Nafion®–polybenzimidazole (PBI) composite membranes for DMFC applications

Nafion®–polybenzimidazole (PBI) composite membranes for DMFC applications

Solid State Ionics 178 (2007) 581 – 585 Nafion® –polybenzimidazole (PBI) composite membranes for DMFC applications Alar A...

327KB Sizes 4 Downloads 92 Views

Solid State Ionics 178 (2007) 581 – 585

Nafion® –polybenzimidazole (PBI) composite membranes for DMFC applications Alar Ainla, Daniel Brandell ⁎ Department of Material Chemistry, Uppsala University, Box 538, SE-75121, Uppsala, Sweden Institute of Technology, Tartu University, Nooruse 1, 504 11, Tartu, Estonia Received 2 October 2006; received in revised form 4 December 2006; accepted 7 January 2007

Abstract Nafion®–PBI composites were prepared by diffusing synthesized PBI from solution phase into Nafion® membranes, using different concentrations and drying temperatures. In some cases, Nafion® was treated with diethyl amine to screen the –SO3H groups and thereby avoid the strong acid–base interactions between the polymers during diffusion. The presence of PBI in the membranes was characterized with FT–IR spectroscopy. The performance of the membranes was studied by in-plane conductivity and methanol permeability. The performance ratio (the ratio between conductivity and methanol permeability compared to Nafion®) increased by up to 50% for the composite membranes compared to Nafion®. © 2007 Elsevier B.V. All rights reserved. Keywords: DMFC; Nafion®; PBI; Composite membranes; Proton conductivity; Methanol permeability

1. Introduction Alongside the extensive current research on Proton Exchange Membrane Fuel Cells (PEMFCs), Direct Methanol Fuel Cell (DMFC) research is also attracting considerable attention [1]. Since methanol is a liquid fuel and therefore easier to handle, DMFCs are promising power sources for portable devices. However, methanol permeation through the archetypal Nafion® proton exchange membrane (methanol cross-over) is a problem which is still limiting its wider commercialization. In order to overcome this obstacle, various strategies has been employed to solve the problem: synthesis of new polymers (aromatic backbone polymers [2] or polybenzimidazoles [3]) or the improvement of existing ones through the preparation of composites with some organic or inorganic additives, e.g., Nafion®/inorganic acid [4], Nafion®/polyvinyl alcohol [5], Nafion®/polyvinylidene fluoride

⁎ Corresponding author. Department of Material Chemistry, Uppsala University, Box 538, SE-75121, Uppsala, Sweden. Tel.: +46 18 4713701; fax: +46 18 513548. E-mail address: [email protected] (D. Brandell). 0167-2738/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2007.01.014

[6], Nafion®/polypyrrole [7–11], Nafion®/Polyvinylimidazole [12] and Nafion®/silica hybrid [13]) . Operating under harsh condition, fuel–cell membrane materials have to be hydrolytically and oxidatively stable, and only few polymers can meet these requirements. One of the most well known materials is Nafion® [14], which combines superior proton conductivity with stability. Polybenzimidazole (PBI) is another attractive material for DMFC applications due to its high thermal and chemical stability [15]. However, PBI alone exhibits too low proton conductivity, with the result that it must be loaded with acid before use. In this paper, we discuss the synthesis and performance of Nafion®–PBI composites in a DMFC context. The essential idea is to form a PBI sub-network in Nafion® to block the methanol cross-over in the membrane, while still retaining high proton conductivity due to non-vehicular proton transport over the sub-network. Our approach has been inspired by Hobson et al. [16], who worked on PBI coated Nafion® membranes, initially proposed by Deluga et al. [17]. Different synthesis routes [18–20] for PBI have been tested, and the one giving the highest yield was used for further steps [18]. The PBI was then incorporated into Nafion® and the performance of the resulting membranes evaluated.


