New poly(ethylene oxide) soft segment-containing sulfonated polyimide copolymers for high temperature proton-exchange membrane fuel cells

Available online at www.sciencedirect.com

Journal of Membrane Science 313 (2008) 75–85

New poly(ethylene oxide) soft segment-containing sulfonated polyimide copolymers for high temperature proton-exchange membrane fuel cells He Bai a,b , W.S. Winston Ho a,b,∗ a

Department of Chemical and Biomolecular Engineering, The Ohio State University, 140 West 19th Avenue, Columbus, OH 43210-1180, USA b Department of Materials Science and Engineering, The Ohio State University, 2041 College Road, Columbus, OH 43210-1178, USA Received 12 June 2007; received in revised form 26 November 2007; accepted 26 December 2007 Available online 5 January 2008

Abstract The synthesis and characterization of a series of new poly(ethylene oxide) (PEO) soft segment-containing six-member ring sulfonated polyimide (SPI) copolymers are described in this paper. One-step high temperature polymerization method was used to prepare the SPI copolymers from 1,4,5,8-naphthalenetetracarboxylic dianhydride (NTDA), 4,4 -diaminostilbene-2,2 -disulfonic acid (DSDSA), and diamine-terminated poly(ethylene oxide) (PEO-diamine, MW = 1000). The relative ratio of the sulfonic acid-containing hard segments to the PEO-containing soft segments was controlled through varying the molar ratio of DSDSA to PEO-diamine. Flexible, transparent, and mechanically strong free-standing membranes were successfully obtained. The membranes were characterized with ion-exchange capacity, Fourier transform infrared spectra, thermogravimetric analysis, water sorption, proton conductivity, and fuel cell performance measurements. The results showed that the new SPI membranes exhibited desirable mechanical properties and thermal stability as well as better proton conductivities than Nafion® 115 at high relative humidity (RH) levels (>50%) at both 70 and 120 ◦ C. The results from the fuel cell performance measurements indicated that the SPI membrane containing 5 mol% (12.4 wt.%) PEO soft segments had similar fuel cell performance as Nafion® 112 at 70 ◦ C and 80% RH, but better fuel cell performance than Nafion® 112 when the current density was higher than 0.32 A/cm2 at 120 ◦ C and 50% RH. © 2008 Elsevier B.V. All rights reserved. Keywords: Sulfonated polyimide (SPI); Poly(ethylene oxide) (PEO) soft segments; Copolymer; Proton-exchange membrane (PEM); High temperature; Proton conductivity; Fuel cell performance; Fuel cell

1. Introduction In recent years, great progress has been made on the development of proton-exchange membrane fuel cells (PEMFCs) for both mobile and stationary applications, particularly for fuel cell vehicles. Dupont’s Nafion® and other perfluorinated sulfonic acid membranes are currently popular to use for low temperature PEMFCs due to their high proton conductivity as well as desirable mechanical strength and chemical stability. However, some disadvantages, such as high cost, relatively low conductivity at high temperatures (above 100 ◦ C) and low humidities, and high dependence of proton conduction on the water content, seriously limit the industrial application of these membranes. High

∗ Corresponding author at: Department of Chemical and Biomolecular Engineering, The Ohio State University, 140 West 19th Avenue, Columbus, OH 43210-1180, USA. Tel.: +1 614 292 9970; fax: +1 614 292 3769. E-mail address: [email protected] (W.S.W. Ho).

0376-7388/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.memsci.2007.12.062

temperature operations can increase the anode’s tolerable level of CO in the fuel and accelerate the reaction rates at the anode and cathode. Low humidity operations can facilitate the water management of the fuel cell system. Therefore, it is desirable for a PEMFC to operate at high temperatures (above 100 ◦ C) and low relative humidities (below 50% RH). As a result, the development of competitive and less expensive PEMs that have efficient performance at high temperatures is crucial for fuel cell applications. Many efforts have been initiated to synthesize costeffective and thermally stable alternative membranes, including sulfonated poly(arylene ether sulfone) copolymers [1–3], sulfonated poly(aryl ether ketone) copolymers [4,5], phosphoric acid-soaking polybenzimidazole (PBI) [6,7], and radiationgrafted membranes [8,9]. Several membranes have continued to attract interest. Recently, sulfonated polyimides (SPIs) have been shown to be promising materials for PEMs mainly because of their excellent mechanical and thermal properties as well as their chemical

76

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

stability. Pineri and his coworkers first synthesized five-member and six-member ring SPIs from 4,4 -diaminobiphenyl-2,2 disulfonic acid (BDSA), 4,4 -oxydianiline (ODA), and two types of dianhydrides—oxydiphthalic dianhydride (OPDA) and 1,4,5,8-naphthalenetetracarboxylic dianhydride (NTDA) [10]. Fuel cell experiments performed with the six-member ring NTDA-based SPI membrane revealed reasonably good performance (similar to Nafion® 117 at 70 ◦ C and a fully hydrated state) and the stability of at least 3000 h. In contrast, the fivemember ring OPDA-based SPI membrane was not stable during fuel cell measurements, which was due to the hydrolysis of imide rings from the sulfonated imide sequence, thus leading to chain scissions. McGrath et al. also prepared a series of five-member and six-member ring SPIs using different sulfonated diamines, two commercially available sulfonated diamines—BDSA and 2,5-diamino benzene sulfonic acid (DABSA) [11,12] and two self-synthesized sulfonated diamines—sulfonated bis(3aminophenyl)phenyl phosphine oxide (SBAPPO) and 3,3 disulfonic acid-bis[4-(3-aminophenoxy)phenyl]sulfone (SADADPS) [13,14]. The SA-DADPS-based six-member ring SPI membranes with the IEC values of about 1.6 mmol/g showed lower conductivities than Nafion® 1135 at 80 ◦ C and all RH levels [15]. The membranes were also used for direct methanol fuel cell (DMFC) measurements, with the results showing similar fuel cell performance as Nafion® 117 at 80 ◦ C with the methanol feed concentration of 0.5 M [15]. However, no high temperature H2 /O2 fuel cell performance data were reported. Okamoto and his coworkers successfully synthesized two types of six-member ring SPIs (main-chain type and side-chain type) to study the “structure–property” relationship of SPIs systematically. They synthesized aromatic sulfonated diamines, such as 4,4 -diaminodiphenyl ether2,2 -disulfonic acid (ODADS) [16], 9,9 -bis(4-aminophenyl) fluorene-2,7-disulfonic acid (BAPFDS) [17], 4,4 -bis(4aminophenoxy)biphenyl-3,3 -disulfonic acid (BAPBDS) [18], and 2,2 -bis(4-aminophenoxy)biphenyl-5,5 -disulfonic acid (oBAPBDS) [19], and prepared several series of main-chain type sulfonated polyimides by their copolymerization with NTDA and common diamines. They also synthesized a new type of side-chain SPIs using their self-developed sulfonated diamines, such as 2,2 -bis(3-sulfopropoxy)benzidine (2,2 -BSPB) [20–22], 3,3 -bis(3-sulfopropoxy)benzidine (3,3 BSPB) [21,22], 3-(2 ,4 -diaminophenoxy)propane sulfonic acid (DAPPS) [23], 3,5-diamino-3 -sulfo-4 -(4-sulfophenoxy) benzophenone (DASSPB) [24], 3,5-diamino-3 -sulfo-4 -(2,4disulfophenoxy)benzophenone (DASDSPB) [24], and bis[4(4-aminophenoxy)-2-(3-sulfobenzoyl)]phenyl sulfone (BAPSBPS) [25]. The results showed that at low temperatures (i.e., 50 or 60 ◦ C), the conductivities of these new SPI membranes were lower than Nafion® 117 at low RH levels and similar or higher than Nafion® 117 at high RH levels. Moreover, at the same RH value, the membrane conductivity improved with the increase of temperature due to the activation energy effect. Further research in Okamoto’s group showed that the BAPBDS-based main-chain type and BSPB-based side-chain type SPI membranes displayed fairly good conductivities

