LoLiPEM: Long life proton exchange membrane fuel cells

LoLiPEM: Long life proton exchange membrane fuel cells

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4 Available online at www.sciencedirect.com S...

2MB Sizes 0 Downloads 15 Views

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

LoLiPEM: Long life proton exchange membrane fuel cells G. Barbieri a,*, A. Brunetti a, M.L. Di Vona b, E. Sgreccia b, P. Knauth c, H.Y. Hou c, R. Hempelmann d, F. Arena d,1, L.D. Beretta e, B. Bauer f,  g,2, L.F. Vega g,h,3 M. Schuster f, J.O. Osso a

National Research Council e Institute on Membrane Technology (ITMeCNR), Via Pietro BUCCI, c/o The University of Calabria, cubo 17C, 87036 Rende CS, Italy b  Roma “Tor Vergata”, Dip. Scienze e Tecnologie Chimiche, 00133 Roma, Italy Universita c Aix Marseille Universite, CNRS, Madirel (UMR 7246), Campus St Jer^ome, 13397 Marseille, France d Saarland University, Physical Chemistry, D-66123 Saarbru¨cken, Germany e EDISON e Centro Ricerca e Sviluppo di Trofarello, via G. LA PIRA 2, 10028 Trofarello TO, Italy f FuMA-Tech GmbH, Am Grubenstollen 11, 66386 St. Ingbert, Germany g MATGAS Research Center, Campus UAB, 08193 Bellaterra, Barcelona, Spain h  licos, Air Products Group, C/ Aragon 300, 08009 Barcelona, Spain Carburos Meta

article info

abstract

Article history:

This paper presents the main results obtained during the European project (FCH-JU)

Received 3 September 2015

“LoLiPEM e Long-life PEM-FCH & CHP systems”. The paper describes significant improve-

Received in revised form

ments in the polymer electrolyte by tailored heat treatments for cross-linking of Sulfonated

13 October 2015

Poly(ether ether ketone) (SPEEK), obtained without any addition of cross-linker species. The

Accepted 23 October 2015

reported properties of the ionomers include mechanical properties, gas permeability and

Available online 11 November 2015

ionic conductivity.

Keywords:

Pt catalyst; the fuel cell currentevoltage characteristics are reported with Nafion and

Cross-linked SPEEK

SPEEK-based binder.

Innovative gas-diffusion electrodes are fabricated by the electrochemical deposition of

Hydrogen cross-over

The fuel cell performances at 80  C of membrane-electrode-assemblies containing a

MEA preparation

SPEEK membrane with a cross-linking degree of 32% are among the best in the literature

SPEEK binder

compared with the PEMFC using membrane alternative than Nafion.

Gas Diffusion Electrode

Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: þ39 0984 492029; fax: þ39 0984 402103. E-mail addresses: [email protected], [email protected] (G. Barbieri). 1 Current address: MECADI GmbH, Industriegebiet in der Kolling, In der Kolling 9, 66450 Bexbach, Germany. 2 Current address: BASF New Business GmbH, BNB e Lu-Benckiserplatz BE 1, 67059 Ludwigshafen, Germany. 3  Empresarial, C/Tres Creus, 236, 08203 Sabadell, Barcelona, Current address: Alya Technology & Innovation, Centre de Promocio Spain. http://dx.doi.org/10.1016/j.ijhydene.2015.10.096 0360-3199/Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

1922

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

Introduction Stationary power generation and combined heat and power systems (SPG&CHP) based on fuel cells (FC) can help to reduce petrol dependence and CO2 emissions in the atmosphere. They can also be useful to decrease pollution, especially in large towns, and in the case of electrical black-out. These cogeneration systems can use very thin and flexible ionomer membranes, which exhibit high proton conductivity at relatively low temperature without any addition of mineral acids. The specific characteristics of the ionomer membranes facilitate the development of small co-generation systems, very suitable for small buildings. Although the potential of fuel cells is well established, the enthusiasm for this technology has decreased in recent years for various reasons, including its limited durability and the too high total price [1e6]. Research to further accelerate their deployment on the market should be focused on these two issues: increasing the durability and reducing the cost of the current systems [7,8]. It is therefore important to develop membranes with high durability and low price as well as new catalytic electrodes that are more stable at the operating temperature of the fuel cells. The proton exchange membrane (PEM) is typically phaseseparated into a percolating network of hydrophilic nanopores embedded in a hydrophobic polymer-rich phase domain [9e13]. The hydrophilic nano-pores contain acidic moieties, which ensure the proton conductivity. The hydrophobic phase domain provides mechanical strength to the membrane. Nowadays extensive research is devoted to finding ionomer membranes with high durability and low cost that can work at a higher temperature [14e17]. Sulfonated Aromatic Polymers (SAPs [11,18,19]) can be valid materials for this purpose, provided the degradation problems that affect this class of polymers are solved. The main distinctive feature of SAPs is the fact that the water filled channels are narrow and tortuous with a small separation between hydrophilic and hydrophobic domains. One positive effect is the permeability reduction for reactants, such as methanol or hydrogen. Conversely, the distance between adjacent sulfonic groups is large and, for this reason, SAPs need a higher ion exchange capacity (IEC) compared with perfluorinated ionomers to achieve the required conductivity [13,20]. However, the high IEC leads to morphological instability and large swelling at high humidity. Water uptake increases with the degree of sulfonation (number of SO3H groups per repeat unit) thereby improving the conductivity of the hydrated membrane, but highly polar water molecules act also as a plasticizer, undermining the electrostatic interactions between SAP macromolecular chains and favoring membrane swelling. Highly sulfonated aromatic polymers swell rather strongly in water and become even soluble if the sulfonation degree is high enough [21,22]; this has made the long-term stability of highly sulfonated membranes questionable until now. Cross-linking reactions are one of the most powerful ways to control and improve the properties of polymeric materials [23e27], such as swelling and mechanical behavior [28e30]. However, reticulation often relies on the presence of cross-

linker species and special procedures, such as grafting by irradiation [23,31e35]. Furthermore, some cross-linker species might be attacked under the harsh conditions of an operating fuel cell [16]. The possibility of achieving reticulation between chains via sulfone bridges, directly by thermal treatment of cast membranes using sulfonic acid groups already present in SAP, is an appealing way to obtain stable ionomeric membranes [36,37]. This method also allows overcoming problems associated with the insolubility of cross-linked polymers in common solvents that make the casting procedure difficult. Furthermore, cross-linking by thermal treatment without any addition of cross-linker molecules can be performed in a costeffective mode from an industrial point of view. Another objective is the development of more efficient noble metal electrocatalysts, reducing the platinum loading of the electrodes significantly [38,39] and, in this way, the overall cost of the fuel cell. The common preparation technique for catalyst layers in fuel cell electrodes starts from carbon black covered with platinum (pre-catalysation) [38e46]. Thus only low catalyst utilization can be obtained, because a considerable fraction of the catalyst material is not in direct contact with the three-phase boundary. The objective of our work is to increase the catalyst utilization by localizing the catalysts particles exclusively in the three-phase boundary. This is only possible a posteriori: a platinum precursor salt is brought into the microlayer during the layer preparation and platinum is site-selectively electrochemically deposited in situ on the carbon surface, without expensive outer plating baths. With improvements in the deposition route described in the following sections, non-aggregated platinum nanoparticles with diameters down to 2 nm can be prepared directly in the microporous layer (MPL) of Gas Diffusion Layers (GDLs), thus transforming GDLs into Gas Diffusion Electrodes (GDEs) [50]. In this way, both noble metal costs and processing costs can be saved, because the electrochemical route to GDEs exhibits fewer processing steps than the conventional route. A further advantage is that in the same way platinum alloy catalysts can easily be prepared, even in medium-throughput. This paper presents examples of membrane improvement by thermal cross-linking of Sulfonated Poly(ether ether ketone) (SPEEK), including mechanical, hydrolytic, electrical and mass transport properties, the innovative preparation of GDEs by electrodeposition of platinum catalyst as well as characteristics of the assembled fuel cells.

