Synthesis and fuel cell performance of phosphonated hybrid membranes for PEMFC applications

Journal of Membrane Science 466 (2014) 200–210

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Synthesis and fuel cell performance of phosphonated hybrid membranes for PEMFC applications Julien Souquet-Grumey, Renaud Perrin, Julien Cellier, Janick Bigarré n, Pierrick Buvat CEA-DAM Le Ripault, BP16, F-37260 Monts, France

art ic l e i nf o

a b s t r a c t

Article history: Received 8 October 2013 Received in revised form 3 April 2014 Accepted 5 April 2014 Available online 13 April 2014

The present work reports on the synthesis of novel organic–inorganic hybrid composites for proton exchange membranes. These original membranes are based on anionically synthesized phosphonic acid polymers, grafted to functionalized silica nanoparticles, and then dispersed in a matrix of poly (vinylidenefluoride-co-hexafluoropropylene), noted poly(VDF-co-HFP). In a first step, poly(vinylphosphonic acid) with different molecular weights (3.2 and 40 kg/mol) were synthesized from commercially available diethylvinylphosphonate and then grafted onto silica. In a second step, various amounts of phosphonic grafted silica nanoparticles, noted SiO2-g-PVPA, were dispersed in the poly(VDF-co-HFP) matrix to prepare membranes by solvent casting. Membranes with phosphonated silica particle loadings from 20 to 63 wt% exhibited proton conductivities from 23 to 54 mS/cm in immersed conditions at 80 1C. The highest values were obtained for the membrane with the highest silica content. Interestingly, the corresponding composite membrane, with a loading of 63 wt%, displayed a power density of 800 mW/cm2 (60 1C, 1.5 A/cm2) in single fuel cell tests. & 2014 Elsevier B.V. All rights reserved.

Keywords: Anionic polymerization Proton exchange membranes fuel cells Fluoropolymers Phosponated polymer Hybrid membrane

1. Introduction New alternative conversion technologies are of primary importance today, as various applications are dependent on fossil fuels. Since the directive tends to reduce power source emissions, the fuel cell technology has been considered an attractive energy source as a result of it offering a high energy conversion efficiency, a high power density and low greenhouse gas emissions [1,2]. The industrial scale development of proton exchange membrane fuel cells (PEMFCs) faces the improvement of the intrinsic performance of the fuel cell core, which particularly concerns the polymer electrolyte membrane (PEM). This element is a key component of the PEMFCs, acting as a solid electrolyte proton conductor as well as a separator to gas permeation from the anode to the cathode. The membrane must exhibit good thermal, chemical and mechanical stabilities, low gas permeability, high proton conductivity and finally a long life span [3]. In this context, polyperfluorosulfonic acids (Nafions, Aquivions, Aciplexs, Flemions, 3MIonomers, Hyflons) have been widely studied and have become references as PEM materials thanks to their excellent properties [4]. However, these membranes present high production costs as well as limited performances at operating temperatures above 100 1C, leading to a drastic decrease of the proton conductivity.

n

Corresponding author. Tel.: þ 33 24 734 4865; fax: þ 33 24 734 5168. E-mail address: [email protected] (J. Bigarré).

http://dx.doi.org/10.1016/j.memsci.2014.04.006 0376-7388/& 2014 Elsevier B.V. All rights reserved.

Indeed, a sulfonic membrane requires hydration to an extent that is directly connected to the total number of water molecules per sulfonate group (denoted λ), limiting the operation temperature to below 80 1C [5]. Under such conditions, serious problems appear such as carbon monoxide poisoning of the catalyst layer, a decrease of the cathode kinetics and the gas transport to the electrode, complicated heat and water management and finally lower stability and durability due to the excess of liquid water [6]. To this end, the U.S. Department of Energy has established a guideline [7] of 120 1C and 20% RH as target operation conditions, for a desired conductivity higher than 0.1 S/cm. During the past decade, significant research efforts have been aimed at developing new alternative membrane materials with enhanced properties and lower cost, able to operate under low humidity and temperatures above 100 1C [8]. With this in mind, several non-fluorinated and fluorinated polymer electrolytes have been developed including sulfonic polysulfone, sulfonic poly(phenylene oxide), sulfonic poly(ether ether ketone), sulfonic poly(aryl ether ketone), perfluorophosphonic acids [9–14], and more recently, terpolymers based on vinylidene fluoride (VDF) and hexafluoropropylene (HFP), grafted by aryl sulfonic acids [15]. In order to reach higher proton conductivities at elevated temperatures, phosphoric acid-doped polybenzimidazole (PBI/H3PO4) blends have been developed (0.03 S/cm at 150 1C). However, these systems present serious issues of stability over time [16]. To avoid these drawbacks, heterocyclic molecules (imidazole, benzimidazole, pyrazole) have been anchored to polymeric systems via spacers [8,17].

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

Heterocycles present the great advantage of a high boiling point allowing operation under drastic conditions. This way, phosphonic acids are protogenic groups which are considered as almostanhydrous proton-conducting moieties due to the high degree of proton self-dissociation arising from their intrinsic amphoteric character and their high dielectric constant. These entities exhibit fast hydrogen bond cleavage and formation allowing the proton transport via anhydrous conduction mechanism, known as the Grotthuss mechanism [18,19]. It has also been proposed that the conductivity of the phosphonate under nearly zero humidity may mainly be due to the presence of small amounts of water during the self-condensation reaction between phosphonic acid groups [20]. Considering this result, sulfonic acid groups, phosphonic acids groups and imidazole groups have been compared at low humidity (RH o40%) and temperatures between 120 and 160 1C by Schuster et al. [21] The authors concluded that phosphonic acids groups were the most suitable proton conductors under such conditions and suggested that they were interesting as protogenic groups for conducting polyelectrolytes. The proton can be transported through an anhydrous conduction mechanism under low humidity conditions or through the dynamics of water, at elevated water contents. Different types of phosphonated membranes have been developed during the last few years. However, phosphonated monomer polymerization is not easy to set up as very few monomers are commercially available and often have to be prepared [22]. Moreover the polymerization is sometimes difficult, leading to low molecular weights products. One efficient way is to perform polymerization and/or copolymerization of diethylvinylphosphonate (DEVP) monomer via an anionic polymerization process described by Jannasch et al. [23–25]. These living polymerizations are very attractive since they render it possible to easily obtain size-controlled and high molecular weight polymers. Thus, interesting properties could be obtained in such regular materials. These membranes exhibit a good proton conductivity at high levels of hydration but the conductivity quickly shuts down as water contain is decreasing. In order to enhance the water retention, some authors have proposed to introduce silica nanoparticles in poly(vinylphosphonic acid) (PVPA) from 5 to 20 wt% [26]. The best conductivity was obtained with a loading of 10 wt% (0.08 S/cm at 100 1C and 50% HR) but mechanical properties were too poor to prepare thin membranes. Consequently, their relevance for PEMFC applications was limited. Another way to prepare hybrid membrane is to use the sol–gel process from epoxycyclohexylethyltrimethoxysilane (EHTMS) as a silica precursor and amino trimethylene phosphonic acid (ATMP) as a phosphonic precursor [27]. The prepared membranes exhibited a relatively high proton conductivity value at 140 1C under anhydrous conditions (45.5 mS/cm). Nevertheless, the conductivity at 20 1C was low (0.088 mS/cm) and no fuel cell test has been reported. This can be due to a reduced performance of the membrane during starting conditions and a low mechanical strength. Moreover, sol–gel processes are usually expensive and sometimes not easy to set up. Organic–inorganic hybrid composites based on an inert polymer matrix and loaded with inorganic nanoparticles have attracted considerable attention. The organic phase provides the mechanical stability whereas the inorganic one manages the proton conduction as well as the water retention. As in the case of Nafions, hydrophobic and hydrophilic nanophases are present and provide a better performance of the membrane [28]. Indeed, the main-chain-type acid functionalized polymers often display lower proton conductivity compared to side-chain-type acid functionalized polymers because of the less distinct separation between hydrophilic and hydrophobic domains [29]. In the present approach, the non-conducting copolymer involved is the poly(vinylidenfluoride-co-hexafluoropropene) noted poly(VDF-co-HFP) because of its physical and electrical

