Sectional electrochemical impedance analysis of a high temperature polymer electrolyte membrane fuel cell with three types of flow-fields

Electrochimica Acta 112 (2013) 342–355

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Sectional electrochemical impedance analysis of a high temperature polymer electrolyte membrane fuel cell with three types of flow-fields C. Siegel a,∗ , I. Buder b,1 , A. Heinzel c a

Siegel Schleimer Ingénieurs-conseils s.à r.l., 150-159, rue Waassertrap, L-4408 Belvaux, Luxembourg Erneuerbare Energien und Elektro Mobility, Fakultät Kommunikation und Umwelt, Hochschule Rhein-Waal, Südstraße 8 D-47475, Kamp-Lintfort, Germany c Zentrum für BrennstoffzellenTechnik (ZBT) GmbH, Carl-Benz-Straße 201, D-47057 Duisburg, Germany b

a r t i c l e

i n f o

Article history: Received 8 March 2013 Received in revised form 22 August 2013 Accepted 23 August 2013 Available online 11 September 2013 Keywords: HTPEM fuel cell PBI/H3 PO4 Sectional measurements Current density distribution Segmented EIS

a b s t r a c t This work reports sectional electrochemical impedance (EIS) measurements of a high temperature polymer electrolyte membrane (HTPEM) fuel cell. These measurements were taken using a high temperature stable polybenzimidazole membrane electrode assembly (MEA) doped with phosphoric acid (PBI/H3 PO4 ). Three different types of flow-fields (namely, a six channel parallel serpentine flow-field, a parallel straight flow-field, and a mixed-type flow-field) were analyzed in gas counter-flow and co-flow configuration under selected operating conditions. The current density distribution was also measured, and the results were presented to highlight the dominant factors that lead to inhomogeneous distributions. The results showed that the distribution mainly depended on the availability of oxygen, overlapped by the fluid-flow distribution. The situation changed once a carbon monoxide (CO) enriched gas was used at the anode side. In this case, the current density distribution decreased close to the cathode inlet and increased near the anode inlet. At high CO concentrations, the highest current densities were found close to the anode inlet. Sectional EIS measurements supported these observed trends and confirmed that the segments close to the anode inlet had a lower charge transfer resistance with a higher charge transfer resistance at segments closer to the anode outlet. © 2013 Elsevier Ltd. All rights reserved.

1. Introduction Fuel cells are considered a promising technology for energy conversion systems. This technology offers the benefits of a high efficiency, significantly reduced emissions, and considerable power density, which makes it a candidate for applications such as (micro)-combined heat and power ((␮)-CHP) units, for powering cars, or for portable applications. As stated in [1], 2011 was by far the most successful year to date in the history of fuel cells with the annual megawatts shipped exceeding 100 MW for the first time as commercialization of the industry took hold. Of the various types of fuel cells, the low temperature PEM (LTPEM) fuel cell, which operates at approximately 70 ◦ C, and the high temperature PEM (HTPEM) fuel cell, which operates at approximately 160 ◦ C, can potentially be used for the aforementioned systems and applications. The higher operating temperature of HTPEM fuel cells helps overcome some major drawbacks. Both water flooding and water management become less critical, carbon monoxide (CO) poisoning of the platinum (Pt) catalyst becomes less prominent, and the

electrode kinetics are generally faster to name a few. Approximately 15 years ago, polybenzimidazole (PBI) doped with phosphoric acid (H3 PO4 ) was presented as a proton conducting membrane capable of operating at temperatures of approximately 160 ◦ C [2–4]. Since then, single cells, fuel cell stacks, and complete HTPEM fuel cell based systems have continuously improved over the years with the result that several products are commercially available today, see e.g. [5–10]. Surprisingly, little information is currently available on the internals of a working HTPEM fuel cell. In this work, the current density distributions of three types of flow-fields were measured along with sectional EIS measurements. The focus was on determining the dominant factors leading to inhomogeneous distributions in the HTPEM fuel cell. In fact, inhomogeneous current density distributions could create hot spots in the cell, which possibly causes local degradation. Therefore, the operating conditions and flow-field designs that lead to inhomogeneous distributions should be prevented. Sectional EIS measurements are one option for identifying not only inhomogeneous current densities but also their causes. 1.1. Literature survey

