Electrochemistry Communications 8 (2006) 1125–1131 www.elsevier.com/locate/elecom
Nanoporous composite, low cost, protonic membranes for direct methanol fuel cells F. Croce a, J. Hassoun b, C. Tizzani b, B. Scrosati a
b,*
Dipartimento di Scienze del Farmaco, Universita` ‘d’Annunzio’, Via dei Vestini 31, 66013 Chieti, Italy b Dipartimento di Chimica, Universita` ‘La Sapienza’, P.le A. Moro, 5 00185 Roma, Italy Received 28 March 2006; received in revised form 3 May 2006; accepted 10 May 2006 Available online 13 June 2006
Abstract New types of nanoporous, composite membranes, prepared by readapting a procedure successfully used in the lithium battery technology, are here described and evaluated. The membranes are based on a polyvinylidene fluoride polymer matrix containing dispersed SiO2 ceramic powder at nanoparticles size. The unique preparation method confers an extended porosity which favors the swelling of the acid solutions to provide a high proton conductivity. The properties of these membranes can be monitored by properly controlling the amount of the dispersed ceramic filler, to finally obtain samples which combine good conductivity with low methanol permeability. Due to these features, the selected membrane samples can be profitably used as separators in ambient temperature direct methanol fuel cells, DMFCs. Ó 2006 Elsevier B.V. All rights reserved. Keywords: Polymer; Membrane; Composite; Proton; Fuel cells
1. Introduction Vehicle transportation represents a significant portion of world energy consumption and contribute considerably to atmospheric pollution. Thus, the development of zero or controlled emission cars is an important goal from both economical and environmental point of view. Direct methanol fuel cells, DMFCs, are suitable systems for power generation in electrotraction [1]. Similar to internal combustion engines, also DMFCs utilize liquid fuel to deliver continuous power with much higher utilization efficiency and intrinsically lower polluting emission [2]. However, in order to be competitive within the transportation market, the proposed DMFCs must be reasonably cheap and capable of delivering high power. At the present, there are still few challenging problems to be solved for reaching this goal. Among the major issues, those associ-
*
Corresponding author. Tel.: +39 6 4462866; fax: +39 6 491769. E-mail address:
[email protected] (B. Scrosati).
1388-2481/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.elecom.2006.05.006
ated to the electrolyte membrane play a crucial role. In fact, the ideal properties of a DMFC membrane are a high proton conductivity associated to a low methanol crossover and a low cost. These properties are not totally met by the commonly used membranes which are mainly of the perfluorosulphonate type, a typical example of which is NafionÒ. These membranes suffer by a low thermal stability and a low methanol selectivity. In addition, they are the components which, together with the noble metals in the catalysts, mostly contribute to the cost of DMFCs. Thus, great R&D effort is presently devoted for a breakthrough in the development of membranes alternative to Nafion in terms of cost and methanol permeability [3]. Following this trend, we have directed our attention to new types of porous membranes prepared by readapting synthesis procedures proved in the nineties by Bellcore Laboratories to be successful in the lithium battery technology [4]. These involve first the cast of a slurry formed by a poly(vinylidene) fluoride-chloro tetrafluoro ethylene, PVdF-CTFE, copolymer with dispersed ceramic filler and with the addition of dibutylphthalate, DBP. The DBP is
1126
F. Croce et al. / Electrochemistry Communications 8 (2006) 1125–1131
then removed by diethyl ether extraction to promote flexibility and porosity. This porous composite membrane is finally activated by swelling it with acid aqueous solutions. Clearly, these membranes have a structure and a conductivity mechanism totally different than those of Nafion: in the latter the proton transport is assisted by the sulfonate groups while in the former is provided by the acid solution entrapped in the polymer matrix [5]. In addition, these PVdF-based porous membranes have a prospected cost which is order of magnitude lower than that of Nafion. The applicability of these membranes as fuel cells separators has been originally proved by Peled and co-workers [6]. In this work we confirm their practical relevance by demonstrating the high conductivity, the selectivity towards methanol and the good response when tested in laboratory DMFCs prototypes.