201

properties, its low permeability to gases and finally its low cost which make it highly beneficial for fuel cell applications [30]. On the other hand, the inorganic phase is incorporated using functionalized fumed silica nanoparticles noted A390 which exhibit interesting advantages such as high specific area, small particle size (7 nm), hydrophilic surface (high water retention) and low cost. The idea is to combine the properties of organic and inorganic materials instead of polymers with both properties, which involves complex synthesis procedures. As a matter of fact, quite a few articles report on the use of poly(VDF-co-HFP) in composite materials to elaborate fuel cell membranes. Among them, Pereira et al. [31] synthesized an organic–inorganic hybrid membrane of poly(VDFco-HFP) and mesostructured silica containing sulfonic acids groups using the sol–gel process. However, the ion-exchange capacity (IEC) was quite low (less than 0.4 meq/g) giving the fuel cells a lower performance of about 100 mV compared to Nafion 112s. In order to significantly increase the IEC value, Niepceron et al. [32] fabricated original sulfonated hybrid membranes based on heterogeneous poly (VDF-co-HFP)/nanosilica modified by poly(stryrenesulfonic acid) and validated the use of core-shell like silica particles for protonconducting membranes. These composite membranes exhibited proton conductivity values of 15–95 mS/cm at 20 1C and a power density of 1.0 W/cm2 at 70 1C in single cell fuel cell tests with nonhydrated gas feeds. In order to obtain membranes performing at high temperature, Labalme et al. [33] developed hybrid membranes based on poly(VDF-co-HFP) and phosphonic polymers grafted onto silica nanoparticles. The authors performed the 4-chloromethylstyrene radical polymerization, followed by the post-phosphonation of the polymers by the Mickaelïs–Arbuzov reaction. The reaction was initiated by the chloromethylphenyltrimethoxysilane rendering it possible to graft the polymer onto the silica surface. The proton conductivity of the membrane based on 40 wt% loading in modified silica reached 65 mS/cm at 80 1C in liquid water and the corresponding hybrid membrane exhibited a good mechanical stability. However the main disadvantages of these procedures include the multi-step synthesis and the difficulty of controlling the polymerization. Phosphonated membranes are mainly developed for automotive or for stationary applications which must be able to start at room temperature and operate in a broad range of temperatures. Indeed, above 100 1C important benefits may be generated for the fuel cell system: less sensibility to CO poisoning of the catalyst, less complex auxiliary system, better performance. But, very few studies present fuel cell tests and polarization curves of phosphonated membranes at these intermediate temperatures. This could be due to the difficulty of starting the fuel cell at room temperature and increasing the temperature until its operating conditions. Some authors have proposed to use sulfonic and phosphonic acid bi-functionalized membranes in order to cover the temperature range [34]. However, the performance in fuel cells was found to be very low. Recently, triazole-based membrane doped with 13 wt% phosphoric acid was seen to operate in a wide temperature range (25–150 1C) with little humidification [35]. Specific gas diffusion electrodes with triazole-grafted polysiloxane electrolyte were fabricated in order to favor the membrane performance at high temperature. Here, the power density of the assembly was 180 mW/cm2 at 80 1C and 210 mW/cm2 at 150 1C under dry H2/O2 gases. The present article reports on a simple and fast preparation of hybrid membranes based on poly(VDF-co-HFP) and phosphonic polymers grafted onto fumed silica. As previously described, polymerization of diethylvinylphosphonate (DEVP) monomer can be obtained via anionic polymerization. Hübner et al. [36] recently demonstrated that chlormethylbenzyl groups can be used as termination agents. The idea presented in this work involved grafting the living polymers on chloromethylbenzyl-modified silica. This process should provide a much higher grafting density

202

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

than other “grafting to” methods. The structure of the final hybrid membrane should promote the nanophase-separation of phosphonic acid aggregates, which could serve as pathways for proton conduction. The first part of this work will focus on the feasibility and the relevance of the synthesis based on anionic polymerization in order to obtain various targeted molecular weight phosphonic polymers grafted on silica. Then, several poly(VDF-co-HFP)-based composite membranes with different loadings of functionalized silica will be presented and characterized via thermal analysis, ionic exchange capacity (IEC), water uptake and swelling. The second part presents proton conductivity and fuel cell tests. Standard gas diffusion electrodes with Nafion were used to perform the fuel cell tests. No hot pressing was carried out during the preparation of the membrane electrode assemblies. Thus, electrode limitation should appear at high temperatures due to Nafions losing water. Consequently, the fuel cell tests were performed at temperatures below 80 1C and in dry conditions in order to study the starting conditions of these phosphonic membranes. Their performance at temperatures above 80 1C will be investigated in future work with specific electrodes.

2. Experimental section 2.1. Materials Fumed silica (Sigma, particle size: 0.007 nm, surface area: 390 740 m2/g), denoted A390, was dried at 100 1C for 24 h under vacuum under magnetic stirring. ((Chloromethyl)phenylethyl) trimethoxysilane (CPMS, ABCR), diethylvinylphosphonate (DEVP, Aldrich, 97%), 1,1-diphenylethylene (DPE, Aldrich, 99%), aqueous HCl (Aldrich, 37%) and a 2.5-M solution of n-butyllithium in hexane (n-BuLi, Aldrich) were used as received, as were toluene (Aldrich, anhydrous, 99.8%), tetrahydrofuran (THF, Aldrich, anhydrous, 99.8%), methanol (VWR, rectapur), ethanol (VWR, rectapur, absolute), and dimethyl sulfoxide (DMSO, Aldrich, anhydrous, 99.8%). Millipores water was utilized in the preparation of aqueous solutions of the silica particles as well as for characterizing the composite membranes. A high molecular weight grade of a poly(VDF-co-HFP) copolymer from Solvay (Solefs 21216) was selected for the preparation of the composite membranes. 2.2. Characterizations

equation: IECth: ¼ IECPDEVP %PDEVP y

ð1Þ

where IECPDEVP is the non-grafted polymer PDEVP IEC (equal to 9.26 meq/g considering one proton), %PDEVP is the PDEVP weight ratio in A390-g-PDEVPx and y is the A390-g-PDEVP weight loading ratio in the hybrid membrane. A390-g-PDEVPx compositions were calculated from the thermogravimetric analysis of A390-g-PDEVPx (silica-grafted polymers) and PDEVP (non-grafted polymer), according to the equations: %PDEVP ¼

100  W ðA390  gPDEVP@550Þ 100  W ðPDEVP@550Þ

%A390 ¼ 100  %PDEVP

ð2Þ ð3Þ

where W(A390-g-PDEVP@550) is the weight ratio of the remaining A390g-PDEVPx at 550 1C, W(PDEVP@550) is the weight ratio of the remaining PDEVP at 550 1C, %A390 is the weight ratio of A390 in A390-g-PDEVPx and %PDEVP is the weight ratio of PDEVP in A390-g-PDEVPx. 2.2.3. Experimental ion-exchange capacity (IECexp.) Experimental ion-exchange capacities (IECexp.) of the membranes were evaluated by acid–base titration using a pH meter (Thermo scientific ORION 4 star). The membranes in their protonated form were soaked in an aqueous 2-M NaCl solution for 72 h, and the solutions were titrated against a 0.01-M NaOH solution to obtain a titration curve. The solutions were then back-titrated with 0.01-M HCl aqueous solution to get a reverse titration curve as described in [37]. 2.2.4. Water uptake and hydration number measurements Weights were measured using a Mettler balance with 0.1 mg precision. The membranes were equilibrated in Milli-Q water for 24 h at room temperature before recording the sample mass (Wwet). Dry state weights (Wdry) were determined after drying the membranes in a vacuum oven at 70 1C for 24 h. The water uptake (Wwater) was calculated as follows: W water ¼