Experimental Membrane preparation Sulfonated Poly(ether ether ketone) (SPEEK) was prepared by reaction of PEEK (Victrex 450P, MW ¼ 38,300 g/mol) with concentrated sulfuric acid (96%) in ratio 1:35 (g mL1) under N2 at 50  C for 2e4 days in order to obtain a large degree of sulfonation. After this time, the solution was poured, under continuous stirring, into an excess of ice-cold water thereby obtaining a white precipitate [47]. After resting overnight, the precipitate was filtered and washed in a dialysis membrane (Sigma-Aldrich D9402) until neutral pH in order to eliminate the residual sulfuric acid completely. The sulfonated polymer

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

1923

(SPEEK) was then dried at 60e80  C. In this way, SPEEK was obtained with a degree of sulfonation (DS) in the range 0.75e0.90. The polymer was then dissolved in dimethylsulfoxide (DMSO). The polymer/DMSO ratio was around 1:100 (g/mL). After evaporation to around 1/3 of the original volume, the solution was spread on a glass plate using a doctor-blade type equipment and then put in the oven at 120  C for 24 h. After cooling to room temperature, the resulting membranes were peeled off and used for the thermal cross-link. Pilot-scale production of SPEEK membranes from DMSO solutions was achieved on the continuous membrane production line at FuMA-Tech GmbH. The production process included coating of the polymer solution onto a carrier foil, followed by a threestage drying process. The process of pilot-scale membrane production is fully compatible with standard large-scale membrane production enabling the rapid transfer and integration of the new technology into the standard operation of manufacturing processes.

determined in a closed glass container after removal of excess of water on the surface by carefully wiping the membranes with filter paper. The mechanical properties were determined with an Adamel-Lhomargy traction machine (M250-2.5CT) at room temperature and humidity. The tensile tests were carried out using samples with 25 mm length and 5 mm width. The crosshead speed was 5 mm/min. Special adhesive tape was used to hold the samples and assure rupture in the center of the samples. The proton conductivity was determined at 25  C in fully humidified conditions by impedance spectroscopy (EG&G model 6310) with an AC amplitude of 20 mV at frequencies between 1 Hz and 100 kHz. The wet membrane samples were sandwiched between two stainless steel electrodes (throughplane configuration) in a closed Swagelok cell. The membrane resistance was determined from the real axis intercept of the electrode arc and converted into the proton conductivity s using the equation:

Membrane cross-linking



Membrane

The thermal cross-linking procedure is as follows: the cast membrane containing 1e2 DMSO molecules per sulfonic acid group was treated in an oven at 180  C for 3, 7 or 14 h. After this treatment, all membranes were immersed in H2O2 during 1 h, then in 5 M H2SO4 during 2 h and finally rinsed with water. The sample treated for 14 h was further immersed for 24 h at 100  C in a solution of 5% DMSO in water. The degree of cross-link (DXL) can be calculated from the ion exchange capacity before (IEC ) and after (IEC) thermal treatments, determined by titration: DXL ¼

IEC0  IEC IEC0

(1)

Membrane characterization The degree of sulfonation was measured after swelling the membrane in water at 100  C for 5 h to fully remove DMSO that could modify the equivalent weight. The titration was performed as follows: acid-form samples with given dry weight were soaked in 1.5 M NaCl solution one day to exchange Hþ with Naþ and the protons were back-titrated with 0.02 M NaOH solution. The ion exchange capacity (IEC) was calculated using the dry weight mdry of the sample and the quantity of exchanged protons. The dry weight was measured after storing the membranes for at least 3 days over P2O5. The water uptake coefficient (hydration numbers, l) was measured by weighting membrane samples in dry and wet conditions: l¼

mwet  mdry mdry $IEC$18

(2)

The membrane was immersed in water contained in a closed Teflon vessel at various temperatures for at least 24 h and then stabilized at 25  C for 24 h. The wet weight mwet was

Thickness

Membrane

Resistance

$AElectrode

(3)

The electrode area was 0.19 cm2. The proton conductivity of the sample with a DXL ¼ 32, showing the best fuel cell performances (see below), was also measured as a function of temperature at 90% relative humidity. In this case the conductivity measurements were done in a home-made apparatus [37]. The mass transport properties of the membranes were determined by gas permeation measurements with H2, N2 and O2. Given the dependence of the membrane performance on the relative humidity and temperature, a new approach for the systematic evaluation of the mass transport properties was proposed, introducing a protocol [48] for permeation measurements to compare accurately the transport properties of different membranes as a function of the operating conditions. Summarizing, the permeance and selectivity of membranes were investigated feeding pure H2, O2 and N2 at different temperatures and trans-membrane pressure differences, in conditions as close as possible to the real application of the membrane in the MEA. Each gas was fed at different values of relative humidity (50e100%) in a temperature range of 80e120  C. The experimental set up used for the permeation measurements is described in detail elsewhere [48]. Once the permeating flow rate, retentate and permeate pressure were measured, the mass transport properties of the membranes were calculated using the following equations: Permeating fluxi ¼

Permeate flow ratei 1 ; LðSTPÞm2 h membrane area

Driving force ¼ DPTM ¼ PRetentate  PPermeate ; bar i i i Permeancei ¼

Permeating fluxi 1 1 ; LðSTPÞm2 h bar DPTM i

(4) (5)

(6)

1924

Permeabilityi ¼

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

Permeating fluxi 1 1 thicknessMembrane ; LðSTPÞmm2 h bar DPTM i

The standard temperature and pressure (STP) are 0  C and 100 kPa (1 bar). Eqs. (4) and (5) describe the permeating flux and the corresponding driving force for the ith species permeating through the membrane. The permeance, defined as the ratio of the permeating flux and the driving force, can be calculated by Eq. (6). The permeability is defined as the permeance multiplied by the membrane thickness (Eq. (7)). In this work, the permeance was considered instead of the permeability, in order to take into account the slight membrane thickness changes that can occur according to the relative humidity, temperature and trans-membrane pressure difference. However, the permeability was used for comparison with Nafion 117, owing to the different thickness of the membrane samples. In this calculation, the thickness of the dry membrane was assumed as reference value.

New concept Gas Diffusion Electrodes (GDEs) Non-woven carbon fibre paper TGP-H-90 from Toray was employed as GDL, offering a thickness of 280 mm and 78% porosity. The preparation of the 50 cm2 GDEs implied the following steps: a) ink preparation, b) ink coating, c) Pt precursor impregnation, d) pulsed potentiostatic electrodeposition, and e) leaching of the GDE.

Ink preparation For SPEEK containing inks (called MPL-S) typical solid fractions amounted to 2e2.5 wt%. In a first step, half of the total amount of the dispersion agent isopropyl alcohol was poured into a plastic beaker. The PTFE dispersion (60 wt%, from Dupont) and the high surface area carbon Ensaco 350G (Timcal) were added. An initial dispersion was done by ultrasonication for 5 min. After adding the remaining isopropyl alcohol and the SPEEK solution (5 wt% in DMAc), the ink was vigorously stirred at 20,000 rpm for 45 min using a dispersing unit from IKA (T 25 digital ULTRA-TURRAX®). To avoid thermally induced evaporation of the dispersion agent, the ink container was cooled to room temperature utilizing a thermoset water bath; in addition a specific silicon sealed lid was employed. After 45 min the stirring was diminished to 10,000 rpm and maintained until the further processing of the ink. The amounts during the ink preparation were chosen in order to obtain a 7.5:1.5:1 weight ratio among carbon, SPEEK and PTFE with regard to the dry MPL [49]. Nafion containing inks (called MPL-N) were prepared analogously to the aforementioned MPL-S preparation. In contrast, typical solid fractions amounted to 7%. Furthermore, the Nafion/carbon/PTFE ratio amounted preferably to 3.5:5.5:1.0 [50].