where Ap is the total surface area and W the weight of the sample. The total porosity, Ptot (%) was given by the product: P tot ¼ qB V tot 100 where qB is the bulk density in g/cm3 and Vtot is the cumulative volume in cm3/g. The morphology of the samples was investigated by Scanning Electron Microscopy, SEM, using a LEO 1450VP instrument. All the membrane samples were activated by swelling them in 6 M H2SO4 aqueous solution. To achieve complete swelling, the samples were immersed in the acid solution for about 3–4 h at 80 °C. The swelling was monitored in terms of percentage of weight change, DP, determined by the relation: DP ¼ ðP f P i Þ=P i 100
2. Experimental The nanoporous membranes developed in this work are based on matrices formed by a PVdF-CTFE copolymer with dispersed SiO2 ceramics. The synthesis procedure involved the following steps. First, the PVdF-CTFE copolymer (SolefÒ 32008) was intimately mixed in a ball miller with the required amount of the ceramic powder (SiO2 fumed silica, 99.8% Aldrich, particle size 14 nm, surface area 200 m2/g ± 25 m2/g, Catalog No. S5505). Separately, a plasticizer, porogen component, i.e. dibutylphthalate (DBP, Aldrich) was dissolved in acetone. The resulting solution was then added to the PVdF-CTFESiO2 mixed powder and magnetically stirred for 16 h at room temperature to achieve complete dissolution of PVdF-CTFE and of DBF, such as to obtain a homogeneous slurry with dispersed SiO2. The slurry was then poured on a glass substrate and cast into a 100 lm thick film by Doctor Blade. After drying the film membranes were repeatedly washed with diethyl ether to extract DBP, to finally produce flexible and highly porous membranes. Several membrane samples, differing by the SiO2 content, were prepared. Table 1 lists the samples studied in this work and their related composition. The pore size and their distribution in the membranes was determined by mercury porosimetry (Porosimeter 2000, FINSON Instruments). The pores specific surface area, As (m2 g1) was determined by using the equation: As ¼ Ap =W
Table 1 PVdF-CTFE composite membranes examined in this work Sample
SiO2 content (Wt%)
PVdF-0 PVdF-10 PVdF-20 PVdF-40 PVdF-60
0 10 20 40 60
where Pi is the initial weight of the membrane before immersion in the acid solution and Pf is the final weight at the end of the immersion period. The through-plane conductivity of the swelled membranes was obtained by impedance spectroscopy run on symmetric Pt/membrane sample/Pt cells in a 1 Hz–1 MHz frequency range using a computer controlled Solarton 1260 FRA. The thermal gravimetric analysis, TGA, was performed using a Perkin–Elmer instrument at a scan rate of 5 °C min1 in the 25–180 °C temperature range. For the fuel cell tests, a membrane electrode assembly, MEA, was fabricated following a procedure similar to that adopted for the preparation of the electrolyte membrane samples. Accordingly, a monolayer was obtained by first intimately mixing a blend of Super P carbon and Pt black (6:4 weight ratio) with PVdF powder (6020 Solvay-Solef Binder) in a 20% total weight ratio. The blend was dispersed in acetone and added with a Teflon emulsion in a 1:1 weight ratio. The final resulting suspension was mixed with DBF in a 1:2 weight ratio. The slurry was dried for 15 min at 70 °C. This procedure gave a highly viscous paste which was pressed at 70 °C and 1 ton/cm2 to obtain an homogeneous, thin, membrane. This membrane was finally washed with diethyl ether to extract DBF and thus, promote porosity. The Pt loading in this porous, electrode membrane was 4 mg/cm2. Two of these monolayer electrode membranes, one at the anode side and the other at the cathode side, were combined with the selected electrolyte membrane and pressed together to obtain the final MEA. The current–voltage curves of the cells were obtained by a PAR 273A potentiostat under air flux at the cathode side and a 2 M methanol aqueous solution flux, acidified by H2SO4, at the anode side. The cell tests were run at room temperature. The methanol cross over was determined using a U-shaped cell having two compartments separated by the given membrane sample. One compartment was filled by water and the other by a methanol aqueous solution. At
F. Croce et al. / Electrochemistry Communications 8 (2006) 1125–1131
fixed time intervals, samples at the water side were analyzed by gas chromatography to monitor the methanol crossing through the membrane. The chromatographic stationary phase was polyethylene glycol (Carbowax).66.
1127 16 PVdF-0
PVdF-40
12
PVdF-20
10
3. Results and discussion
8 6
100
0
30
Porosity / (nm)
600
4300
20000
2
100000
600000
4
1,00E+07
Fig. 1 shows scanning electron microscopy, SEM, images of PVdF-0 (Fig. 1A) and of PVdF-10 (Fig. 1B) samples. The high and uniformly distributed porosity is clearly visible in both samples. The presence of this extended porosity, involving a sequence of empty cavities of nanometric size (198 nm), is essential for favouring the entrapment of the acid liquid solution. In addition, as it will be discussed in a following part of this paper, the low dimension of the pores prevents the diffusion of the liquid phase across the membrane, this being beneficial in terms of reducing methanol crossover. Apparently, there are not major differences in the morphology of the ceramic-free PVdF-0 sample and that of the composite PVdF-10 sample. However, data obtained by mercury porosimetry reveal that the dispersion of silica has a finite influence on both the dimension and the distribution of the pores. This is clearly shown by Table 2, which lists the characteristics of the various samples studied in this work. It is evident that the presence of the silica filler does increase the dimension of the pores and of the specific surface area. From Fig. 2, which reports the relative distribu-
Relative Distribution / %
14 PVdF-10
Fig. 2. Relative distribution of the porosity in function of the size of the pores in various PVdF-based, composite membrane samples. For sample identification see Table 1.