∗ Corresponding author. Tel.: +352 26378737; fax: +352 26378747. E-mail address: [email protected] (C. Siegel). 1 ISE Member. 0013-4686/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.electacta.2013.08.142

The PBI/H3 PO4 membrane was introduced as a proton conductor in fuel cells by Wainright et al. [2], Samms et al. [3],

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

Fig. 1. CAD-files of the different types of flow-field (overlapping anode and cathode side flow-field displayed) (a); segmented cathode side bipolar-plates used for all measurements (b); non-segmented anode side bipolar-plates used for all measurements (c).

and Wang et al. [4] among others. In [11–14], different membrane casting methods were investigated including the sol–gel process. Recently, Bandlamudi [15] systematically characterized several types of HTPEM fuel cell membranes. A good overview of the ongoing HTPEM fuel cell activities can be found in [11,13]. EIS is a well established technique for investigating LTPEM fuel cells [16,17]. Because HTPEM fuel cell technology is relatively new, little EIS data has been reported. Jalani et al. [18] used EIS measurements to obtain a detailed view of the PBI/H3 PO4 sol–gel membrane processes under different operating conditions including variable temperature, anode and cathode stoichiometry, and anode humidification. A high frequency intercept of 0.1  cm2 was reported. By varying the cathode stoichiometry, the observed change in performance and impedance behaviour indicated that a variation of oxygen partial pressure from inlet to outlet had a profound effect on the global responses in

343

Fig. 2. Modified cathode side aluminum endplate and PEEK insulation plate (connector plate (contact holder) for all gold-plated spring contact probes) (a); segmented cathode side bipolar-plate, anode side endplate, PEEK insulation plate, and the 50 cm2 PBI/H3 PO4 MEA with the sealing on top of the anode side bipolar-plate (b).

voltage and impedance. The impedance signatures developed during fuel starvation showed a 45◦ line with a similar signature to when the cell was electrically shorted. Zhang et al. [19] used EIS and cyclic voltammetry (CV) to obtain the exchange current densities and activation energies, respectively, for both half-cell reactions. Herein, the ohmic contribution was slightly below 0.1  cm2 when operating the cell at 160 ◦ C. The used equivalent circuit consisted of two resistor capacitor (constant phase element, CPE) pairs and one resistor to analyze the charge and gas transfer resistances. Additionally, different methods for improving the gas diffusion processes in the HTPEM fuel cells were highlighted. Andreasen et al. [20] took extensive EIS measurements to examine poisoning effects when running an HTPEM fuel cell using reformate gas. The impedance was evaluated at different temperatures, currents, and anode gas compositions using equivalent circuit modeling. The model used consisted of two resistor CPE pairs (exponent fixed to 0.85), one resistor capacitor pair, one resistor inductance pair and one resistor to differ between the high, intermediate, and

344

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

low frequency loop. The parameters of numerous operating conditions were discussed in great detail. A value of approximately 0.2  cm2 was reported for the high frequency resistance at 160 ◦ C. The same author also published a study on characterizing HTPEM fuel cell stacks using EIS [21]. Equivalent circuit modeling was performed, and the parameters and transfer functions were discussed. Mamlouk and Scott [22] characterized the effect that the electrode parameters such as Pt catalyst wt%, acid doping in both PBI and PTFE based electrodes and catalyst heat treatment had on the kinetic and mass transport characteristics. The influence of the load current, temperature and oxidant gas on the response was demonstrated and interpreted using different equivalent circuits. Bandlamudi [15] used EIS to analyze the impedance of HTPEM fuel cells under different compressions and discussed the Nyquist plots of highly doped and poorly doped membranes under different load currents. The evolution of various resistances was studied at different temperatures and load currents using a simplified Randles model. Hu et al. [23,24] performed EIS measurements over a 500 h continuous test and found that the main degradation was caused by platinum particle agglomeration. Schaltz et al. [25] and Jespersen et al. [26] reported the EIS measurements and performed equivalent circuit modeling using two resistor CPE pairs and one