A. Ainla, D. Brandell / Solid State Ionics 178 (2007) 581–585

2. Experimental

Table 1 Preparation of the Nafion®–PBI composite membranes

2.1. Materials


Preparation procedure

M1 M2

Original Nafion® Membrane dipped for 15 min into a 0.1% PBI solution, after which it was dried briefly. The procedure was repeated 4 times, after which the membrane was dried at 200 °C for 1 h Membrane was immersed in a solution of diethyl amine (∼ 50% at DMSO) for 4 h, followed by a 1% PBI solution for 2.5 h. Finally, the membrane was dried at 100 °C for 2 h Membrane dipped for 15 min into a 0.1% PBI solution, after which it was dried at 200 °C for 5 min, followed by immersion in a 1% Nafion® solution (in DMSO) for 5 min, after which the membrane was dried at 200 °C for 5 min. The entire cycle was repeated 4 times before drying at 200 °C for 1 h Membrane was dipped for 60 min into a 0.1% PBI solution, after which it was briefly dried. This cycle was repeated four times, after which the membrane was dried at 200 °C for 1 h Membrane immersed into a solution of diethyl amine (∼ 50% at DMSO) for 5 h, followed by a 0.1% PBI solution for 1.5 h. Finally, the membrane was dried at 100 °C for 20 h Membrane immersed in a mixture of solutions of PBI, Nafion® and diethyl amine (5 mL 1% Nafion® solution, 5.5 mL diethyl amine, 5 mL 0.1% PBI solution) for 15 min; then dried at 100 °C for 3 h and at 200 °C for 1 h Membrane dipped for a 15 °min into 1% PBI solution, after which it was dried at 100 °C for 5 h

Diaminobenzidine (DAB) (99%, Aldrich), isophthalic acid (99%, Aldrich), polyphosphoric acid (PPA) (reagent grade, 115%), ammonia solution (purum, 24% NH3 in H2O, Fluka), diethyl amine (Fluka), dimethyl sulfoxide (DMSO) (99%, Aldrich) were all purchased from Sigma–Aldrich and used as received. Nafion® 117 was purchased from E.I. du Pont de Nemours and purified using a procedure used by Ludvigsson et al. [21] whereby the membranes were first boiled for 1 h first in 10% H2O2, then for 1 h in water, followed by 1 h in 0.2 M H2SO4 and finally for 1 h in water.




2.2. Synthesis of polybenzimidazole (PBI) 1 g DAB (1 in Fig. 1), 776 mg isophthalic acid 2 and 20 g PPA were weighed before mortaring and mixing thoroughly to obtain a homogeneous dispersion. The dispersion was loaded into a round-bottom flask fitted with a magnetic stirrer and septum in a slow flow of N2-gas. After stirring for 17 h at 200 °C, the reaction mixture turned dark red and became viscous. The reaction was quenched by pouring the hot solution into a large amount of cold water. The fibers formed were stirred for 48 h in water and 7 h in 300 mL of a 10% NH4OH solution, and then dried in an oven at 100 °C for 10 h. The 80% product 3 was obtained as dark brown solid. The presence of PBI was shown by NMR and FT–IR. 2.3. Preparation of composite membranes Prior to composite preparation, the Nafion® membranes were dried for at least 24 h at 105 °C to remove free water. PBI solution was prepared by fitting a round-bottom flask to a magnetic stirrer and a reflux condenser. The flask was charged with DMSO and 1 wt. % (relative to DMSO) ground PBI. The mixture was refluxed until all PBI was dissolved (∼ 1 h); 1 wt.% PBI solution was obtained as an orange liquid. Composite Nafion®–PBI membranes were prepared by immersing Nafion® 117 films with an area of 10 cm2 into PBI solutions. Different solution concentrations, immersion times, drying temperatures, and Nafion® types (acidic or complexed with diethyl amine) were used (Table 1). Before characterization, all membranes were cleaned using the method described in Section 2.1. The cleaning also served as a preliminary stability test. 2.4. Characterization NMR spectra of PBI were recorded on a JEOL Eclipse + 400 (1H 400 MHz; 13C 100 MHz) in deuterated DMSO with the




