CHEMSUSCHEM FULL PAPERS DOI: 10.1002/cssc.201402209

Promising Aquivion Composite Membranes based on Fluoroalkyl Zirconium Phosphate for Fuel Cell Applications Anna Donnadio,*[a] Monica Pica,[b] Surya Subianto,[c] Deborah J. Jones,[c] Paula Cojocaru,[d] and Mario Casciola[a]

Layered zirconium phosphate (ZP) that bears fluorinated alkyl chains bonded covalently to the layers (ZPR) was used as a nanofiller in membranes based on a short-side-chain perfluorosulfonic acid (PFSA) to mechanically reinforce the PFSA hydrophobic component. Compared to the pristine PFSA, membranes with a ZPR loading up to 30 wt % show enhanced mechanical properties, and the largest improvement of elastic modulus (E) and yield stress (sY) are observed for the 10 wt % ZPR membrane: DE/E up to 90 % and DsY/sY up 70 % at 70 8C

and 80 % relative humidity (RH). In the RH range 50–95 %, the in-plane conductivity of the composite membranes reaches 0.43 S cm1 for 10 wt % ZPR at 110 8C and is on average 30 % higher than the conductivity of the pristine PFSA. The 10 wt % ZPR membrane is as hydrated as the neat PFSA membrane at 50 % RH but becomes progressively less hydrated with increasing RH both at 80 and 110 8C. The fuel cell performance of this membrane, at 80 8C and 30 % RH, is better than that of the unmodified PFSA.

Introduction The current trend towards environmental protection and more efficient power production has shifted interest from conventional fuels and internal combustion engines toward alternative fuels and power sources. Much interest is focused on the development of proton-exchange membrane fuel cells (PEMFCs), which use a polymer membrane as the electrolyte.[1] The future application of this type of technology depends greatly on the enhancement of membrane stability. To address these issues, early research activity was focused on different types of proton-conducting membrane based on polymer matrices endowed with high chemical inertia such as polybenzimidazole doped with phosphoric acid,[1c, 2] sulfonated aromatic polymers,[1c–d, 3] and perfluorosulfonic acids (PFSAs), which include Nafion, the Dow polymer, and Aquivion.[4] Besides chemi-

[a] Dr. A. Donnadio, Prof. M. Casciola Department of Chemistry, Biology and Biotechnologies University of Perugia Via Elce di Sotto 8, 06123 Perugia (Italy) Fax: (+ 39) 0755855566 E-mail: [email protected] [b] Dr. M. Pica Department of Pharmaceutical Sciences University of Perugia via Fabretti 48, 06123 Perugia (Italy) [c] Dr. S. Subianto, Dr. D. J. Jones Institute Charles Gerhardt, Agrgats, Interfaces et Matriaux pour l’Energie UMR CNRS 5253, Universit Montpellier 2 Place Eugne Bataillon, 34095 Montpellier cdex 5 (France) [d] Dr. P. Cojocaru Solvay Specialty Polymers Viale Lombardia, 20, 20021 Bollate (MI) (Italy) Supporting Information for this article is available on the WWW under http://dx.doi.org/10.1002/cssc.201402209.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

cal inertia, mechanical stability is another necessary requirement to improve the overall membrane stability. In the past, the mechanical properties of fuel cell membranes have not been of the most critical importance, mainly because initial efforts used 50–200 mm thick membranes. However, recent developments have been geared towards the use of thinner membranes (< 50 mm) because of the advantages they confer, such as lower membrane resistance and improved water transport. If we consider the mechanical stresses the membrane has to withstand in terms of hydration–dehydration or variation in stack compression, improvement in the membrane mechanical properties becomes an important goal to increase the resistance of the membrane to premature failure.[5] Improved mechanical properties can be achieved either through the modification of the polymer structure or by the incorporation of inorganic filler materials or reinforcements. Organic–inorganic composite membranes based on hydrophilic inorganic fillers,[6] such as oxides, clays, and zirconium phosphate,[1b–d, 7] have been explored intensively and remain one of the most interesting routes in the preparation of mechanically reinforced electrolytes for application in PEMFCs.[5] Hydrophilic fillers are of interest because, besides providing the necessary mechanical strength, they are expected to improve the membrane hydration, especially at low relative humidity (RH), which leads to enhanced proton conductivity. However, only a few recent papers report the use of organically modified inorganic materials as hydrophobic fillers designed purposely to interact with the perfluorinated PFSA backbone. Despite the few data available to date, all composite membranes made of PFSAs filled with oxide-based hydrophobic fillers[8] show a surprising improvement of proton conductivity toChemSusChem 0000, 00, 1 – 10

