Development of a reference electrode for a PEMFC single cell allowing an. evaluation of plate potentials

Transcription

1 Development of a reference electrode for a PEMFC single cell allowing an evaluation of plate potentials Johan ANDRE 1, Nicolas GUILLET 2, Jean-Pierre PETIT 3, Laurent ANTONI 4, * 1 CEA LITEN 17 rue des Martyrs 3854 Grenoble cedex 9 johan_g_andre@yahoo.fr 2 CEA LITEN 17 rue des Martyrs 3854 Grenoble cedex nicolas.guillet@cea.fr 3 LEPMI/ENSEEG/UMR CNRS/INPG/UJF 5631 Domaine Universitaire, BP 75 F-3842, 3842 St Martin d'hères Cedex jean-pierre.petit@enseeg.inpg.fr 4 CEA - LITEN 17 rue des Martyrs 3854 Grenoble cedex laurent.antoni@cea.fr Abstract Increasing lifetime and performance is critical for PEMFC using stainless steel plates. A good compromise between passivity and electrical contact resistance of the plate material is required. Measuring the potential of each plate during fuel cell operation is of paramount importance to lead to relevant ex situ tests in order to investigate new materials. From a review on methods used for potential measurement, the present work focused on the realization and use of a Dynamic Hydrogen Electrode (DHE) device as a reference electrode in a PEMFC single cell, its evaluation in terms of accuracy and drift. With classic reference electrodes introduced into the flow field, measurements were shown to be irrelevant because of the impossibility to ensure good and stable ionic conductivity between the reference electrode and the plate when operating the cell. Several examples of DHE found in the literature were reviewed and used to realize a DHE which showed correct accuracy and stability of its potential under fully humidified conditions. The experimental device was shown to be reliable and easily adaptable for different single cells. It was used to investigate transient phenomena while cycling a cell, but needs some improvement when the cell is operated with unsaturated gases. Keywords Reference electrode; DHE; PEMFC; corrosion; bipolar plate; fuel cell 1

2 1. Introduction Among the new technologies for energy, fuel cells and in particular PEMFC (Proton Exchange Membrane Fuel Cell) represent an attractive solution, convenient for many applications [1]. However, some technical issues still have to be solved: increasing lifetime while maintaining performance is of major interest in the challenge for PEMFC development. Bipolar plate is a key component of PEMFC, with electrical, chemical, thermal, and mechanical functions [2-5]. Some efforts are focused on finding a compromise between passivity and electrical contact resistance of passive films formed on stainless steel bipolar plates [6]. Cathodic and anodic bipolar plate voltages are usually considered to reach respectively about.74 to.84 V/SHE (for standard use) and.4 to.1 V/SHE [7-9]. Nevertheless, based on previous studies of Andreaus et al. [1,11] about humidification aspects and analysis of performance losses at high current densities, Kuhn et al. [12] claimed that anode could contribute to some extent to the cell voltage variation. To perform electrochemical studies which reproduce as well as possible fuel cell operating conditions, it is critical to determine precisely these values. Little information about reference electrodes suited for PEMFC was published elsewhere [13-2]. The kinds of reference electrode presented in this paper were selected in order to get a device with a reduced or null perturbation to the cell, and an improved accuracy. An original version is the DHE (Dynamic Hydrogen Electrode) [18,21] which requires two platinum wires. At the surface of one of them, hydrogen is produced by electrolysis. If stable, the current enables good hydrogen coverage of the cathodic wire and a stable electrode potential. First studies about DHE date back to 1964 (Giner et al. [22]). As main advantages, this kind of electrode was shown to be convenient to use and non intrusive, because of the absence of salt bridge and foreign ions. A simpler version of hydrogen electrode named RHE (Reversible Hydrogen Electrode) was shown to be useful [13], but hydrogen coverage is not steady on its surface, and adsorption of contaminants is possible. Unlike pseudo reference electrodes [12,17] presented elsewhere, the potential of a DHE can be referred to a classical 2