ðW wet  W dry Þ 100% W dry

ð4Þ

The hydration number (λ) was calculated using the equation:

λ¼

W water ðM water  IECexp: Þ

ð5Þ

where Mwater is the molecular weight of water (18 g/mol). 2.2.1. Characterization by NMR and TGA of the silica-grafted polymers Nuclear magnetic resonance (NMR) spectroscopy data of both 29 Si, 13C and 31P in solid samples were collected with a Bruker Avance DSX 300 spectrometer using a cross-polarization technique coupled with magic angle spinning, commonly referred to as CP-MAS, at a frequency of 59.6 MHz, 75.5 MHz and 121.5 MHz, respectively. Liquid phase 1H, 13C and 31P NMR data were collected with a Bruker Avance DPX 200 spectrometer at frequencies of 200 MHz, 50.3 MHz and 80.8 MHz. The chemical shifts are reported in units of δ. The weight loss of the different copolymers was evaluated by thermogravimetric analysis (TGA) on a Q500 analyzer from TA Instruments. The data were collected from 100 to 600 1C after the samples had been kept at 120 1C for 20 min. The samples were analyzed under an argon atmosphere at a heating rate of 10 1C/min. 2.2.2. Theoretical ion-exchange capacity (IECth.) The theoretical ion-exchange capacities (IECth.) were determined from the A390-g-PDEVPx (where x is the polymer DPn determined by 1 H NMR spectroscopy, see paragraph 3.1) composition according to the

2.2.5. Water swelling measurements The water swelling (Swater) was obtained by measuring the wetand dry-state thickness of the membranes. A Mitutoyo digital micrometer was used for the thickness measurements. The dispersion on three measurements was around 5%. First, dry-state thicknesses (Tdry) were determined after keeping the membranes in a vacuum oven at 70 1C for 24 h. Then, the membranes were equilibrated in Milli-Q water for 24 h at room temperature. The excess water was removed using tissues before recording the wet thicknesses (Twet). Finally, the water swelling was calculated according to the equation: Swater ¼

ðT wet  T dry Þ 100% T dry

ð6Þ

2.2.6. Proton conductivity measurements The proton conductivity was evaluated by electrochemical impedance spectroscopy (EIS) using a Solartron SI 1255 dielectric analyzer equipped with a Novocontrol broadband Dielectric Converter and quatro system for temperature control. Prior to the

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

after which the temperature was increased to 60 1C. After 30 min at 60 1C, the current density was increased to 1.0 A/cm2 and the cell was kept under steady state during 24 h. The polarization curves were recorded after 30 min at the studied temperature by increasing the current density between OCV to 1.8 A/cm2.

24000 22000 20000 18000 16000

- Z" (Ohm)

203

14000

2.3. Phosphonated silica nanoparticle preparation

12000 10000

The work presented herein proposes a powerful process based on the “grafting to” approach to access new hybrid materials. The synthesis [38] was carried out in two steps (Fig. 2). Firstly, the polymer was prepared by anionic polymerization as described by Jannasch's group [23–25]. Secondly, CPMS-modified silica was used as a quenching agent for the anionic polymerization.

8000 6000 4000 2000 0 0

10000

20000

30000

40000

Z' (Ohm) Fig. 1. Example of an impedance diagram recorded with the EIS cell in order to calculate the proton conductivity of the membrane (U ¼50 mV).

conductivity measurements, the membranes were first stored 24 h in deionized water at room temperature. Measurements were carried out with wet-state membranes (5  20 mm2) using a laboratory-built cell composed of two parallel wire electrodes of platinum separated by 10 mm. The proton conductivity was measured between 20 and 80 1C in water. Impedance data were gathered over the frequency range of 10  1–107 Hz at a voltage amplitude of 50 mV and were analyzed with the software WinDeta (Novocontrol). A typical impedance diagram measured with the EIS cell is presented in Fig. 1. The proton conductivity (s) was calculated with the following equation:

s¼

d eLR

ð7Þ

here d is the distance between the two electrodes, e is the thickness of the wet membrane, L is the width of the wet membrane and R is the diameter of the half-circle in the impedance diagram Z0 , Z″ (Fig. 1). The activation energy (Ea) between 20 to 80 1C was calculated using the Arrhenius law according to the equation: log ðsÞ ¼  log ðs0 Þ

Ea RT

ð8Þ

where R¼8.314 J/mol and T is the temperature in Kelvin. 2.2.7. MEA preparation and fuel cell tests The membranes were characterized as fuel cell parts on a H2/O2 single-cell test station from Areva Energy Storage. Laboratorymade gas diffusion electrodes (GDE) composed of a microporous layer including 2.0 mg/cm2 of Pt and 1.0 mg/cm2 Nafion ionomer on woven web were used. This high level of platinum guaranteed a low level of activation loss. Losses at high current densities (4 0.5 A/cm2) can be considered to be mainly due to the membrane and the interface resistances. Before being set up between the gas diffusion electrodes, the membranes were immersed in a 1-M HCl solution during one minute at room temperature, rinsed twice in deionized water at room temperature and quickly dried using tissues in order to remove the surface water. Membrane electrode assemblies (MEAs) were prepared by simply placing the wet membrane between both electrodes. No specific procedures, such as for instance hot-pressing, were used to integrate the electrodes to the membranes. They were merely held together by the clamping pressure applied at the current collectors. The gases were dry and the pressure was 2 absolute bars. The cell was started at room temperature and maintained there until it reached 0.5 A/cm2,

2.3.1. Surface modification of silica particles (CPMS-grafted silica) In a first step, fumed silica particles were surface-modified with CPMS using a procedure similar to the one described by Niepceron et al. [32] in order to prepare the quenching-group grafted silica [36,39]. Toluene (500 mL) was added to fumed silica particles (6 g, equivalent to approximately 0.02 mol of surface silanols) in a trinecked 1-L round-bottomed flask under a flow of argon. Modified silica was further dispersed by sonification. The dispersion was then brought to reflux and CPMS (10 g, 0.036 mol) was added to the reaction mixture. The reaction was allowed to proceed for 4 h. After isolation, the surface-modified silica particles were thoroughly purified by two successive dispersion and centrifugation cycles in ethanol and acetone. The CPMS-grafted nanoparticles were finally dried in a vacuum oven at 80 1C overnight. 13 C CP/MAS ( 75.5 MHz, δ) δ 145–138 ppm (C1); δ 125 ppm (C2); δ 60 ppm (C3); δ 45 ppm (C4); δ 28 and 15 ppm (C5 and C6). 29Si CP/MAS (59.6 MHz, δ) δ  55 ppm (T1); δ  65 ppm (T2); δ  77 ppm (T3); δ  100 ppm (Q3); δ  55 ppm (Q4) (Fig. A.1). 2.3.2. PDEVP synthesis and grafting onto CPMS-modified silica (A390-g-PDEVP) A dried 250-mL round-bottomed glass reactor was equipped with a gas inlet and outlet, a thermometer, a septum and a magnetic stirrer. The reactor was filled with anhydrous THF (100 mL) and then cooled to  78 1C. The system was purged with argon at least seven times and left under a pressure of argon. Then, a desired amount of DPE (0.5  10  3 mol, 0.108 g) was introduced in the reactor and stirred during 15 min. Using a carefully dried syringe, a volume of nBuLi was first added to consume the small quantities of remaining impurities in the THF. Subsequently, a known quantity of n-BuLi (0.5  10  3 mol, 0.2 mL) was further loaded in order to activate the DPE, until a deep red color appeared. Next, DEVP (15.2 10  3 mol, 2.5 g) was added dropwise, slowly enough to keep the temperature below  55 1C. The reaction mixture was then colored in light yellow and kept at  78 1C for 1.5 h. At the end of this period, a small aliquot of the reaction mixture was extracted and quenched in methanol in order to obtain a sample of the poly(DEVP) for further analysis. Finally, the solution was transferred to a second reactor containing the dispersed CPMS-grafted nanoparticles (1 g) in anhydrous THF (100 mL), at

Fig. 2. Synthetic route for the preparation of poly(vinylphosphonic acid) PVPAgrafted silica nanoparticles (A390-g-PVPA).