&1&

These are not the final page numbers! ÞÞ

CHEMSUSCHEM FULL PAPERS gether with a better dimensional stability with respect to the neat PFSA. Recently, both microcrystals and nanoparticles of a-zirconium phosphate (Zr(HPO4)2, hereafter ZP), which bear dodecyl groups bonded covalently to the layer surface, have been prepared by reaction of ZP with 1,2-epoxydodecane solutions.[9, 10] These materials are not expected to be suitable fillers for composite membranes for PEMFCs because the alkyl chains can undergo oxidative degradation easily under the PEMFC working conditions. Therefore, in this study, the layer surface of ZP nanoparticles has been modified by reaction with a fluoroalkyl epoxide ([2,2,3,3,4,4,5,5,6,6,7,7,8,9,9,9-hexadecafluoro-8-(trifluoromethyl) nonyl]oxirane, hereafter C12H5F19O) to generate POC covalent bonds, and the hydrophobic ZP thus obtained has been employed as a filler for proton-conducting membranes based on the short-side-chain PFSA ionomer Aquivion. These membranes have been characterized by TEM and their proton conductivity, water uptake, and mechanical properties have been determined. The dependence of the proton conductivity and mechanical properties of the composite membranes on filler loading, temperature, and RH has allowed the selection of the optimal composition of the membrane to be tested in a PEMFC.

Results and Discussion Filler synthesis and characterization According to Ref. [9], the reaction between ZP and 1,2-epoxydodecane leads to the formation of POC bonds that allow the alkyl chain to be anchored covalently to the phosphate groups (Scheme 1).

www.chemsuschem.org

Figure 1. TGA curves for a) ZP and ZPRx compounds synthesized with different C12H5F19O/Zr molar ratios: b) 0.5, c) 1, d) 2, e) 4.

lated in the interlayer region of the lamellar solids or adsorbed on the particle surface. The second step centered around 320 8C is associated with the decomposition of the fluoroalkyl chains bonded covalently to the P atoms of ZP through POC bridges. The third step starts above 350 8C and can be attributed to the condensation of the monohydrogen phosphate groups of ZPRx. However, the weight loss in this step appears to be greater in all cases than that expected for the formation of ZrP2O7 from ZP, which indicates that the removal of the organic component is not complete in the second step and needs a temperature above 350 8C. The number of fluoroalkyl chains per Zr atom was calculated from the weight loss that corresponds to the second and third steps of the TGA curve. The value of x increases almost linearly with the C12H5F19O/ZP molar ratio and reaches 1.6 for C12H5F19O/ZP = 4 (Table 1).

Table 1. x as function of C12H5F19O/ZP for ZPRx compounds.

Scheme 1. Formation of POC bonds that allow the alkyl chain to be anchored covalently to the phosphate groups, in which Z = alkyl chain.

Therefore, a similar reaction was used to functionalize nanosized ZP with the fluorinated alkyl chains of the epoxide C12H5F19O. Experiments were performed by mixing ZP gels in propanol with different volumes of 0.1 m C12H5F19O solutions in diethyl ether so that the C12H5F19O/ZP molar ratio was in the range of 0.5–4. The resulting samples with the general formula Zr(HPO4)2x(RPO4)x·n S (in which R =C12H6F19O and S is the intercalated solvent) are indicated as ZPRx. The samples were analyzed by thermogravimetry to calculate the number of fluoroalkyl chains (x) per Zr atom. The total weight loss of the ZPRx samples is higher than that of pristine ZP and increases with the increasing C12H5F19O/ZP molar ratio (Figure 1). In all cases, the thermogravimetric analysis (TGA) curves are characterized by three steps. The first weight loss below 200 8C ranges from 1.3 % (for C12H5F19O/ZP = 4) to 7 % (for C12H5F19O/ ZP = 0.5) and is assigned to the removal of the solvent interca 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

C12H5F19O/ZP

x

0.5 1 2 4

0.31 0.65 0.89 1.6

Thus the complete functionalization of all the phosphate groups (i.e., x = 2) is not obtained even in the presence of a large excess of C12H5F19O in solution, probably because of the steric hindrance of the terminal group CF(CF3)2 of the alkyl chain. Different from the parent ZP, the powder XRD patterns of the ZPRx samples (Figure 2) show only a few broad reflections of low intensity, which indicate that the main structural modification induced by the layer functionalization is, in all cases, a dramatic loss of crystallinity. A common feature of all these diffraction patterns is a weak reflection at 2 q = 33.88. The same reflection is also present in the pattern of pristine ZP with a Miller index of 0 2 0 and the associated interplanar distance is half the separation between neighboring Zr atoms in the a layer (5.297 ). Consequently, the position of the 0 2 0 reflection of ZP remains unaltered if ChemSusChem 0000, 00, 1 – 10

&2&

These are not the final page numbers! ÞÞ

CHEMSUSCHEM FULL PAPERS

www.chemsuschem.org Composite membranes

Figure 2. XRD patterns of ZPRx compounds and microcrystalline ZP.