[18,20–22] and excellent water stability (more than 1000 h in boiling water) [26]. So they were more suitable for the real fuel cell applications, particularly for direct methanol fuel cells (DMFCs) because of their low methanol permeabilities [27]. The BAPBDS-based SPI membranes showed comparable fuel cell performance to Nafion® 112 at 90 ◦ C and nearly 100% RH and the short-term stability up to 50 h in the H2 /O2 fuel cell measurements [28]. Okamoto et al. also prepared a series of branched/crosslinked SPI membranes by using their self-synthesized 1,3,5-tris(4-aminophenoxy)benzene (TAPB) as a crosslinker, and the resulting membranes showed similar conductivities and fuel cell performance as the BAPBDSbased SPI membranes but much improved water stability and mechanical properties under the accelerated aging treatment [29,30]. In this article, we report the synthesis of a series of new soft segment-containing six-member ring SPI copolymers using NTDA, DSDSA, and PEO-diamine. Hydrophilic soft segments of PEO were copolymerized into SPIs to not only increase water retention in the membranes (particularly at high temperatures and low RHs), but also improve the membrane mechanical properties. The resulting free-standing membranes were characterized with ion-exchange capacity, Fourier transform infrared spectra, thermogravimetric analysis, water sorption, proton conductivity, and fuel cell performance measurements, and their potential applications for high temperature PEMFCs were explored. 2. Experimental 2.1. Materials Triethylamine (TEA, 99.5%), m-cresol (99%, bp = 203 ◦ C, d = 1.034 g/ml), and benzoic acid (99.5+%), all from Aldrich (Milwaukee, WI), were used as received without further purification. 1,4,5,8-Naphthalenetetracarboxylic dianhydride (NTDA, 99+%, Aldrich) and 4,4 -diaminostilbene-2,2 -disulfonic acid (DSDSA, 94+%, TCI America (Portland, OR)) were dried in a vacuum oven at 150 ◦ C overnight before the reaction. Diamine terminated poly(ethylene oxide) (PEO-diamine, MW = 1000, 100%) was donated by Kawaken Fine Chemicals Co., Ltd. (Tokyo, Japan), and was dried in vacuum at 120 ◦ C overnight prior to use. 2.2. Synthesis of new soft segment-containing six-member ring SPI copolymers A typical procedure for preparation of new soft segmentcontaining six-member ring sulfonated polyimide copolymers is described below using the copolymer of NTDA/DSDSAEt3 N (95 mol%)/PEO-diamine (5 mol%) as an example. In the preparation, 3.744 g (9.5 mmol, 94+%) of DSDSA, 2.307 g (22.8 mmol) of TEA, and 44 g of m-cresol were charged into a 250-ml, completely dried 4-neck flask equipped with a mechanical stirring device and a nitrogen inlet. The mixture was heated to 80 ◦ C under stirring with a nitrogen flow until DSDSA-Et3 N was completely dissolved, and a transpar-

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

77

ent orange solution was obtained. Then, to the reactor, 0.500 g (0.5 mmol) of PEO-diamine, 2.682 g (10.0 mmol) of NTDA, and 2.442 g (20.0 mmol) of benzoic acid were added successively, and additional 66.7 g of m-cresol was added to obtain a solid concentration of 7.23 wt.%. The reaction was continued at 80 ◦ C for 4 h and 180 ◦ C for 25 h with the nitrogen purge. Finally, a very viscous brown solution was observed. When the copolymer solution was cooled down, more solvent m-cresol, about 93.4 g, was added to dilute the highly viscous solution to obtain the final copolymer concentration of 4.05 wt.%. The copolymer solution was poured into 350 ml of acetone with vigorous stirring, and the precipitate was filtered off and washed thoroughly with acetone. The resulting orange, fiber-like copolymer was collected after drying at 120 ◦ C in vacuum overnight. Finally, 8.233 g of product was obtained, yield = 95.5%.

2.4.3. Thermogravimetric analysis Thermogravimetric analysis (TGA) was performed to estimate the thermal stability of the membranes with a Pyris 1 TGA thermogravimetric analyzer (PerkinElmer, Shelton, CT) at a heating rate of 20 ◦ C/min in N2 in the temperature range of 30–390 ◦ C. All of the specimens were dried in vacuum at 90 ◦ C overnight before measurements.

2.3. Membrane preparation

where Wd and We are the weights of dry and corresponding water-equilibrated film sheets, respectively. Three sheets of each membrane composition (20–30 mg per sheet) were measured by the above method, and the average value was calculated.

