Proton conductive polyimide Lonomer membranes: Effect of NH, OH, and COOH groups

Article abstract of DOI:10.1002/pola.24061

A new series of sulfonated polyimide (SPI) copolymers containing NH, OH, or COOH groups were synthesized by the polycondensation of 1,4,5,8- naphthalnetetracarboxylic dianhydride, 3,3′-bis(sulfopropoxy)-4,4′- diaminobiphenyl, and 3-(4-aminophenyl)-5-(3-aminophenyl)-1H-1,2,4-triazole (SPI-8m), 3,5-bis(4-aminophenyl)-1H-1,2,4-triazole (SPI-8-p), 3,6-diaminocarbazole (SPI-9), 3,5-diamino-1H-1,2,4-triazole (SPI-10), bis(3-aminopropyl)-amine (SPI-11), 2,6-diaminopurine (SPI-12), 2,4-diamino-6-hydroxyprymidine (SPI-13), or 3,5-bis(4-aminophenoxy)benzoic acid (SPI-14). The obtained SPIs were soluble in polar organic solvents and gave tough and flexible mem-branes by solution casting. The SPI membranes having NH and COOH groups showed high thermal (decomposition temperature ≈200 °C) and mechanical (maximum stress >22 MPa) stability. Introducing NH groups, especially triazole and carbazole groups, was effective in improving proton conductive properties of SPI membranes at low humidity.

Full text of DOI:10.1002/pola.24061

Proton Conductive Polyimide Ionomer Membranes: Effect of NH, OH,
and COOH Groups
JUMPEI SAITO,1 MANABU TANAKA,2 KENJI MIYATAKE,2,3 MASAHIRO WATANABE2
1Interdisciplinary Graduate School of Medicine and Engineering, University of Yamanashi, 4 Takeda, Kofu 400-8510, Japan
2Fuel Cell Nanomaterials Center, University of Yamanashi, 6 Miyamae, Kofu 400-0021, Japan
3Division of Fuel Cell Research, Clean Energy Research Center, University of Yamanashi, 4 Takeda, Kofu 400-8510, Japan
Received 1 March 2010; accepted 4 April 2010
DOI: 10.1002/pola.24061
Published online in Wiley InterScience (www.interscience.wiley.com).
ABSTRACT: A new series of sulfonated polyimide (SPI) copoly-
mers containing NH, OH, or COOH groups were synthesized
by the polycondensation of 1,4,5,8-naphthalnetetracarboxylic
dianhydride, 3,30-bis(sulfopropoxy)-4,40-diaminobiphenyl, and
3-(4-aminophenyl)-5-(3-aminophenyl)-1H-1,2,4-triazole (SPI-8-
m), 3,5-bis(4-aminophenyl)-1H-1,2,4-triazole (SPI-8-p), 3,6-di-
aminocarbazole (SPI-9), 3,5-diamino-1H-1,2,4-triazole (SPI-10),
bis(3-aminopropyl)-amine (SPI-11), 2,6-diaminopurine (SPI-12),
2,4-diamino-6-hydroxyprymidine (SPI-13), or 3,5-bis(4-amino-
phenoxy)benzoic acid (SPI-14). The obtained SPIs were soluble
in polar organic solvents and gave tough and flexible mem-
branes by solution casting. The SPI membranes having NH
and COOH groups showed high thermal (decomposition tem-
perature ꢀ200 ꢁC) and mechanical (maximum stress >22 MPa)
stability. Introducing NH groups, especially triazole and carba-
zole groups, was effective in improving proton conductive
C
V
properties of SPI membranes at low humidity.
2010 Wiley
Periodicals, Inc. J Polym Sci Part A: Polym Chem 48: 2846–
2854, 2010
KEYWORDS: fuel cells; ionomers; polyelectrolytes; polyimides;
proton conduction
INTRODUCTION Polymer electrolyte membrane fuel cells
(PEMFCs) have attracted considerable attention as an alter-
native energy source because of its high-energy efficiency
and no emissions of pollutants. PEMFCs are expected for
the applications to electric vehicles and stationary power
sources.1 However, improvement of performance and dura-
bility, and reduction of cost are needed before the practical
applications.
mine comonomers. High thermal stability and low gas per-
meability are attracting properties, whereas low proton con-
ductivity under low humidity conditions and hydrolytic
instability are major drawbacks of the SPI membranes. Oka-
moto and coworkers revealed that SPIs bearing pendant sul-
fonic acid groups are somewhat more proton conductive and
more stable to hydrolysis.14 However, SPIs require further
improvement in the conductivity and stability.15 The relation-
ship between chemical structure and properties should be
investigated in more detail.
Proton-conducting polymer electrolyte membranes are one
of the key components in PEMFCs.2–4 Polymer electrolyte
membranes have to have high proton conductivity, high
chemical and physical stability, and low gas permeability.