reactions and/or intercalation processes, which occur in the interlayer region of ZP, do not alter the ZrOP framework of the a layer. We can reasonably assume that in ZPRx too, the reflection at 2 q = 33.88 is associated with the ZrZr distance, and we can conclude that the reaction between C12H5F19O and ZP leaves the structure of the a layer substantially unaltered. Apart from the reflection at 2 q = 33.88, only ZPR0.65 shows three other weak reflections (at 2 q = 11.6, 19.8, and 24.98) that are in common with the pattern of ZP, which indicates that a small fraction of unreacted ZP is left after functionalization. The absence of these reflections in the pattern of the sample with the lowest content of fluoroalkyl chains (ZPR0.31) could be surprising, but can be justified if we consider that the presence of unreacted ZP depends mainly on how the fluoroalkyl chains are accommodated in the interlayer region: the same number of chains can either be distributed uniformly in the interlayer region to give rise to only one disordered phase (with a low content of chains) or they can be arranged to form a chainrich disordered phase so as to leave ZP unreacted.[11] To investigate the stability of the POC bond in an acidic environment similar to the aqueous nanophase of the sulfonated ionomers, ZPR0.89 was immersed in a solution of 5 m H2SO4 and kept at 80 8C for 24, 72, and 144 h. Then, the powder was separated from the solution and washed with propanol and diethyl ether to remove the free alcohol (C12H7F19O2) that could be formed by the hydrolysis of the alkyl phosphate. The residual solid was dried in an oven at 80 8C and analyzed by TGA. In the temperature range in which the decomposition of the alkyl chains occurs, the difference between the weight loss of untreated and H2SO4-treated samples was of the order of the measurement error (Figure S1). More specifically, in the range 200–1200 8C, untreated ZPR0.89 lost 65.1 % of its weight at 200 8C, and the corresponding loss of the samples treated for 72 and 144 h is 64 and 63 %, respectively. These results show that ZPRx compounds possess good hydrolytic stability in an acidic environment and can be used as fillers for sulfonated ionomers.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

To maximize the hydrophobic interaction between the chains of the polymer and those of the filler, the degree of functionalization of the filler should be neither too low nor too high because, in the latter case, the dense packing of the filler chains would restrict the polymer–filler interaction to the terminal groups of the filler chains. The most favorable situation is that in which the degree of functionalization allows an efficient interdigitation of the filler and polymer chains. Based on these considerations, ZPR0.89 (hereafter ZPR), in which the alkylphosphate groups are diluted by a similar number of monohydrogen phosphate groups, was selected for the preparation of composite membranes based on EW 700 Aquivion with a filler loading in the range 5–30 wt %. These membranes will be hereafter indicated as C70/ZPR-y, in which y represents the ZPR loading (y = 5, 10, 20, 30 wt %).

TEM analysis The TEM image of the composite film that contains 10 wt % ZPR (Figure 3) reveals the presence of platelike particles with a planar dimension of approximately 20 nm that forms agglomerates with a size of 100–200 nm. The morphology of the filler particles is similar to that of the gel particles[12] used for the membrane preparation.

Figure 3. TEM image of a C70/ZPR-10 composite membrane.

Stress–strain tests The mechanical properties of C70/ZPR composite membranes, which contain 5–30 wt % filler, were investigated at ambient temperature/RH conditions, and at 70 8C/80 % RH. In both cases, the presence of the filler does not modify the profile of the stress–strain curve of the parent ionomer, as shown in Figure 4 in which the stress–strain curves of the composite ChemSusChem 0000, 00, 1 – 10

&3&

These are not the final page numbers! ÞÞ

CHEMSUSCHEM FULL PAPERS

www.chemsuschem.org

Figure 5. a) Young’s modulus and b) yield stress versus ZPR mass fraction at T = 20 8C and RH = 53 %, and T = 70 8C and RH = 80 %. Figure 4. Stress–strain curves for C70 and C70/ZPR-10 membranes.

membrane with 10 wt % ZPR are compared with those of a membrane (C70) cast from the D70 dispersion. The Young’s modulus (E) and the yield stress (sY) derived from the analysis of the stress–strain curves are shown in Figure 5 for membranes with different filler loadings. Although all the composite membranes show a strong proportional increase in the Young’s modulus in comparison with the pristine PFSA membrane (from a minimum of ~ 50 % for 5 wt % ZPR at 70 8C, to a maximum of ~ 120 % for 20 wt % ZPR at 70 8C), the yield stress benefits from the presence of the filler mainly for 10 wt % loading. For this membrane composition, the E and sY values at 70 8C/80 % RH reach a maximum that is ~ 90 and ~ 70 % higher, respectively, than the corresponding value of the neat PFSA membrane. Interestingly, for ZPR loadings up to 20 wt %, the elongation at break of the composite membranes is in all cases higher than 200 %, but decreases below 100 % for 30 wt % ZPR. These findings prove that the ZPR acts effectively as a stiffener of the polymer matrix that keeps a good ductile behavior up to a loading of 20 wt %. To investigate the influence of the enhanced mechanical properties on the degree of hydration (and, therefore, on the swelling) of the composite membranes, water uptake (WU) measurements were performed at 70 8C/80 % RH. The water content of all the composite membranes was in the range 18– 20 wt % (Figure S2), which is much lower than that of the un 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