The SPI copolymers (in triethylammonia salt form) were dissolved in m-cresol to form about 3.0 wt.% solutions. The membranes with a controlled thickness (dry membrane thickness = 25 ␮m) were prepared by casting the solutions using a GARDCO adjustable micrometer film applicator (Paul N. Gardner Company, Pompano Beach, FL) onto clean glass plates. The as-cast films were dried in a hood overnight, followed by drying in an oven at 80 ◦ C for 12 h and 120 ◦ C for 16 h. Then, the membranes were carefully removed from the glass substrates, and were soaked for the purpose of proton-exchange treatment in 1.0 M hydrochloric acid at room temperature for more than 24 h. The resulting free-standing membranes (in acid form) were thoroughly washed with de-ionized water and dried in vacuum at 90 ◦ C overnight. 2.4. Membrane characterization The resulting dry membranes (in acid form) were characterized with ion-exchange capacity, Fourier transform infrared spectra, thermogravimetric analysis, water sorption, proton conductivity, and fuel cell performance measurements. 2.4.1. Ion-exchange capacity The ion-exchange capacity (IEC) was measured by means of a classical titration method. A membrane sample of about 0.200 g was soaked in 50 ml of 1.0 M NaCl solution for 2 days. Released proton concentration was titrated using a 0.01 M NaOH solution. 2.4.2. Fourier transform infrared spectra Fourier transform infrared (FTIR) spectroscopy spectra were recorded on a Nexus 470® FTIR Spectrometer (Thermo Nicolet, Madison, WI), using Smart MIRacleTM Single Reflection Horizontal ATR (attenuated total reflectance). Each of the membrane samples was put in appropriate physical contact with the sampling plate of the spectrometer accessory, yielding high quality and reproducible spectra.

2.4.4. Water sorption The membranes were vacuum-dried at 90 ◦ C overnight, weighed (Wd ) and immersed in de-ionized water at room temperature for 48 h. The wet membranes were wiped dry with tissue papers and quickly weighed again (We ). The water uptake of the membranes was calculated in weight percent as follows: Water uptake (%) =

(We − Wd ) × 100 Wd

(1)

2.4.5. Proton conductivity measurement The method for proton conductivity measurements was based on a four-point-probe electrochemical impedance spectroscopy (EIS) technique [17,23]. The existing fuel cell (25 cm2 effective membrane area) hardware (ElectroChem, Inc., Woburn, MA), gas paths, and test station (Model 890C, Scribner Associates, Inc., Southern Pines, NC) were used for proton conductivity measurements. As shown in Fig. 1a, a conductivity cell (BekkTech LLC, Loveland, CO) was installed between the anode and cathode plates of the fuel cell hardware, and the membrane conductivity was measured through the connected four probes. Fig. 1b shows the inside structure of the conductivity cell, which consisted of a Teflon block, a membrane clamp, and four platinum wires. The two outside platinum wires were used as working and counter electrodes to apply a current to the sample membrane (2.5 cm × 0.5 cm) through the two connected platinum gauzes, and the two inside platinum wires at 0.425 cm apart were used as reference electrodes. The membrane sample was equilibrated by the incoming hydrogen gas with a specific humidity level through the existing gas path. The relative humidity level of the hydrogen gas was controlled by the set temperature of the self-designed humidifier. The temperature of the membrane sample was maintained by the fuel-cell test station, and the back pressure regulator was used to adjust the system pressure. The AC impedance measurements at various cell temperatures and relative humidity levels were carried out over the amplitude of 300 mV and the frequency range from 10 kHz to 1 Hz using a Solartron 1260 frequency response analyzer and a Solartron 1287 potentiostat (Solartron Analytical, Houston, TX). The resistance value associated with the membrane conductance was determined from the intercept of the impedance with the real axis using the Zplot/Zview software [31]. From the measured membrane resistance, the proton conductivity σ of the membrane was calculated using the following

78

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

Fig. 1. The pictures of the apparatus for proton conductivity measurements: (a) outside appearance of the assembled conductivity cell and (b) inside structure of the conductivity cell.

equation: σ=

L RDW

(2)

where L is the distance between the two reference electrodes, R is the measured membrane resistance value, and D and W are the thickness and width of the sample membrane at the ambient conditions, respectively. The possible effects of membrane swelling on the changes of the dimensions during measurements were not considered. 2.4.6. Fuel cell performance measurement The membrane electrode assembly (MEA) preparation and fuel cell performance measurements were conducted at the University of California, Riverside. The investigation of fuel cell performance was carried out in a 5-cm2 single-cell test fixture (Arbin Instruments, College Station, TX). The electrodes (GDL 25CC, SGL Carbon Group, St. Marys, PA) had 0.2 mg/cm2 Pt loading by applying 20 wt.% Pt on carbon support Vulcan XC72 (20 wt.% Pt/C, E-TEK, Natick, MA) and were impregnated with 1.0 mg/cm2 Nafion® using 5 wt.% Nafion® solution via a spraying technique. The MEA was prepared by assembling the membrane between two electrodes and hot pressing at 135 ◦ C under 140 atm for 3 min. The single-cell MEA was installed in a fuel-cell test station (FCTS-50W-50A-5V-1/2slpm, Arbin Instruments, College Station, TX). Each of the H2 and O2 gas flow rates was 200 ml/min. Measurements of the polarization curves (voltage vs. current density) were conducted at 70 ◦ C/80% RH/2 atm and 120 ◦ C/50% RH/3 atm, respectively. 3. Results and discussion A series of new soft segment-containing six-member ring SPI copolymers were prepared from NTDA, DSDSA, and PEOdiamine. BDSA and DSDSA are the only two commercially available sulfonated diamines so far. As reported in the literature