solvent proton as an internal standard. IR spectra of PBI and composite membranes were recorded on a Perkin Elmer Spectrum One FT–IR Spectrometer with an ATR module. The spectral range was 650–4000 cm− 1. The conductivity of the membrane was measured at room-temperature (20 °C) with a four-point probe and a Solartron analytical impedance spectrometer Impedance/Gain–Phase Analyzer 1260. Before measurements all membranes were kept in water to maintain maximum swelling. The impedance was constant and real in the frequency range 1 kHz–1 MHz, and could be interpreted directly as the resistance of the membrane. The methanol permeability was measured using a diffusion cell [22,23] and a Raman spectrometer. The diffusion cell consisted of two cavities: one was filled with MeOH solution and the other with pure water. Slow stirring was applied. The MeOH concentration in the water cavity was monitored with a Renishaw Ramascope Raman spectrometer, where intensities of C–H vibrations (2945 cm − 1 , 2836 cm − 1 ) and O–H vibrations (3400 cm− 1) were used for concentration evaluation. The membrane thickness was measured with Mitutoyo Absolute micrometer.

Fig. 1. Synthesis of PBI.

A. Ainla, D. Brandell / Solid State Ionics 178 (2007) 581–585


3. Results and discussion 3.1. Synthesis of PBI and membrane preparation The synthesized PBI was first characterized. The proton chemical shifts of PBI were: 1H NMR (400MHz, (CD3)2SO) δH 13.28, 9.16, 8.34, 7.84, 7.80, 7.69, 7.63. End-group analysis was used to estimate the average chain length, which was found to be approximately 40 repeat units. The strongest characteristic FT–IR absorption peaks were at 801 cm− 1, 1286 cm− 1 and 1443 cm− 1. During the cleaning, all membranes, except M4, maintained their appearance. Membrane M4 lost some material, indicating instability in the Nafion®–PBI–Nafion® composite structure. The membranes could be divided visually into three groups: colourless, e.g. the intrinsic Nafion® (M1, M3, M7), slightly yellowish with interference effects, indicating a strongly inhomogeneous distribution of the PBI additive (M2, M4, M5, M8), and finally a homogeneous pale yellow membrane, showing no interference (M6). ATR FT–IR was used to confirm the presence of PBI in the membranes (Figs. 2 and 3). The characteristic PBI absorption peak at 1443 cm− 1 was used to establish the presence of PBI. Untreated Nafion® had no absorption peak in this range, whereas the composites had. The strongest absorption – and therefore the highest PBI content – corresponds to membranes M2, M5, M6 and M8, while M3 and M7 showed slightly smaller peaks. 3.2. Membrane performance Membrane performance was evaluated as the ratio between conductivity and methanol permeability (Fig. 4). It is noticeable that all membranes which exhibited a change in their visual appearance (a yellowish colour or an interference effect on the surface) gave better performance (Table 2). The PBI IR absorption also correlates well with the improved performance. This all constitutes a clear indication of the presence of PBI, either on the surface or within the membranes, which can reduce

Fig. 3. Reference spectrum of the characteristic PBI phenyl benzimidazole peak (pure PBI).

the methanol permeation considerably. The only exception was M4; on the other hand, this displayed poor mechanical stability. Generally, use of diethyl amine in the synthesis procedure of the composite reduced the methanol cross-over significantly. This is an indication of good PBI blockage, but might also be due to remanent diethyl amine. However, the diethyl amine treatment also caused the conductivity to drop, which gave poor overall performance. The reason for this could either be the incomplete removal of diethyl amine during purification after preparation, or different distributions of PBI in the membrane. The first hypothesis is supported by the fact that diethyl amine– Nafion® complexes can be relatively stable due to the strong acid–base interaction between them. We have observed similar interactions when mixing PBI and Nafion® solutions, leading to cross-linking and gel formation. Nevertheless, we could not detect any specific IR signal arising from the presence of diethyl amine. Only a slight shift (10 cm − 1 ) towards higher wavenumber was observed for the PBI peak at 1443 cm− 1. On the other hand, even a very small amount of diethyl amine could have a strong effect on conductivity, since it binds the protons and blocks the –SO3H proton-jumping sites.