modified PFSA (30 wt %). Under the same conditions of temperature and RH, the WU of the filler is 8 % of its anhydrous weight. If it is assumed that both filler and ionomer in the composite membrane keep the same hydration as they have as separate components, then the water content of the composite membranes should decrease linearly from 28.7 to 23.3 wt % with increasing filler loading from 5 to 30 wt % (Figure S2). Even in the extreme case of a completely anhydrous filler, the water content of the composite membranes lies in the range 28.3–20.9 wt %, which is again higher than that found experimentally especially for the lowest filler loadings (Figure S2). On the basis of these considerations, it must be inferred that the ionomer hydration in the presence of the filler is lower than that of the neat ionomer. These findings are explained qualitatively if we consider that the amount of water taken up by osmosis is limited by the membrane internal pressure that is generated as a result of swelling and acts on the aqueous nanophase confined in the membrane.[13] The fact that the internal pressure is proportional to the Young’s modulus of the membrane is consistent with the observed decrease in water content with the increasing Young’s modulus. Proton conductivity The in-plane conductivity of C70/ZPR membranes with filler loadings from 0 to 30 wt % was determined in the RH range ChemSusChem 0000, 00, 1 – 10

&4&

These are not the final page numbers! ÞÞ

CHEMSUSCHEM FULL PAPERS

www.chemsuschem.org Table 2. Water uptake [wt %] as a function of RH at 80 and 110 8C for C70 and C70/ZPR-10. RH [%]

C70 T = 80 8C

C70/ZPR10 T = 80 8C

C70 T = 110 8C

C70/ZPR10 T = 110 8C

50 65 80 95

9.6 18.7 27.7 41.0

11.3 15.3 20.1 26.2

14.0 21.7 34.4 45.2

16.0 18.7 27.1 33.1

Table 3. Hydration (expressed as l) as a function of RH at 80 and 110 8C for C70 and C70/ZPR-10.

Figure 6. Conductivity at 80 and 110 8C as a function of RH for C70/ZPR membranes with the indicated filler loadings.

50-95 % at 80 and 110 8C (Figure 6). At both temperatures and for all RH values, the conductivity shows a maximum for the composite membrane filled with 10 wt % ZPR. The highest conductivity of this membrane is 0.43 S cm1 at 110 8C. The dependence of the conductivity on filler loading at 80 8C, for 50 and 95 % RH is shown in Figure S3. In comparison with the C70 membrane, the conductivity of the C70/ZPR-10 membrane is higher by approximately 30 % at both RH values, whereas the conductivity of ZPR-30 is lower by 67 and 85 % at 95 and 50 % RH, respectively. The decrease in conductivity with the increasing filler loading above 10 wt % is simply because the filler is less conductive than the neat ionomer: for 95 % RH, the conductivity of a pellet of ZPR is 1.5  103 S cm1 at 80 8C and 2.3  103 S cm1 at 110 8C, whereas the conductivity of C70 is 0.26 and 0.35 S cm1, respectively. An equally straightforward explanation cannot be provided for the higher conductivity of C70/ZPR-10 with respect to C70. Besides the lower conductivity of the filler, the WU data determined at 70 8C/80 % RH suggest that the composite membranes are likely to be less hydrated than C70 under the temperature and RH conditions of the conductivity measurements: in this case, the higher conductivity of C70/ZPR-10 with respect to C70 could not be explained based on a higher water content. To clarify this point unambiguously, WU determinations for C70 and C70/ZPR-10 were performed at 80 and 110 8C as a function of RH. At the same temperature, the WU of the  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

RH [%]

C70 T = 80 8C

C70/ZPR10 T = 80 8C

C70 T = 110 8C

C70/ZPR10 T = 110 8C

50 65 80 95

3.7 7.3 10.8 16.0

4.9 6.6 8.7 11.3

5.5 8.4 13.4 17.6

6.9 8.1 11.7 14.3

composite membrane is similar to that of C70 for 50 % RH but becomes progressively lower for higher RH values (Table 2). In the following, the WU is expressed in terms of l, that is, the number of water molecules per sulfonic group. The l values of C70, calculated based on its ion-exchange capacity (IEC; 1.41  0.01 meq g1), are collected in Table 3. With increasing RH in the range 50–95 %, l increases from 3.7 to 16.0 at 80 8C and from 5.5 to 17.6 at 110 8C. Strictly speaking, the same calculation cannot be made for the composite membrane because of the difficulty to separate the WU component associated with the ionomer (WUi) from that associated with the filler (WUf). Nevertheless, l can be estimated by excess by assuming WUi ~ WU. The l values thus calculated (Table 3) are close to those of C70 at low RH but become progressively lower with increasing RH. The decrease in hydration is more evident at 80 8C: at 95 % RH, the l values for C70/ZPR-10 are 11.3 at 80 8C and 14.3 at 110 8C, whereas those for C70 are 16.0 and 17.6, respectively (Table 3). Therefore, at least at high RH, the ionomer hydration in the composite membrane is definitely lower than that of the neat ionomer. At the same RH, the composite C70/ZPR-10 is more conductive than C70, which is at the same time less hydrated. Consequently, proton conduction seems to be assisted by water more efficiently in the composite membrane than in the neat ionomeric membrane (Figure 7). Interestingly, although unmodified ZP[14] is more conductive than ZPR by at least a factor of five, no significant increase in conductivity is reported in the literature[7] either for short- or long-side-chain composite PFSA membranes loaded with ZP: rather, in most cases, the presence of ZP makes the composite membrane less conductive than the neat ionomer especially at low RH. An explanation of the different behavior of the composite PFSA membranes filled with organically modified or neat ZP can be suggested by taking into account the possible interacChemSusChem 0000, 00, 1 – 10