Perfluorinated sulfonic acid (PFSA) ionomers such as Nafion
NRE212 (Du Pont) have been most used. Although PFSA
membranes are superior in terms of proton conductivity and
durability, environmental inadaptability and production cost
are issues. Therefore, there has been a great demand for
nonfluorinated alternative PEMs.5
We have previously reported synthesis and properties of sul-
fonated polyimides (SPI-8) containing 1,2,4-triazole groups
in the main chains.16 The introduction of triazole groups
contributed to the improvement of thermal stability and me-
chanical properties. In addition, the SPI ionomers containing
triazole groups showed higher proton conductivity than that
of nontriazole-containing SPIs due to the increased water
uptake. The membrane was durable for 5000 h in single fuel
cell operation at 80 ꢁC.17,18 The objective of this research is
to investigate the effect of NH, OH, and COOH groups on the
properties of SPI membranes. A series of new SPI mem-
branes containing triazole, carbazole, aliphatic amine,
hydroxyl, and carboxylic acid groups have been synthesized.
Their proton conductivity and gas permeability as well as
Sulfonated polyimides (SPIs) are one of the promising candi-
dates as well as sulfonated polyethers.6–13 They are readily
available via typical polycondensation reaction of presulfo-
nated diamine monomers and aromatic tetracarboxylic dia-
nhydrides. Ion exchange capacity (IEC) of the SPIs can be
easily controlled by copolymerizing with nonsulfonated dia-
Additional Supporting Information may be found in the online version of this article. Correspondence to: K. Miyatake (E-mail: miyatake@yamanashi.
ac.jp) or M. Watanabe (E-mail: m-watanabe@yamanashi.ac.jp)
C
V
Journal of Polymer Science: Part A: Polymer Chemistry, Vol. 48, 2846–2854 (2010) 2010 Wiley Periodicals, Inc.
2846
INTERSCIENCE.WILEY.COM/JOURNAL/JPOLA

ARTICLE
SCHEME 1 Synthesis of SPIs 8–14.
thermal, mechanical, and chemical stability have been mea-
sured and compared to one another.
ble 1). Other SPI copolymers (SPI-11, -12, -13, and -14) were
relatively low molecular weight. The molecular weight of
SPI-14 increased significantly from 9.6 kDa to 56.0 kDa in
Mn or from 24.9 kDa to 174.8 kDa in Mw by drying under
RESULTS AND DISCUSSION
ꢁ
vacuum at 150 C (SPI-14H), indicating possible crosslinking
Synthesis and Characterization of SPI Copolymers
reaction via dehydration among carboxylic acid and/or sul-
fonic acid groups. 1H NMR and IR spectra were compared
between pristine SPI-14 and thermally treated SPI-14H. The
intensity of symmetric C¼¼O stretching vibration ascribed to
carboxylic acid groups (1715 cmꢃ1) in the IR spectrum
decreased slightly after the thermal treatment, whereas no
Synthetic procedure of the SPIs investigated in this study is
shown in Scheme 1. p-APTAZ and m-APTAZ were synthesized
by the condensation reaction of 4-aminobenzohydrazide with
4- or 3-aminobenzonitrile, respectively, under basic condi-
tions. The purity of each APTAZ was confirmed by 1H NMR
and LC/MS analyses to be high enough for the polymeriza-
tion reactions. Other nonsulfonated diamine monomers for
SPI-9 to SPI-14 were commercially available. The polycon-
densation reaction of TCND, BSPA, and the nonsulfonated
diamine comonomers was carried out in m-cresol in the
presence of excess triethylamine (TEA). Molar percentage of
BSPA to the nonsulfonated comonomers (X) was set as X ¼
50 (SPI-8-p, 8-m, and -9), 46 (SPI-11 and -12), 45 (SPI-13),
and 56 (SPI-14) in order that the obtained copolymers have
similar IEC values (ꢂ1.7 mequiv gꢃ1). As SPI-10 was only
slightly soluble in m-cresol, DMSO, and NMP, when X ¼ 43
(IEC ¼ 1.77 mequiv gꢃ1), higher composition of BSPA (X ¼
50) and therefore higher IEC (1.95 mequiv gꢃ1) were
adopted for SPI-10 with better solubility. All SPI copolymers,
except for SPI-11, were obtained as fibers. GPC analyses
revealed that SPI-8-m, -8-p, -9, and -10 copolymers were
high molecular weight (Mw > 88.1 kDa, Mn > 44.9 kDa; Ta-
1
changes were observed in the H NMR spectra between SPI-
14 and SPI-14H (see Supporting Information Figs. S1 and
S2).
The SPI copolymers were characterized by 1H NMR spectra.
1
The H NMR spectra are summarized in Figure 1, except for
SPI-11, of which NMR analysis was not available due to its
insufficient solubility in DMSO-d6 and the other deuterated
solvents. For the other SPIs, peaks were well-assigned to the
supposed chemical structures. The peaks at 7.5–8.8 ppm
were aromatic protons of the polyimides and the peaks at
1.8–4.3 ppm were aliphatic protons of the pendant groups.
The two strong and sharp peaks designated with asterisk
were due to ethanol used for washing the membranes. There
were no evidences of side reactions (such as the reaction
with NH, OH, and COOH groups) and/or residual amic acid
groups. Copolymer composition and IEC value were
PROTON CONDUCTIVE POLYIMIDE IONOMER MEMBRANES, SAITO ET AL.