Ink coating The inks prepared in this way were coated on the GDL using a spiral blade (200 mm). After evaporation of the remaining

(7)

solvent at 80  C for 30 min, the sample was allowed to cool down to room temperature.

Pt precursor impregnation A further step to the preparation of GDEs was the impregnation of MPL-coated GDL preferably with a quadrivalent Pt precursor such as Hexachloroplatinic Acid (HCPA). For this purpose, HCPA was dissolved in a first step in the appropriate amount of isopropyl alcohol (HPLC grade). The Pt precursor solution was poured into a leveled vessel. The sample to be impregnated was then dipped with the MPL side first into the precursor solution. The sample was dried overnight under vacuum (~104 bar). To avoid humidity induced migration of the Pt ions, the samples were stored in vacuum desiccators until the deposition process. Immediately prior to the electrochemical deposition, the samples were removed from the desiccator and were stored over n-hexane before mounting them in the deposition cell.

Pulsed potentiostatic electrodeposition The Pt catalyst was electrodeposited starting from the precursor-impregnated MPL. Therefore, the coated GDL was arranged as Working Electrode and was pressed in a fuel-celllike setup onto a half MEA consisting of Nafion 115 and a Johnson Matthey Counter Electrode (1.5 mg Pt cm2) acting as Hydrogen Depolarized Anode. During the electrodeposition process, the hydrogen depolarized anode was fed continuously with 40 mL min1 of humidified hydrogen (dew point ¼ 25  C) and served thus as both counter electrode and, in the sense of a Dynamic Hydrogen Electrode, as reference electrode. The protons produced were transported through the PEM to the cathode side where the reduction of the Pt precursor took place. To avoid water induced migration of the Pt ions during deposition, which would lead to undesired catalyst agglomerates, the cathode side was purged with 100 mL min1 of dry nitrogen. Furthermore, the technique implied the use of a potentiostatic double pulse including a nucleation phase (50 mV versus Dynamic Hydrogen Electrode for 1 ms) and a growth phase (150 mV versus Dynamic Hydrogen Electrode for 100 ms); subsequently, the potential was set for 50 ms to the Open Circuit Potential. Further experimental details concerning this pulsed potentiostatic electrodeposition for precursor reduction by a Hydrogen Depolarized Anode were previously described [50]. The related Pt loading was determined by Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES) measurements. For this purpose, 2 cm2 of the deposited GDE were reduced to ashes. Acid digestion of the residue led to a clear solution, which was analyzed after appropriate dilution and addition of an internal standard (Zn 10 ppm).

GDE leaching After the deposition procedure and the dismounting of the cell, the so-prepared GDE was removed, whereas the semi-

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

MEA was reused for further electrodepositions. In all cases the activation of the Pt surface, as well as the removing of remaining non-reduced Pt precursor, was attained by several purification steps after the catalyst deposition procedure. For this purpose, the GDE was alternately leached in 1 M sulfuric acid, in 2 M isopropyl alcohol and washed in deionized water at room temperature. A fundamental problem related to the conventional SPEEK containing GDE preparation is the coating of the electrocatalyst nanoparticles by a thin SPEEK layer causing a significant decrease in fuel cell performance. This problem was avoided using this technique and fuel cell performance results were obtained similar to the state-of-the-art GDEs containing Nafion as ionomer.

MEA Preparation General Setup The experimental hot pressing setup for MEA preparation consisted of a customized Enerpac hydraulic workshop press based on a VLP256PAT1 pressing frame. The force between upper and lower pressing plate was adjusted manually using a P-392 hand pump. The plate temperature was independently set (up to 250  C) using a 3216 process-controller from Eurotherm.

Nafion-based MEAs For MEAs utilizing Nafion as PEM, the sandwich to be assembled was introduced in the pre-heated hot press. Hot pressing was carried out at 125  C at 500 N cm2 for 6 min. Cooling down to room temperature was done using a water cooling system with a specific cooling rate of 30  C min1, while the force between the plates was perpetuated to avoid delamination phenomena.

Cross-linked SPEEK-Based MEAs Prior to the hot pressing of cross-linked SPEEK, the membrane was reinforced by the use of a self-adhesive polymer frame (Fumatech). For MEAs based on cross-linked SPEEK, an optimized hot pressing temperature of 190  C was applied. The corresponding pressure of 500 N cm2 was maintained for a net pressing time of 6 min, which was measured from the time that the pressing temperature was reached. Pre-heating from room temperature to the set pressing temperature was done including the cross-linked SPEEK-based sandwich under applied pressure and was denoted as slow heating. Whereas slow cooling implied cooling down to room temperature without any cooling liquid; the associated cooling rate amounted to 1  C min1.

MEA preconditioning prior fuel cell testing Chronopotentiometric preconditioning is the activation of the MEA by use of both water vapor and in situ formed water on the cathode side due to the electrochemical reactions. Therefore, the fuel cell housing was set to 60  C and the current density was increased within 4 h from zero to 1 A cm2. After 2 h of operation at 1 A cm2, the operating temperature was increased every 2 h by 10  C increments up to 80  C. In the last step, the current density was reduced to zero within 1 h

1925

using constant current density steps. During the experiment, the relative humidity for both inlet gases (H2; grade: 5.0 and O2; grade: 4.5) was held at 95%. Stoichiometry was constant during the preconditioning step and the accounted ratios of the provided to the consumed hydrogen and oxygen were 1.1 and 1.2, respectively.

Fuel cell testing The MEAs were assembled into a single fuel cell housing (50 cm2) to test their performance. Polarization experiments were carried out using a modular test bench consisting of an electronic load (Electrochem), gas mass flow controllers (Sierra Instruments) and a temperature controller (Micromega). All components were controlled using a LabView®-based homemade software. Reactant stoichiometry was kept constant during the experiments. Typically employed ratios of the provided to the consumed hydrogen and oxygen were inbetween 1.0e1.5 and in-between 1.2e2.0, respectively. Hydrogen (grade: 5.0) was used as fuel and oxygen (grade: 4.5) as oxidant. Both reactant gases were fed humidified to the fuel cell. The setup for gas humidification consisted of a boiling water reservoir, a condensing unit (MC-tech) and a heated line (Horst) connected to the fuel cell. The condenser units were kept at constant temperature using cooling thermostats (Lauda & Julabo); the temperature of the heated lines was preferably 5  C higher than the corresponding dew point in order to exclude condensation inside the supply lines. Steam related parameters such as relative humidity and dew point were checked using a dew point mirror from Michell. Polarization curves were collected preferably in a potential range between Open Circuit Potential and 400 mV. Galvanostatic operation mode was used for all polarization curves. Typical current steps amounted to 2e5 A with corresponding time steps in the range of 300e600 s. For a given set of operating parameters, quasi-stationarity of the currentevoltage characteristics (CVC) was reached after 4e5 cycles, corresponding to an operation time of at least 10 h. For mid-term testing, the aforementioned test bench was employed. Short-term tests were based on polarization experiments; mid-term fuel cell testing (up to 25 h) was based on a chronopotentiometric technique. Further details of the test bench can be found in a previous paper [50].

Results and discussion Thermally cross-linked SPEEK membranes The recently described cross-linking mechanism [28,51] is an electrophilic aromatic substitution, where sulfonium ions attack the aromatic rings of adjacent macromolecular chains. Fig. 1 shows the schematic formula of cross-linked SPEEK. Table 1 reports the characteristics of the studied membranes, including IEC, equivalent weight, degree of crosslinking and hydration numbers. Obviously the degree of cross-linking increases with the time of thermal treatment and the ionic exchange capacity and hydration number decrease. Table 2 reports the mechanical properties and the proton conductivity before and after various thermal cross-

1926

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

HO3S O

HO3S

O

O O

O O

O

S

O

O O

Fig. 2 e Relation between elastic modulus E and equivalent weight M of cross-linked SPEEK membranes.

