A new supramolecular sulfonated polyimide for use in proton exchange membranes for fuel cells

Article abstract of DOI:10.1039/c0cc02325f

Uracil-terminated telechelic sulfonated polyimides (SPI-U) were transformed into noncovalent network membranes through biocomplementary hydrogen bonding recognition in the presence of an adenine-based crosslinking agent. SPI-U membrane exhibited dramatica

Full text of DOI:10.1039/c0cc02325f

View Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/chemcomm | ChemComm
A new supramolecular sulfonated polyimide for use in proton exchange
membranes for fuel cellsw
Yun-Sheng Ye, Yao-Jheng Huang, Chih-Chia Cheng and Feng-Chih Chang*
Received 2nd July 2010, Accepted 24th August 2010
DOI: 10.1039/c0cc02325f
Uracil-terminated telechelic sulfonated polyimides (SPI-U) were
transformed into noncovalent network membranes through
biocomplementary hydrogen bonding recognition in the presence
of an adenine-based crosslinking agent. SPI-U membrane
exhibited dramatically improved methanol permeability, oxidative
stability, proton conductivity, and selectivity relative to those of
the standard SPI.
external crosslinking agent.9 In self-associative systems, inter-
chain crosslinking (e.g., dimerization) provides great control
over the network structure, but the degree and strength of the
crosslinks are difficult to control. In the second category, the
degree and strength of the crosslinks can be tuned by varying
the amount or type of crosslinking agent. Recently, we
demonstrated that it is possible to form DNA-like side-chain
polymers by choosing biocomplementary hydrogen bonding
recognition units (thymine–adenine; TÁÁÁA),10 thereby providing
a potential route to the design of new molecular architectures
in supramolecular polymers.
Nonfluorinated polymeric materials are attracting increasing
attention as alternatives to perfluorinated polymer membranes,
such as Nafion or Aciplex, for use as proton exchange
membranes (PEMs) in fuel cells, because of their advantages
in terms of cost, monomer toxicity, ease of synthesis, and
structural diversity.1 Although nonfluorinated PEMs can
achieve high proton conductivities when they feature a high
content of sulfonic acid groups, such systems tend to
exhibit poor mechanical strength and low permeability. The
development of more efficient membranes exhibiting improved
proton conductivity and reduced methanol crossover, without
detrimentally affecting the mechanical and chemical stabilities,
remains a challenge.2 One effective way to enhance PEM
performance is to distinctly separate the hydrophilic sulfonic
acid group regions from the hydrophobic polymer main chains
by locating the sulfonic acid groups on side chains through
grafting onto the polymer main chains.3 In addition, tethering
of N-heterocycles to a polymer backbone, followed by blend-
ing with a sulfonic acid polymer, is an attractive strategy to
achieve high PEM performance through acid–base interactions
between the sulfonic acid groups in one (acidic) polymer and
the N-atom-containing groups in the other (basic) polymer.4
Another strategy for improving the PEM performance is to
generate uniformly distributed hydrophilic conductive sites
and then control the hydrophobic nature by minimized cross-
linking over the sulfonated polymer structure.5
In this study, we prepared a uracil-terminated telechelic
sulfonated polyimide (SPI-U) and transformed it into
noncovalent network membranes through biocomplementary
hydrogen bonding recognition in the presence of an adenine-
based crosslinking agent (SMA-A; Scheme 1).
We synthesized the SMA-A crosslinking agent (polymer 7)
from monomer 1 in four steps (Scheme S1, ESIw). The
synthesis of the SPI-U (polymer 8) involved three steps
(Scheme S2, ESIw): an anhydride-terminated sulfonated
polyamic acid (PAA) was synthesized using a slight excess of
NTDA in the copolymerization reaction, the terminal anhydride
groups were end-capped with uracil (monomer 4) to produce a
uracil-terminated PAA, and then the SPI-U was obtained after
thermal imidization. The theoretical molecular weight of the
SPI-U was calculated by comparing the integrals of the signals
1
of the uracil and naphthene groups in its H NMR spectrum
(25 1C, DMSO-d6). The copolymer composition (n : m) was
confirmed by the relative H NMR spectroscopic absorption
1
peak areas of signals j and i in Fig. S7 (ESIw). The properties
of the SPI-U and the standard SPI containing 80% hydro-
philic units (n : m = 8 : 2) is the focus of this Communication.