2847

JOURNAL OF POLYMER SCIENCE: PART A: POLYMER CHEMISTRY DOI 10.1002/POLA
TABLE 1 Molar Percentage, Molecular Weight, IEC, and Density of SPI Membranes
Mn
Mw
IECb
(mequiv gꢃ1
IECc
(mequiv gꢃ1
IECd
(mequiv gꢃ1
Density
(g cmꢃ3
Polymer
Xa
(kDa)
(kDa)
Mw/Mn
)
)
)
)
Membrane
SPI-8-p
SPI-8-m
SPI-9
50
50
50
50
46
46
45
56
56
56.6
120.2
56.1
44.9
13.1
4.8
191.8
323.8
109.3
88.1
3.4
2.7
1.9
2.0
1.8
2.2
2.7
2.6
3.1
1.70
1.70
1.78
1.95
1.76
1.76
1.76
1.76
1.76
1.68
1.64
1.58
1.83
1.76
1.81
1.76
2.12
1.49
1.75
1.60
1.47
Turbid, flexible
Turbid, flexible
Transparent, flexible
Turbid, flexible
SPI-10
SPI-11
SPI-12
SPI-13
SPI-14
SPI-14He
a
f
f
f
23.6
–
–
–
Turbid, brittle
f
f
10.7
–
1.72
1.77
1.64
1.57
–
Transparent, brittle
Transparent, brittle
Transparent, rigid
Transparent, rigid
f
f
5.6
15.0
–
–
f
9.6
24.9
1.51
1.45
d
–
56.0
174.8
1.42
Molar percentage of the BSPA containing component.
Calculated from the copolymer composition.
Determined by titration.
Obtained from the 1H NMR spectrum.
After thermal treatment at 150 ꢁC.
Not measured.
b
c
e
f
1
calculated from the integral ratio in the H NMR spectra. For
example, the IEC value of SPI-8(50)-m was determined from
the integral ratio of the peaks, f: (j þ k þ l) in Figure 1. The
obtained IEC values were in fair agreement with the ones
expected from the feed monomer ratios. The experimentally
obtained IEC of SPI-14H was slightly lower than that of SPI-
14, supporting the above claim that a small amount of cross-
linking was contained in the SPI-14H via thermal dehydra-
tion reactions of acid groups. Casting DMSO solutions of SPI-
8-m, -8-p, -9, -10, and -14H gave membranes with the thick-
ness of ꢂ50 lm. The obtained membranes were of brown
color typical for polyimides with toughness and flexibility.
SPI-11, -12, and -13 gave rather brittle membranes due to
their insufficient molecular weight, which made them
unavailable for the properties analyses. Similar to SPI-8-p,16
SPI-8-m and -10 containing 1H-1,2,4-triazole groups gave
somewhat turbid membranes, whereas the SPI-9 and -14H
membranes were transparent. The linear and rod-like struc-
ture of triazole groups in the polymer backbone would be
responsible for possibly higher crystallinity of SPI-8-p, SPI-8-
m, and -10 membranes. The IEC values obtained by titration
were slightly lower than the ones obtained from the feed
monomer ratios and 1H NMR spectra. The results may sug-
gest that a small portion of sulfonic acid groups interact
with NH and COOH groups.
share the same hydrophilic component (BSPA). In the DTA
curves of SPIs, no glass transition temperatures were
observed below the decomposition temperatures.
Oxidative and hydrolytic stability of the SPI membranes
were tested under accelerated testing conditions. In both
experiments, changes in weight, IEC, and molecular weight
of the membranes were plotted as a function of IEC in Fig-
ures 3 and 4. For comparison, the stability data are also
given for NRE212 membrane; NRE212 membrane was intact
even under the harsh conditions. The oxidative stability of
SPI membranes were evaluated in hot Fenton’s reagent for 1
h. The SPI membranes showed relativity good stability to ox-
idation with only minor weight losses. Although SPI-14H
membrane showed large decrease in IEC, the other mem-
branes lost less than 12% of the pristine IEC values. All SPI
membranes investigated showed large loss in molecular
weight. Among them, SPI-10 was the most stable. The results
indicate that the oxidative degradation involves main chain
scission and 1H-1,2,4-triazole groups directly attached to im-
ide linkages at 3,5-positions contributed to improving the
oxidative degradation.
The hydrolytic stability of SPI membranes was evaluated in
pressurized water at 140 ꢁC for 24 h. The SPI-8-m, -8-p, -9,
and -14H membranes showed high hydrolytic stability with
small changes in weight and IEC value. Unlike the above oxi-
dative stability, SPI-10 membrane showed larger loss in mo-
lecular weight. This is probably because SPI-10 membrane
with the higher IEC (1.95 mequiv gꢃ1) and lower molecular
weight than other ionomers dissolved into water (100%
weight loss) and was hydrolyzed faster in aqueous solution.
In these accelerated stability testings, major degradation
mechanism was the main chain scission via hydrolysis of im-
ide bonds. NH- and COOH-containing groups in the polymer
backbone seemed to have minor effect on the stability of SPI
membranes under wet conditions.