tion of the porosity in function of the size of the pores, one sees that the ceramic filler also promotes a homogeneous distribution of the pores. Obviously, a large porosity favours liquid adsorption: this is indeed confirmed by Fig. 3 which illustrates the percentage of swelling of an aqueous acid solution in the hosting membrane as function of the silica content. A large swelling is welcome in terms of conductivity (see later); however, it is important to point out that an increase in porosity affects also the bulk density (see Table 2) and
Fig. 1. SEM pictures of PVdF-0 (A) and of PVdF-10(B) membrane samples.
Table 2 Characteristics of PVdF-based, composite membrane samples determined by mercury porosimetry Sample
Total cumulative volume (mm3/g)
Specific surface area (m2/g)
Pore radius average (nm)
Bulk density (g/cm3)
Sample total porosity (%)
PVdF-0 PVdF-10 PVdF-20 PVdF-40
28.18 87.84 150.22 656.96
6.09 16.59 26.81 64.65
40 40 170 590
1.71 1.49 1.29 0.76
4.81 13.08 19.37 49.92
1128
F. Croce et al. / Electrochemistry Communications 8 (2006) 1125–1131
100
250
A
Linear Fit
200
80 Weight %
Average Swelling, wt%
300
150 100
60 PVdF-0 PVdF-30 PVdF-60
50 0
10
20
30 40 wt% SiO2
50
40
60
0
Fig. 3. Average swelling of an aqueous sulphuric acid solution as a function of the SiO2 content in the membrane samples.
50
100 150 Temperature / ˚C
200
0 B
Derivate weight %
-10
-20
-30 PVdF-0 PVdF-30 PVdF-60
-40 40
80
120
160
200
Temperature / ˚C Fig. 4. Thermal gravimetric analysis of silica-free and of silica-added membrane samples swelled by an aqueous acid solution (A) and related derivate curve (B). For sample identification see Table 1.
-1
Conductivity / Scm-1
this lowers the mechanical properties of the membranes. Therefore, the final choice of the membrane for practical application must necessarily be based on a compromise. This aspect will be further discussed in the last part of this paper. Fig. 4A shows the thermal gravimetric analysis, TGA, of a silica-free, PVdF-0 sample in comparison with that of silica-added, PVdF-30 and PVdF-60 samples, all swelled by a H2SO4 solution. The plots reveal that all samples experience weight losses, this being associated to the release of the liquid phase. The percentage of loss increases for the membranes having a progressively increasing silica content, since the filler favours liquid uptake, see Fig. 3. From the TGA derivate plot of Fig. 4B, the velocity of the process of the liquid release can be estimated. The velocity reaches its highest value in correspondence of the maxima, where a change of slope does occur. The figure shows that an increase in silica content reflects in a shift toward higher temperature of the peak of the related membrane. This finally confirms that the dispersed silica not only favours a high swelling of the membrane but also appears to contribute significantly to hinder an apparent mass-transport-limited weight loss of the liquid phase entrapped within the polymer matrix. The ceramic filler, by favouring liquid adsorption, has a key role in promoting the proton conductivity in the composite membranes. Indeed, considering that these membranes transport by a ‘‘free acid’’ mechanism, it is expected that to an increase in liquid content should correspond an increase in conductivity. This is indeed the case, as shown by Fig. 5 which reports the conductivity in function of silica content in the tested membrane samples. The plots reflect the trends observed in Fig. 3. To be noted that the conductivity increases of about one order of magnitude passing from sample PVdF-0 to sample PVdF-60. This conductivity enhancement is associated to the dispersion of the ceramic filler which, in virtue of its hydrophilic properties and extended surface area, is capable to adsorb the acid solution and to retain it into the
10
-2
10
-3
10
0
10
20
30
40
50
60
% SiO2 Fig. 5. Conductivity in function of silica content for membranes swelled with a H2SO4 aqueous solution.
polymer matrix. This combined effect favours the proton transport in the membranes which, as already pointed out, is associated to the liquid phase trapped in the pores.