Before conducting crosslinking studies, we first investigated
(Fig. 1) the biocomplementary (UÁ Á ÁA) hydrogen bonding
recognition between the SPI-U and the adenine compound
Noncovalent crosslinking offers several novel strategies for
materials design, because noncovalent crosslinking is inherently
reversible, highly tunable, avoids potential side reactions
(e.g., chain degradation), and provides unlimited processability.6
Hydrogen bonding is the most common noncovalent interaction
used to produce reversible crosslinked polymeric networks.7–9
Generally, polymeric networks based on hydrogen bonding
can be classified into two broad classes: those formed through
self-association8 and those formed through the addition of an
Institute of Applied Chemistry, National Chiao-Tung University,
Hsin-Chu, Taiwan. E-mail: changfc@mail.nctu.edu.tw;
Tel: +886 03 57131512
w Electronic supplementary information (ESI) available: Details of
experiment, NMR spectra and membrane properties. See DOI:
10.1039/c0cc02325f
Scheme 1 Graphical representation of a crosslinked SPI-U/SMA-A
complex.
c
7554 Chem. Commun., 2010, 46, 7554–7556
This journal is The Royal Society of Chemistry 2010

View Online
blends (Fig. S10, ESIw). These two sets of blends exhibited
opposite trends, confirming that hydrogen bonding was indeed
occurring in this biocomplementary system.
Thermogravimetric analysis (TGA) revealed two-step weight
losses for sulfonated polymer membranes (Fig. S11, ESIw).
We observed slight increases of the first decomposition
temperatures for the SPI/SMA-A and SPI-U/SMA-A
membranes relative to pure SPI, implying that the sulfonic
acid groups were more tightly held by the SPI/SMA-A and
SPI-U/SMA-A membranes. We tested the oxidative stability
of the prepared membranes using Fenton’s reagent (30 wt%
H2O2 and 30 ppm FeSO4) at 30 1C (Table S1, ESIw). The
oxidative stability of the blended membranes improved upon
adding the adenine-based polymer because the acid–base
interaction between the sulfonated polymer and SMA-A3,10
retarded the permeation of H2O2 through the membrane and
decreased the rate of decomposition of the backbone aromatic
units. Therefore, the formation of the supramolecular cross-
linked structure resulted in improved oxidative stability relative
to that of the unmodified SPI.
Fig. 1 DSC curves of SPI and SPI-U3250 samples containing various
amounts of A-C16.
A-C16 using differential scanning calorimetry (DSC). To
monitor the thermal transitions more clearly, we used an
SPI-U of lower theoretical molecular weight [SPI-U3250
;
Fig. S7 (a), ESIw] in these studies. The value of Tm of the
pure A-C16 was 123 1C; neither SPI nor SPI-U3250 exhibited a
value of Tm. For the complex formed from SPI-U3250 and
15 wt% A-C16, we did not observe a value of Tm for A-C16,
indicating that the uracil groups of SPI-U3250 were highly
complementary to the adenine groups of A-C16. The use of
30 wt% A-C16 resulted in a slightly visible peak at 128.6 1C,
suggesting the presence of a slight excess of A-C16 in the
SPI-U3250/A-16C complex. We obtained similar results after
converting the SPI-U3250/A-16C complex into its highly
dissociable sulfonic acid form, indicating that SPI-U3250 and
A-C16 formed highly stable hydrogen-bonded complexes in
the bulk state. In contrast, when the SPI was blended with
30 wt% A-C16, we obtained an obviously lower value of Tm.
Wide-angle X-ray diffraction (WAXD) patterns of pure SPI
and pure SPI-U display (Fig. S8, ESIw) several amorphous
halos (26.01 for SPI; 26.0, 15.5, 8.3, and 6.61 for SPI-U). When
UÁ Á ÁA interactions were present, the characteristic peak of
A-C16 disappeared and the amorphous halo of SPI-U3250
(2y = 26.01) shifted to a lower value; in contrast, the
characteristic peaks of A-C16 (2y = 3.2 and 3.51) remained
in the case of the SPI/A-C16 complex.
The type of crosslinking agent, the crosslinking density, and
the microstructural change that occurs after crosslinking all
have dramatic effects on the water uptake, the state of water,
and the proton conductivity of crosslinked membranes.