Thermal, Oxidative, and Hydrolytic Stability
The thermal stability of SPI copolymers (SPI-8-m, -9, -10,
and -14H) was investigated by TG/DTA-MS analyses under
dry Ar atmosphere (Fig. 2). In the TGA curves, two-step
weight losses were observed from room temperature to 180
ꢁC and above 200 ꢁC. By combining TG with MS chromato-
grams, the first gradual weight loss was assigned to the
evaporation of hydrated water (H2O ¼ 18 m/z) and the sec-
ond more rapid weight loss was due to the decomposition of
SPIs. The major fragment ions observed in the second weight
loss were SO ¼ 48 and SO2 ¼ 64 m/z, indicating that the
thermal decomposition commenced by the loss of sulfonic
acid groups in the side chains. The TG curves of SPI-8-m, -9,
-10, and -14H were similar to that of SPI-8-p,16 because they
Mechanical Properties
Mechanical properties of the SPI membranes were evaluated
ꢁ
with tensile testing at 85 C and 93% RH and are shown in
2848
INTERSCIENCE.WILEY.COM/JOURNAL/JPOLA

ARTICLE
FIGURE 1 1H NMR spectra of SPI-8-m, -9, -10, -12, -13, and -14 in DMSO-d6. *Peaks derived from residual solvent (ethanol).
PROTON CONDUCTIVE POLYIMIDE IONOMER MEMBRANES, SAITO ET AL.
2849

JOURNAL OF POLYMER SCIENCE: PART A: POLYMER CHEMISTRY DOI 10.1002/POLA
FIGURE 4 Hydrolytic stability of SPI and NRE212 membranes.
and interaction with acid groups. The calculation results
revealed that the NH-containing units in SPI-8-p, -8-m, -9,
and -10 have compact structures and COOH-containing unit
of SPI-14 is distorted and bulky due to the flexible ether
bonds (see Supporting Information Fig. S3). The binding
energy of these components with methanesulfonic acid, sup-
posing that the atoms having the highest elctoro-negativity
interact with sulfonic acid groups, is in the order of SPI-14H
> SPI-10 > SPI-8-m > SPI-8-p > SPI-9 (see Supporting In-
formation Table S1). Except for SPI-14H, the order is in good
agreement with the results of Young’s moduli. As carboxylic
acid group can interact strongly with sulfonic acid group by
participating two points hydrogen bondings, the binding
energy of the model component of SPI-14H was higher than
that of the others. However, in the bulk polymer membrane,
this would not be the case due to the tortuosity of the struc-
ture of SPI-14.
On the other hand, density of the membranes was not in the
same order (Table 1); SPI-8-m (1.75 g cmꢃ3) > SPI-9 (1.60 g
cmꢃ3) > SPI-8-p (1.49 g cmꢃ3) > SPI-10 (1.47 g cmꢃ3) >
SPI-14H (1.42 g cmꢃ3). The results may indicate that, in
some cases such as SPI-9, intermolecular interactions (either
p–p and/or NH-acid interaction) are relatively weak even
when the polymer chains are densely packed.
FIGURE 2 TG/DTA and MS curves of SPI membranes under Ar
atmosphere.
Figure 5. SPI membranes exhibited better mechanical prop-
erties, higher maximum stress (>22 MPa) and lower elonga-
tion at break (<52%) than those of NRE212 membrane. The
SPI-8-p membrane showed the highest maximum stress. The
SPI-8-m and SPI-10 membranes showed the highest Young’s
modulus (1.5 GPa), whereas that of SPI-9 and -14H mem-
branes was rather low (0.95 and 0.61 GPa, respectively; Ta-
ble 2). These two membranes were transparent and prob-
ably less crystalline than SPI-8-m, -8-p, and -10 membranes.
Gas Permeability
Temperature dependence of dry hydrogen and oxygen per-
meability coefficient (Q) of SPI-8-p, -8-m, -14H, and NRE212
membranes is shown in Figure 6. Results of SPI-9 and -10
could not be obtained because the membranes were not
strong enough and broke during the measurements. The SPI-
8-p, -8-m, and -14H membranes showed lower gas perme-
ability for both hydrogen and oxygen than that of NRE212.
SPI membranes showed lower activation energy of the gas
permeability than that of NRE212 as is obvious from the
slope in the Arrhenius plots [Ea (H2) ¼ 39 and Ea (O2) ¼ 53
kJ molꢃ1 (NRE212), Ea (H2) ¼ 17 and Ea (O2) ¼ 14 kJ molꢃ1
(SPI-8-p), Ea (H2) ¼ 18 and Ea (O2) ¼ 19 kJ molꢃ1 (SPI-8-
m), Ea (H2) ¼ 18 and Ea (O2) ¼ 15 kJ molꢃ1 (SPI-14H),
respectively]. The lower gas permeability of SPI membranes
was more pronounced at higher temperature. Compared
with SPI-8-p and SPI-14H membranes, SPI-8-m membrane
showed slightly lower gas permeability. The results are rea-
sonably explained by the density data. SPI-8-m membrane
has higher density (1.75 g cmꢃ3) than that of SPI-8-p (1.49 g
cmꢃ3) and SPI-14H (1.42 g cmꢃ3) membranes. As gases
Then, we carried out the DFT calculation of the NH- and
COOH-containing units to estimate roughly their structure
FIGURE 3 Oxidative stability of SPI and NRE212 membranes.