&5&

These are not the final page numbers! ÞÞ

CHEMSUSCHEM FULL PAPERS

www.chemsuschem.org Tests in a single-cell configuration Based on the mechanical and conduction properties, the composite membrane C70/ZPR-10 was selected for fuel cell tests with membrane electrode assemblies (MEAs) based on a catalyst-coated membrane (CCM) and characterized at 70 8C/100 % RH and 80 8C/30 % RH. For comparison, current–voltage (I–V) curves of a neat C70 membrane and a C70 membrane filled with 10 wt % ZP were recorded under the same conditions. It can be seen that at 70 8C/100 % RH, all three MEAs have a similar performance (Figure 8), whereas at 80 8C/30 % RH the MEA based on C70/ZPR-10 performs better than that based on the ZP composite membrane and the pristine ionomer (Figure 9). At 0.8 A cm2, the power density for C70/ZPR-10 (0.34 W cm2) is more than double the power density for C70 (0.14 W cm2). At 80 8C/30 % RH the open-circuit voltage (OCV) values are 0.91 V for the two composite membranes and 0.85 V for C70. Moreover, the MEA resistance per square centimeter, estimated from the slope of the linear region of the polarization curves (~ 0.47 W cm2 for C70/ZPR-10 and 0.6 W cm2 for C70), is higher than the corresponding membrane resistance (0.07 and 0.1 W cm2, respectively) calculated by extrapolating the data shown in Figure 6 to 30 % RH, which indicates that the polarization curves are dominated to a large extent by the resistance of the electrode–electrolyte interface. As at 80 8C/30 % RH, the

Figure 7. Conductivity (s) as a function of hydration (l) for C70 and C70/ ZPR-10 at 80 and 110 8C.

tions of ZP or ZPR with the ionomer matrix. As a result of the surface POH groups, neat ZP nanoparticles are likely to interact with the aqueous nanophase of the ionomer, in which proton transport takes place. The presence of ZP in the conduction pathways is expected to give rise to a decrease in the ionomer conductivity as ZP is much less conductive than the aqueous nanophase, the conductivity of which should not be far from that of a solution of sulfuric acid with the same proton concentration (i.e., 1.2–1.6 S cm1 for l values in the range 5–20 at 100 8C).[15] If the filler surface bears long alkyl chains, it is reasonable to suppose that the filler particles interact preferably with the hydrophobic moieties of the ionomer and are located mainly outside the aqueous nanophase. According to this model, the filler should not affect the mobility of the proton carriers in the conduction pathways. Consequently, to justify the higher conductivity of the composite membranes, it can be suggested that the presence of the filler induces modifications in the ionomer microstructure (in terms of nanophase separation, tortuosity/connectivity of the conduction pathways), which lead to enhanced ionomer conductivity.

Figure 8. Polarization and power density curves for CCM MEAs assembled with C70 and C70 composite membranes filled with ZPR and ZP.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

ChemSusChem 0000, 00, 1 – 10

&6&

These are not the final page numbers! ÞÞ

CHEMSUSCHEM FULL PAPERS

www.chemsuschem.org humidity values. This infers that the filler induces modifications of the membrane microstructure that renders the composite membrane more conductive than the neat ionomer. The composite membrane with 10 wt % filler performs better in a hydrogen/air fuel cell at 80 8C/30 % relative humidity than the corresponding neat ionomer membrane and than a composite membrane filled with 10 wt % unmodified zirconium phosphate, which confirms the positive influence of the organically modified filler on the properties of the composite membrane.

Experimental Section Chemicals and materials A 20 wt % Aquivion dispersion in water (D70-20B, ionomer equivalent weight = 700 g/equiv.) was supplied by Solvay Specialty Polymers, Italy. Zirconyl propionate (ZrO1.26(C2H5COO)1.49, MW = 220 Da) was supplied by MEL Chemicals, UK. Concentrated orthophosphoric acid (85 %, 14.8 m) was purchased from Fluka, and anhydrous propanol was from Carlo Erba. All other reagents were from Aldrich and used without further purification.

Filler preparation

Figure 9. Polarization and power density curves for CCM MEAs assembled with C70 and C70 composite membranes filled with ZPR and ZP.

MEAs based on the two composite membranes have the same OCV value and approximately the same slope in the Ohmic region, the better performance of C70/ZP-ZPR must be ascribed to a lower activation overpotential for this kind of MEA. In particular, from the extrapolation to zero current of the linear region of the polarization curves, it can be estimated that the overpotential of the C70/ZPR-10 MEA is ~ 77 mV lower than that of the C70/ZP10-MEA. Finally, preliminary wet/dry cycling tests at the OCV performed at 80 8C show that the OCV decay rate is significantly slower for C70/ZPR than for C70 (Figure S4).