Fig. 1 e Formula of cross-linked SPEEK. linking times. The equivalent weight and the elastic modulus increase with the time of thermal treatment showing progressive cross-linking.

Mechanical properties The enhancement of the elastic modulus can be explained by the lower average distance between macromolecular chains, owing to the SO2 cross-linking bridges, which enhances the van der Waals interactions [52]. The relation between the elastic modulus and the equivalent weight of cross-linked polymers was established in early work by Flory [27] and Gregor [53] using a statistical thermodynamic approach. Applied to our membranes, this relation can be expressed as: E ¼ 3RTrMx1=3

(8)

In this equation, R is the gas constant, T the absolute temperature, r the density of the dry polymer (r(SPEEK) z 1.3 g/cm3) and x the volume fraction of polymer in

Table 1 e Ionic exchange capacity (IEC), equivalent weight (M), degree of cross-linking (DXL) and hydration numbers (l) at 100  C for investigated SPEEK membranes. Samples As received 180  C 3 h 180  C 7 h 180  C 14 ha a

IEC, meq g1

M, g eq1

DXL

l

2.70 2.30 2.07 1.84

370 434 483 543

0 15 23 32

∞ 910 100 44

With DMSO as plasticizer.

the hydrated membrane. The plot of the elastic modulus E vs. the equivalent weight M (Fig. 2) should then be a straight line with a slope of the order of 3RTrx1/3, which is in remarkable agreement with the experiment. The linearity shows that an approximately constant polymer volume fraction x (means also a similar hydration number) can be assumed for all samples. This is not surprising, as Gregor [54] showed a long time ago that the water uptake below 75% of relative humidity is very similar for various degrees of cross-linking. The influence of cross-linking is felt only above this point, as in our case when the membranes are fully hydrated by immersion in liquid water. In the mechanical tests, the relative humidity was always in the 40e60% range. Large changes of the elastic modulus are in fact observed when the hydration numbers vary strongly, as shown previously for the case of Nafion [55] and SPEEK [52]. The enhanced tensile strength (Table 2) is related to the formation of supplementary strong covalent cross-linking bonds. This is a good point for reduced swelling under high humidity and, together with dimensional stability related to dimensional swelling and water uptake, can imply better durability under changing humidity conditions. Conversely, the elongation at break decreases with the reticulation (Table 2): the polymer is stiffer and less plastic. This is a critical point and careful operation is necessary during the hot pressing procedure and in the fuel cell in order to avoid membrane fracture. One can modulate the mechanical properties, maintaining a good hydrolytic stability but reducing the brittleness, by the use of a plasticizer: DMSO is an excellent candidate because it shows a good compatibility with membranes and is homogeneously sorbed. Samples thermally

Table 2 e Mechanical properties and proton conductivity at 25  C (full humidification) of SPEEK membranes thermally treated at 180  C for various times. Samples Untreated 180  C 3 h 180  C 7 h 180  C 14 h 180  C 14 ha a

Young modulus (E), MPa 1080 1160 1480 1640 920

With DMSO as plasticizer.

± 100 ± 100 ± 100 ± 100 ± 20

Tensile strength (TS), MPa 43 46 50 52 43

± 10 ±5 ±5 ±2 ±1

Elongation at break (ε), %

Proton conductivity, S cm1

140 ± 10 110 ± 10 80 ± 50 90 ± 10 e

0.014 0.018 0.024 0.020 e

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

Fig. 3 e Proton conductivity (s) of cross-linked SPEEK as function of the hydration number (l). Experimental values (o: this work, D: ref. [58]); calculated values according to ref. [58] (solid line).

treated for 14 h and afterwards immersed in a solution of H2O/ DMSO show intermediary mechanical properties (Table 2), but are stable in boiling water in contrast with uncross-linked membranes. The plasticizer is used during the MEA assembly for fuel cell tests.

Proton conductivity As previously shown [56,57], the amount of water present in the membrane is decisive for the membrane conductivity, because the proton mobility shows a strong power law dependence on the proton concentration. The memory effect, which is the membrane ability to “remember” the water uptake reached at high temperature also at lower temperature, was exploited in order to achieve high values of conductivity. The proton conductivity after hydration at 100  C is shown as a function of the hydration number (l) at 25  C in Fig. 3; the intermediate maximum of conductivity corresponds to the best compromise of proton concentration and proton mobility (which decreases strongly with increasing proton concentration [56,57]). The maximum proton conductivity at 25  C

(a)

1927

calculated using this dependence is above 0.02 S cm1 at a hydration number around 100 [58]. This is actually very near the hydration number obtained for 7 h heat treatment (Table 1) and the observed high proton conductivity is consistent with the calculation. For larger hydration numbers (e.g. l ¼ 910 in Table 2), the conductivity is lower. The distinctly lower conductivity of the as-received sample is simply related to the lower temperature of hydration (25  C): this uncrosslinked membrane cannot be hydrated at 100  C, because it dissolves under these conditions. For larger cross-linking times, the hydration number is lower and the proton conductivity decreases (Table 1). Overall, calculated and measured proton conductivity values are in very good agreement (Fig. 3). Using the “memory effect” of the membranes, hydration at high temperature is thus a very powerful way to keep a high conductivity (and actually improve it with respect to the untreated membrane), although some sulfonic acid groups were lost by the cross-linking reaction. One can also assume that the cross-linking process reduces the tortuosity of the membranes [57]. The conductivity values measured at the operating temperature of fuel cells tests (see below) at 90% RH were 0.045 S cm1 at 80  C, 0.05 S cm1 at 90  C and 0.052 S cm1 at 105  C.

Mass transport properties Firstly, the mass transport properties of the membrane were measured only for H2 as a function of the relative humidity at the three investigated temperatures. At 80  C, the permeance was significantly reduced as the relative humidity increased (Fig. 4). A low relative humidity corresponds to a low water uptake, thus, in other words to less free volume occupied by water molecules which hinders the passage of the other smaller molecules. As the relative humidity increases, the greater water uptake provokes a change in the polymer matrix, with higher fraction of free volume occupied by water, and thus, a reduced diffusive transport of the gas. As expected, the hydrogen permeance increased with the temperature given that the permeation of gaseous species through the membrane is an activated process that follows the Arrhenius law. Given that the membrane is a dense medium,

(b)

Fig. 4 e H2 permeance of a cross-linked SPEEK membrane (14 h) as a function of (a) relative humidity at 80  C and (b) temperature at relative humidity ¼ 100%.

1928

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

(a)

(b)

Fig. 5 e Comparison of different membranes H2 permeability (XL-SPEEK 14 h, Nafion 117) as a function of (a) relative humidity at 80  C and (b) temperature at relative humidity ¼ 100%. the mass transport through it follows the solution diffusion mechanism. This means that the permeability is defined as the product of solubility and diffusivity and the permeation of gases is thus function of temperature, gas type, and affinity between the gas and the polymer. It must be emphasized that during the permeation measurements, the cross-linked SPEEK (14 h) membrane withstood drastic changes in operating conditions and exhibited a significant resistance, since, apart from the hydrothermal changes induced by the variation of relative humidity and temperature, it was exposed to a supplementary stress induced by the high pressure, necessary to create the driving force responsible for the permeation, and not present in the fuel cell. This is an indication of dimensional stability related to dimensional swelling and water uptake which, together with the good mechanical strength, suggest a good durability of the membranes under stressed conditions. The transport properties of the cross-linked SPEEK membrane were compared with those of Nafion 117, which is currently the most-used polymer electrolyte in PEMFC and, thus, a reference material. As already mentioned, the results were analyzed in terms of permeability instead of permeance, owing to the different thicknesses of the two membranes.