2850
INTERSCIENCE.WILEY.COM/JOURNAL/JPOLA

ARTICLE
FIGURE 6 Temperature dependence of dry hydrogen and oxy-
gen permeability of SPI-8-p, -8-m, -14H, and NRE212 mem-
branes. [Color figure can be viewed in the online issue, which
is available at www.interscience.wiley.com.]
FIGURE 5 Stress versus strain curves of SPI and NRE212 mem-
branes at 85 ꢁC and 93% RH. [Color figure can be viewed in the
online issue, which is available at www.interscience.wiley.com.]
would mainly permeate through free volume in membranes,
it is considered that the SPI-8-m membrane with higher den-
sity showed the lowest gas permeability. SPI-14H membrane
showed lower gas permeability, especially oxygen, than that
of SPI-8-p. Possible explanation would be that heterocyclic
triazole moieties have affinity to oxygen and could solubilize
more oxygen molecules.
also showed high proton conductivity. In contrast, the SPI-8-
m and -9 membranes exhibited higher proton conductivity at
low humidity than that of SPI-8-p in spite of the formers’
lower water absorbability.
Water Uptake and Proton Conductivity
Humidity dependence of water uptake and proton conductiv-
ꢁ
ity were measured at 80 C for SPI membranes (Fig. 7). The
measurement was carried out by repeating sorption and de-
sorption processes and the data in Figure 7 were obtained
during the sorption process. The SPI-8-m, -9, and -10 mem-
branes showed slightly lower water uptake than that of SPI-
8-p membrane at all humidities, and the water uptake of
SPI-14H membrane was higher than that of SPI-8-p (except
at high humidity). The high water absorbability of SPI-14H,
despite of its low IEC and partly crosslinked structure, is
presumably due to weaker intermolecular interaction as sug-
gested by the mechanical properties, DFT calculation, and
density measurement.
The proton conductivity of SPI membranes was not in the
same order as their water uptake. The SPI-8-p membrane
showed the highest proton conductivity at high humidity
(>60% RH). SPI-14H membrane showing high water uptake
TABLE 2 Mechanical Properties of SPI Membranes at 85 8C
and 93% RH
Maximum
Elongation
Young’s
Membrane
Stress (MPa)
at Break (%)
Modulus (GPa)
SPI-8-p
SPI-8-m
SPI-9
103
50
37
22
29
28
7
1.2
1.5
FIGURE 7 Humidity dependence of the water uptake and pro-
ton conductivity of SPI membranes (IEC, mequiv gꢃ1 in paren-
theses) at 80 ꢁC. [Color figure can be viewed in the online
issue, which is available at www.interscience.wiley.com.]
19
3
0.95
1.5
SPI-10
SPI-14H
52
0.61
PROTON CONDUCTIVE POLYIMIDE IONOMER MEMBRANES, SAITO ET AL.
2851

JOURNAL OF POLYMER SCIENCE: PART A: POLYMER CHEMISTRY DOI 10.1002/POLA
amino-6-hydroxyprymidine (>98%, TCI), and 3,5-bis(4-ami-
nophenoxy)benzoic acid (>95%, TCI) were used as received.
3,30-Bis(sulfopropoxy)-4,40-diaminobiphenyl (BSPA) was syn-
thesized from 3,30-dihydroxy benzidine (98%, TCI) and 1,3-
propanesultone (Kanto Chemical) according to the litera-
ture.20 3,5-Bis(4-aminophenyl)-1H-1,2,4-triazole (p-APTAZ)
was synthesized according to the literature.16 These mono-
mers were characterized by 1H NMR spectroscopy and
stored in the dark. TEA (Aldrich) and benzoic acid (Kanto
Chemical) were used as received. m-Cresol (Kanto Chemical)
was dried over molecular sieves 3 Å before use. Other chem-
icals were of commercially available grade and used as
received.
3-(4-Aminophenyl)-5-(3-aminophenyl)-1H-
1,2,4-triazole (m-APTAZ)
m-APTAZ was synthesized from 4-aminobenzohydrazide
(98%, TCI) and 3-aminobenzonitrile (98%, TCI) according to
the literature.21 This monomer was characterized by liquid
chromatography/mass spectroscopy (LC/MS) and 1H NMR
spectroscopy, and stored in the dark.
FIGURE 8 Proton diffusion coefficient of SPI membranes as a
function of water volume fraction at 80 ꢁC. [Color figure can be
viewed in the online issue, which is available at www.
interscience.wiley.com.]
Polymerization
A typical procedure for the copolymerization is as follows.