[18,26] and from our previous research [34], the BDSA-basedSPI membranes had very poor stability in water, which were highly swollen or dissolved in water even at low temperatures. Thus, DSDSA was used in this work. Our previous research [34] and this work showed that DSDSA-based-SPI membranes had very good stability in water even at high temperatures (no swollen or dissolution occurred). Definitely, other selfsynthesized sulfonated diamines that had been proven to have desirable water stability, such as SA-DADPS [13] and BAPBDS [18], could also be used. Moreover, the usage of DSDSA provides the possibility for post-modification of the membranes, for example, UV or chemical crosslinking of C C to further strengthen the membrane stability or chemical modification based on C C to further improve the membrane performance. One-step high temperature copolymerization was carried out in m-cresol in the presence of triethylamine and benzoic acid (Fig. 2). This method has been extensively employed in the literature [10–30]. The resulting copolymers consisted of the hard segments of sulfonated polyimide and the soft segments of poly(ethylene oxide). The hard segments had sulfonated functional groups that could make the membrane conductive. The soft segments were very hydrophilic and flexible, which could not only increase water retention of the membrane even at high temperatures and low RHs but also improve the membrane mechanical properties. The relative ratio of the hard segments to the soft segments was controlled by varying the molar ratio of DSDSA to PEO-diamine. DSDSA had poor solubility in every solvent. The triethylammonium salt form produced after a reaction with the highly basic tertiary amine (TEA) at 80 ◦ C was soluble in m-cresol, which not only allowed its solution to be prepared for copolymerization but also produced the free, much more reactive non-Zwitterion amino groups that could react favorably with the dianhydride. The resulting SPI copolymers generally showed high viscosity with 2.5–3.0 wt.% solutions in m-cresol, indicating that high molecular weight copolymers were synthesized. The soft segment-containing SPI copolymers in their salt form

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

79

Fig. 2. Synthesis of the new soft segment-containing six-member ring sulfonated polyimide copolymers.

were soluble in m-cresol or dimethyl sulfoxide (DMSO), but insoluble in other common dipolar aprotic solvents, e.g., Nmethyl-2-pyrrolidinone (NMP) and N,N-dimethylformamide (DMF). Flexible and mechanically strong free-standing membranes were successfully obtained using the casting method described earlier, and the resulting membranes were acidified for proton-exchange treatment. 3.1. Ion-exchange capacity values The measured ion-exchange capacity (IEC) values for the membranes investigated are listed in Table 1. As shown in this table, the measured values were lower than the calculated ones (about 10%). This might be caused by the incomplete acidification and/or incomplete drying of the membranes. 3.2. Infrared spectra The FTIR spectra of the new SPI membranes with various amounts of PEO soft segments (2–20 mol%) are shown in Fig. 3. The characteristic absorption bands of the imide carbonyl groups (C O) were observed around 1710 cm−1 (asymmetric) and 1680 cm−1 (symmetric), respectively. The out-of-phase bending of the imide rings was observed around 715 cm−1 . The C N C stretching vibration of the imide rings was observed

Fig. 3. FTIR spectra of the new SPI membranes with various amounts of PEO soft segments: (a) NTDA/DSDSA (80 mol%)/PEO-diamine (20 mol%), (b) NTDA/DSDSA (90 mol%)/PEO-diamine (10 mol%), (c) NTDA/DSDSA (95 mol%)/PEO-diamine (5 mol%), and (d) NTDA/DSDSA (98 mol%)/PEOdiamine (2 mol%).

Table 1 IEC values and water uptake of the new SPI membranes Membranes

Calculated IEC (mmol/g)

Measured IEC (mmol/g)

Water uptake (%)

NTDA/DSDSA (80 mol%)/PEO-diamine (20 mol%) NTDA/DSDSA (90 mol%)/PEO-diamine (10 mol%) NTDA/DSDSA (92.5 mol%)/PEO-diamine (7.5 mol%) NTDA/DSDSA (95 mol%)/PEO-diamine (5 mol%) NTDA/DSDSA (98 mol%)/PEO-diamine (2 mol%) Nafion® 115

2.20 2.70 2.85 3.00 3.19 0.91

2.02 2.35 2.45 2.56 2.87 0.91

42.9 50.0 55.6 63.6 65.5 22.9 [32]

80

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

around 1340 cm−1 . These peaks could confirm the formation of polyimides. No peak at 1780 cm−1 (corresponding to poly(amic acid)) was observed, which suggested complete imidization of the SPI membranes [29]. The absorption bands of the sulfonic acid groups (S O) appeared at 1080 cm−1 (asymmetric) and 1020 cm−1 (symmetric), respectively [14]. In the IR measurements, all membranes used had a very similar membrane thickness of about 25 ␮m. In Fig. 3, the imide peaks at 1710, 1680, 1340, and 715 cm−1 of different membranes had very similar peak heights since these membranes had a similar imide content even the PEO amount was changed from 2 to 20 mol%. Thus, these imide peaks can be regarded as the standard unchanging peaks. The absorption bands at 1120 and 870 cm−1 were assigned to the C O C and C N groups in the PEO-imide soft segments, respectively. From Fig. 3, it can be seen that with the increase of the PEO amount, these two peaks significantly increased. 3.3. Thermal stability Thermal stability of the new SPI membranes was investigated by thermogravimetric analysis (TGA) measurements. All the membrane samples were completely dried in a vacuum oven at 90 ◦ C overnight before measurements. Fig. 4 shows the TGA curves of the soft segment-containing SPI membranes with various amounts of PEO soft segments (2–20 mol%), and their thermal properties are listed in Table 2. The initial small amount of weight loss before 200 ◦ C was caused by the evaporation of the sorbed water within the membranes. The amount of initial weight loss of these new SPI membranes was in the order of 2 mol% PEO > 5 mol% PEO > 10 mol% PEO > 20 mol% PEO (Table 2). The reason is that with the decrease of the PEO amount, the SPI membrane contained more sulfonic acid groups and more bonded water to these groups. As shown in Table 2, with the increase of the PEO amount (5–20 mol%), the new SPI membrane showed an earlier first decomposition (244–231 ◦ C), which might have been due to the loss of the PEO soft segments, and a later second decomposition (308–318 ◦ C), which was due to the loss of the sulfonic acid groups. For the SPI membrane with 2 mol% PEO soft segments, only one apparent decomposition at around 302 ◦ C was observed. Moreover, higher PEO contents resulted in less final residues at 390 ◦ C (Table 2), which is reasonable since those PEO soft segments were lost before 390 ◦ C. Generally, the TGA curves clearly indicate that the new SPI membranes have fairly good thermal stability, with the great potential for high temperature applications.