Cross-linked SPEEK membranes exhibited generally a lower permeability than Nafion 117 at all the investigated operating conditions, except for a relative humidity of 50% at 80  C, where the H2 permeability was higher than the Nafion 117 one (Fig. 5a). The permeability was significantly reduced at higher relative humidities, whereas the Nafion 117 permeability was quite constant in the whole considered relative humidity range. Similar trends were obtained at 100  C. In addition, the permeability of both cross-linked SPEEK and Nafion increases (Fig. 5b) with temperature. In conclusion, cross-linked SPEEK showed a lower permeability than Nafion 117 indicating better barrier properties and, thus, reduced crossover affecting the PEMFC performance [59].

Fig. 6 e XRD pattern related to (microporous layer) MPL-S with Miller indices of the lattice planes corresponding to Pt.

Fig. 7 e TEM image of the deposited electrocatalyst on high surface area carbon support.

SPEEK containing GDEs SPEEK containing (MPL-S) and Nafion containing (MPL-N) GDEs were prepared following the route described in Section 2.4. For the MPL-S-based GDEs an average Pt loading of (0.52 ± 0.06) mg cm2 was determined via ICP-OES. Further XRD (Fig. 6) and TEM (Fig. 7) analysis revealed an average crystallite size of 9 nm. In this context, the Scherrer analysis

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

was performed using the Pt [220] reflection at 2 Q ~ 67.7 . For Nafion containing MPL, a typical average Pt crystallite size of 7 nm was observed [50]. Conventionally employed Pt/C powders contain Pt in a crystallite size range of 3e5 nm. Larger catalyst particles are in general less active, because of the decreased catalytically active surface area for a given Pt loading. In contrast, smaller catalyst crystallites are more subject to growth during fuel cell operation especially at temperatures >100  C [60]. An additional TEM analysis of MPLS confirmed the aforementioned results obtained from XRD. For the investigated MPL, a low agglomeration degree with agglomerates in the range of 20 nm was detected consisting of crystallites with sizes in the range 5e10 nm (Fig. 7). Furthermore, the MPL-S surface morphology was investigated by means of Scanning Electron Microscopy (SEM) (Fig. 8). From the SEM images, the typical mud-cracked structure of the coated MPL is revealed. In this context, the mud-cracked structure is thought to improve the transport of reactant gases as well as of reaction water through the MPL and therefore to reduce the mass transport related voltage losses at high current densities [61]. Typical crack lengths were in the range of 20e40 mm. The typical MPL thickness amounted to 18e25 mm; it was investigated by means of cross-sectional SEM images of an entire MEA (Fig. 9). The new concept GDEs manufactured in this way were evaluated by means of polarization experiments, using the electrodeposited MPL-S- and MPL-N-based GDEs as anode. Together with a commercial Nafion containing cathode from Johnson Matthey (ELE0162; 0.4 mg Pt cm2) and a Nafion 212 membrane, the MEAs were prepared via a hot pressing process (see Section 2.5). Fig. 10 shows the fuel cell performance results in the temperature range between 60-up to 110  C. Each currentevoltage curve represents the quasi-stationary state, which was generally reached after 4e5 h of operation at the corresponding operating temperature. Further operating conditions related to the experiment are mentioned in the header as well as in the legend of the corresponding graph. For the investigated MPL-S, a loss in performance was observed with increasing temperature, especially for

Fig. 8 e SEM image of the prepared (microporous layer) MPL-S with a 100-fold magnification. Related information dealing with acceleration voltage, magnification factor and scale are summarized in the bottom line.

1929

Fig. 9 e Cross sectional SEM image of Nafion 212 membrane assembled with SPEEK containing GDEs. Corresponding information concerning utilized acceleration voltage, magnification factor and scale are summarized in the bottom line of the image. temperatures >100  C. This was mainly attributed to the low relative humidity of the reactants leading to dehydration of the ionomer membrane and therefore to a significant increase of the ohmic resistance and to the low reactant partial pressure in absence of back-pressure. MPL-S offered a nearly identical fuel cell performance at 60  C and 80  C indicating good mass transport properties at high current densities. The comparison of the SPEEK containing MPL-S with the Nafion containing MPL-N at 80  C revealed the competitive performance of the SPEEK containing GDE. It has to be noticed that the comparison in Fig. 10 concerns the binder polymer material in the microporous layer. Standard (commercial) GDEs were designed to “cooperate” with Nafion membranes; therefore the electro-catalyst nanoparticles are “glued” to the GDL by Nafion originating from a Nafion solution/dispersion (not originating from the membrane). Nafion has a very high permeability for oxygen and

Fig. 10 e Polarization (circles) and CWC (squares) in the temperature range 60e110  C with Nafion 212 membrane. Ratio of the provided to the consumed hydrogen ¼ 1.2, ratio of the provided to the consumed oxygen ¼ 2.0. The black curves are related to the Nafion containing benchmark microporous layer. dp is the dew point. The color code and the operating conditions of each curve are summarized in the headline and the legend.

1930

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

hydrogen gas; therefore, even if the electro-catalyst nanoparticles are completely coated with Nafion binder, the gas molecules can reach them by permeation. In this sense, the high permeability of Nafion represents an advantage, but it is a severe disadvantage for the membrane, because it is the reason for a decrease of the OCV and it accelerates aging processes. Oxygen gas on the anode side leads to peroxy radicals, which are harmful for the polymer. SPEEK, in contrast, has very low gas permeability, which is an advantage for the membrane but it is a severe problem for the catalyst layer: electro-catalyst nanoparticles which are completely coated by SPEEK binder are less accessible for hydrogen and oxygen molecules. Therefore, our efforts were directed to the development of a route for the catalyst loading of the GDL which avoids this problem. With electro-catalyst nanoparticles in a SPEEK binder matrix we reach the same performance as in a Nafion binder matrix. In the next step of development, see Section 3.3, we then proceed to use a SPEEK membrane. It is important to have the same polymer materials in the membrane and in the microporous layer of the GDE to avoid delamination during long term operation, and we reached this goal in two steps: first step Section 3.2, second step Section 3.3. Our results show the successful transfer of the pulsed potentiostatic electrodeposition technique to SPEEK containing systems, which is an innovation with regard to the disadvantageous conventional process, where the noble metal catalyst is already added in the catalyst ink. In this context, the developed pulsed potentiostatic electrodeposition technique offers a high degree of versatility with regard to the ionomer utilized within the MPL. The separation of the MPL coating from the catalyst deposition (separate process steps) offers numerous advantages, such as:  increase in the degree of catalyst utilization,  prevention of the covering of catalyst nanoparticles with a thin layer of SPEEK amplifying the mass transport limitation, a common problem for the conventional preparation route,  independent optimization of the MPL including hydrophobicity, micro- and macro-porosity. In view of the promising fuel cell performance, additional reproducibility tests were carried out using three independently prepared SPEEK containing GDEs. The related

Table 3 e Current densities at an operating point of 650 mV in the temperature range 60e110  C for MEAs based on four independently prepared GDEs containing SPEEK as ionomeric binder MPL-S(a)e(d). T,  C

60 80 90 100 110

Current density at 650 mV, A cm2

Deviation

MPL- MPL- MPL- MPL- Mean ± deviation S(a) S(b) S(c) S(d)

%

0.78 0.78 e 0.23 0.10

0.80 0.70 0.39 e 0.15

0.74 0.62 0.45 0.30 0.15

0.73 e e e e

0.76 0.70 0.42 0.27 0.13

± 0.03 ± 0.08 ± 0.04 ± 0.05 ± 0.03

3.9 11.4 9.5 34.8 30.8

polarization curves were measured in a temperature range between 60 up to 110  C. For all measurements the reactant gas stoichiometry was kept constant and the accounted ratios of the provided to the consumed hydrogen and oxygen were 1.2 and 2.0, respectively for oxygen. Reactant gas humidification was set preferably to 100% relative humidity. For operating temperatures >85  C, a dew point of 85  C, representing the upper limit of the gas humidifier, was maintained. As index for the reproducibility of the overall GDE preparation process, the average current density and the related standard deviation for a given operating temperature were determined at the reference potential 650 mV. The results are summarized in Table 3. The polarization experiments employing independently prepared SPEEK anodes revealed the reproducibility of the entire GDE preparation process offering a FC performance in the same range as Nafion containing GDEs. In the operating temperature range of 60e90  C, relative span in the order of 10% were determined: a satisfying result in view of the nonautomated preparation process. For operating temperatures >100  C, the reasonable deviation in the range of 30% was attributed to the non-ideal operating conditions of the MEA. The non-appropriate humidification level of the inlet reactant gases is thought to induce the so-called dry out phenomenon of the PEM; furthermore, the reactant gas partial pressures were reduced due to the limitation to 1 bar total pressure.