3 reference electrode so that not only overvoltage values can be determined, but also absolute values of potentials i.e. referable to any reference electrode. Some recommendations [17-19] were given by researchers for the use of both kinds of electrodes such as providing a good alignment between cathode and anode and/or using several reference electrodes [21], positioning the electrode in an area of stable potential (at least further than 3x membrane thickness), and in an edge-type configuration rather than in a sandwich-type. In order to get an easy-to-use reference device, Siroma et al. [2] even proposed a DHE device hot pressed within the MEA (membrane electrode assembly), but such a system, with the reference electrode very close to the MEA, implies the use of a high electrolysis current and an increased difference from the equilibrium potential. Operation of commercial electrodes was tried in order to get directly the potential of bipolar plates versus a classical reference electrode. Then, with the help of previous works, a DHE easily adaptable for different single cells was developed. 2. Experimental 2.1 Single cell configuration and use Single cell tests were performed with AISI 316L monopolar plates (thickness: 8mm, width x length: 9 mm x 8 mm with machined single serpentine channel. Homemade 25cm² 3-layers membrane electrodes assemblies (MEA) were employed using a Nafion 112 membrane. Pt loading was about.3 mg/cm² for each electrode (anode / cathode). Gas diffusion layers were prepared with 1% Teflon Freudenberg felt and homemade micro porous layer. Unless stated different, all experiments were conducted feeding the cell by pure and fully humidified hydrogen and oxygen or air gases. In order to get information about potential evolution during transient states, a current cycle was defined with some rapid variations (increase and decrease) of the current load with five minute steps, as shown on Figure 1. 3

4 2.2 Commercial reference electrodes A first idea was to introduce a reference electrode into a hydrogen flow channel. Due to the limited space available, microelectrodes were used: WPI Flexref with flexible PTFE body and WPI RC-6 silver chloride half cell with glass body. The first model presents a limited leakage rate (5.7*1-8 ml/h) and is quite easy to handle, whereas the second one shows an excessive brittleness but no leakage. The electrode is introduced through a tight connector shown in Figure 2, close to an outlet of the cell. Because the fuel cell is fed with fully humidified gases, and due to the short distance (<1mm) between the reference electrode and the bipolar plate, ohmic drop is supposed to be negligible. 2.3 Realization of a Dynamic Hydrogen Electrode During the next step of our work, two models of DHE were realized with PTFE insulated platinum wires (125µm diameter). In the first version, the extremities of the platinum wires were naked whereas they were coated with Pt 6wt%/C black ink in the second one in order to improve the quality of the Nafion / DHE interface [2]. A 13A Marconi Adret source was used to provide a stable current for DHE operation. A special device allowed good contact between each Pt wire and the membrane, and offered the possibility to connect the wires on the side of the plates, in order to reduce tightness problems. DHE was positioned far enough from the active area to reduce perturbations due to fuel cell operation (hydration and temperature gradients ) taking into account the single cell geometry, ohmic drop, and risks of membrane drying (distance superior to 3 * membrane thickness, according to Li et al. [18] recommendations). Before using the single cell, tightness was verified under air, as well as the absence of shortcircuit between Pt wires and monopolar plates, and between Pt wires themselves. Figure 3 gives an overview of the experimental device. 4

5 2.4 Evaluation of our Dynamic Hydrogen Electrode device. The reference electrode is in contact with the protonic membrane, itself in contact with the active layer; so that we get rid of Donnan potentials [23] encountered with classical reference electrodes (with internal reference solution). Moreover, the assembly Active Layer / Microporous Layer / Gas Diffusion Layer is electrically conductive and in contact with the monopolar plate so that the potential of one of these components is representative for the others, with a small difference due to ohmic drop (Figure 4). However, if important, the electrolysis current used to ensure hydrogen coverage induces a non negligible deviation from the equilibrium potential (hydrogen production over voltage) so that the reference electrode potential may differ slightly from that of SHE (Standard Hydrogen Electrode). A Biologic VMP2 multipotentiostat was used to measure DHE potential versus MSE (Mercurous Sulfate Electrode) in a.5 M sulphuric acid solution for different values of current. The device shown in Figure 5 was operated at 6 C to reproduce fuel cell conditions. The 5-layer MEA (membrane, electrodes and gas diffusion layers) was immersed into sulphuric acid. In order to generate proton reduction at the surface of the DHE, this electrode was used as counter electrode while the other Pt wire was the working electrode. Counter electrode potential was recorded vs. MSE. 3. Results and discussion 3.1 Use of commercial microelectrodes Figure 6 represents a polarization curve with the continuous acquisition of anodic and cathodic potentials vs. Flexref. The evolution of cathodic potential is quite abnormal: for small currents, the cathodic potential is close to the anodic one then increases up to 9mV at high current. The same kind of abnormal potential variation is observed with dry RC-6 reference electrodes. Measurements may be disturbed by gas flow at the surface of reference electrode. These classical commercial microelectrodes do not seem to be suitable for fuel cell use. Figure 7 shows the evolution of Open Circuit Voltage, cathodic and anodic potentials vs. RC-6 Reference electrode with time in presence of residual hydrogen and oxygen. Low 5