BSPA (1.0 mmol), 3,6-diaminocarbazole (1.0 mmol), TEA
(10.8 mmol), and 10 mL of m-cresol were placed in a four-
neck round-bottomed flask equipped with a magnetic stir-
ring bar and N2 inlet. The mixture was stirred at 100 ꢁC
until dissolved. After clear solution was obtained, the mix-
ture was cooled to room temperature and stirred under a
stream of N2. TCND (2.0 mmol) and benzoic acid (4.0 mmol)
were then added into the mixtꢁure. The mixture was heated
For more detailed investigation, the proton conductivity was
plotted as a function of number of absorbed water molecules
per sulfonic acid (k), which takes into account both water
uptake and IEC, for SPIs and compared with that of our pre-
vious SPI (SPI-5)19 containing aliphatic segments in place of
NH- or COOH-containing groups (see Supporting Information
Fig. S4). Except for SPI-14H membrane, the new SPI mem-
branes showed higher proton conductivity at low k value (k
< 5) than that of SPI-5. The most pronounced effect was
observed in SPI-10, which contains 1H-1,2,4-triazole groups
directly attached to imide linkages at 3,5-positions.
ꢁ
at 175 C for 15 h and at 195 C for 3 h. After the reaction,
the mixture was poured dropwise into a large excess of ace-
tone. A yellow powder of the precipitate was filtered, washed
ꢁ
with acetone, and dried at 80 C under reduced pressure for
The proton diffusion coefficient (Dr) of SPI membranes was
calculated with the Nernst-Einstein equation regarding water
volume and concentration of proton charge carrier in poly-
mer electrolyte membranes. Dr value of SPI membranes was
plotted in Figure 8 as a function of water volume fraction
(U) at 80 ꢁC. The proton diffusivity of NRE212 was much
higher than those of the SPI membranes, and the difference
between NRE212 and SPIs became larger at lower U. The
results suggest that NRE212 membrane utilizes water more
efficiently for the proton conduction at low water content. At
lower U, Dr of the current SPI membranes was higher than
that of SPI-5. Among the SPIs, SPI-8-m, -9, and -10 mem-
branes showed relativity higher Dr values at lower U. The
NH groups, either within triazole or carbazole rings, were
most effective in improving the proton transport properties
in the SPI membranes.
12 h to obtain SPI-9(50) as triethylammonium form. The
other copolymers, SPI-8-m(50), -8-p(50), -10(50), -11(46), -
12(46), -13(45), and -14(56), were synthesized in a similar
manner.
Membrane Preparation
The copolymers (0.4 g) were dissolved in 10 mL of dimethyl
sulfoxide (DMSO) and cast onto a flat glass plate. After dry-
ing at 55 ꢁC, crude membrane in triethylammonium form
was immersed in ethanol containing 1 N HNO3 for 30 min at
ꢁ
50 C. The acidification procedure was repeated three times
followed by washing with pure ethanol. After drying at 80
ꢁC under reduced pressure for 12 h, the copolymer mem-
branes in acid form (ꢂ50 lm thick) were obtained.
Measurements
1H (400 MHz) NMR spectra were recorded on a Bruker
AVANCE 400S spectrometer with deuterated dimethyl sulfox-
ide (DMSO-d6) or deuterated water (D2O) as a solvent and tet-
ramethylsilane (TMS) as an internal reference. LC/MS experi-
ments were performed on a Shimadzu SPD-M10Avp (column:
Shodex VP-ODS) and a Shimadzu LCMS-2010A. Molecular
weight was measured with gel permeation chromatography
EXPERIMENTAL
Materials
1,4,5,8-Naphthalenetetracarboxylic
(>97%, Lancaster), 3,6-diaminocarbazole (>98%, TCI), 3,5-
diamino-1H-1,2,4-triazole (>97%, TCI), bis(3-aminopropyl)-
amine (>98%, TCI), 2,6-diaminopurine (>97%, TCI), 2,4-di-
dianhydride
(TCND)
2852
INTERSCIENCE.WILEY.COM/JOURNAL/JPOLA

ARTICLE
(GPC) equipped with two Shodex KF-805 columns and a Jasco
805 UV detector set at 300 nm with DMF containing 0.01 M
LiBr as eluent. Molecular weight was calibrated with standard
polystyrene samples. IEC of the SPI membranes was deter-
mined by 1H NMR spectra and titration. In the titration
method, a piece of membrane samples was equilibrated in
jected to the gas chromatography to quantify the test gas
permeated through the membrane. The gas permeability
coefficient, Q (cm3 (STD) cm cmꢃ2 sꢃ1 cmHgꢃ1), was calcu-
lated according to the following equation:
273 1
1 1
Q ¼
B l
ꢁ
T A t 76
large excess of 0.1 M NaCl aq solution at 40 C for one day.
The amount of HCl released from the membrane samples was
measured by titration with 0.01 N NaOH aq solution. Density
of dry SPI membranes was measured by pycnometer (Ultra-
pycnometer1000, Quantachrome Inc.) using helium as test gas
at room temperature. Density at 80 ꢁC was calibrated using
the volume expansion coefficient of perfluoroalkyl polymer
(PFA: 1.3 ꢄ 10ꢃ4 Kꢃ1) and polyimide (PI: 4.0 ꢄ 10ꢃ5 Kꢃ1), as
the coefficient of NRE212 and SPIs were not available.
where T (K) is the absolute temperature of the cell, A (cm2)
is the permeation area, B (cm3) is the volume of test gas
permeated through the membrane, t (s) is the sampling
time, and l (cm) is the thickness of the membrane.