resistor to distinguish between the charge and gas transfer resistance. A high frequency resistance of 0.2  cm2 was reported at 160 ◦ C. Kurz [27] performed EIS measurements on an HTPEM fuel cell stack under different operating conditions. Segmented EIS measurements of LTPEM fuel cells were published by Andreaus et al. [28] and Brett at al. [29] who presented a method for measuring the spectroscopy response of the localized electrochemical impedance over a wide frequency range using a Solartron 1260 hardware, a multiplexer, a rack of 10 loads, and individual current lines. Hakenjos and Hebling [30] presented segmented current density and EIS measurement results for an LTPEM fuel cell. These measurements were performed using a multichannel frequency response analyzer with a multichannel potentiostat. In [31], a very similar measurement setup was used (Solartron 1254 frequency response analyzer with two 1251 multichannel extensions) and allowed for the simultaneous measurement of the impedance spectra for single cells in a fuel cell stack. Schneider et al. [32–35] published many good and detailed works on segmented EIS measurements in LTPEM fuel cells, which included investigations on the water balance under different operating conditions, direct measurements of the local cell current density at submillimeter resolution, local cyclic voltammetry, and segmented EIS measurements in the

Fig. 3. Simplified diagram of the experimental setup for performing sectional EIS measurements.

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

channel and land areas. Hogarth et al. [36] used a 20 cm2 segmented fuel cell to investigate the performance of hydrogen-air using two different flow-fields. Each segment was connected to a separate current line and voltage sensor, and all measurements were taken after setting the local potential of each segment to 0.4 V to prevent any cross currents in the electrode. Gerteisen et al. [37] published a 50 channel characterization system for PEM fuel cells. This system is capable of both traditional electrochemical measurements and EIS measurements. The system relies on dedicated potentiostats to control the current and voltage with independent frequency response analyzers. In another work [38], the same author used a segmented cell consisting of 49 segments to measure the local current density and high frequency resistance. A parametric study was used to investigate the effects of the cell voltage, inlet relative humidity and flow rate and configuration using a three-channel serpentine flow-field. Based on the experimental results published by Schneider et al. [39,40], Kulikovsky [41] recently introduced a model for a segmented electrode for LTPEM fuel cells and discussed the high and low frequency arcs in the impedance spectra. It was shown that slight differences in the local impedance spectra could occur when exciting only one segment while all other segments were not excited. Reshetenko et al. [42–46] published very interesting study on LTPEM fuel cell discussing the effects of back pressure, gas stoichiometry, gas humidification, media material porosity, gas composition, and flow-field design on the current density distribution. They used different equivalent circuit models to discuss the parameters. Measuring the segmented current density distribution and EIS of HTPEM fuel cells is new and only few studies are currently available. Bergmann et al. [47] performed segmented current density and EIS measurements on an HTPEM fuel cell (25 cm2 MEA) using a 50 channel (synchronized) potentiostat-impedance spectrometer. Various operating conditions were analyzed. The results showed that the current density distribution mainly depended on the cathode stoichiometry. Additionally, the Nyquist plots contained more pronounced second arcs at cells towards the cathode outlet. The high frequency resistance did not vary with the cathode stoichiometry and did not vary between the investigated segments. Similar results were presented in [48] for an HTPEM fuel cell with straight flow-fields under various operating conditions. Several current density distributions were discussed and the EIS spectra fitted using equivalent circuits. Schaar [49] used a commercially available measurement system from the company S++ [50] to measure the current density distribution of HTPEM fuel cells over a large active area of 200 cm2 . In [51], a two channel parallel serpentine flow-field was divided into several segments. The current density was measured over an active area of 10 cm2 for various operating conditions. Recently, Lobato et al. [52,53] presented different studies comparing the current density distributions of different flow-fields to each other. Segmented current density measurements of HTPEM fuel cells using a 50 cm2 MEA were presented in [54]. In [55,56], the influence of the fluid-(gas)-phase on the solidphase temperature distribution was investigated using numerical simulations and segmented measurements.