6 voltage values were attributed to slight gas leakages. In these conditions, the difference between cathodic and anodic potentials gives a value close to the cell voltage with an error about 4 to 5mV, whereas with the gas valve aperture, potentials (particularly the anodic one) suddenly drift to abnormally high values, which seems to confirm the idea that measures seem impossible in presence of gas flow (modifications in ionic conduction and transport of species). 3.2 Preliminary realization of a DHE Figure 8 shows first results obtained with a DHE, fed with a voltage source (potentiostatic control). The difference between cathodic and anodic potentials gives the cell voltage with an error lower than 2 mv. Nevertheless, a strong instability about 15 mv was noticed on each potential, invisible on the polarization curve. This phenomenon is due to the difficulty encountered to generate a stable current during cell operation. The membrane resistance changes so that maintaining a particular voltage does not allow to use a stable current and a continuous hydrogen coverage of the DHE. 3.3 Optimization of the existing device Figure 9 and Figure 1 are representations of the polarization curves after previously described modifications (especially use of a current source for galvanostatic control) respectively under oxygen and under air. The difference between cathodic and anodic potential corresponds now to the cell voltage with an error lower than 1 mv. And above all, the very small and stable current used for electrolysis (5 µa) shows that anodic potential at OCV is close to V/DHE, which means that the reference potential is very close to that of SHE. A clearly different evolution of anodic potential vs. current density is revealed under air and oxygen. In a relative sense, anodic contribution to the global voltage loss is higher under oxygen. This phenomenon is put in evidence in Figure 11. Indeed, under pure oxygen, the quantity of oxygen available is so high that gas diffusion is not the limiting step: voltage loss is due to charge transfer and ohmic drop whose contribution is mainly membrane-dependent. 6

7 On the other hand, under air at high current, nitrogen hinders oxygen access to cathodic sites, and oxygen diffusion becomes more limiting than other phenomena (charge transfer, and ohmic losses), resulting in a stable anodic potential. 3.4 Characterization of the DHE With the device described in Figure 5, an ex situ measurement of the DHE potential vs. MSE was performed. Results are reported on Figure 12. Using an electrolysis current of 5 µa, the hydrogen production over voltage associated with possible differences in proton and hydrogen activity vs. SHE are responsible for a difference about -1 mv compared with SHE. This seems compatible with the polarization curve presented on Figure 8. Moreover, in situ use of the potentiostat was helpful to see the evolution of the voltage between platinum wire electrodes vs. electrolysis current. This diagnostic should be done before and during use to make sure of the initial quality of the DHE device and then to control its possible drift. 3.5 Use of the DHE in investigating PEMFC transient phenomena Figure 13 and Figure 14 show respectively the cell voltage evolution during current cycling of a single cell under oxygen and under air. Cathodic and anodic potentials were followed vs. DHE. Under oxygen, the anodic potential remains steady (33 to 55 mv), whereas cathode is mainly responsible for cell voltage variations. No particular evolution of electrode potentials was noticed during current changes. On the other hand, because of the time required for air flow regulation under air, abnormal evolution of cell voltages were evidenced during a sudden current rise. Figure 15 shows that a quick consumption of fuel and air (when starting to produce current) induces a cell voltage drop from 91 to 31 mv for some seconds. During this short period, cathodic potential decreases strongly, and anodic potential increases slightly. This means that electrode potentials (and so bipolar plate potentials) become closer to (or remains into) the passivity domain of 316L stainless steel, as shown on Figure 16. Transient phenomena do not seem to play a detrimental role in corrosion of bipolar plate materials. 7