204

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

which point the colored solution turned to uncolored. The system was kept under stirring overnight. The solvent and the remaining monomer were evaporated and the solution was precipitated in cold heptane. The yield of the polymerization of DEVP was calculated by comparing the weight of PDEVP and the weight of the monomers charged to the reactor considering that all the silica was recover during the purification process. The amount of PDEVP in the nanoparticles was determined by weighing the dry product and removing the amount of silicate. The PDEVP molecular weight was determined by 1H NMR spectroscopy on the aliquot of the mixture reaction. The products were denoted A390-g-PDEVPx where x corresponded to the DPn. PDVEP: 1H NMR (200 MHz, DMSO-d6, δ) δ 7 ppm (m, 10H, HAr); δ 4 ppm (s, 4H, HP–O–CH2 –CH3 ); δ 1-2.5 ppm (aliphatic region); δ 1.2 ppm (s, 6H, HP–O–CH2 –CH3 ). 13C CP/MAS (50.3 MHz, DMSO-d6, δ) δ 145–125 ppm (CAr); δ 68 ppm (CP–O–CH2 –CH3 ); δ 40–20 ppm (Caliphatic); δ 19 ppm (CP–O–CH2 –CH3 ). 31P NMR (80.8 MHz, δ); δ 32 ppm (s, 1P-phosphonate). A390-g-PDEVP: 13C CP/MAS (75.5 MHz, δ) δ 145–125 ppm (CAr); δ 68 ppm (CP–O–CH2 –CH3 ); δ 40–20 ppm (Caliphatic); δ 19 ppm (CP–O–CH2 –CH3 ). 2.4. Membrane preparation Dispersions of A390-g-PDEVPx and poly(VDF-co-HFP) in DMSO were obtained by first dispersing a known quantity of lyophilized A390-g-PDEVP in DMSO for 24 h before adding a known quantity of poly(VDF-co-HFP). The dispersions were then stirred for 24 h prior to casting, affording composite membranes with loadings ranging from 22 to 84 wt%. Composite hybrid membranes were obtained by solvent evaporation on a glass plate using a laboratory-scale handcoater set at 55 1C. The membranes were dried during at least 24 h and could be removed from the glass plates by immersion in a water-bath. The membranes were then stored at room temperature after drying between two absorbent sheets. The phosphonated membranes were named A390-gPDEVPx-y%, with x the DPn of PDEVP and y the equivalent A390-g-PDEVPx weight loading ratio. To obtain the acidic form, the membranes were immersed in a reflux of HCl for 7 days under slow stirring. These acidic membranes were named A390-gPVPAx-y% following the same rules as mentioned previously. The membranes were rinsed with distilled water until pH ¼7. Proton conductivity and fuel cell performances were compared to those of the Nafions NRE212 CS membrane. Nafions NRE212 CS thicknesses were 51 mm in the dry state and 61 mm in the wet state. Thus, water swelling was 20% and water uptake was 30 wt%. The ionic exchange capacity (IEC) was around 0.95–1.01 meq/g.

Anionic polymerization of polydiethylvinyl phosphonate (PDEVP) was previously exposed using diphenylethylene (DPE) in the presence of nBuLi in order to form 1,1diphenylhexyllithium to initiate the polymerization of the DEVP monomer [29,30,36]. Although it was recently demonstrated that the polymerization can occur without DPE [25] in the case of polymer grafting, the use of DPE is necessary for the homopolymerization of vinylphosphonate monomers in order to avoid uncontrolled side reactions [41]. With the aim of demonstrating the feasibility of the synthesis and the easy access to both low and high molecular weight compounds, two polymers with different targeted molecular weights (i.e., 5.0 kg/mol and 50.0 kg/mol) were prepared via anionic polymerization. The living polymers were quenched onto the modified silica as described in Fig. 2. Due to the polymer aggregating on the SEC column, SEC could not be performed on these polymers to determine their molecular weight. Rather, 1H NMR was carried out on the aliquot extracted before the quenching step to determine the molecular weight by integrating the aromatic signal from the 1,1-DPE around 7.4 ppm and the signal of the phosphonated ethyl ester groups at 4.3 ppm (Fig. 3). It was then possible to determine a DPn equal to 30 for A390-gPDEVP30 and to 370 for A390-g-PDEVP370, as well as a molecular weight equal to 4.9 kg/mol and 61 kg/mol under the ester form. These results were very close to the targeted molecular weight expected, demonstrating the relevance of anionic polymerization. The obtained DPn implied a molecular weight equal to 3.2 kg/mol and 40 kg/mol for the respective acid forms. These results should be compared to the polymerization by atom transfer radical polymerization (ATRP) by Labalme et al. [33]. Indeed, low molecular weights were obtained (3.8 kg/mol) whereas the author expected 15 kg/mol. Furthermore, the complete reaction occurred with excellent yield, contained between 90 and 95%. Considering these observations, anionic polymerization proved to be an effective alternative for realizing a controlled reaction. To completely characterize the material, particles of A390-gPDEVP30 and A390-g-PDEVP370 were hydrolyzed during 5 days in reflux HCl. The acidic forms were labeled A390-g-PVPA30 and A390-g-PVPA370. The structures of the polymer were also confirmed by 31P NMR and 13C NMR (Figs. 4 and 5) for the ester form and acid form according to the literature [23–25]. For the ester form, a typical phosphorus signal at 32 ppm was observed corresponding to the phosphonated group. The 13C NMR spectrum of A390-g-PDVEP30 showed shifts from the diethyl ester carbons P–O–CH2–CH3 at δ 68 ppm and from P– O–CH2–CH3 at δ 19 ppm. The structure of the hydrolyzed polymer

3. Results and discussion

Bu 3.1. Synthesis and characterization of silica nanoparticles grafted with poly(diethylvinylphosphonate) Fumed silica particles were surface-modified as already described by Niepceron et al. [32]. Two alkoxysilane molecules were added per silanol for the condensation reaction in order to reach an optimum surface modification with a number of silanol groups equal to 4.6 OH/nm (Kiselev–Zhuravlev function) [40]. The modified particles were purified by several centrifugation cycles in acetone/ethanol to ensure the complete removal of the nonreactive component. As has been previously demonstrated [32], 29 Si and 13C CP-MAS NMR spectroscopy studies confirmed the successful modification of the surface silanols by CPMS molecules and the absence of residual CPMS was highlighted by TGA.

n O P O O Ha

Hc

Hb

Hb Ha Hc

10

8

6

4

2

0

δ (ppm) Fig. 3. 1H NMR spectrum of PDEVP. The data were recorded in a DMSO-d6 solution.

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

205

Modified A390 with grafted polymer (A390-g-PDEVP) Modified A390 with chloromethylbenzyl group

Si

Cl

a

100

80

60

40

20

a

0

δ (ppm) Fig. 4.

31

P NMR spectrum of A390-g-PDEVP30.

P

300

OH

C b H 2 Ca H3

60

40

20

0

100

Ca

Cb

A390-g-PDEPV 30

80

60

40

20

0

13

C solid state CP-MAS NMR spectra of A390-g-PDVEP30 and A390-g-PVPA30.