2. Experimental 2.1. Developed segmented HTPEM fuel cell Fig. 1 shows the bipolar-plate of a segmented cathode with different types of flow-fields, namely a six channel parallel serpentine flow-field, a parallel straight flow-field, and a mixed-type flow-field, used in the test cell along with the respective nonsegmented anode bipolar-plate. The cathode of the HTPEM fuel cell had to be significantly modified for the segmented measurements. It consisted of 36 segments (4 × 9 segments of 1.34722 cm2

345

Fig. 4. Used equivalent circuit model of the HTPEM fuel cell (print screen from the SIM software).

each) embedded in a PEEK insulation matrix. The cathode aluminum endplate had 36 pinholes drilled through it to insert the 36 PT-100 resistive thermal devices (RTD) (electrically fully insulated) and gold-plated spring contact probes. A second temperature and chemically stable PEEK plate was used to fix the contact probes and 36 PT-100 devices to ensure good contact between the tips and bipolar-plate segments. Finally, the segmented cathode allowed the current density and temperature distributions to be measured over the entire MEA area of 50 cm2 . Sectional EIS was performed by contacting 36 segments from the backside using the above mentioned probes. The anode aluminum endplate was equipped with two 200 W heating elements to heat the cell to the desired solid-phase temperature. Other parts used on the anode side were the PEEK plate, the gold-plated copper current collector and the high temperature stable bipolar-plate. Fig. 2 shows the setup of the segmented HTPEM fuel cell with the sealings and the MEA. A commercially available PBI/H3 PO4 sol–gel MEA (BASF Celtec® -P 1000 series) [4] with an active area of approximately 50 cm2 was used for all measurements. As shown in Fig. 2, the MEA and both bipolar-plates were sandwiched between two PEEK plates to minimize the direct influence of the 2 heating elements on the local solid-phase temperature distribution. Before operating, the entire assembly was thermally insulated against the environment to minimize heat loss and ensure good qualitative measurements. 2.2. The simplified setup for sectional EIS measurements The segmented HTPEM fuel cell offered the possibility to perform sectional EIS measurements (sequentially scanned) to provide deeper insight into fuel cell operation. By combining such measurements along with the current density measurements it is possible to pinpoint the limiting factors under various operating conditions. Fig. 3 depicts a simplified diagram of the experimental setup for performing sectional EIS measurements. The hardware consisted of a high-precision electrochemical workstation, IM6eX (IM6 EPC-42), from Zahner Elektrik, Germany [57] and the respective electric load EL300. To choose the correct excitation signal, dummy measurements were performed for different signals (1, 5, 10 mV signals and 200, 500, and 1000 mA signals), and the spectra analyzed. No noticeable differences were observed, and all of the spectra were recorded with 200 and 500 mA. Depending on the operating conditions, the modulation frequency ranged from 20 mHz up to 10 kHz. Experimental artifacts were

346

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

Fig. 5. Measured current density distribution (a) and local spectra obtained from sectional EIS measurements (b) for type I flow-field (gas counter-flow) at stoichiometric flow rates, a set-temperature of 157 ◦ C, and 15 A load current.

Fig. 6. Fitted EIS model parameters for type I flow-field (counter-flow) at stoichiometric flow rates, a set-temperature of 157 ◦ C, and 15 A load current. High (a) and low frequency resistance (b).

minimized using twisted sense wires, short cable lengths, and best possible cable connectors.

non-segmented MEA, lateral currents flow between adjacent segments might disturb the current density measurement itself.