Fig. 4. TGA curves of the new SPI membranes with various amounts of PEO soft segments: (a) NTDA/DSDSA (80 mol%)/PEO-diamine (20 mol%), (b) NTDA/DSDSA (90 mol%)/PEO-diamine (10 mol%), (c) NTDA/DSDSA (95 mol%)/PEO-diamine (5 mol%), and (d) NTDA/DSDSA (98 mol%)/PEOdiamine (2 mol%).

3.4. Water uptake The water uptake values of the new SPI membranes with various amounts of PEO soft segments are also listed in Table 1, and the water uptake of Nafion® 115 [32] is listed in the same table. With the decrease of the PEO amount (20–2 mol%), the sulfonic acid concentration (IEC value) of the membrane increased (2.02–2.87 mmol/g). Thus, the membrane water uptake value increased (42.9–65.5%). This is in agreement with the TGA measurements (initial weight loss). The water uptake of these new SPI membranes was much higher than Nafion® 115 (22.9%) [32], which was due to their higher IEC values. Moreover, the water uptake of these new PEO-containing SPI membranes was much higher than the other non-PEO-containing SPI membranes (below 20.0%) as reported in literature [32]. 3.5. Proton conductivities The proton conductivity of the new soft segment-containing SPI membranes as a function of relative humidity was measured at 70 and 120 ◦ C, respectively.

Table 2 Thermal properties of the new SPI membranes Membranes

Initial weight loss before 200 ◦ C

First decomposition starting temperature

Second decomposition starting temperature

Residues at 390 ◦ C

NTDA/DSDSA (80 mol%)/PEO-diamine (20 mol%) NTDA/DSDSA (90 mol%)/PEO-diamine (10 mol%) NTDA/DSDSA (95 mol%)/PEO-diamine (5 mol%) NTDA/DSDSA (98 mol%)/PEO-diamine (2 mol%)

0.4% 0.6% 1.6% 2.0%

231 ◦ C 237 ◦ C 244 ◦ C 302 ◦ C

318 ◦ C 314 ◦ C 308 ◦ C

77.6% 79.1% 79.7% 80.3%

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

81

Fig. 5. Proton conductivity as a function of relative humidity for new SPI membranes in comparison with Nafion® 115 at 70 ◦ C and atmospheric pressure.

3.5.1. Proton conductivities at 70 ◦ C The membrane conductivities at 70 ◦ C and different RHs were measured at ambient pressure. For comparison, the conductivities of Nafion® 115 were measured at the same conditions (see Fig. 5), in which the Nafion® samples were used without any treatment prior to being measured. The measured conductivities of Nafion® at 70 ◦ C were consistent with past research [17,33], which appears to support the reliability of our measurements. As shown in Fig. 5, the proton conductivity of the new SPI membranes significantly increased with relative humidity at 70 ◦ C and atmospheric pressure. The proton conductivity also strongly depended on the membrane IEC value. The SPI membrane with 2 mol% PEO soft segments had the highest conductivities at 70 ◦ C because it had the highest IEC value. With the increase of the PEO amount (2–20 mol%), the sulfonic acid concentration (IEC value) of the membrane decreased (2.87–2.02 mmol/g) and the water uptake value of the membrane also decreased (65.5–42.9%). Thus, the membrane conductivities at 70 ◦ C decreased in the same trend. Compared to Nafion® 115, the new SPI membrane with 2 mol% PEO soft segments had lower conductivities than Nafion® at low RH levels (<50%) at 70 ◦ C and atmospheric pressure. This may be caused by the fact that Nafion® has unique ion-rich channels (clusters), which are favorable for proton transport, particularly at low RH levels [19]. In contrast, the SPI membrane has a homogeneous structure (no clear ionic domains are separated from the polymer matrix). However, at high RH levels (>50%), the conductivities of this new SPI membrane (2 mol% PEO soft segments) were higher than Nafion® . The reason may be that the SPI membrane has a higher water content at high RH levels than Nafion® due to its higher sulfonic acid concentration (IEC value), which could be helpful for the increased proton transport capability.

Fig. 6. Proton conductivity as a function of relative humidity for new soft segment-containing SPI membranes in comparison with the SPI membranes without soft segments at 70 ◦ C and atmospheric pressure (the conductivities of the SPI membranes without soft segments were reported in our previous work [34]).

Furthermore, the new soft segment-containing SPI membranes (in this paper) had much higher conductivities than the other non-soft segment-containing SPI membranes (in previous work) [34] at the same temperature and pressure as well as the similar relative humidity conditions. As shown in Fig. 6, at 70 ◦ C and atmospheric pressure, the membrane of NTDA/DSDSA (98 mol%)/PEO-diamine (2 mol%) had much higher conductivities than the membrane of NTDA/DSDSA (90 mol%)/ODA (4,4 -oxydianiline, 10 mol%) at all RH levels even though they had very similar IEC values (2.87 mmol/g vs. 2.84 mmol/g). Similarly, the membrane of NTDA/DSDSA (92.5 mol%)/PEOdiamine (7.5 mol%) had much higher conductivities than the membrane of NTDA/DSDSA (70 mol%)/ODA (30 mol%), both with similar IEC values (2.45 mmol/g vs. 2.43 mmol/g). The reason is that the PEO soft segments could increase water retention of the SPI membranes, which is helpful for the proton transport. The introduction of the PEO soft segments could not only improve the membrane conductivity but also improve the membrane mechanical properties. The SPI polymer with the composition of NTDA/DSDSA (100 mol%) was also prepared in the same method, but the resulting membrane was very brittle and highly swollen in water, which could not be used for conductivity or fuel cell performance measurements. After introducing 10 mol% unsulfonated diamine ODA [34], the membrane of NTDA/DSDSA (90 mol%)/ODA (10 mol%) (IEC = 2.84 mmol/g) was still brittle, and its mechanical properties were still not desirable for fuel cell applications. To further improve the membrane mechanical properties, more relative ratio of unsulfonated diamine ODA was used, thus, the membrane IEC value and proton conductivity decreased [34].