8 Because anode contribution to cell voltage variations under classic conditions seems light, we tried to increase it modifying the fuel stoichiometric factor, in order to induce fuel starvation and see its effect on anodic potential. Results are shown on Figure 17. Under the test conditions, anode contribution becomes critical only for a stoichiometric factor inferior to.8 when cell voltage is already very low (<4 mv). In this case, anode potential reaches 1 mv. This is not the case under usual operating conditions (St H 2 = 1.2): for a cell voltage about 385 mv, cathode reaches 42 to 444 mv and anode 17 to 59 mv respectively, instead of 476 and 91 mv. Figure 18 and Figure 19 show the evolution of cell voltage, and each electrode voltage during an on-off cycle (-2A, steps of 5 min during 1 h), respectively with 1%RH and 5%RH gases. Whereas Figure 19 reveals that the actual device is not operative under reduced humidity conditions (strong and fast drift), Figure 18 shows that cycling does not induce sharp voltage variations of electrode potentials. The kickup effect shown to be problematic by Mitsuda et al. [24] in the field of PAFC, is not evidenced for PEMFC, even with reduced fuel or oxygen stoichiometry. Conclusions Several solutions of reference electrodes suited for PEMFC were reviewed. Commercial reference microelectrodes inserted directly into the flow field channel were shown inadequate for this application because of the impossibility to ensure a good ionic conduction even under optimum hydration conditions. A DHE device, developed from the state-of-the-art available data, was characterized and optimized to distinguish anodic and cathodic contributions to cell voltage evolution. Investigation of transient states did not show potential variations which could increase corrosion of bipolar plate materials. 8

9 Tracking electrolysis voltage during DHE operation may help in drift analysis, and some improvement in cell conception to retain water in contact with the reference electrode could enable to use of this device with dry gases and/or higher temperatures. References [1] W. Vielstich, A. Lamm, H. A. Gasteiger, Handbook of fuel cell: fuel cell fundamentals, technology, and applications part 1 (23) [2] J. S. Cooper, J. Power Sources 129 (24) [3] A. Kumar, Materials, design, and modelling for bipolar/end plates in polymer electrolyte membrane fuel cells, A Dissertation. Department of Metallurgical and Materials Engineering, University of Alabama (24) [4]. V. Mehta and J. S. Cooper, J. Power Sources 114 (23) [5]. R. Mosdale, in Techniques de l'ingénieur traité Génie électrique D 557 [6] A. Agneaux, M.H. Plouzennec, L. Antoni, J. Granier, 2nd Deutschland Fuel Cell Conference FDFC Belfort (24) [7] M. Li et al., Corros. Sci. 46 (24) [8] R. F. Silva et al., Electrochim. Acta 51 (26) [9] H. Wang and J. A. Turner, J. Power Sources 128 (24) [1] B. Andreaus and G. G. Scherer, Solid State Ionics 168 (24) [11] B. Andreaus, A. J. McEvoy, G. G. Scherer, Electrochim. Acta 47 (22) [12] H. Kuhn, B. Andreaus, A. Wokaun, G. G. Scherer, Electrochim. Acta 51 (26)

10 [13] A. Küver, I. Vogel, W. Vielstich, J. Power Sources 52 (1994) 77-8 [14] V. Paganin, E. Sitta, T. Iwasita, W. Vielstich, J. Appl. Electrochem. 35 (25) [15] A. Taniguchi, T. Akita, K. Yasuda, Y. Miyazaki, J. Power Sources 13 (24) [16] S. B. Adler, J. Electrochem. Soc. 149 (22) E166-E172 [17] W. He and T. V. Nguyen, J. Electrochem. Soc. 151 (24) A185-A195 [18] G. Li and P. G. Pickup, Electrochim. Acta 49 (24) [19] Z. Liu, J. S. Wainright, W. Huang, R. F. Savinell, Electrochim. Acta 49 (24) [2] Z. Siroma et al., J. Power Sources 156 (26) [21] G. Li and P. G. Pickup, Electrochem. Solid-State Lett. 9 (26) A249-A251 [22] J. Giner, J. Electrochem. Soc. 111 (1964) [23] A. J. Bard and L. R. Faulkner, Electrochemical methods: fundamentals and applications John Wiley & sons, inc., Ed. 2 (2) [24] K. Mitsuda and T. Murahashi, J. Appl. Electrochem. 21 (1991)

11 1 current density (ma/cm²) time (min) Figure 1 : Current cycle used for transient phenomena characterization Figure 2 : Photographies of the modified experimental device with a Flexref reference microelectrode 11

12 DHE in contact with Nafion membrane Voltage source # 2V 1 µa Figure 3 : Schematic representation of the experimental DHE device 316L GDL/MPL/AL +water Pt wire+h2+ H2O+Nafion Φ E Φ1 Φ2 Φ due to contact resistance Φref Figure 4 : measuring chain between monopolar plate and reference electrode 12