2.3. Note on the experimental setup

2.4. Equivalent circuit modeling

In the absence of hardware for the simultaneous (parallel) measurement of the local spectra, the sectional EIS measurements were performed sequentially rather than simultaneously. This major drawback increases the recording times, and the EIS measurements were only performed at selected segments as a direct consequence. Another drawback was that only one segment was excited at a time, which could lead to erroneous interpretations associated with drift or transients between measurement cycles [42]. Hakenjos and Hebling [30] and Gerteisen et al. [37] discussed the benefits of multichannel characterization systems for segmented measurements in great detail and included the potential controls for each segment to minimize the potential differences between adjacent segments, and to eliminate current leaks along the highly conductive gas diffusion layer (GDL). However, the drawbacks of such a complex setup were discussed by Reshetenko et al. [42]. A relatively simple segmented cell design minimizes the impact of the setup (e.g., removal of the uniform voltage control limitation). Additionally, the use of standard testing procedures and the flexibility associated with the use of a single load is sometimes beneficial. In this work, for the current density measurements, the hardware was expected to slightly influence the current density distribution. Due to the

According to [26], a schematic representation of the typical Nyquist plot is easy to understand but may not be accurate. The loop in the high frequency range and low frequency range cannot be separated in this way because events on the anode side may influence the cathodic loss region and vice versa. In this work, the equivalent circuit model shown in Fig. 4 was used. This model was taken from the SIM software (Zahner [57]) model library and modified accordingly [58]. This model is able to simulate the impedance spectra over a wide range of operating conditions. As stated in [26], losses in the various parts of the model should not be considered completely independent of each other but the general tendencies can be derived and considered valid. The total ohmic resistance of the cell was described using the resistive element R . The impedance is independent of the frequency, and the phase angle is 0◦ . It accounts for all of the ohmic losses within the HTPEM fuel cell, e.g., membrane resistance, ohmic resistance for the different cell components, and contact resistance. The inductive element had the same features as a coil, and its impedance increased with increasing frequency. The phase angle is 90◦ . This element accounts for the wiring and cables in the setup. The capacitive element has the same features as an electronic

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

347

Fig. 7. Measured current density distribution for type II flow-field (co-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, 20 A load current, a = 1.3, c = 4 (a); a = 1.3, c = 2.5 (b); a = 1.3, c = 2 (c); a = 1.1, c = 1.4 (d).

capacitor. Its impedance decreases with increasing frequency, and its phase angle is −90◦ . This element accounts for the double layer capacity within the MEA. The high frequency loop within the spectra was represented by the high frequency resistor capacitor pair shown in Fig. 4. The most important contribution generally comes from the sluggish oxygen reduction reaction with a minor contribution from the hydrogen oxidation reaction and capacitive charge storage in the porous media [26]. The low frequency loop was related to the gas transfer resistance, which accounts for the convective transport behaviour in the flow-field and the transfer of the gases through the porous media towards the reaction layer [26]. In addition to the resistor capacitor pair, the Nernst resistance accounted for the contribution of the diffusion process as part of the charge transfer reaction (equivalent to the general Warburg impedance) [57].

ZN =

W



j·ω

 · tanh

j·ω kN



2.5. Analysis of the EIS data The software package Thales 4.12 USB was used to evaluate the recorded data. The SIM software was used to fit the experimental data to an equivalent circuit. A feature of the used software was the stringent error treatment through all of the data sampling and processing steps. The measurement algorithm provides uncertainty estimates for each impedance sample [57]. For the fitting procedure itself, the recorded EIS data were first loaded into the software and then automatically smoothed. Next, the ZHIT transform recreated the impedance data from the phase and frequency data to judge the steady state and treat steady state violations in the experimental data using the following equation [57]:

  2 ln H(ω0 ) ≈ const. + ×

ω0 ϕ(ω) · d ln ω +  ·



(2)

ωs

(1)

The first input parameter, W, is the Warburg impedance, and the second parameter, kN , characterizes the relative reach of diffusion compared to a finite length. The diffusion length is finite because the concentration is assumed to be constant at a certain distance from the electrode. The Nyquist plot exhibits the same shape as the (special) Warburg impedance diagram in the high frequency part and is similar to a resistor capacitor pair in the low frequency part [57]. Due to possible interactions between the Nernst impedances, it was decided to use a capacitive element instead of a CPE within the equivalent circuit. The porous electrode described the simulated impedance of a system with homogenous pores. It was noted that the contribution to the impedance was minor; however, it was used within the equivalent circuit due to its slightly lower fitting error. It must be noted, that for the low frequency loop, Schneider et al. [39,40] performed segmented measurements using an LTPEM fuel cell and suggested that the origin of the low frequency loop arises from oscillations in the gas partial pressures throughout the flow-field. As stated in [20], at low frequencies the oscillations in the reactant concentrations due to changes in the drawn load current extend into the flow channel. The oscillations are then compounded throughout the flow-field, resulting in a low frequency arc.