(A390-g-PVPA30) was then confirmed by 13C NMR results. Indeed, the shift from the diethyl ester carbons P–O–CH2–CH3 at δ 68 ppm totally disappeared thus confirming the complete hydrolysis. However, a signal at δ 19 ppm was still observed but in this case the remaining shift at δ 19 ppm corresponded to the methylene carbons from the polymer backbone (CH3–CH2–CH2–CH2–(VPA)n) [23]. Based on these results, the anionic polymerization of DEVP was validated, and it thus represents a fast and easy route to obtain sized-controlled polymers from low to high molecular weights. 13 C CP-MAS NMR spectroscopy analyses were then used to confirm the reaction between the living polymer chains and the chloromethylbenzyl groups grafted on the silicate surface (Fig. 6). The 13C NMR spectrum of the CPMS-modified A390 shows a shift (a) from the chloromethylene carbon at δ ¼ 46 ppm (Cl– CH2–). This signal completely disappears after the grafting of living PDEVP onto the silica nanoparticles, high-lighting the quantitative quenching of the anionic polymerization by a nucleophilic substitution reaction onto chloromethylene carbon. As no more chloromethylene carbons were present, all potential grafting sites had become occupied, implying a high grafting density. Moreover, Niepceron et al. [32] demonstrated that approximately one chloromethylbenzyl molecule/nm2 was anchored onto the surface of the CPMS-modified A390. Considering the total disappearance of the chloromethylene carbon shift in the 13C CPMAS NMR spectroscopy analysis and according the calculation by Niepceron et al. it can be assumed that one polymer chain/nm2 was successfully attached to the inorganic particles. These results imply a silica nanoparticles/PDEVP ratio of 1/1.6 for A390-gPDEVP30 and 1/7.3 for A390-g-PDEVP370.

Weight (%)

200 180 160 140 120 100 80 δ (ppm) Fig. 5.

0

Fig. 6. 13C solid state CP-MAS NMR spectra of CPMS-modified A390 and CPMSmodified A390 grafted with PDEVP.

200 180 160 140 120 100 80 δ (ppm) O

100

δ (ppm)

A390-g-PVPA 30

P

200

60

A390-g-PDEVP30 A390-g-PVPA 30 40

100

200

300

400

500

600

Temperature (°C) Fig. 7. Thermogravimetric analyses (TGA) of A390-g-PDEVP30 and A390-g-PVPA30 under argon flux at a heating rate of 10 1C/min.

3.2. Thermal stability of the polymer-grafted silica nanoparticles Before blending with poly(VDF-co-HFP), the thermal stability of the ester and acid forms of the polymer grafted on the modified silica was investigated by TGA under argon at 10 1C/min (Fig. 7). These analyses were also performed to determine the polymer/ silica ratio and thus the theoretical IEC (see Section 2). Table 1 presents the main characteristics of the hybrid material. As shown in Fig. 7, the ester form of the polymer highlights a first degradation at 300 1C corresponding to the loss of diethyl ester groups [42] and a second degradation that appears around 480 1C. The remaining weight at 550 1C corresponds to the non-degradable part of PDEVP plus silica contained in the compound. A first analysis of raw PDEVP gives an estimation of the non-degradable part of PDEVP at 550 1C corresponding to 28% of the initial weight (solvent excluded). Therefore, it is possible to determine the polymer/silicate weight ratio

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

in the A390-g-PDEVP30 sample equal to 62 wt% and to 88 wt% in the sample A390-g-PDEVP370. Consequently, the IEC was evaluated to respectively at 4.8 and 7.7 meq/g according to the polymer, considering only one proton per acid group. TGA was also performed on the acidic form of the polymer (A390-g-PVPA). A continuous decrease of weight was here observed between 130 1C and 470 1C, corresponding to the water released, generated by the P-O–P bond formation [24], and followed by the degradation step around 470 1C.

3.3. Preparation and properties of the hybrid membranes 3.3.1. Thermal characterization, IEC, water uptake and swelling The composite membranes were obtained from the dispersion of polymer-grafted silica and poly(VDF-co-HFP) in DMSO. The experiments demonstrated that the use of the acid form of the polymer led to inhomogeneous membranes, which can be explained by the important interactions existing between the acid groups and the solvent. Nevertheless, hybrid membranes could be prepared from the ester form of the polymer, offering homogeneous materials. Finally, the conductive polymer electrolyte was further obtained by immersing the membranes in refluxing HCl for 7 days. The thickness ranged from 30 to 50 μm. As summarized in Table 2, six membranes were fabricated, two of them from the A390-g-PDEVP370 nanoparticles and four from the A390-g-PDEVP30 nanoparticles. To determine the exact ratio between poly(VDF-co-HFP) and polymer-grafted silica nanoparticles and then calculate the theoretical IEC values, TGA analyses were carried out under the ester and the acid forms of the membranes. As described for the A390g-PDEVP30 particles, the membrane also exhibited a clear weight loss at 300 1C corresponding to the ethyl-degradation (Fig. 8). Depending on the poly(VDF-co-HFP)/A390-g-PDEVP30 ratio, a second weight loss can be observed around 420 1C corresponding to the poly(VDF-co-HFP) degradation and a last one at 480 1C for the 2nd degradation of PDEVP. The temperature range of the two last degradations hinders an easy separation of the weight losses for each membrane. Nevertheless, a previous TGA of raw poly(VDF-co-HFP) enables us to determine the remaining weight at 550 1C equal to 20 wt% under such conditions. Finally, considering the remaining weight of poly(VDF-co-HFP) and that of A390-g-PDEVPx, it was possible to determine the ratio of hybrid nanoparticles, and thus the

composition of the membranes. Theoretical IEC values which should be attained ranged from 1.1 to 4.3 meq/g as shown in Table 2. The experimental IEC was evaluated by direct acid–base titration. Depending on the molecular weight of the polymer grafted onto the silica, the titration curves showed different behaviors as can be seen in Fig. 9 for membrane A390-g-PVPA30-63% and membrane A390-g-PVPA370-56%. For the membranes prepared with the A390-g-PVPA30 particles, the initial pH was around 6.0 after 72 h in 2-M NaCl solution and quickly increased to 8.5 for only 0.5 mL of NaOH which corresponded to an IEC of 0.2 meq/g for the membrane loaded at 63% (theoretical IEC for this membrane is 3.0 meq/g). Then, the exchange coefficient with Na was very low after 72 h (only 6.7%). The increase in exchange time up to 120 h did not lead to a change of the exchange coefficient. For membranes prepared with the A390-g-PVPA370 particles, the initial pH was lower (around 5) and the equivalent volume was 1.62 mL. This gave an IEC around 0.71 meq/g for the theoretical IEC calculated at 4.3 meq/g. Thus, the exchange coefficient (around 16.3%) was improved but remained low. Moreover, only one step was seen in the titration curve for direct NaOH neutralization. The curves indicated a pKa value around 8.0. This behavior has already been described for the PSU-co-PVPA copolymer in [24]. The authors explained that the close proximity of the acid units only allowed the dissociation of one proton for electrostatic reasons. 100

80

Weight (%)

206

60

b a c d

40

a. A390-g-PDEVP loading 71%

Table 1 Molar weight for PDEVP and PVPA, silica content for the hybrid polymer and respective ion exchange capacity (IEC) for the acid form. Polymer

Mna (kg/mol)

Silica ratio (wt%)

IECb (meq/g)

A390-g-PDEVP30 A390-g-PVPA30 A390-g-PDEVP370 A390-g-PVPA370

4.9 3.2 61 40

38 48 12 17

– 4.8 – 7.7

a b

b. A390-g-PVPA loading 63%

20

c. A390-g-PDEVP d. A390-g-PVPA

e

loading 62% loading 56%

e. PVDF film

0 100

200

300

400

500

600

Temperature (°C) Fig. 8. Thermogravimetric analyses of the hybrid membranes with different charge loadings in their ester form (PDEVP), acid form (PVPA) and PVDF film, under argon at a heating rate of 10 1C/min.

Determined by 1H NMR. Determined by TGA.