SPEEK as PEM in MEAs The increase in the degree of cross-linking has beneficial effects on the thermal, mechanical and chemical properties of SPEEK membranes. Unpublished MEA performance tests revealed that increasing the degree of cross-linking made them more susceptible for incisions at the PEM/GDE border caused by the induced stiffness and brittleness of the material. The arising dilemma turned out to become amplified by an unsuitable preparation/operation of the MEA. In order to minimize mechanical stress and therefore to maximize MEA lifetime and performance, the main parameters were identified, screened and optimized. In this context, a reinforcing polymeric frame, an appropriate preconditioning technique, an optimized hot pressing procedure as well as modified fuel cell testing conditions were developed to reduce the induced mechanical stress prior and during the FC operation. Fig. 11 shows the time dependent CVCs and a mechanical failure leading to the End-of-Life of an MEA based on crosslinked SPEEK. After the initial increase of the fuel cell performance (within the first 6 h of operation), an abrupt loss in maximum current density and Open Circuit Potential is observed. Noticeably, the performance loss did not go along with an increase in ohmic resistance indicating constant membrane conductivity without any dry out phenomena and the absence of any delamination-related degradation of the MEA. Dismounting of the MEA revealed an incision at the GDE/ PEM border, identified as the cause for the observed behavior and probably also for the relatively low initial OCV. Mid-term FC performance tests were carried out using cross-linked SPEEK membranes (thickness: 30 mm) with thermal treatment times of 7 h and 14 h (see Table 1). Test MEAs were prepared assembling the cross-linked SPEEK membranes with

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

Fig. 11 e Polarization curves as a function of the operating time for an MEA based on cross-linked SPEEK employing commercial Nafion-based GDEs. Ratio of the provided to the consumed hydrogen ¼ 1.1, ratio of the provided to the consumed oxygen ¼ 1.2. The color transition from yellow to orange to red indicates a progressing operating time. In this context yellow represents the initial situation and red the End-of-Life. Detailed parameters related to the polarization experiment are given on top of the graph. Top right of the graph all measured values for the applied current density (green), the voltage (blue), and the MEA ohmic resistance (orange) are represented as a function of time. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 12 e Chronopotentiometric measurement of an MEA based on a cross-linked SPEEK membrane treated 7 h at 180  C. Ratio of the provided to the consumed hydrogen ¼ 1.1, ratio of the provided to the consumed oxygen ¼ 1.2 Current (red curve) was increased stepwise with time, while the related voltage response of the FC was measured (blue curve). Furthermore, the FC operating temperature was varied in-between 80e95  C. dp is the dew point. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

1931

Fig. 13 e Chronopotentiometric measurement of an MEA based on a cross-linked SPEEK membrane treated 14 h at 180  C. Ratio of the provided to the consumed hydrogen ¼ 1.1, ratio of the provided to the consumed oxygen ¼ 1.2. Current (red curve) was increased stepwise with time, while the related voltage response of the FC was measured (blue curve). Furthermore, the FC operating temperature was varied in-between 80e105  C. dp is the dew point. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

two commercial Nafion-containing GDEs from Johnson Matthew (ELE0162; 0.4 mg Pt cm2) via hot pressing. Preconditioning prior to FC testing was done using the chronopotentiometric routine (5e6 h). FC testing included an initial chronopotentiometric experiment (20e24 h) followed by polarization measurements (20e30 h). In both cases, the operating temperature was varied. Figs. 12 and 13 show the results for the initial chronopotentiometric experiments. The current density (red curve) was increased stepwise before the operating temperature was raised and the related voltage signal was collected (blue curve). The dew points on the anode and cathode were adjusted to maintain 95% relative humidity up to an operating temperature of 90  C. At 95  C a dew point of 90  C was used as the limiting dew point of the gas humidifier system working at atmospheric pressure. Up to 90  C a constant fuel cell performance was observed up to a maximum current density of 1.5 A cm2. In order to prevent the MEA from drying out, the operating temperature was increased under a high current load in order to use the produced water on the cathode side to maintain high humidity and therefore high conductivity of the cross-linked SPEEK material. A further increase of operating temperature to 95  C led to a significant voltage loss combined with high voltage fluctuations. It is well known that the fuel cell performance is strongly affected by the operating temperature and humidity; in this work, these two parameters were systematically studied in order to trace the performance trends. An additional experiment was carried out using an MEA based on highly cross-linked SPEEK (14 h). In the initial chronopotentiometric measurement (Fig. 13) the influence of the operating temperature in the range of 60-up to 110  C and of the relative humidity of the inlet gases was studied with focus

1932

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

on fuel cell performance and lifetime. For this purpose, the current density was kept constant for a certain time period and the corresponding voltage signal was recorded. Various operating conditions were studied in order to determine optimal conditions assuring the best performance. Up to a temperature of 97.5  C the project aim with regard to fuel cell performance (0.7 A cm2 at 650 mV) was accomplished. Operation at higher temperatures was accompanied by a significant loss of voltage for comparable current densities. This was mainly due to the low relative humidity of the inlet gases caused on the one hand by the dew point limitation of the gas humidifier and on the other hand by the limit of maximum water content in the feed gases for operation at atmospheric pressure. In addition comparative polarization experiments were carried out at 80  C and 90  C. For this purpose the aforementioned MEAs based on cross-linked SPEEK were compared with a Nafion-based MEA. Identical commercial electrode materials were utilized for all MEAs. The corresponding results are represented in Fig. 14. The excellent good performance of the cross-linked SPEEK-based MEAs was revealed from the polarization experiments at 80  C. In the potentiostatically-deposited GDE used in this work, the result of the specific platinum surface was of 516 cm2 mg1, and, analogously to what observed in a previous work [50], it was ca. 20% higher than the values of the electrodeposited GDE, but the fuel cell performance of the potentiostaticallyelectrodeposited GDE was around 10% better than this reference GDE. As detailed in [50], these differences in the fuel cell activity can be attributed to better contact to the electron conducting phase.

Fig. 14 e Polarization results for MEAs based on crosslinked SPEEK membranes and a Nafion 212 membrane (triangles). The experiment was carried out in an operating temperature range of 80e90  C applying the parameters summarized in top of the graph. Annealing time of the utilized cross-linked SPEEK samples amounted to 7 h (squares) and 14 h (circles), respectively. Information related to the experimental conditions is summarized in the headline of the graph. Ratio of the provided to the consumed hydrogen ¼ 1.2, ratio of the provided to the consumed oxygen ¼ 2.0. dp is the dew point.