F. Croce et al. / Electrochemistry Communications 8 (2006) 1125–1131
Fig. 6A shows the time evolution of the conductivity of various membrane samples at 25 °C and Fig. 6B the related Arrhenius plots. For the best membranes, the room temperature conductivity assumes values of the order of 102 S cm1 and remains unchanged for various days and this demonstrates that at this temperature the membranes are stable with no appreciable release of the liquid phase. The Arrhenius plots of Fig. 6B reveals some decays in conductivity for the silica-free, PVdF-0 sample at temperatures exceeding 70 °C, thus showing that the liquid retention cannot be assured beyond this temperature levels. The result is in agreement with that obtained by TGA which indeed demonstrated weight losses at progressively increasing temperatures (see Fig. 4B). Finally, Fig. 6B further evidences the role of the dispersed ceramic filler in improving and stabilizing the conductivity of the membranes. On the basis of these results, one may conclude that composite membranes at medium–high ceramic content,
1129
are the most appropriate for application in fuel cells designed for operation in the 60–70 °C temperature range. To confirm this prevision, we have examined the response of laboratory prototype fuel cells using a series of PVdF-based composite membranes as electrolyte separators. Fig. 7A shows the current–voltage curves of laboratory type DMFCs obtained at room temperature. To assure electrolyte/electrode compatibility, these cells were assembled using monolayer, PVdF-based electrodes, see experimental part. Fig. 7A again evidences the key role of the ceramic filler in retaining the liquid component into the polymer matrix. In fact, the cell which uses a membrane with a low ceramic content, i.e. 10% of SiO2, shows very a poor performance (see inset), probably because of a fast leaching out of the acid solution. On the contrary, cells based on higher ceramic contents show a much better response, i.e. a power density of about 1.4 mW cm2 and currents of the order of 10–20 mA cm2 are delivered by the cell using a PVdF-40 sample. These values are appreciable for cells operating at room temperature and in fact,
A -2
0.8
Cell voltage / V
0.6
PVdF-0 PVdF-10
10
0.5
0.20 PVdF-10 0.15 0.10 0.05
0.1
0.00 0.4 -2
Current Density / mA cm
2
4
6
8
10
12
1.0
0.3 0.2
0.5
0
5
Temperature / ˚C 80
70
60
50
40
30
20
2.5 Nafion 117 membrane
0.7
Cell Voltage / V
-1
Conductivity / Scm
1.5
0.4 1.0
0.3 0.2
PVdF-0
0.5
0.1
PVdF-10
0.0
PVdF-40
10
0.5
0
-4
2,8
2,9
3,0
3,1
3,2
3,3
3,4
-1
1000 / T (K ) Fig. 6. Conductivity of membrane samples swelled by H2SO4 aqueous solution as function of time at 25 °C (A) and as function of temperature (B). For sample identification see Table 1.
5
10 15 20 Current Density / mA cm-2
Power Density / mW cm-2
-3
B
2.0
0.6 -2
0.0 25
10 15 20 Current Density / mA cm-2
0.8
B
10
1.5
0.4
0.0
Time / days
10
2.0
0.1
0
10
0.3
PVdF-10 PVdF-30 PVdF-40
PVdF-40
-3
-1
0.2
2.5 A
Power Density / mW cm-2
Conductivity / Scm
-1
0.7
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.0
Power Density / mW cm-2
Cell voltage /V
10
0.0 25
Fig. 7. Current–voltage and power density curves at room temperature of a laboratory DMFC using composite PVdF-based membranes activated by H2SO4 solution (A) and Nafion 117 (B) as the electrolyte. The inset of Fig. 7 enlarges the response of a cells based on sample PVdF-10. Sample identification see Table 1.