Water Uptake and Proton Conductivity
Water uptake and proton conductivity of the SPI membranes
were measured with a Bel Japan solid electrolyte analyzer
system MSB-AD-V-FC equipped with a chamber, a magnetic
suspension balance, and a four-point probe conductivity cell.
For water uptake measurement, membraneꢁ samples (40–60
mg) were set in a chamber and dried at 80 C under vacuum
for 3 h until constant weight as dry materials was obtained.
The membrane was equilibrated with N2 gas at given tem-
perature and humidity for at least 1 h before the gravimetry
was done. For the proton conductivity measurement, mem-
brane samples (1.0 cm wide and 3.0 cm long) were set in
the chamber where the temperature and the humidity were
controlled by flowing humidified N2. The membrane was
equilibrated with N2 gas at the set temperature and humid-
ity for at least 1 h before the measurement. The samples
were contacted with two gold wire outer current-carrying
electrodes and two gold wire inner potential-detecting elec-
trodes. Impedance measurements were carried out with a
Solartron 1255B frequency response analyzer and a Solar-
tron SI 1287 potentiostat. The instrument was used in
potentiostatic mode with an ac amplitude of 300 mV with
the frequency range from 10 to 100,000 Hz.
Thermal Properties
Thermal analyses were performed via TG/DTA-MS with a Mac
Science TG/DTA 2000 equipped with a Bruker MS 9600 mass
spectrometer at a heating rate of 5 ꢁC minꢃ1 under argon.
Oxidative Stability
A small piece of membrane sample was soaked in Fenton’s
ꢁ
reagent (3% H2O2 containing 2 ppm FeSO4) at 80 C for 1 h.
The stability was evaluated by changes in IEC, weight, and
molecular weight of the test samples.
Hydrolytic Stability
A small piece of membrane sample was treated in water at
140 ꢁC for 24 h. The stability was evaluated by changes in
IEC, weight, and molecular weight of the test samples.
Mechanical Strength
Tensile testing was performed with a Shimadzu universal
testing instrument Autograph AGS-J500N equipped with a
chamber, in which temperature and humidity were con-
trolled by flowing humidified air with a Toshin Kogyo tem-
perature control unit Bethel-3A. Stress versus strain curves
were obtained at a stretching rate of 10 mm minꢃ1 for sam-
ples cut into a dumbbell shape [DIN-53,504-S3, 35 mm ꢄ 6
mm (total) and 12 mm ꢄ 2 mm (test area)].
From the proton conductivity and density data, the proton
diffusion coefficient (Dr) was calculated according to the fol-
lowing equations:
RT
F2 cðHþÞ
r
Gas Permeability
Dr ¼
Hydrogen and oxygen permeability through the ionomer
membrane was measured with a GTR-Tech 20XFYC gas per-
meation measurement apparatus equipped with a Yanaco
G2700T gas chromatography. The gas chromatography con-
tained a Porapak-Q column and a TCD detector. Argon and
helium were used as a carrier gas for the measurement of
hydrogen and oxygen, respectively. A membrane sample (u
¼ 30 mm) was set in a cell that has a gas inlet and outlet
on both sides of the membrane. Placing it in an oven con-
trolled the cell temperature. On one side of the membrane,
dry test gas (hydrogen or oxygen) was supplied at a flow
rate of 30 mL minꢃ1, whereas on the other side of the mem-
brane, the same gas as the carrier used in the gas chromato-
graph (flow gas) was supplied at a flow rate of 30 mL
minꢃ1. Before each measurement was done, the membrane
was equilibrated with the gases at the given temperature for
at least 3 h. Then, 3.7 mL of flow gas was sampled and sub-
where R is the gas constant, T is the absolute temperature
(K), F is Faraday constant, and c(Hþ) is the concentration of
the proton charge carrier (mol Lꢃ1).
Computational Calculation
The computational calculations were carried out for opti-
mized geometries, atomic charges, and ground-state energy
with Gaussian 03W program. All calculations were performed
at the Density Functional Theory (DFT) method with the
B3LYP (B3:Becke’s three-parameter, LYP: Lee-Yang-Parr) func-
tional and with a reasonably large basis set of 6-31G(d,p).
CONCLUSIONS
We have synthesized a new polyimide ionomers (SPI-8 – 14)
containing NH, OH, or COOH groups in the polymer
PROTON CONDUCTIVE POLYIMIDE IONOMER MEMBRANES, SAITO ET AL.