82

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

However, in this work, the introduction of only 2 mol% PEO soft segments made the membrane very flexible and water stable. The membrane of NTDA/DSDSA (98 mol%)/PEO-diamine (2 mol%) (IEC = 2.87 mmol/g) was not soluble in water, which is a key point for the membrane stability. Moreover, this membrane was very flexible even after measurements at 70 ◦ C and various RHs. Thus, the mechanical properties of the SPI membranes were much improved by the introduction of small amounts of PEO soft segments. This allowed the SPI membranes with higher IEC values to be prepared without the sacrifice of the membrane mechanical properties. 3.5.2. Proton conductivities at 120 ◦ C The membrane conductivities at 120 ◦ C and different RHs were also measured (the corresponding high pressures were applied). For comparison, the conductivities of Nafion® 115 were measured at the same conditions (see Fig. 7), in which the Nafion® samples were used without any treatment prior to being measured. In a similar trend as at 70 ◦ C, the proton conductivity of the new SPI membranes significantly increased with relative humidity at 120 ◦ C. As shown in Fig. 7, in the range of 20–5 mol% PEO soft segments, with the decrease of the PEO amount, the sulfonic acid concentration (IEC value) of the membrane increased. Thus, the membrane conductivities at 120 ◦ C increased, which was in accordance with the trend at 70 ◦ C. However, with a further decrease of the PEO amount from 5 to 2 mol%, the membrane conductivity had no further increase, but a slight decrease, which showed a reverse trend compared to that at 70 ◦ C. This meant that the hydrophilicity of the SPI membranes at low temperatures was not always in accordance with that at high temperatures. At 70 ◦ C, the SPI membrane with 2 mol% PEO soft segments

showed higher conductivities than that with 5 mol% PEO since the former had a higher IEC value and a higher liquid water uptake value (Table 1). However, at 120 ◦ C (above the water boiling point), the SPI membrane with 5 mol% PEO soft segments might be more hydrophilic and could retain more water vapor than that with 2 mol% PEO since the latter contained too small amount of hydrophilic PEO soft segments. Thus, the SPI membrane with 5 mol% PEO soft segments showed higher conductivities even though its IEC value was lower than that with 2 mol% PEO. In this paper, the water uptake (Table 1) was measured in liquid water at room temperature, which might only reflect the water retention ability of the membranes at low temperatures (e.g., 70 ◦ C). No equipment for measuring the water vapor uptake values at high temperatures (e.g., 120 ◦ C) was available in this work. In the range of 20–5 mol% PEO soft segments, with the decrease of the PEO amount, the water vapor uptake value of the SPI membrane at 120 ◦ C might increase, which was in accordance with the order of the liquid water uptake values measured in this work. However, at 120 ◦ C, the water vapor uptake value of the SPI membrane with 5 mol% PEO soft segments might be higher than that with 2 mol% PEO, which was in the reverse order compared to the liquid water uptake values measured in this work. Considering the fact that the SPI membrane with 5 mol% PEO soft segments had a higher conductivity than that with 2 mol% PEO even though the latter had a higher IEC value, we believe that the PEO soft segments played a more important role in water retention and proton conductivity of the SPI membranes at high temperatures. In a similar trend as at 70 ◦ C, the new SPI membrane with 5 mol% PEO soft segments had lower conductivities at low RH levels (<60%) and higher conductivities at high RH levels (>60%) than Nafion® 115 at 120 ◦ C. These could be explained by the same reason as that at 70 ◦ C. In addition, for the SPI membranes without PEO soft segments, no conductivity data at high temperatures and low RHs were ever reported in the literature [10–30]. Thus, the high temperature and low humidity conductivity results reported in this paper are very promising, which explore the potential applications of the new soft segmentcontaining SPI membranes for high temperature PEMFCs. 3.6. Fuel cell performance

Fig. 7. Proton conductivity as a function of relative humidity for new SPI membranes in comparison with Nafion® 115 at 120 ◦ C (the corresponding high pressures were applied).

The membrane electrode assembly (MEA) was fabricated from a SPI membrane (5 cm × 5 cm) sample containing 5 mol% PEO soft segments (see Fig. 8) using the MEA preparation method described earlier since this membrane composition showed the best conductivities at high temperatures. Then, the MEA was used for fuel cell performance measurements at different conditions (70 ◦ C/80% RH and 120 ◦ C/50% RH, respectively). The thickness of the SPI membrane was about 25 ␮m. For comparison, the fuel cell performance of Nafion® 112 (membrane thickness = 50 ␮m) was measured at the same conditions (see Figs. 9 and 10). The polarization curves of the new SPI membrane and Nafion® 112 at 70 ◦ C and 80% RH are shown in Fig. 9. The conditions for the measurements were

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

Fig. 8. MEA prepared from the new SPI membrane (NTDA/DSDSA (95 mol%)/PEO-diamine (5 mol%)). Photo taken by Zhongwei Chen (the University of California, Riverside).

Tcell = 70 ◦ C, Tanode/cathode = 65 ◦ C/65 ◦ C, each of H2 and O2 flow rates = 200 ml/min, and the pressure of the system = 2 atm. The figure shows that the new soft segment-containing SPI membrane had quite similar fuel cell performance as Nafion® 112 at 70 ◦ C and 80% RH. The polarization curves of the new SPI membrane and Nafion® 112 at 120 ◦ C and 50% RH are shown in Fig. 10. The conditions for the measurements were Tcell = 120 ◦ C, Tanode/cathode = 100 ◦ C/100 ◦ C, each of H2 and O2 flow rates = 200 ml/min, and the pressure of the system = 3 atm. The figure shows that the new SPI membrane had a lower open circuit voltage (OCV) than Nafion® 112 (0.72 V vs. 0.92 V) at

83

Fig. 10. Polarization curves of the new SPI membrane (NTDA/DSDSA (95 mol%)/PEO-diamine (5 mol%)) and Nafion® 112 at 120 ◦ C and 50% RH (pressure = 3 atm). Data provided by Zhongwei Chen (the University of California, Riverside).