Although the addition of SMA-A to SPI-U resulted in reduced
ion exchange capacity (IEC), water uptake (WU), and proton
conductivity at 30 1C and 90% RH (Table S2, ESIw), the
SPI-U/SMA-A exhibited higher proton conductivity
(0.15 S cmÀ1) at 80 1C and 90% RH relative to those of the
SPI (0.14 S cmÀ1) and the SPI/SMA-A (0.12 S cmÀ1) under the
same conditions. According to the Arrhenius law, the
temperature dependence of conductivity is associated with
the activation process of proton conduction. The activation
energies (Ea) for the SPI, SPI/SMA-A, and SPI-U/SMA-A
membranes at 90% RH were 10.8, 16.8, and 14.8 kJ molÀ1
,
respectively; when the RH decreased from 90 to 50%, the
values of Ea of both blended membranes (SPI/SMA-A and
SPI-U/SMA-A) for unblended native SPI nearly doubled.
These results indicate that a Grotthuss-type mechanism is
predominant in both of the blended membranes under 50%
RH conditions. Moreover, at 50% RH, both blended
membranes displayed proton conductivities higher than those
of SPI at temperatures between 30 and 80 1C (Fig. 2), implying
that the incorporation of the adenine-based SMA-A into the
membranes improved proton conduction at low RH, due to
the hopping of protons between the sulfonic acid groups of the
sulfonated polymer and the nitrogen atoms of SMA-A.4,10,11
We obtained similar results from proton conductivity
measurements performed under anhydrous conditions (Fig. S12,
ESIw). Notably, the SPI-U/SMA-A membrane exhibited a
more significant increase in proton conductivity upon increasing
temperature at the various RH levels relative to those of the
SPI and SPI/SMA-A membranes. This behavior arose
presumably because the minimized network structure suppressed
ionic cluster formation and resulted in a homogeneous
distribution of the conductive sites over the polymer
backbone.5 TEM micrographs of the SPI and SPI-U/SMA-A
membranes (Fig. 3) verified that the addition of SMA-A
resulted in a better distribution of ionic clusters.
After conducting these preliminary biocomplementary
hydrogen bonding recognition studies, we tested the cross-
linking of SPI-U with the adenine-based crosslinking agent
SMA-A. The solution of SPI-U (at 30 wt% solid content) was
a low-viscosity fluid that readily flowed when we inverted the
vial; in contrast, the SPI-U crosslinked with 20 wt% SMA-A
was an elastic solid. The left-hand image in Fig. S9 (ESIw)
presents the vial-inversion experiment performed using 10 wt%
SMA-A; the gel did not flow after several minutes at room
temperature, but it did flow gradually after several hours. The
right-hand image in Fig. S9 (ESIw) presents the crosslinked
SPI-U/SMA-A fluid after heating at 80 1C, a temperature at
which the gel fluid was transformed into a low-viscosity fluid.
The variation in flow behavior indicates that the degree of
crosslinking can be tuned by varying the amount of the
crosslinking agent. Similar results were verified in viscosity
measurements of the SPI/SMA-A and the SPI-U/SMA-A
c
This journal is The Royal Society of Chemistry 2010
Chem. Commun., 2010, 46, 7554–7556 7555

View Online
times higher than that of the standard SPI and Nafion 117,
respectively.
In summary, we have synthesized a new type of PEM
comprising an SPI hydrogen-bonded with an adenine-based
crosslinking agent. Compared with the pure SPI, both the
SPI/SMA-A and SPI-U/SMA-A membranes exhibited
improved properties: higher proton conductivities at high
temperature and low RH, lower methanol permeabilities,
and higher oxidative stabilities; the latter blend (SPI-U/SMA-A)
is more favorable for membrane performance in DMFCs. We
believe that this technique based on noncovalent interactions
should be applicable to other hydrocarbon-based polymer
membranes, thereby providing a useful method for improving
the properties of PEMs.
Fig. 2 Temperature dependence of the proton conductivities for
the SPI, SPI/SMA-A, and crosslinked SPI-U/SMA-A membranes at
50 and 90% RH.
Notes and references
1 (a) M. A. Hickner, H. Ghassemi, Y. S. Kim, B. R. Einsla and
J. E. McGrath, Chem. Rev., 2004, 104, 4587; (b) B. C. H. Steele and
A. Heinzel, Nature, 2001, 414, 345; (c) C. Y. Wang, Chem. Rev.,
2004, 104, 4727.
2 (a) Y. Woo, S. Y. Oh, Y. S. Kang and B. Jung, J. Membr. Sci.,
2003, 220, 31; (b) Y. S. Ye, W. Y. Chen, Y. J. Huang, M. Y. Cheng,
Y. C. Yen, C. C. Cheng and F. C. Chang, J. Membr. Sci., 2010,
362, 29; (c) H. Tang, M. Pan, S. Lu, J. Lu and S. P. Jiang, Chem.