2853

JOURNAL OF POLYMER SCIENCE: PART A: POLYMER CHEMISTRY DOI 10.1002/POLA
6 Vallejo, E.; Pourcelly, G.; Gavach, C.; Mercier, R.; Pineri, M. J
backbone. Among them, SPI-8-m, -8-p, -9, -10, and -14H con-
taining triazole, carbazole, and carboxylic acid groups gave
tough and flexible membranes by solution casting. These SPI
membranes were thermally stable up to 200 ꢁC without
glass transition and decomposition. Oxidative and hydrolytic
stability was confirmed by accelerated testings. All SPI mem-
branes showed relativity good oxidative stability, whereas
the effect of the NH and COOH-containing groups seemed
limited since the major degradation mechanism was scission
of the imide linkages. SPI-8-p, -8-m, and -10 membranes con-
taining triazole groups exhibited high mechanical strength
due to their possibly high crystallinity. Hydrogen and oxygen
permeability of SPI-8-p, -8-m, and -14H membranes was
much lower than that of NRE212 at wide range of tempera-
ture. Triazole and carbazole groups had an impact on the
proton conducting properties of SPI membranes at low hu-
midity (or low water content). However, the conductivity
was still lower than that of NRE212 and therefore, further
molecular consideration such as block copolymer architec-
ture would be needed.
Membr Sci 1999, 160, 127–137.
7 Genies, C.; Mercier, R.; Sillion, B.; Cornet, N.; Gebel, G.;
Pineri, M. Polymer 2001, 42, 359–373.
8 Fang, J.; Guo, X.; Harada, S.; Watari, T.; Tanaka, K.; Kita, H.;
Okamoto, K.–I. Macromolecules 2002, 35, 9022–9028.
9 Guo, X.; Fang, J.; Watari, T.; Tanaka, K.; Kita, H.; Okamoto,
K.-I. Macromolecules 2002, 35, 6707–6713.
10 Miyatake, K.; Zhou, H.; Watanabe, M. Macromolecules 2004,
37, 4956–4960.
11 Matsumoto, K.; Nakagawa, T.; Higashihara, T.; Ueda, M. J
Polym Sci A: Polym Chem 2009, 47, 5827–5834.
12 Roy, A.; Hickner, M. A.; Einsla, B. R.; Harrison, W. L.;
McGrath, J. E. J Polym Sci A: Polym Chem 2009, 47, 384–391.
13 Miyatake, K.; Yasuda, T.; Watanabe, M. J Polym Sci A:
Polym Chem 2008, 46, 4469–4478.
14 Sutou, Y.; Yin, Y.; Hu, Z.; Chen, S.; Kita, H.; Okamoto, K.-I.;
Wang, H.; Kawasato, H. J Polym Sci A: Polym Chem 2009, 47,
1463–1477.
This work was partly supported by the New Energy and Indus-
trial Technology Development Organization (NEDO) through
HiPer-FC Project, and the Ministry of Education, Culture,
Sports, Science and Technology (MEXT), Japan, through a
Grant-in-Aid for Scientific Research. The authors thank K. Yagi
of University of Yamanashi for his help with computational
calculation.
15 Borup, R.; Meyers, J.; Pivovar, B.; Kim, Y. S.; Mukundan, R.;
Garland, N.; Myers, D.; Wilson, M.; Garzon, F.; Wood, D.; Zele-
nay, P.; More, K.; Stroh, K.; Zawodzinski, T.; Boncella, J.;
McGrath, J. E.; Inaba, M.; Miyatake, K.; Hori, M.; Ota, K.; Ogumi,
Z.; Miyata, S.; Nishikata, A.; Siroma, Z.; Uchimoto, Y.; Yasuda,
K.; Kimijima, K.-I.; Iwashita, N. Chem Rev 2007, 107, 3904–3951.
16 Saito, J.; Miyatake, K.; Watanabe, M. Macromolecules 2008,
41, 2415–2420.
17 Kabasawa, A.; Saito, J.; Yano, H.; Miyatake, K.; Uchida, H.;
REFERENCES AND NOTES
Watanabe, M. Electrochim Acta 2009, 54, 1076–1082.
1 Kordesch, K.; Simader, G. Fuel Cells and Thier Applications;
18 Kabasawa, A.; Saito, J.; Miyatake, K.; Uchida, H.; Watanabe,
VCH: Weinheim, Germany, 1996.
M. Electrochim Acta 2009, 54, 2754–2760.
2 Appleby, A. J.; Foulkes, F. R. Fuel Cell Handbook; Van Nos-
19 Asano, N.; Aoki, M.; Suzuki, S.; Miyatake, K.; Uchida, H.;
trand Reinhold: New York, 1989.
Watanabe, M. J Am Chem Soc 2006, 128, 1762–1769.
3 Carrette, L.; Friedrich, K. A.; Stimming, U. Fuel Cells 2001, 1,
20 Zhou, H.; Miyatake, K.; Watanabe, M. Fuel Cells 2005, 5,
5–39.
296–301.
4 Steele, B. C. H.; Heinzel, A. Nature 2001, 414, 345–352.
21 Yeung, K.-S.; Farkas, M. E.; Kadow, J. F.; Meanwell, N. A.
5 Miyatake, K.; Watanabe, M. Electrochemistry 2005, 73, 12–19.
Tetrahedron Lett 2005, 46, 3429–3432.
2854
INTERSCIENCE.WILEY.COM/JOURNAL/JPOLA