A gel of a-ZP nanoparticles in propanol, which contained  8 wt % ZP, was synthesized following the procedure reported in Ref. [12]. Subsequently, a suitable volume of a solution of 0.1 m C12H5F19O in diethyl ether was added to the ZP gel with C12H5F19O/ZP molar ratios in the range 0.5–4. The samples, in Teflon bottles, were stirred magnetically overnight at RT and then heated at 80 8C in an oven for 24 h. The products were separated from the solution by centrifugation at 3000 rpm and washed under vigorous stirring three times with a suitable volume of diethyl ether to remove unreacted C12H5F19O and then twice with propanol. The resulting gels contained  10–12 wt % solid phase.

Composite membrane preparation The D70–20B Aquivion dispersion was heated to dryness at 80 8C. Dry ionomer (1 g) was dissolved in propanol (20 mL), and the resulting solution was mixed with a weighed amount of the ZPR gel. The mixture was stirred overnight and then cast onto a glass support by using an automatic film coater (Elcometer 4340 Applicator). After solvent evaporation (RT for 24 h), the composite membrane was heated at 90 8C for 2 h to remove the residual solvent, washed with 1 m HCl at RT for approximately 3 h, and finally thermally treated at 160 8C in an oven for 1 h. All composite membranes were  30 mm thick.

Conclusions Nanosized zirconium phosphate was modified organically by bonding fluorinated alkyl chains covalently to the phosphate groups and used as a filler for a low equivalent weight shortside-chain perfluorosulfonic acid ionomer. In comparison with the pristine perfluorosulfonic acid membrane, the composite membranes showed improved mechanical properties and proton conductivity for filler loadings up to 10 wt %. Water uptake determinations performed under the conditions of the conductivity measurements revealed that the higher conductivity of the composite membranes is associated with a lower degree of hydration of the polymer component at high relative  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Characterization Powder XRD patterns were collected by using a Panalytical X’Pert PRO diffractometer and a PW3050 goniometer equipped with an X’Celerator detector using a CuKa radiation source with a 2 q step size of 0.01708 and a step scan of 60 s. The LFF ceramic tube was operated at 40 kV and 40 mA. To minimize preferred orientations, the powder samples were carefully side-loaded onto a glass sample holder, and rectangle-shaped film stripes were loaded onto an aluminum sample holder. TGA was performed by using a NETZSCH STA 449 Jupiter thermal analyzer connected to a NETZSCH TASC 414/3 A controller at ChemSusChem 0000, 00, 1 – 10

&7&

These are not the final page numbers! ÞÞ

CHEMSUSCHEM FULL PAPERS

www.chemsuschem.org

a heating rate of 10 8C min1 with an air flow of approximately 30 mL min1. TEM analysis was performed by using a JEOL 1200 EXII microscope operating at 120 kV. Samples are embedded in Agar low-viscosity resin, dried at 80 8C for 24 h, and prepared by using a vibrating blade microtome. The samples were deposited onto a “lacey carbon” grid. Stress–strain mechanical tests were performed by using a Zwick Roell Z1.0 testing machine with a 200 N static load cell and equipped with a climatic chamber that operated in the RH range of 30–95 % ( 0.5 %) and in the temperature range of 10–80 8C ( 0.5 8C). The Young’s modulus (the slope of the stress–strain curve in the elastic-deformation region), yield stress (maximum stress that can be developed without causing plastic deformation), stress at break (the tensile stress at the breaking point of the specimen), and elongation at break (the percentage increase in length that occurs before the sample breaks) were measured on rectangleshaped film stripes, obtained by using a cutting machine, the length and width of which were 100 and 5 mm, respectively. Before the tests at RT, samples were equilibrated for seven days in vacuum desiccators at 53 % RH and RT (20–23 8C), whereas the high-temperature tests were performed after equilibration of the samples in the climate chamber at 70 8C and 80 % RH for one day. The thickness of the film stripe, determined with an uncertainty of 1 mm, was in the range of 25–30 mm. An initial grip separation of 10.000  0.002 mm and a crosshead speed of 30 mm min1 were used. At least five replicate film stripes were analyzed. The data were elaborated by the TestXpert V11.0 Master software. The IEC of the membranes was determined by acid–base titrations by using a Radiometer automatic titrimeter (TIM900 TitraLab and ABU91 Burette) that operated with the equilibrium point method. Before titration, the membranes were heated at 120 8C for 3 h to eliminate water and to determine the weight of the anhydrous sample. The dry membranes were equilibrated overnight at RT with a 0.1 m NaCl aqueous solution to exchange Na+ ions for the membrane protons. The solutions were then titrated with 0.1 m NaOH without removing the membranes. The in-plane conductivity of the membranes was determined on 5 cm  0.5 cm membrane strips in the frequency range 10 Hz– 100 kHz with 100 mV signal amplitude by four-probe impedance measurements by using an Autolab, PGSTAT30 potentiostat/galvanostat equipped with a frequency response analysis (FRA) module as described in Ref. [16]. The RH was controlled by using stainlesssteel sealed-off cells that consist of two communicating cylindrical compartments held at different temperatures. The cold compartment contained water, and the hot compartment housed the membrane under test. RH values were calculated from the ratio between the pressures of saturated water vapor (p) at the temperatures of the cold (Tc) and hot (Th) compartment [Eq. (1)]. RH ¼ pðT c Þ=pðT h Þ  100