Fig. 2. IR spectra of different Nafion®–PBI composites.


A. Ainla, D. Brandell / Solid State Ionics 178 (2007) 581–585

Fig. 4. Comparison of characteristics of Nafion®–PBI composite membranes.

The quantitative PBI distribution could not be assessed on the basis of our experimental data. However, when using diethyl amine, no significantly larger overall PBI content could be observed with IR compared to that for the membranes prepared without diethyl amine. This is somewhat surprising, since one would expect that blockage of the sulphonate groups would lead to better PBI penetration into the Nafion®. However, in no case where diethyl amine was used, could we observe any interference effect on the surface of the composite. This indicates that, when Nafion® is not treated with diethyl amine, PBI forms a thin layer bound to its surface, while it otherwise diffuses into the membrane to form a more homogenous composite. This could be the reason for the better methanol blockage. The most exceptional membrane in this respect was M6 which, from visual observations (it had a yellow colour without any interference), seemed to acquire the highest and most homogeneous PBI content inside Nafion®. M6 showed improved performance, although the conductivity was almost half that of untreated Nafion®. It is still unclear why higher concentration and longer overall exposure time cause lower PBI content in the final membrane (M6 versus M3). A possible explanation is that higher concentration causes PBI to form agglomerates which are less able to diffuse into the Nafion® membrane. This can also be an Table 2 Experimental results and comparison with Nafion® (M1) Membrane Conductivity relative MeOH cross-over Relative performance to Nafion® (%) relative to Nafion®(%) ratio (%)

M1 M2 M3 M4 M5 M6 M7 M8




100 105 42 91 92 55 39 117

100 70 58 83 69 44 43 97

100 151 73 110 133 125 90 121

Specific conductivity of pure Nafion® (M1) was found to be 7.15(38) S/m (similar to previous findings [24]) and the methanol diffusion constant was 0.01583 mm2/s. Uncertainties in performance is in the range 2–8%, except for M4 which was 15%.

effect of PBI chain length, i.e., longer chains diffuse slower than shorter ones. In agreement with Hobson et al. [16], we also find a decrease in methanol permeation by N 50% and, in some cases, also a slight increase in conductivity. They found that the best preparation strategy was screen printing with minimal PBI penetration into the Nafion® film. Even though their preparation methods were different, the resulting membranes should resemble most closely our M2, M5 and M8 membranes, which exhibited interference, indicating a surface layer of PBI. The difference between our and Hobson's materials can be explained by different preparation methods (solvents, temperatures, etc.) which cause differences in membrane swelling and PBI diffusion. However, on the basis of our membrane M6, we may also suggest that, despite some conductivity drop, a more dispersed distribution of PBI could also lead to slight improvements in performance. Composite preparation through recasting the polymer mixture could be an interesting approach in further studies, since it would probably yield a more homogeneous PBI distribution in the membrane. Recasting from solutions also facilitates the use of arbitrary polymer mixture ratios in the composite. To obtain homogeneous PBI–Nafion® solutions, our observations suggest treatment of Nafion® with diethyl amine to avoid acid–base reactions between PBI and Nafion®, leading to gel-formation. Our first efforts following this approach have not been evaluated here, since the cast composite membranes were found to be too uneven in thickness, even after hot-pressing. 4. Conclusions Nafion®–PBI composite membranes have been prepared and their performance evaluated. Membranes which appeared to contain PBI also exhibited improved performance. Interestingly, not only was the MeOH cross-over reduced but, in some cases, an unexpected conductivity increase was observed. The maximum overall performance of the composite membranes investigated was 150% of that of Nafion®. In summary; we can conclude that the Nafion®–PBI membranes have the potential to improve the performance of DMFCs.