Both cross-linked SPEEK membranes showed a higher fuel cell performance than the Nafion-based system. Noticeably, the cross-linked SPEEK (14 h) with higher cross-linking degree gave a better result in terms of fuel cell performance. At 650 mV a current density amounting to 1.3 A cm2 was measured. The cross-linked SPEEK (7 h) offered a current density of 1.1 A cm2, whereas the Nafion-based system showed the lowest performance with 1.0 A cm2. Increasing the operating temperature to 90  C the fuel cell performance decreased, especially in the high current density region of the CVC, indicating a non-appropriate structure of the utilized MPLs. For both cross-linked SPEEK-based membranes, a current density in the order of 0.8 A cm2 was determined. In addition to the increased FC performance, the preparation/operating parameter optimization increased significantly the lifetime of the cross-linked SPEEK-based MEAs in short- and mid-term FC tests. Therefore, the End-of-Life time could be significantly increased from initially 5e6 h to a point where no End-of-Life was observed within 40 h of operation. The chronopotentiometric experiments revealed the heavy dependence of the FC performance on the humidification level of the inlet gases, especially at temperatures close to and higher than 100  C. The water-assisted proton conduction of the cross-linked SPEEK membranes can be significantly increased by increasing the relative humidity of the reactant gases, which necessitates the use of operating pressures higher than 1 atm.

Conclusions Thermal cross-linking of SPEEK membranes was achieved by heat treatments at 180  C for 3e14 h following the membrane casting. The best results were obtained for a cross-linking degree of 32%, corresponding to a thermal treatment of 14 h. This procedure, combined with the addition of a small quantity of DMSO acting as plasticizer, leads to improved mechanical properties, so that the membranes show less swelling and better durability under changing humidity. The elastic modulus dependence on the equivalent weight is in remarkable agreement with the theoretical prediction. The cross-linked membranes can be hydrated in boiling water and, using the memory effect, this is the way to keep (or actually even to improve) the proton conductivity of crosslinked membranes, although sulfonic acid groups are lost during the cross-linking reaction. These are major results of the LoLiPEM project, which show the way to further SAP membrane improvement in the future. A second major result is the fabrication of gas-diffusion electrodes (GDE) by electrochemical deposition of Pt catalyst directly on the membrane. The main advantage is the excellent good utilization of the catalyst so that its quantity can be reduced. Furthermore, innovative GDE containing SPEEK instead of Nafion as ion-conducting binder were also prepared. The fuel cell characteristics with SPEEK as binder are comparable to or even better than those with Nafion. The fuel cell tests using an optimized cross-linked SPEEK membrane also give excellent interesting results up to 80  C; at a temperature up to 110  C, the low relative humidity and low partial pressure of reactants, owing to the experimental

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

apparatus limitation on atmospheric pressure, shows leads to an increase of the membrane resistance and a decrease in the fuel cell characteristics. This problem might be solved by using a higher cell back pressure conditions.

Acknowledgment The project has received funding from the European Union Seven Programme for Research, Technological Development and Demonstration through FCH-JU under GA no. 245339 “LoLiPEM e Long-life PEM-FCH &CHP systems at temperatures higher than 100  C”.

references

[1] Gamburzev S, Appleby AJ. Recent progress in performance improvement of the proton exchange membrane fuel cell (PEMFC). J Power Sources 2002;107:5e12. [2] Devanathan R. Recent developments in proton exchange membranes for fuel cells. Energy Environ Sci 2008;1:101e19. [3] Costamagna P, Srinivasan S. Quantum jumps in the PEMFC science and technology from the 1960s to the year 2000: part I. Fundamental scientific aspects. J Power Sources 2001;102:242e52. [4] Li QF, He RH, Jensen JO, Bjerrum NJ. Approaches and recent development of polymer electrolyte membranes for fuel cells operating above 100  C. Chem Mater 2003;15:4896e915. [5] Shao Y, Yin GP, Wang ZB, Gao YZ. Proton exchange membrane fuel cell from low temperature to high temperature: material challenges. J Power Sources 2007;167:235e42. [6] Barbir F. PEM fuel cells: theory and practice. 2005. p. 1e433. [7] Wu JF, Yuan XZ, Martin JJ, Wang HJ, Zhang JJ, Shen J, et al. A review of PEM fuel cell durability: degradation mechanisms and mitigation strategies. J Power Sources 2008;184:104e19. [8] Marrony M, Barrera R, Quenet S, Ginocchio S, Montelatici L, Aslanides A. Durability study and lifetime prediction of baseline proton exchange membrane fuel cell under severe operating conditions. J Power Sources 2008;182:469e75. [9] Kreuer K, Paddison S, Spohr E, Schuster M. Transport in proton conductors for fuel-cell applications: simulations, elementary reactions, and phenomenology. Chem Rev 2004;104:4637e78. [10] Mauritz KA, Moore RB. State of understanding of nafion. Chem Rev 2004;104:4535e85. [11] Park CH, Lee CH, Guiver MD, Lee YM. Sulfonated hydrocarbon membranes for medium-temperature and lowhumidity proton exchange membrane fuel cells (PEMFCs). Prog Polym Sci 2012;36:1443e98. [12] Kreuer KD. On the development of proton conducting materials for technological applications. Solid State Ion 1997;97:1e15. [13] Kreuer KD. On the development of proton conducting polymer membranes for hydrogen and methanol fuel cells. J Membr Sci 2001;185:29e39. [14] Steininger H, Schuster M, Kreuer KD, Kaltbeitzel A, Bingoel B, Meyer WH, et al. Intermediate temperature proton conductors for PEM fuel cells based on phosphonic acid as protogenic group: a progress report. Phys Chem Chem Phys 2007;9:1764e73. [15] Alberti G, Casciola M, Massinelli L, Bauer B. Polymeric proton conducting membranes for medium temperature fuel cells (110e160 C). J Membr Sci 2001;185:73e81.

1933

[16] Collier A, Wang HJ, Yuan XZ, Zhang JJ, Wilkinson DP. Degradation of polymer electrolyte membranes. Int J Hydrogen Energy 2006;31:1838e54. [17] Jannasch P. Recent developments in high-temperature proton conducting polymer electrolyte membranes. Curr Opin Colloid Interface Sci 2003;8:96e102. [18] Bauer B, Jones DJ, Roziere J, Tchicaya L, Alberti G, Casciola M, et al. Recent developments in high-temperature proton conducting polymer electrolyte membranes. J New Mater Electrochem Syst 2000;3:93e8. [19] Roziere J, Jones DJ. Non-fluorinated polymer materials for proton exchange membrane fuel cells. Annu Rev Mater Res 2003;33:503e55. [20] Liu B, Kim DS, Guiver MD, Kim YS, Pivovar BS. Sulfonated poly(aryl ether)-type polymers as proton exchange membranes: synthesis and performance. In: Peinemann Klaus-Viktor, Nunes Suzana Pereira, editors. Membrane for energy conversion. Geesthacht: Wiley-VCH; 2008. p. 1e45. [21] Xing PX, Robertson GP, Guiver MD, Mikhailenko SD, Wang KP, Kaliaguine S. Synthesis and characterization of sulfonated poly (ether ether ketone) for proton exchange membranes. J Membr Sci 2004;229:95e106. [22] Silva VS, Ruffmann B, Silva H, Gallego YA, Mendes A, Madeira LM, et al. Proton electrolyte membrane properties and direct methanol fuel cell performance: I. Characterization of hybrid sulfonated poly (ether ether ketone)/zirconium oxide membranes. J Power Sources 2005;140:34e40. [23] Kerres JA. Blended and cross-linked ionomer membranes for application in membrane fuel cells. Fuel Cells 2005;5:230e47. [24] Park KT, Chun JH, Kim SG, Chun BH, Kim SH. Synthesis and characterization of crosslinked sulfonated poly(arylene ether sulfone) membranes for high temperature PEMFC applications. Int J Hydrogen Energy 2011;36:1813e9. [25] Han MM, Zhang G, Li MY, Wang SA, Zhang Y, Li HT, et al. Considerations of the morphology in the design of proton exchange membranes: cross-linked sulfonated poly(ether ether ketone)s using a new carboxyl-terminated benzimidazole as the cross-linker for PEMFCs. Int J Hydrogen Energy 2011;36:2197e206. [26] Hou HY, Di Vona ML, Knauth P. Building bridges: crosslinking of sulfonated aromatic polymers e a review. J Membr Sci 2012;423:113e27. [27] Flory PJ, Rehner J. Statistical mechanics of cross linked polymer networks. J Chem Phys 1943;11:521e6. [28] Di Vona ML, Sgreccia E, Licoccia S, Alberti G, Tortet L, Knauth P. Analysis of temperature-promoted and solventassisted cross-linking in sulfonated poly(ether ether ketone)(SPEEK) proton-conducting membranes. J Phys Chem B 2009;113:7505e12. [29] Sgreccia E, Chailan JF, Khadhraoui M, Di Vona ML, Knauth P. Mechanical properties of proton-conducting sulfonated aromatic polymer membranes: stressestrain tests and dynamical analysis. J Power Sources 2010;195:7770e5. [30] Hou HY, Di Vona ML, Knauth P. Durability of sulfonated aromatic polymers for proton-exchange-membrane fuel cells. ChemSusChem 2011;4:1526e36. [31] Chen XW, Gholamkhass B, Han X, Vamvounis G, Holdcroft S. Polythiophene-graft-styrene and polythiophene-graft(styrene-graft-C60) copolymers. Macromol Rapid Commun 2007;28:1792e7. [32] Mikhailenko SD, Robertson GP, Guiver MD, Kaliaguine S. Properties of PEMs based on cross-linked sulfonated poly (ether ether ketone). J Membr Sci 2006;285:306e16. [33] Chen JH, Maekawa Y, Asano M, Yoshida M. Double crosslinked polyetheretherketone-based polymer electrolyte