Table 2 Membrane compositions and properties: nanoparticles ratio, IEC, water uptake and hydration number (λ). Membranes

Nanoparticles ratio (wt%)a

PVDF ratio (wt%)

Silica ratio (wt%)

PVPA ratio (wt%)

IECa (meq/g)

Water uptake (wt%)

Total hydration number (λtotal)

PVPA hydration number (λPVPA)

A390-g-PVPA370-44% A390-g-PVPA370-56% A390-g-PVPA30-22% A390-g-PVPA30-27% A390-g-PVPA30-46% A390-g-PVPA30-63%

44 56 22 27 46 63

56.0 44.0 78.0 73.0 54.0 37.0

7.5 9.5 10.6 13.0 22.1 30.2

36.5 46.5 11.4 14.0 23.9 32.8

3.4 4.3 1.1 1.3 2.2 3.0

56 65 26 31 53 71

9.7 8.4 14.4 14.4 14.0 13.6

7.3 6.9 7.3 7.5 7.3 7.2

a

Determined by TGA.

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

12

80

11

207

A390-g-PVPA370 A390-g-PVPA30

70

10

60

Water Uptake (%)

9 8

pH

7 6 5

50 40 30 20

4 3

10

NaOH back titration HCl titration

A390-g-PVPA -63%

2 A390-g-PVPA

1

0

NaOH back titration HCl titration

-56%

2

4

6

10

20

30

40

50

60

70

80

90

100

Nanoparticles ratio (wt. %)

0 0

0

8

10

Vol (mL) Fig. 9. Curves of the direct NaOH titration and back titration with HCl for the experimental IEC measurements.

However, the curves of the back titration with HCl showed two steps with a Pka1 of approximately 8.0 and a Pka2 of approximately 5.5. For membranes with a low molecular weight polymer (for example membrane A390-g-PVPA30-63%), the first step was very wide and the equivalent volume was close to the second step. For membranes with a high molecular weight polymer (e.g., A390-gPVPA370-56%), the HCl pH curve showed two well-separated steps but the equivalent volumes were not double as would be expected if all of the acid units were exchanged. Consequently, long-chain polymers grafted on silica allowed a better dissociation of the two acid units but the second acid unit was not completely dissociated. For short polymer chains, the sterical exclusion around silica seemed to highly disrupt the dissociation and the ion exchanges of the second acid unit. Thus, the direct and back acid–base titrations were not well adapted to those composite membranes for the measurement of the experimental IEC since the exchanges were very slow and not complete due to sterical exclusion or electrostatic reasons around silica. More work has to be done to study the effect of the length of the polymer chain on the exchange coefficient and exchange kinetics. Consequently, at the moment, only the theoretical IEC values can be used to characterize the composite membrane. The water uptake as a function of the loading increased linearly with the nanoparticle ratio except in the case of the high ratio of the A390-g-PDEVP30 particle (Table 2 and Fig. 10). In this system, only the hybrid nanoparticles contributed to the water uptake due to the poly(VDF-co-HFP) matrix being hydrophobic. The hybrid nanoparticles were composed of silica and grafted PVPA. Nanoparticles of silica are well known to absorb large quantities of water but a large part of their surface is occupied by the grafted polymer. Also PVPA is hydrophilic and takes up water. In contrast to what has been described in a similar nanocomposite sulfonated system [32], no percolation threshold could be observed in Fig. 10. This indicates that the nanoparticles were well dispersed within the matrix and that they all contributed to the water uptake. In this loading range, the water uptake was independent of the nanoparticle composition (silica ratio and polymer molar weight). Thus, it can be considered that silica and PVPA had approximately the same water uptake ability. The experimental data of Fig. 10 are well fitted by using 1.15 wt% of water uptake per wt% of silica or PVPA. Classically, the hydration number λ is defined by the number of water molecules per acidic unit [H2O]/[PO3H2]. If the membrane is

Fig. 10. Water uptake at 20 1C versus nanoparticles ratio for the two molar weight nanoparticles.

composed of a single polymer, the hydration number can be easily calculated with the water uptake and the IEC according to Eq. (5). If this equation is directly used for hybrid membranes, λ ranges between 9.3 and 13.2. These values are higher than those estimated by Pereira et al. for PVPA [43]. Indeed, they were experimentally determined for a poly(styrene-co-vinyl phosphonic acid) membrane a hydration number equal to 6.3. Moreover, by using density functional theory (DFT) calculation, the authors obtained a convergence for λ equal to 7. In fact, the water uptake of the poly(VDF-co-HFP)/A390-g-PVPA membranes was due to PVPA and silica. If solely the water uptake due to PVPA was taken into account (by using the coefficient 1.2 wt % of water uptake per wt% of PVPA), the λPVPA would then be equal to 7.1 for high molecular weight and 7.2 for low molecular weight. Consequently, PVPA can in these hybrid membranes be considered as completely hydrated. These values are sufficient for proton conduction as demonstrated by [43]. However, they are much lower than those obtained for hybrid membrane based on silica/ styrenic sulfonic polymers (λ up to 41) [32], which may imply a different proton conduction mechanism. Finally, for all membranes, the water swelling was not measurable with the Mitutoyo digital micrometer which has a precision of 1 μm. The dispersion on 5 thickness measurements was around 5%. We thus considered the swelling to be less than 5%. No swelling was observed in the lateral dimensions. This behavior demonstrates that the P(VDFco-HFP) matrix contained hydroscopic stresses due to the water uptake. This low swelling is an advantage since it is favorable for long fuel cell durations in cycling conditions.

3.3.2. Proton conductivity under immersed conditions The performance of a membrane for PEMFC applications is closely dependent on the proton conductivity. Higher proton conductivities potentially lead to decreased ohmic losses and consequently to increased power densities. Only the PVPA contributes to the proton conductivity but the quantity of water can improve the conductivity by changing the conduction mechanism. Fig. 11 displays the proton conductivity of the composite membranes at 20 1C and 80 1C in water as a function of the PVPA ratio for both A390-g-PVPA30 and A390-g-PVPA370. In line with the IEC and the water uptake evolution, the proton conductivity increased almost linearly with the hybrid nanoparticle ratio, reaching a value of 33 mS/cm at 20 1C for the A390-gPVPA30-63% membrane, which is close to that of Nafions NRE212 CS. These results prove the easy accessibility of the acid sites. However, a low polymer molar weight presented a higher proton conductivity due to the larger silica ratio and higher water uptake

208

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

T(°C)

at 20°C A390-g-PVPA 370

50

-1.0

20 10

0

20

40

40

20

-1.5

NRE212 A390-g-PVPA 30 -63%

60

A390-g-PVPA 370-56%

Nafion NRE212

100 90

-2.0 2.7 2.8

-1

Conductivity (mS cm )

60

Nafion NRE212

30

0

80

A390-g-PVPA 30

40

log(σ (S.cm-1))

-1

Conductivity (mS cm )

60

2.9

3.0

3.1

3.2

3.3

3.4

3.5

1000/T (K-1)

50

Fig. 12. Arrhenius plot of proton conductivity in liquid water.

40 30 20

at 80°C A390-g-PVPA370

10 0

Table 3 Activation energy (Ea) of the proton conductivity in water immersed conditions for Nafions NRE212 CS, A390-g-PVPA30  x% and A390-g-PVPA370  x% membranes as determined from the Arrhenius plot.

A390-g-PVPA30 0

20

40

60

PVPA ratio (wt. %) Fig. 11. Influence of the PVPA ratio and temperature on the membrane proton conductivity in liquid water.

associated to silica. This behavior seems to demonstrate that water associated to silica can contribute to the increase in conductivity of the grafted polymer. As expected, raising the temperature leads to an improved proton conductivity (54 mS/cm), due to the better mobility of both water molecules and phosphonic polymer chains. The proton conductivity at 80 1C is also better for a low polymer molecular weight. The conductivity increases with temperature following an Arrhenius behavior as shown in Fig. 12. The activation energy (Ea) calculated from the Arrhenius law is presented in Table 3. The Ea values of the phosphonated membranes were half those of Nafions NRE212 CS, which means that the temperature dependence was lower. The range of Ea values (between 6.6 to 10.5 kJ/ mol) suggests that the conductivity mechanism combines both vehicular and Grotthuss mechanisms. Additionally, considering that the proton conductivity measurements were performed in water for a significant duration, it is interesting to note that no striking elution of the particles occurred at 80 1C, presumably prevented by the entanglements of the phosphonic polymer with the poly(VDF-co-HFP).