A. Ainla, D. Brandell / Solid State Ionics 178 (2007) 581–585

To get a better overview of how the different parameters varied in the synthesis routines effect the membrane performance, a more thorough investigation is needed. We also plan to investigate the composites in a prototype fuel cell. Acknowledgements We wish to acknowledge Professors Josh Thomas and Michel Armand for many fruitful discussions. We also gratefully acknowledge the EU-FP5 “Portapower” project, the Nordic Fuel Cell Network and the Archimedes Foundation for their financial support. References [1] N.W. DeLuca, Y.A. Elabd, J. Polym. Sci. B. 44 (2006) 2201. [2] B. Smitha, S. Shridhar, A.A. Khan, J. Memb. Sci. 259 (2005) 10. [3] J.A. Asensio, S. Borrós, P. Gómez-Romero, J. Polym. Sci. A. 107 (2003) 7827. [4] P. Dimitrova, K.A. Friedrich, U. Stimming, B. Vogt, Solid State Ionics 150 (2002) 115. [5] Z.-G. Shao, I.-M. Hsing, Electrochem. Solid-State Letters 5 (2002) A185. [6] M.-K. Song, Y.-T. Kim, J.M. Fenton, H.R. Kunz, H.-W. Rhee, J. Power Sources 117 (2003) 14. [7] E.B. Easton, B.L. Langsdorf, J.A. Hughes, J. Sultan, Z. Qi, A. Kaufman, P.G. Pickup, J. Electrochem. Soc. 150 (2003) C735.


[8] M.A. Smit, A.L. Ocampo, M.A. Espinosa-Medina, P.J. Sebastian, J. Power Sources 124 (2003) 59. [9] B.L. Langsdorf, J. Sultan, P.G. Pickup, J. Phys. Chem. B. 107 (2003) 8412. [10] B.L. Langsdorf, B.J. MacLean, J.E. Halfyard, J.A. Hughes, P.G. Pickup, J. Phys. Chem. B. 107 (2003) 2480. [11] N. Jia, M.C. Lefebure, J. Halfyard, Z. Qi, P.G. Pickup, Electrochem. Soc. Solid-State Letters. 3 (2000) 529. [12] B. Bae, H. Yong Ha, D. Kim, J Electrochem. Soc. 153 (2005) A1366. [13] N. Miyake, J.S. Wainright, R.F. Savinell, J. Electrochem. Soc. 148 (2001) A905. [14] W.G. Grot, Symposia 82 (1994) 161. [15] J.-T. Wang, J.S. Wainright, R.F. Savinell, M. Litt, J. Appl. Electrochem. 26 (1996) 751. [16] L.J. Hobson, Y. Nakano, H. Ozu, S. Hayase, J. Power Sources 104 (2002) 79. [17] G. Deluga, B.S. Pivovar, in: O. Savadogo (Ed.), Proceedings of the 3rd International Symposium on New Materials for Electrochemical Systems, Montreal, Que., Canada, July 4–8 1999, p. 132. [18] L. Xiao, H. Zhang, E. Scanlon, L.S. Ramanathan, E.-W. Choe, D. Rogers, T. Apple, B.C. Benicewicz, Chem. Mater. 17 (2005) 5328. [19] H.-J. Kim, S.J. An, J.-Y. Kim, J.K. Moon, S.Y. Cho, Y.C. Eun, H.-K. Yoon, Y. Park, H.J. Kweon, E.-M. Shin, Macromol. Rapid Commun. 25 (2004) 1410. [20] F.L. Hedberg, C.S. Marvel, J. Polym. Sci. Polym. Chem. Ed. 12 (1974) 1823. [21] M. Ludvigsson, J. Lindgren, J. Tegenfeldt, J. Mater. Chem. 11 (2001) 1269. [22] Y. Woo, S.Y. Oh, Y.S. Kang, B. Jung, J. Memb. Sci. 220 (2003) 31. [23] B. Baradie, J.P. Dodelet, D. Guay, J. Electroanal. Chem. 489 (2000) 101. [24] P. Dimitrova, K.A. Friedrich, U. Stimming, B. Vogt, Solid State Ionics 150 (2002) 115.