120 ◦ C and 50% RH. This might be caused by the hydrolysis of the SPI membrane, which was well studied and widely reported particularly at high temperatures in the literature [26,35]. During the measurements at high temperatures, the hydrolysis of the SPI copolymers might occur, which could increase hydrogen crossover and result in the lower OCV. However, the exact reason is not yet known at this time, and further research is still ongoing. However, the slope of the polarization curve (the second linear Ohmic polarization region) for the new SPI membrane was much lower than Nafion® 112, which meant that the resistance of the SPI membrane was much lower than Nafion® 112 at 120 ◦ C and 50% RH. The reason is that water retention in the Nafion® membrane was reduced at high temperatures and low RHs, which resulted in its high resistance since water is very important for the proton transport. In contrast, the new soft segment-containing SPI membrane could still retain water because of its high water retention. Thus, its resistance was much lower. When the current density was higher than 0.32 A/cm2 , the new SPI membrane showed a higher voltage than Nafion® 112 because of the low resistance of this SPI membrane. For example, at the current density of 0.6 A/cm2 , Nafion® 112 showed a voltage of 0.44 V. However, the new SPI membrane showed a voltage of 0.5 V. Therefore, the new SPI membrane has the great potential for high temperature and low relative humidity applications. 4. Conclusions

Fig. 9. Polarization curves of the new SPI membrane (NTDA/DSDSA (95 mol%)/PEO-diamine (5 mol%)) and Nafion® 112 at 70 ◦ C and 80% RH (pressure = 2 atm). Data provided by Zhongwei Chen (the University of California, Riverside).

A series of new sulfonated polyimide (SPI) copolymers were successfully synthesized based on 4,4 -diaminostilbene2,2 -disulfonic acid (DSDSA), a new commercially available sulfonated diamine. Hydrophilic soft segments of poly(ethylene oxide) (PEO) were copolymerized into SPIs to not only increase water retention in the membranes particularly at high tempera-

84

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85

tures and low RHs, but also improve the membrane mechanical properties. The resulting new soft segment-containing SPI membranes were transparent, tough, ductile, mechanically strong, and thermally stable. They exhibited better conductivities than Nafion® 115 at high RH levels (>50%) at both 70 and 120 ◦ C. The SPI membrane containing 5 mol% PEO soft segments showed similar fuel cell performance as Nafion® 112 at 70 ◦ C and 80% RH but better fuel cell performance than Nafion® 112 when the current density was higher than 0.32 A/cm2 at 120 ◦ C and 50% RH, indicating potential applications of the new soft segment-containing SPI membranes, particularly for high temperature and low humidity PEMFCs. Acknowledgments We would like to thank Zhongwei Chen and Dr. Yushan Yan (the University of California, Riverside) for the membraneelectrode-assembly preparation and fuel cell performance measurements of our SPI membranes. We would also like to thank Kawaken Fine Chemicals Co., Ltd. (Tokyo, Japan) for giving us the diamine-terminated poly(ethylene oxide) (PEOdiamine, MW = 1000). We are grateful to the Ohio Department of Development (Wright Center of Innovation Grant No. 3420561) and the Ohio State University for the financial support of this work. References [1] F. Wang, M. Hickner, Y.S. Kim, T.A. Zawodzinski, J.E. McGrath, Direct polymerization of sulfonated poly(arylene ether sulfone) random (statistical) copolymers: candidates for new proton exchange membranes, J. Membr. Sci. 197 (2002) 231–242. [2] W.L. Harrison, M.A. Hickner, Y.S. Kim, J.E. McGrath, Poly(arylene ether sulfone) copolymers and related systems from disulfonated monomer building blocks: synthesis, characterization, and performance—a topical review, Fuel Cells 5 (2) (2005) 201–212. [3] H. Ghassemi, J.E. McGrath, T.A. Zawodzinski, Multiblock sulfonatedfluorinated poly(arylene ether)s for a proton exchange membrane fuel cell, Polymer 47 (2006) 4132–4139. [4] M. Gil, X. Ji, X. Li, H. Na, J.E. Hampsey, Y. Lu, Direct synthesis of sulfonated aromatic poly(ether ether ketone) proton exchange membranes for fuel cell applications, J. Membr. Sci. 234 (2004) 75–81. [5] P. Xing, G.P. Robertson, M.D. Guiver, S.D. Mikhailenko, S. Kaliaguine, Sulfonated poly(aryl ether ketone)s containing naphthalene moieties obtained by direct copolymerization as novel polymers for proton exchange membranes, J. Polym. Sci., Part A: Polym. Chem. 42 (2004) 2866–2876. [6] O. Savadogo, B. Xing, Hydrogen/oxygen polymer electrolyte membrane fuel cell (PEMFC) based on acid-doped polybenzimidazole (PBI), J. New Mater. Electrochem. Syst. 3 (2000) 345–349. [7] Q. Li, R. He, J.O. Jensen, N.J. Bjerrum, PBI-based polymer membranes for high temperature fuel cells—preparation, characterization and fuel cell demonstration, Fuel Cells 4 (3) (2004) 147–159. [8] L. Gubler, S.A. Gursel, G.G. Scherer, Radiation grafted membranes for polymer electrolyte fuel cells, Fuel Cells 5 (3) (2005) 317–335. [9] M. Shen, S. Roy, J.W. Kuhlmann, K. Scott, K. Lovell, J.A. Horsfall, Grafted polymer electrolyte membrane for direct methanol fuel cells, J. Membr. Sci. 251 (2005) 121–130. [10] S. Faure, N. Cornet, G. Gebel, R. Mercier, M. Pineri, B. Sillion, Sulfonated polyimides as novel proton exchange membranes for H2 /O2 fuel cells, in: Proceedings of the Second International Symposium on New Materials for Fuel Cell and Modern Battery Systems, Montreal, Canada, July 6–10, 1997, pp. 818–827.