F. Croce et al. / Electrochemistry Communications 8 (2006) 1125–1131
they compare well with those of similar cells using Nafion 117 as the electrolyte, see Fig. 7B. Although the PVdFbased composite membranes themselves and the whole cell structure clearly require further optimization, the values of Fig. 7A and their comparison with those of Fig. 7B, are quite promising in suggesting that these new types of membranes are indeed suitable for application in low-rate, room temperature DMFCs. A key parameter for rating membranes to be proposed for DMFC application is the value of the methanol crossover. The latter is indeed a key aspect in fuel cell technology, since it affects various cell parameters, including energy efficiency. Fig. 8 compares the methanol crossover level of various membrane samples investigated in this work. Clearly, the crossover increases for the membranes having progressively higher ceramic content. This is expected since the increase in ceramic content reflects in an increase in porosity, see Fig. 2, which in terms favours membrane permeability. Accordingly, the lower permeability level is obtained by the membrane having the lower content of SiO2. Fig. 8 also reports the methanol permeability of a typical Nafion 117 membrane. Clearly, the methanol crossover of the low silica containing PVdF-based composite membranes is consistently lower than that observed in Nafion. To be noticed that this favourable difference is kept for membranes having 10% or even 20% SiO2 content. The above stressed effect of the content of the ceramic filler is further evidenced by Fig. 9 which compares the time evolution of the open circuit voltage of DMFCs using two composite membranes, i.e. sample PVdF-10 and sample PVdF-30, respectively, both activated by swelling with the same sulphuric acid solution. The occurrence of methanol crossover is revealed by OCV decays. In fact, by reaching the cathode side, methanol is oxidized, this giving rise to a mixed potential which, by subtracting the active specie, i.e. oxygen, lowers the overall cell potential.
0.9
Methanol Cross-Over / mmol cm-2
0.8 0.7
Nafion 117
0.6 0.5 0.4 0.3
Membrane + SiO2
0.2 0.1 0 0 5
10
% SiO2
20 40
Fig. 8. Methanol permeability through PVdF-based composite membranes. The permeability of a typical Nafion 117 membrane is also reported for comparison purpose. Results obtained by gas chromatography.
1.0
10% SiO2 30% SiO2
0.8
Cell voltage / V
1130
0.6
0.4
Flow starting 0.2
0.0 0
100
200
300
400
500
600
Time / s Fig. 9. Time evolution of the open circuit voltage of DMFCs using two different membrane samples having different SiO2 content, i.e. PVdF-10 and PVdF-30 samples. For sample identification see Table 1. Room temperature.
Clearly, this effect is lower for the membrane having the lower SiO2 concentration. On the other hand, membranes with low ceramic content, i.e. below 10% SiO2, have low conductivity, see Fig. 5, and particularly, very low cell performance, see inset of Fig. 7A. Thus, membranes having a moderately higher filler content, e.g. around 20–30%, have to be chosen as the preferred electrolytes for DMFC application. This is an acceptable choice since these membranes, which perform well in practical cells, are still characterized by levels of crossover which are lower than that observed for Nafion 117 membranes, see Fig. 8B and by relatively good physical properties, see Table 2. 4. Conclusion The results reported in this work show that new types of proton conducting nanoporous membranes can be obtained by readapting a procedure used in the lithium battery technology. The performance of the membranes can be optimized by the dispersion of SiO2 ceramic powder at nano particle size. Indeed, nanoporous, composite membranes so obtained have promising features, such as high conductivity combined with a low methanol permeability. Since activated by aqueous acid solutions, these membranes have a limited thermal stability due to the release of the liquid component. However, the transport and the physical properties, associated with a prospected very low cost, make these membranes attractive for some special applications, such as low temperature fuel cells for portable electronics. Preliminary tests performed in this work demonstrate that these membranes are indeed suitable for room temperature DMFCs where they compare well with conventional, much more expensive materials, such as Nafion. Moreover, it is worth to mention that the eventual liquid acid release from membranes, as it
F. Croce et al. / Electrochemistry Communications 8 (2006) 1125–1131
could occur during extended fuel cell operation, can be properly managed by a well engineered stack recycling circuit. Acknowledgement This work was performed in the framework of a project titled ‘‘Development of composite membranes and of innovative electrode configurations for polymer electrolyte fuel cells’’ sponsored by the Italian Ministry of University and Research, MIUR, FISR 2001.
1131
References [1] A.S. Arico`, S. Snirisavasan, V. Antonucci, Fuel Cells 2 (2001) 1. [2] A.S. Arico`, V. Baglio, P. Creti, A.D. Blasi, V. Antonucci, J. Brunca, A. Chapot, A. Bozzi, J. Schoemans, J. Power Sources 123 (2003) 107. [3] A.S. Arico`, P. Bruce, B. Scrosati, J.-M. Tarascon, W. van Schalkwijk, Nat. Mater. 4 (2005) 366–377. [4] J.M. Tarascon, A.S. Gozdz, C. Schumtz, F. Shokoohi, P.C. Warren, Solid State Ion. 86–88 (1996) 49. [5] S. Panero, F. Ciuffa, A. D’Epifanio, B. Scrosati, Electrochim. Acta 48 (2003) 2009. [6] E. Peled, T. Duvdevani, A. Ahoron, A. Melman, Electrochem. SolidState Lett. 3 (2000) 525.