ð1Þ

The WU at controlled temperature and RH was determined as described in Ref. [17]. Specifically, the cell used to measure the WU has the same size and shape as the conductivity cell and differs from that mainly because the MEA holder is replaced by a glass container that hosts the membrane sample (  0.5 g). The cell is equipped with a device that allows the sample container to be closed with a Teflon plug without opening the cell. After a suitable equilibration time (usually a day) at the desired temperature and RH, the sample container is closed, extracted from the cell, and  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

weighed. The water content (l) is determined based on the weight of the membrane dried at 120 8C for 5 h and on the amount of water trapped in the sample container at the temperature and RH of the experiment. The error of the determination of l is estimated to be  0.5 at most. Fuel cell tests were performed by using a Fuel Cell Technology test station with 25 cm2 active area single cells. The MEA consisted of a CCM sandwiched, without hot pressing, between Sigracet 10BC gas diffusion layers with a microporous layer on one side. The membrane was coated with the catalyst by the decal method using the D70-24B ionomer dispersion and a TANAKA catalyst (0.25 mg cm2 Pt, 50 %Pt/C for both cathode and anode). The catalyst was transferred to the membrane by applying 1.5 bar pressure for 3 min at 180 8C. The MEA was then incorporated into the fuel cell setup by using a rigid subgasket and conditioned for 14 h at 0.5 A cm2 at 70 8C and 100 % RH under H2/air (H2/O2 = 1.5:2). The polarization curves were collected after 6 h of equilibration at 0.5 A cm2 at the appropriate temperature and RH by setting the cell current density at predetermined points and holding the current density for 3 min at each point. Measurements at 30 % RH were performed under a backpressure of 0.5 bar (1.5 bar absolute pressure) by regulating the backpressure before equilibration and before each 3 min holding step.

Acknowledgements The EU-FP7 (FCH-JU) project “MAESTRO—MembrAnES for STationary application with RObust mechanical properties” (GA 256647) is gratefully acknowledged for cofunding this work. Keywords: mechanical properties · membranes · organic– inorganic hybrids · proton transport · zirconium phosphate [1] a) S. J. Peighambardoust, S. Rowshanzamir, M. Amjadi, Int. J. Hydrogen Energy 2010, 35, 9349 – 9384; b) S. Bose, T. Kuila, T. X. H. Nguyen, N. H. Kim, K. T. Lau, J. H. Lee, Prog. Polym. Sci. 2011, 36, 813 – 843; c) H. Zhang, P. K. Shen, Chem. Rev. 2012, 112, 2780 – 2832; d) A. Chandan, M. Hattenberger, A. El-Kharouf, S. Du, A. Dhir, V. Self, B. G. Pollet, A. Ingram, W. Bujalski, J. Power Sources 2013, 231, 264 – 278. [2] a) L. Qingfeng, H. A. Hjuler, N. J. Bjerrum, J. Appl. Electrochem. 2001, 31, 773 – 779; b) J. Mader, L. Xiao, T. J. Schmidt, B. C. Benicewicz, in Fuel Cells II Advances in Polymer Science, Vol. 216, Springer, Heidelberg, 2008, pp. 63 – 124; c) C. H. Shen, L. C. Jheng, S. L. C. Hsu, J. T. W. Wang, J. Mater. Chem. 2011, 21, 15660 – 15665; d) C. Y. Chen, W. H. Lai, J. Power Sources 2010, 195, 7152 – 7159. [3] a) J. Rozire, D. J. Jones, Annu. Rev. Mater. Res. 2003, 33, 503 – 555; b) C. H. Park, C. H. Lee, M. D. Guiver, Y. M. Lee, Prog. Polym. Sci. 2011, 36, 1443 – 1498. [4] W. Groth, Fluorinated Ionomers, Elsevier, Oxford, 2011. [5] S. Subianto, M. Pica, M. Casciola, P. Cojocaru, L. Merlo, G. Hards, D. J. Jones, J. Power Sources 2013, 233, 216 – 230. [6] a) G. Alberti, M. Casciola, Annu. Rev. Mater. Res. 2003, 33, 129 – 154; b) D. J. Jones, J. Rozire in Fuel Cells I, Springer, Heidelberg, 2008, pp. 219 – 264; c) R. K. Nagarale, W. Shin, P. K. Singh, Polym. Chem. 2010, 1, 388 – 408. [7] a) C. Yang, S. Srinivasan, A. B. Bocarsly, S. Tulyani, J. B. Benziger, J. Membr. Sci. 2004, 237, 145 – 161; b) F. Bauer, M. Willert-Porada, Solid State Ionics 2006, 177, 2391 – 2396; c) G. Alberti, M. Casciola, D. Capitani, A. Donnadio, R. Narducci, M. Pica, M. Sganappa, Electrochim. Acta 2007, 52, 8125 – 8132; d) M. Casciola, D. Capitani, A. Comite, A. Donnadio, V. Frittella, M. Pica, M. Sganappa, A. Varzi, Fuel Cells 2008, 8, 217 – 224; e) L. C. Chen, T. L. Yu, H. L. Lin, S. H. Yeh, J. Membr. Sci. 2008, 307, 10 – 20; f) M. P. Rodgers, Z. Shi, S. Holdcroft, Fuel Cells 2009, 9, 534 – 546; g) H. L. Lin, S. H. Yeh, T. L. Yu, L. C. Chen, J. Polym. Res. 2009, 16, 519 –