3.3.3. Fuel cell tests Fuel cells tests were performed on the A390-g-PVPA30-63% material exhibiting the highest proton conductivity. This membrane had a thickness of 52 μm, a water uptake around 71 wt% and an immeasurable swelling (o 5%). Moreover, this composite demonstrated good mechanical strength and was easy to set up in the cell. No hot pressing was used to fabricate the membrane electrode assemblies (MEA) but the membrane was set up in the wet state. The gas tightness was tested under nitrogen gas at room temperature. The fuel cell was started at 20 1C, 2 absolute bars

Membranes

IECa (meq/g)

Ea (kJ/mol)

Nafions NRE212 CS A390-g-PVPA370-44% A390-g-PVPA370-56% A390-g-PVPA30-63% A390-g-PVPA30-46% A390-g-PVPA30-27% A390-g-PVPA30-22%

0.95 3.4 4.3 3.0 2.2 1.3 1.1

14.7 6.6 8.3 7.1 5.8 8.2 10.5

a

Determined by TGA.

under H2/O2 dry gases and the current density was set at 0.5 A/ cm2. When the cell potential became stabilized, the temperature was raised to 60 1C. After several minutes of stabilization, the current density was increased to 1.0 A/cm2 for 20 h. The voltage chronogram in Fig. 13 demonstrates the good stability of the MEA. Under the above-mentioned conditions of pressure and current density, the membrane can be considered as completely hydrated. Fig. 14 shows the polarization curves of the A390-g-PVPA30-63% membrane, measured at 60 1C and 80 1C and 2 absolute bars under dry H2/O2 gases. Between 60 and 80 1C, the temperature had little effect on the polarization curves as compared to Nafions NRE212 CS. This result is consistent with the activation energy calculated from the proton conductivity measurements presented in Table 3. The open current voltage (OCV) was about 1.02 V, indicating a good interface in the membrane electrode assembly (MEA) and no crossover phenomena. Interestingly, a maximum power density of 800 mW/cm2 was measured for a current density of 1.5 A/cm2, which constitutes a good result when compared to other research work in the phosphonic membrane field [35]. As compared to Nafions NRE 212 CS, performances of this novel material are very promising. As mentioned before, phosphonic membranes have a lower proton conductivity level than Nafions NRE 212 which may explain the difference in performance, related to the ohmic losses. However, the very low swelling (o5%) could be a great advantage for aging resistance in cycling conditions (start and go). Finally, the membrane A390-g-PVPA30-63% perfectly reached the objectives of the starting conditions, i.e., at a temperature below 80 1C and under anhydrous conditions.

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

Appendix A

700 675

Cell potential (mV)

209

See Fig. A.1

650 625 600

OSi Si

SiO

575

OSi

SiO

OSi

550

Si

OSi

SiO

Si

OH

OSi

OSi

Q4

Q3

OH

OH

Q3

Q2

R

R

525 R

500 0

2

4

6

8

10 12 14 16 18 20

Si

SiO

O Si

SiO

O Si

time (h)

29

Si and

100

HO

Si

Q4

OH

OSi

T2

150 Fig. A.1.

OH

OSi

T3

Fig. 13. Chronogram of cell potential recorded for the A390-g-PVPA30-63% membrane during fuel cell starting and in steady state at 1 A/cm2, 60 1C and 2 absolute bars under dry H2/O2 gases.

Si

Q2

T1

50

0

T2 3 T

-50

-100

-150

13

C solid state CP-MAS NMR spectra of chloromodified silica.

Cell potential (mV)

1000 800 References

600 400

A390-g-PVPA 30 -63% at 60°C A390-g-PVPA 30 -63% at 80°C

200 0 0.0

Nafion NRE212 at 60°C Nafion NRE212 at 80°C

0.4 0.8 1.2 1.6 Current density (A.cm-2)

2.0

Fig. 14. Polarization curves performed at 60 1C, 80 1C and 2 absolute bars under dry H2/O2 gas feeds: comparison between Nafions NRE212 CS and the A390-g-PVPA3063% membrane.

4. Conclusions Poly(vinylphosphonic acid)-grafted silica nanoparticles were successfully synthesized via an anionic polymerization “grafting to” approach, demonstrating that it was an easy and fast method for obtaining sized-controlled phosphonic polymers. These modified nanoparticles were incorporated into a poly(VDF-co-HFP) matrix to develop proton exchange membranes. Proton conductivities of these membranes reached values up to 54 mS/cm at 80 1C in liquid water and demonstrated good stabilities in fuel cell tests. Moreover, these membranes offered very promising performances, 800 mW/cm2 at 1.5 A/cm2, which constitutes an original result. In order to test the fuel cell performances at temperatures above 80 1C, specific electrodes must be developed containing a phosphonated ionomer able to conduct protons at high temperatures and low water contents. Current work is focused on optimizing the chemical structure of phosphonic polymers and studying the membrane formulation in order to improve the final performance.

Acknowledgments The authors thank Pascal Palmas at the LCP of CEA, France for the NMR analyses. This work has been carried out within the framework of the HPAC program (MEMFOS) financed by the L' Agence Nationale de la Recherche (ANR).

[1] F. De Bruijn, The current status of fuel cell technology for mobile and stationary applications, Green Chem. 7 (2005) 132–150. [2] J. Zhang, Z. Xie, J. Zhang, Y. Tang, C. Song, T. Navessin, Z. Shi, D. Song, H. Wang, D.P. Wilkinson, Z.-S. Liu, S. Holdcroft, High temperature PEM fuel cells, J. Power Sources 160 (2006) 872–891. [3] H. Zhang, P.K. Shen, Advances in the high performance polymer electrolyte membranes for fuel cells, Chem. Soc. Rev. 41 (2012) 2382–2394. [4] V. Di Noto, T.A. Zawodzinski, A.M. Herring, G.A. Giffin, E. Negro, S. Lavina, Polymer electrolytes for a hydrogen economy, Int. J. Hydrogen Energy 37 (2012) 6120–6131. [5] K.D. Kreuer, On the development of proton conducting polymer membranes for hydrogen and methanol fuel cells, J. Membr. Sci. 185 (2001) 29–39. [6] A.C. Dupuis, Proton exchange membranes for fuel cells operated at medium temperatures: materials and experimental techniques, Prog. Mater. Sci. 56 (2011) 289–327. [7] L. Gubler, G.G. Scherer, Trends for fuel cell membrane development, Desalination 250 (2010) 1034–1037. [8] K.-D. Kreuer, S.J. Paddison, E. Spohr, M. Schuster, Transport in proton conductors for fuel–cell applications: simulations, elementary reactions, and phenomenology, Chem. Rev. 104 (2004) 4637–4678. [9] E.P. Jutemar, P. Jannasch, Locating sulfonic acid groups on various side chains to poly(arylene ether sulfone)s: effects on the ionic clustering and properties of proton-exchange membranes, J. Membr. Sci. 351 (2010) 87–95. [10] D.J. Jones, J. Rozière, Recent advances in the functionalisation of polybenzimidazole and polyetherketone for fuel cell applications, J. Membr. Sci. 185 (2001) 41–58. [11] G. Alberti, M. Casciola, L. Massinelli, B. Bauer, Polymeric proton conducting membranes for medium temperature fuel cells (110–160 1C), J. Membr. Sci. 185 (2001) 73–81. [12] J.A. Asensio, S. Borrós, P. Gómez-Romero, Sulfonated poly(2,5-benzimidazole) (SABPBI) impregnated with phosphoric acid as proton conducting membranes for polymer electrolyte fuel cells, Electrochim. Acta 49 (2004) 4461–4466. [13] Z.P. Guan, M. Xiao, S.J. Wang, Y.Z. Meng, Synthesis and characterization of poly (aryl ether ketone) ionomers with sulfonic acid groups on pendant aliphatic chains for proton-exchange membrane fuel cells, Eur. Polym. J. 46 (2010) 81–91. [14] R. Tayouo, G. David, B. Améduri, J. Rozière, S.p. Roualdès, New fluorinated polymers bearing pendant phosphonic acid groups. proton conducting membranes for fuel cell, Macromolecules 43 (2010) 5269–5276. [15] R. Souzy, B. Boutevin, B. Ameduri, Synthesis and characterizations of novel proton-conducting fluoropolymer electrolyte membranes based on poly(vinylidene fluoride-ter-hexafluoropropylene-ter-α-trifluoromethacrylic acid) terpolymers grafted by aryl sulfonic acids, Macromolecules 45 (2012) 3145–3160. [16] J.C. Persson, P. Jannasch, Intrinsically proton-conducting benzimidazole units tethered to polysiloxanes, Macromolecules 38 (2005) 3283–3289. [17] M.F.H. Schuster, W.H. Meyer, M. Schuster, K.D. Kreuer, Toward a new type of anhydrous organic proton conductor based on immobilized imidazole, Chem. Mater. 16 (2003) 329–337. [18] N. Agmon, The Grotthuss mechanism, Chem. Phys. Lett. 244 (1995) 456–462. [19] K.D. Kreuer, Fast proton conductivity: a phenomenon between the solid and the liquid state? Solid State Ionics 94 (1997) 55–62. [20] H. Steininger, M. Schuster, K.D. Kreuer, A. Kaltbeitzel, B. Bingol, W.H. Meyer, S. Schauff, G. Brunklaus, J. Maier, H.W. Spiess, Intermediate temperature