1934

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44] [45] [46]

[47]

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 1 9 2 1 e1 9 3 4

membranes prepared by radiation and thermal crosslinking techniques. Polymer 2007;48:6002e9. Zhong SL, Cui XJ, Cai HL, Fu TZ, Zhao C, Na H. Crosslinked sulfonated poly(ether ether ketone) proton exchange membranes for direct methanol fuel cell applications. J Power Sources 2007;164:65e72. Kerres J, Cui W, Disson R, Neubrand W. Development and characterization of crosslinked ionomer membranes based upon sulfinated and sulfonated PSU crosslinked PSU blend membranes by alkylation of sulfinate groups with dihalogenoalkanes. J Membr Sci 1998;139:211e25. Di Vona ML, Sgreccia E, Tamilvanan M, Khadhraoui M, Chassigneux C, Knauth P. High ionic exchange capacity polyphenylsulfone (SPPSU) and polyethersulfone (SPES) cross-linked by annealing treatment: thermal stability, hydration level and mechanical properties. J Membr Sci 2010;354:134e41. Di Vona ML, Alberti G, Sgreccia E, Casciola M, Knauth P. High performance sulfonated aromatic ionomers by solvothermal macromolecular synthesis. Int J Hydrogen Energy 2012;37:8672e80. Ticianelli EA, Beery JG, Srinivasan S. Dependence of performance of solid polymer electrolyte fuel cells with low platinum loading on morphologic characteristics of the electrodes. J Appl Electrochem 1991;21:597e605. Paganin VA, Ticianelli EA, Gonzalez ER. Development and electrochemical studies of gas diffusion electrodes for polymer electrolyte fuel cells. J Appl Electrochem 1996;26:297e304. Cheng X, Yi B, Han M, Zhang J, Qiao Y, Yu J. Investigation of platinum utilization and morphology in catalyst layer of polymer electrolyte fuel cells. J Power Sources 1999;79:75e81. Gasteiger HA, Kocha SS, Sompalli B, Wagner FT. Activity benchmarks and requirements for Pt, Pt-alloy, and non-Pt oxygen reduction catalysts for PEMFCs. Appl Catal B Environ 2005;56:9e35. Taylor EJ, Anderson EB, Vilambi NRK. Preparation of highplatinum-utilization gas diffusion electrodes for protonexchange-membrane fuel cells. J Electrochem Soc 1992;139:L45e6. Eikerling M, Kornyshev AA. Electrochemical impedance of the cathode catalyst layer in polymer electrolyte fuel cells. J Electroanal Chem 1999;475:107e23. Litster S, McLean G. PEM fuel cell electrodes. J Power Sources 2004;130:61e76. Wilson MS. Membrane catalyst layer for fuel cells. Patent US5211984 A. 1993. O'Hayre R, Lee SJ, Cha SW, Prinz FB. A sharp peak in the performance of sputtered platinum fuel cells at ultra-low platinum loading. J Power Sources 2002;109:483e93. Di Vona ML, Sgreccia E, Licoccia S, Khadhraoui M, Denoyel R, Knauth P. Composite proton-conducting hybrid polymers:

[48]

[49]

[50]

[51]

[52]

[53] [54]

[55]

[56]

[57]

[58]

[59]

[60]

[61]

water sorption isotherms and mechanical properties of sulfonated PEEK and substituted PPSU blends. Chem Mater 2008;20:4327e34. Brunetti A, Fontananova E, Donnadio A, Casciola M, Di Vona ML, Sgreccia E, et al. New approach for the evaluation of membranes transport properties for polymer electrolyte membrane fuel cells. J Power Sources 2012;205:222e30. Arena F, Mitzel J, Hempelmann R. Electrocatalyst e membrane interface and fuel cell performance with sulfonated poly ether ether ketone as ionomer. Z fur Phys Chem 2013;227:615e26. Mitzel J, Arena F, Natter H, Walter T, Batzer M, Stefener M, et al. Electrodeposition of PEM fuel cell catalysts by the use of a hydrogen depolarized anode. Int J Hydrogen Energy 2012;37:6261e7. Maranesi B, Hou H, Polini R, Sgreccia E, Alberti G, Narducci R, et al. Cross-linking of sulfonated poly(ether ether ketone) by thermal treatment: how does the reaction occur? Fuel Cells 2013;13:107e17. Hou HY, Maranesi B, Chailan JF, Khadhraoui M, Polini R, Di Vona ML, et al. Crosslinked SPEEK membranes: mechanical, thermal, and hydrothermal properties. J Mater Res 2012;27:1950e7. Lazare L, Sundheim BR, Gregor HP. A model for cross-linked polyelectrolytes. J Phys Chem 1956;60:641e8. Gregor HP, Sundheim BR, Held KM, Waxman MH. Studies on ion-exchange resins. V. Water vapor sorption. J Coll Sci 1952;7:511e34. Alberti G, Narducci R, Sganappa M. Effects of hydrothermal/ thermal treatments on the water-uptake of nafion membranes and relations with changes of conformation, counter-elastic force and tensile modulus of the matrix. J Power Sources 2008;178:575e83. Knauth P, Sgreccia E, Donnadio A, Casciola M, Di Vona ML. Water activity coefficient and proton mobility in hydrated acidic polymers. J Electrochem Soc 2011;158:B159e65. Knauth P, Di Vona ML. Sulfonated aromatic ionomers: analysis of proton conductivity and proton mobility. Solid State Ion 2012;225:255e9. Di Vona ML, Pasquini L, Narducci R, Pelzer K, Donnadio A, Casciola M, et al. Crosslinked sulfonated aromatic ionomers via SO2 bridges: conductivity properties. J Power Sources 2013;243:488e93. Arena F, Mitzel J, Hempelmann R. Permeability and diffusivity measurements on polymer electrolyte membranes. Fuel Cells 2013;13:58e64. Zhang S, Yuan XZ, Hin JNC, Wang H, Friedrich KA, Schulze M. A review of platinum-based catalyst layer degradation in proton exchange membrane fuel cells. J Power Sources 2009;194:588e600. Seland F, Berning T, Børresen B, Tunold R. Improving the performance of high-temperature PEM fuel cells based on PBI electrolyte. J Power Sources 2006;160:27e36.