13 Counter electrode (DHE) Working electrode (other Pt wire) Reference electrode (Mercury Mercurous Sulfate Electrode) 5 layer MEA H 2 O 2 H 2 SO 4,5M Heating stirrer Figure 5 : Device for measuring DHE potential 9 7 V (mv) 5 3 cell voltage (mv) cathodic plate potential/ref AgCl (mv) anodic plate potential/ref AgCl (mv) current density (ma/cm²) Figure 6 : Example of polarization curve with 25cm² single cell test, H 2 /air, Stoichiometric factors 1.2/2. 1%RH, 1.5bar absolute, 6 C 13

14 cell voltage (mv) cathodic plate potential / AgCl ref (mv) anodic plate potential / AgCl ref (mv) Vc-Va (mv) V (mv) gas valve aperture time (min) Figure 7 : E = f(t) with 1% humidified residual gases 25cm² single cell test, H 2 /air, room temperature and atmospheric pressure V (mv) cell voltage (mv) Vc/DHE (mv) Va/DHE (mv) Vc-Va (mv) current density (ma/cm²) Figure 8 : Polarization curve with 25cm² single cell test, H 2 /O 2, Stoichiometric factors 1.2/1.5, 1%RH, 1.5bar absolute, 6 C 14

15 cell voltage (mv) Vc/DHE (mv) Va/DHE (mv) Vc-Va (mv) V (mv) current density (ma/cm²) Figure 9 : Polarization curve with 25cm² single cell test, H 2 /O 2, Stoichiometric factors 1.2/1.5 1%RH, 1.5bar absolute, 6 C cell voltage (mv) Vc/DHE (mv) Va/DHE (mv) Vc-Va (mv) V (mv) current density (ma/cm²) Figure 1 : Polarization curve with 25cm² single cell test, H 2 /air, Stoichiometric factors 1.2/2., 1%RH, 1.5bar absolute, 6 C 15

16 current density (ma/cm²) H under air H2O H2 under O2 H2O under air H2O O2 H2O O 2 under O Va,Vc (mv) / DHE Figure 11 : I = f(e) curves from previous polarization curves under oxygen and under air V / SHE (mv) V counter electrode (mv) /SHE V DHE (mv) /SHE I (µa) I (µa) time (h) Figure 12 : Electrolysis current vs. DHE potential 16

17 1 24 cell voltage, Vc, Va (mv) time (min) current density (ma/cm²) cell voltage (mv) Vc / DHE (mv) Va / DHE (mv) current density (ma/cm²) Figure 13 : Evolution of electrode potentials during current cycling under oxygen 1 24 cell voltage, Vc, Va (mv) time (min) current density (ma/cm²) cell voltage (mv) Vc / DHE (mv) Va / DHE (mv) current density (ma/cm²) Figure 14 : Evolution of electrode potentials during current cycling under air 17

18 1 1 9 cell voltage, Vc, Va (mv) I : =>2 A Vc : 952=>127=>69 mv Va : 41=>112==>4 mv Vcell : 91=>31=>569 mv current density (ma/cm²) time (min) cell voltage (mv) Vc / DHE (mv) Va / DHE (mv) current density (ma/cm²) Figure 15 : Transient behaviour of electrode potentials during current cycling i (µa/cm²) 2 1 passivity domain in fuel cell environment V / SHE (mv) Figure 16 : Polarisation curve of bright annealed AISI 316L SS in a typical PEMFC environment 18

19 cell voltage, Vc, Va (mv) / current density (ma/cm²) for Vcell # 386mV : Vc # 476mV/DHE Va # 91mV/DHE time (min) current density (ma/cm²) cell voltage (mv) Vc / DHE (mv) Va / DHE (mv) H2 stoichiometric factor 1,3 1,1,9,7,5,3,1 hydrogen stoichiometric factor Figure 17 : Effect of fuel starvation on electrode potentials V (mv) I (A) time (s) cell voltage Vc / DHE Va / DHE I Figure 18 : Effect of fuel and oxygen starvation on electrode potentials with fully humidified gases 19

20 V (mv) I (A) time (s) cell voltage Vc / DHE Va / DHE I Figure 19 : Effect of fuel and oxygen starvation on electrode potentials with 5% humidified gases 2