210

[21]

[22]

[23]

[24]

[25]

[26] [27]

[28]

[29]

[30]

[31]

J. Souquet-Grumey et al. / Journal of Membrane Science 466 (2014) 200–210

proton conductors for PEM fuel cells based on phosphonic acid as protogenic group: a progress report, Phys. Chem. Chem. Phys. 9 (2007) 1764–1773. M. Schuster, T. Rager, A. Noda, K.D. Kreuer, J. Maier, About the choice of the protogenic group in pem separator materials for intermediate temperature, low humidity operation: a critical comparison of sulfonic acid, phosphonic acid and imidazole functionalized model compounds, Fuel Cells 5 (2005) 355–365. E. Abouzari-Lotf, H. Ghassemi, A. Shockravi, T. Zawodzinski, D. Schiraldi, Phosphonated poly(arylene ether)s as potential high temperature proton conducting materials, Polymer 52 (2011) 4709–4717. R. Perrin, M. Elomaa, P. Jannasch, Nanostructured proton conducting polystyrene  poly(vinylphosphonic acid) block copolymers prepared via sequential anionic polymerizations, Macromolecules 42 (2009) 5146–5154. J. Parvole, P. Jannasch, Polysulfones grafted with poly(vinylphosphonic acid) for highly proton conducting fuel cell membranes in the hydrated and nominally dry state, Macromolecules 41 (2008) 3893–3903. M. Ingratta, M. Elomaa, P. Jannasch, Grafting poly(phenylene oxide) with poly (vinylphosphonic acid) for fuel cell membranes, Polym. Chem. 1 (2010) 739–746. A. Aslan, K. Gö lcük, A. Bozkurt, Nanocomposite polymer electrolytes membranes based on poly(vinylphosphonic acid)/SiO2, J. Polym. Res. 19 (2012) 1–8. C. Chen, C. Shen, G. Kong, S. Gao, High temperature proton exchange membranes prepared from epoxycyclohexylethyltrimethoxysilane and amino trimethylene phosphonic acid as anhydrous proton conductors, Mater. Chem. Phys. 140 (2013) 24–30. C. Laberty-Robert, K. Valle, F. Pereira, C. Sanchez, Design and properties of functional hybrid organic–inorganic membranes for fuel cells, Chem. Soc. Rev. 40 (2011) 961–1005. E.P. Jutemar, P. Jannasch, Copoly(arylene ether nitrile) and copoly(arylene ether sulfone) ionomers with pendant sulfobenzoyl groups for proton conducting fuel cell membranes, J. Polym. Sci. Part A: Polym. Chem. 49 (2011) 734–745. B. Ameduri, From vinylidene fluoride (VDF) to the applications of VDFcontaining polymers and copolymers: recent developments and future trends, Chem. Rev. 109 (2009) 6632–6686. F. Pereira, A. Chan, K. Vallé, P. Palmas, J. Bigarré, P. Belleville, C. Sanchez, Design of interpenetrated networks of mesostructured hybrid silica and nonconductive poly(vinylidene fluoride)–cohexafluoropropylene (PVdF–HFP) polymer

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40] [41]

[42] [43]

for proton exchange membrane fuel cell applications, Chem. – Asian J. 6 (2011) 1217–1224. F. Niepceron, B. Lafitte, H. Galiano, J. Bigarré, E. Nicol, J.F. Tassin, Composite fuel cell membranes based on an inert polymer matrix and proton-conducting hybrid silica particles, J. Membr. Sci. 338 (2009) 100–110. E. Labalme, G. David, P. Buvat, J. Bigarre, T. Boucheteau, New hybrid membranes based on phosphonic acid functionalized silica particles for PEMFC, J. Polym. Sci. Part A: Polym. Chem. 50 (2012) 1308–1316.0. O. Sel, T. Azais, M. Maréchal, G. Gébel, C. Laberty-Robert, C. Sanchez, Sulfonic and phosphonic acid and bifunctional organic–inorganic hybrid membranes and their proton conduction properties, Chem. – Asian J. 6 (2011) 2992–3000. M.-K. Song, X. Zhu, M. Liu, A triazole-based polymer electrolyte membrane for fuel cells operated in wild temperature range (25–150 1C) with little humidification, J. Power Sources 241 (2013) 219–224. E. Hü bner, J.r. Allgaier, M. Meyer, J.r. Stellbrink, W. Pyckhout-Hintzen, D. Richter, Synthesis of polymer/silica hybrid nanoparticles using anionic polymerization techniques, Macromolecules 43 (2010) 856–867. S.H. Kim, Y.C. Park, G.H. Jung, C.G. Cho, Characterization of poly(styrene-bvinylbenzylphosphonic acid) copolymer by titration and thermal analysis, Macromol. Res. 15 (6) (2007) 587–594. J. Bigarre, P. Buvat, R. Perrin, H. Galiano, Procédé de préparation de particules inorganiques conductrices de protons, PATENT FR2967925, 2010 November 30. K. Viswanathan, T.E. Long, T.C. Ward, Silicon surface modification with trialkoxysilyl-functionalized star-shaped polymers, J. Polym. Sci. Part A: Polym. Chem. 43 (2005) 3655–3666. L.T. Zhuravlev, The surface of amorphous silica. Zhuravlev model, Colloids Surf. A: Physicochem. Eng. Aspects 173 (2000) 1. T. Wagner, A. Manhart, N. Deniz, A. Kaltbeitzel, M. Wagner, G. Brunklaus, W. H. Meyer, Vinylphosphonic acid homo- and block copolymers, Macromol. Chem. Phys. 210 (2009) 1903–1914. N. Inagaki, K. Goto, K. Katsuura, Thermal degradation of poly(diethyl vinylphosphonate) and its copolymer, Polymer 16 (1975) 641. R.P. Pereira, M.I. Felisberti, A.M. Rocco, Intermolecular interactions and formation of the hydration sphere in phosphonic acid model systems as an approach to the description of vinyl phosphonic acid based polymers, Polymer 47 (2006) 1414–1422.