Commun., 2010, 46, 4351.
3 (a) B. R. Einsla and J. E. McGrath, Am. Chem. Soc. Div. Fuel
Chem., 2004, 49, 616; (b) K. Miyatake, Y. Chikashige, E. Higuchi
and M. Watanabe, J. Am. Chem. Soc., 2007, 129, 3879; (c) D. S.
Kim, G. P. Robertson and M. D. Guiver, Macromolecules,
2008, 41, 2126.
Fig. 3 TEM micrographs of the (A) SPI and (B) SPI-U/SMA-A
4 (a) Y. Fu, A. Manthiram and M. D. Guiver, J. Electrochem. Soc.,
2007, 9, 905; (b) Y. Fu, W. Li and A. Manthiram, J. Membr. Sci.,
2008, 310, 262; (c) W. Li, A. Bellay, Y. Z. Fu and A. Manthiram,
J. Power Sources, 2008, 180, 719.
membranes.
Table S1 (ESIw) reveals that both blended membranes had
lower methanol diffusion coefficients relative to that of the SPI
membrane, with an especially significantly lower value for the
crosslinked SPI-U/SMA-A membrane. The incorporation of
the adenine-based SMA-A into the SPI or SPI-U caused the
pendent adenine groups to insert into the sulfonic acid
domains of the SPI or SPI-U, thereby retarding methanol
crossover. Moreover, the presence of the crosslinking agent
between the polymer chains prevented excessive water swelling
while simultaneously retarding the degree of methanol
crossover.4 The methanol permeability of the SPI-U/SMA-A
membrane was only 11% and 7% of that of the SPI membrane
and Nafion 117 under the same conditions, respectively
(0.94 Â 10À7 cm2 sÀ1 for SPI-U/SMA-A, 8.52 Â 10À7 cm2 sÀ1
for SPI, 13.1 Â 10À7 cm2 sÀ1 for Nafion 117). The ratio of the
proton conductivity to the methanol permeability—i.e., the
selectivity (F)—is an effective parameter for evaluating
membrane performance in DMFCs. Table S1 (ESIw) indicates
that the selectivity of the crosslinked supramolecular
SPI-U/SMA-A membrane was approximately five and seven
5 (a) Y. Yin, S. Hayashi, O. Yamada, H. Kita and K. I. Okamoto,
Macromol. Rapid Commun., 2005, 26, 696; (b) Y. S. Oh, H. J. Lee,
M. Yoo, H. J. Kim, J. Han, K. Kim, J. D. Hong and T. H. Kim,
Chem. Commun., 2008, 2028.
6 D. Bogdal, P. Perez, J. Pielichowski and A. Prociak, Adv. Polym.
Sci., 2003, 163, 193.
7 (a) T. Park and S. S. Zimmerman, J. Am. Chem. Soc., 2006, 128,
11582; (b) G. Cooke and V. M. Rotello, Chem. Soc. Rev., 2002,
31, 275.
8 (a) L. L. De Lucca Freitas and R. Stadler, Macromolecules, 1987,
20, 2478; (b) C. Hilger, M. Draeger and R. Stadler, Macro-
molecules, 1992, 25, 2498; (c) L. R. Rieth, R. F. Eaton and
G. W. Coates, Angew. Chem., Int. Ed., 2001, 40, 2153.
9 (a) R. J. Thibault, P. J. Hotchkiss, M. Gray and V. M. Rotello,
J. Am. Chem. Soc., 2003, 125, 11249; (b) T. Kawakami and
T. Kato, Macromolecules, 1998, 31, 4475; (c) K. P. Nair,
V. Breedveld and M. Weck, Macromolecules, 2008, 41, 3429.
10 C. C. Cheng, Y. C. Yen, Y. S. Ye and F. C. Chang, J. Polym. Sci.,
Part A: Polym. Chem., 2009, 47, 6388.
11 (a) J. Saito, K. Miyatake and M. Watanabe, Macromolecules,
2008, 41, 2415; (b) J. Wang, Y. Song, C. Zhang, Z. Ye, H. Liu,
M. H. Lee, D. Wang and J. Ji, Macromol. Chem. Phys., 2008, 209,
1495; (c) E. K. Pefkianakis, V. Deimede, M. K. Daletou,
N. Gourdoupi and J. K. Kallitsis, Macromol. Rapid Commun.,
2005, 26, 1724.
c
7556 Chem. Commun., 2010, 46, 7554–7556
This journal is The Royal Society of Chemistry 2010