[11] N. Gunduz, J.E. McGrath, Wholly aromatic five- and six-member ring polyimides containing pendant sulfonic acid functional groups, Polym. Prepr. 41 (2) (2000) 1565–1566. [12] N. Gunduz, J.E. McGrath, Synthesis and characterization of sulfonated polyimides, Polym. Prepr. 41 (1) (2000) 182–183. [13] H.K. Shobha, M. Sankarapandian, T.E. Glass, J.E. MaGrath, Sulfonated aromatic diamines as precursors for polyimides for proton exchange membranes, Polym. Prepr. 41 (2) (2000) 1298–1299. [14] B.R. Einsla, Y.T. Hong, Y.S. Kim, F. Wang, N. Gunduz, J.E. McGrath, Sulfonated naphthalene dianhydride based polyimide copolymers for proton-exchange-membrane fuel cells. I. Monomer and copolymer synthesis, J. Polym. Sci., Part A: Polym. Chem. 42 (2004) 862–874. [15] B.R. Einsla, Y.S. Kim, M.A. Hickner, Y.T. Hong, M.L. Hill, B.S. Pivovar, J.E. McGrath, Sulfonated naphthalene dianhydride based polyimide copolymers for proton-exchange-membrane fuel cells. II. Membrane properties and fuel cell performance, J. Membr. Sci. 255 (2005) 141–148. [16] J. Fang, X. Guo, S. Harada, T. Watari, K. Tanaka, H. Kita, K. Okamoto, Novel sulfonated polyimides as polyelectrolytes for fuel cell application. 1. Synthesis, proton conductivity, and water stability of polyimides from 4,4 diaminodiphenyl ether-2,2 -disulfonic acid, Macromolecules 35 (2002) 9022–9028. [17] X. Guo, J. Fang, T. Watari, K. Tanaka, H. Kita, K. Okamoto, Novel sulfonated polyimides as polyelectrolytes for fuel cell application. 2. Synthesis and proton conductivity of polyimides from 9,9 -bis(4aminophenyl)fluorene-2,7-disulfonic acid, Macromolecules 35 (2002) 6707–6713. [18] T. Watari, J. Fang, K. Tanaka, H. Kita, K. Okamoto, T. Hirano, Synthesis, water stability and proton conductivity of novel sulfonated polyimides from 4,4 -bis(4-aminophenoxy)biphenyl-3,3 -disulfonic acid, J. Membr. Sci. 230 (2004) 111–120. [19] X. Guo, J. Fang, K. Tanaka, H. Kita, K. Okamoto, Synthesis and properties of novel sulfonated polyimides from 2,2 -bis(4-aminophenoxy)biphenyl5,5 -disulfonic acid, J. Polym. Sci., Part A: Polym. Chem. 42 (2004) 1432–1440. [20] Y. Yin, J. Fang, H. Kita, K. Okamoto, Novel sulfoalkoxylated polyimide membrane for polymer electrolyte fuel cells, Chem. Lett. 32 (4) (2003) 328–329. [21] Y. Yin, J. Fang, T. Watari, K. Tanaka, H. Kita, K. Okamoto, Synthesis and properties of highly sulfonated proton conducting polyimides from bis(3sulfopropoxy)benzidine diamines, J. Mater. Chem. 14 (2004) 1062–1070. [22] Y. Yin, O. Yamada, Y. Suto, T. Mishima, K. Tanaka, H. Kita, K. Okamoto, Synthesis and characterization of proton-conducting copolyimides bearing pendant sulfonic acid groups, J. Polym. Sci., Part A: Polym. Chem. 43 (2005) 1545–1553. [23] Y. Yin, J. Fang, Y. Cui, K. Tanaka, H. Kita, K. Okamoto, Synthesis, proton conductivity and methanol permeability of a novel sulfonated polyimide from 3-(2 ,4 -diaminophenoxy)propane sulfonic acid, Polymer 44 (2003) 4509–4518. [24] Z. Hu, Y. Yin, S. Chen, O. Yamada, K. Tanaka, H. Kita, K. Okamoto, Synthesis and properties of novel sulfonated (co)polyimides bearing sulfonated aromatic pendant groups for PEFC applications, J. Polym. Sci., Part A: Polym. Chem. 44 (2006) 2862–2872. [25] S. Chen, Y. Yin, K. Tanaka, H. Kita, K. Okamoto, Synthesis and properties of novel side-chain-sulfonated polyimides from bis[4-(4-aminophenoxy)2-(3-sulfobenzoyl)]phenyl sulfone, Polymer 47 (2006) 2660–2669. [26] Y. Yin, Y. Suto, T. Sakabe, S. Chen, S. Hayashi, T. Mishima, O. Yamada, K. Tanaka, H. Kita, K. Okamoto, Water stability of sulfonated polyimide membranes, Macromolecules 39 (2006) 1189–1198. [27] K. Okamoto, Y. Yin, O. Yamada, Md.N. Islam, T. Honda, T. Mishima, Y. Suto, K. Tanaka, H. Kita, Methanol permeability and proton conductivity of sulfonated co-polyimide membranes, J. Membr. Sci. 258 (2005) 115–122. [28] O. Yamada, Y. Yin, K. Tanaka, H. Kita, K. Okamoto, Polymer electrolyte fuel cells based on main-chain-type sulfonated polyimides, Electrochim. Acta 50 (2005) 2655–2659. [29] Y. Yin, O. Yamada, S. Hayashi, K. Tanaka, H. Kita, K. Okamoto, Chemically modified proton-conducting membranes based on sulfonated polyimides: improved water stability and fuel-cell performance, J. Polym. Sci., Part A: Polym. Chem. 44 (2006) 3751–3762.

H. Bai, W.S.W. Ho / Journal of Membrane Science 313 (2008) 75–85 [30] Y. Yin, S. Hayashi, O. Yamada, H. Kita, K. Okamoto, Branched/crosslinked sulfonated polyimide membranes for polymer electrolyte fuel cells, Macromol. Rapid Commun. 26 (2005) 696–700. [31] P.D. Beattie, F.P. Orfino, V.I. Basura, K. Zychowska, J. Ding, C. Chuy, J. Schmeisser, S. Holdcroft, Ionic conductivity of proton exchange membranes, J. Electroanal. Chem. 503 (2001) 45–56. [32] W. Jang, C. Lee, S. Sundar, Y.G. Shul, H. Han, Thermal and hydrolytic stability of sulfonated polyimide membranes with varying chemical structure, Polym. Degrad. Stab. 90 (2005) 431–440.

85

[33] Y. Sone, P. Ekdunge, D. Simonsson, Proton conductivity of Nafion 117 as measured by a four-electrode AC impedance method, J. Electrochem. Soc. 143 (4) (1996) 1254–1259. [34] X. Ye, H. Bai, W.S.W. Ho, Synthesis and characterization of new sulfonated polyimides as proton-exchange membranes for fuel cells, J. Membr. Sci. 279 (2006) 570–577. [35] C. Perrot, G. Meyer, L. Gonon, G. Gebel, Ageing mechanisms of proton exchange membrane used in fuel cell applications, Fuel Cells 6 (1) (2006) 10–15.