ChemSusChem 0000, 00, 1 – 10

&8&

These are not the final page numbers! ÞÞ

CHEMSUSCHEM FULL PAPERS 527; h) M. Casciola, G. Bagnasco, A. Donnadio, L. Micoli, M. Pica, M. Sganappa, M. Turco, Fuel Cells 2009, 9, 394 – 400; i) C. Arbizzani, A. Donnadio, M. Pica, M. Sganappa, A. Varzi, M. Casciola, M. Mastragostino, J. Power Sources 2010, 195, 7751 – 7756; j) M. Pica, A. Donnadio, M. Casciola, P. Cojocaru, L. Merlo, J. Mater. Chem. 2012, 22, 24902 – 24908. [8] a) X. Zhang, S. W. Tay, L. Hong, Z. Liu, J. Membr. Sci. 2008, 320, 310 – 318; b) E. Y. Safronova, A. B. Yaroslavtsev, Solid State Ionics 2012, 221, 6 – 10; c) M. L. Di Vona, E. Sgreccia, A. Donnadio, M. Casciola, J. F. Chailan, P. Knauth, J. Membr. Sci. 2011, 369, 536 – 544; d) V. Di Noto, M. Piga, G. A. Giffin, M. Schuster, G. Cavinato, L. Toniolo, S. Polizzi, Chem. Mater. 2011, 23, 4452 – 4458; e) V. Di Noto, N. Boaretto, E. Negro, G. Pace, J. Power Sources 2010, 195, 7734 – 7742; f) V. Di Noto, M. Bettiol, F. Bassetto, N. Boaretto, E. Negro, S. Lavina, F. Bertasi, Int. J. Hydrogen Energy 2012, 37, 6169 – 6181. [9] M. Casciola, D. Capitani, A. Donnadio, G. Munari, M. Pica, Inorg. Chem. 2010, 49, 3329 – 3336. [10] A. Donnadio, M. Pica, D. Capitani, V. Bianchi, M. Casciola, J. Memb. Sci. 2014, 462, 42 – 49.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.chemsuschem.org [11] G. Alberti, U. Costantino in Comprehensive Supramolecular Chemistry, Vol. 7 (Eds.: G. Alberti, T. Bein), Pergamon, Elsevier, Oxford, 1996, Chap. 1. [12] M. Pica, A. Donnadio, D. Capitani, R. Vivani, E. Troni, M. Casciola, Inorg. Chem. 2011, 50, 11623 – 11630. [13] a) K. D. Kreuer, Solid State Ionics 2013, 252, 93 – 101; b) G. Alberti, R. Narducci, M. Sganappa, J. Power Sources 2008, 178, 575 – 583. [14] F. Bauer, M. Willert-Porada, J. Power Sources 2005, 145, 101 – 107. [15] H. E. Darling, J. Chem. Eng. Data 1964, 9, 421 – 426. [16] M. Casciola, A. Donnadio, P. Sassi, J. Power Sources 2013, 235, 129 – 134. [17] A. Donnadio, M. Casciola, M. L. Di Vona, M. Tamivanan, J. Power Sources 2012, 205, 145 – 150.

Received: March 20, 2014 Revised: April 16, 2014 Published online on && &&, 0000

ChemSusChem 0000, 00, 1 – 10

&9&

These are not the final page numbers! ÞÞ

FULL PAPERS A. Donnadio,* M. Pica, S. Subianto, D. J. Jones, P. Cojocaru, M. Casciola && – && Promising Aquivion Composite Membranes based on Fluoroalkyl Zirconium Phosphate for Fuel Cell Applications

Killing two birds with one stone: Organically modified zirconium phosphate (ZPR) with fluorinated alkyl chains is used as a filler with a short-side-chain perfluorosulfonic acid (PFSA) to obtain

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

mechanically reinforced composite membranes with improved conductivity and reduced hydration compared to the PFSA.

ChemSusChem 0000, 00, 1 – 10

&10&

These are not the final page numbers! ÞÞ

Promising aquivion composite membranes based on fluoroalkyl zirconium phosphate for fuel cell applications.

Layered zirconium phosphate (ZP) that bears fluorinated alkyl chains bonded covalently to the layers (ZPR) was used as a nanofiller in membranes based...
881KB Sizes 0 Downloads 3 Views