Lee, J. K.; Kim, H. J.; Kim, T. H.; Lee, C. H.; Park, W. H.; Kim, J.;
Lee, T. S. Macromolecules 2005, 38, 9427−9433.
(18) Okamoto, K.-I.; Tanaka, K.; Muraoka, M.; Kita, H.; Maruyama, Y.
J. Polym. Sci., Part B: Polym. Phys. 1992, 30, 1215−1221.
(19) Thornton, A. W.; Doherty, C. M.; Falcaro, P.; Buso, D.;
Arnenitsch, H.; Han, S. H.; Lee, Y. M.; Hill, A. J. J. Phys. Chem. C 2013,
117, 24654−24661.
(20) Park, H. B.; Jung, C. H.; Lee, Y. M.; Hill, A. J.; Pas, S. J.; Mudie, S.
T.; Van Wagner, E.; Freeman, B. D.; Cookson, D. J. Science 2007, 318,
254−258.
(21) Do, Y. S.; Seong, J. G.; Kim, S.; Lee, J. G.; Lee, Y. M. J. Membr. Sci.
2013, 446, 294−302.
K
dx.doi.org/10.1021/ma501891m | Macromolecules XXXX, XXX, XXX−XXX