dϕ(ω0 ) d ln ω

The fitting samples were then selected exclusively from the Z-HIT data, mostly in the frequency range from 20–50 mHz to 3–5 kHz. Before calling the fitter, the values of the elements in the equivalent circuit were tuned to provide best possible initial conditions. The SIM software used a complex nonlinear regression least square fitting algorithm. This fitting algorithm was used in a very accurate and stable way and was adapted to the individual parameter behaviour for the impedance elements to optimize the model parameters and minimize the deviation between the model transfer function and data set [57]. After returning a solution, all fitting errors (overall error, impedance error, and phase angle error) were checked and the parameters returned for a physical sound and visual examination. The reported uncertainty and significance of all the fitted EIS model parameters were examined. 3. Results and discussion 3.1. Testing protocol The segmented HTPEM fuel cell was operated in a dedicated teststand, equipped with all of the necessary hardware (e.g., differential pressure transducers, mass flow controllers, and valves) and software. For all of the measurements, the data were recorded

348

0.05

0.05

0.04

S17 20 A 1.3 / 2

S15 20 A 1.3 / 4

S15 20 A 1.3 / 2

S11 20 A 1.3 / 4

cm2

S10 20 A 1.3 / 4

0.04

Abs. IM Impedance /

Abs. IM Impedance /

S17 20 A 1.3 / 4

0.03

0.02

0

a)

0

0.01

0.02

0.03

Abs. RE Impedance /

0.04

0.03

0.02

0 0

0.05

0.01

0.02

cm2 0.05

0.05

Abs. IM Impedance /

S10 20 A 1.3 / 2.5 0.03

0.02

0.05

S15 20 A 1.1 / 1.4 0.04

S11 20 A 1.1 / 1.4

cm2

cm2

S11 20 A 1.3 / 2.5

0.04 cm2

S17 20 A 1.1 / 1.4

S17 20 A 1.3 / 2.5

0.04

0.03

Abs. RE Impedance /

S15 20 A 1.3 / 2.5

Abs. IM Impedance /

S10 20 A 1.3 / 2

0.01

0.01

b)

S11 20 A 1.3 / 2

cm2

a)

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

S10 20 A 1.1 / 1.4 0.03

0.02

0.01

0.01

0

0

0

0.01

0.02

0.03

Abs. RE Impedance /

0.04

0.05

cm2

Fig. 8. Local spectra obtained from sectional EIS measurements for type II flow-field (co-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, 20 A load current, a = 1.3, c = 4 (a); a = 1.3, c = 2.5 (b).

(field-points with a resolution of 16 bits), displayed using a graphical user interface (GUI), and evaluated using a spreadsheet application. The heat-up procedure began at 21 ◦ C, and the segmented HTPEM fuel cell was heated to 100–110 ◦ C before drawing power. Starting from 1 A, the load current was increased by 5 A per 30 min until a load current of 20 A was reached. The solid-phase temperature increased to 150–160 ◦ C due to the heat produced by the exothermal electrochemical reactions. Measurements were taken after a 30 min waiting period to ensure steady-state operating conditions. During all of the measurements, the necessary data were logged and continuously checked. 3.2. Type I flow-field The cell was operated in gas counter-flow configuration at 157 ◦ C while drawing 15 A load current. From Fig. 5, it can be observed that the oxygen availability mainly dictated the almost

b)

0

0.01

0.02

0.03

Abs. RE Impedance /

0.04

0.05

cm2

Fig. 9. Local spectra obtained from sectional EIS measurements for type II flow-field (co-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, 20 A load current, a = 1.3, c = 2 (a); a = 1.1, c = 1.4 (b).

linear gradient from the cathode inlet to outlet. In total, 10 selected segments were sequentially scanned, and the data recorded. The light grey spectra represent the segments close to the cathode inlet, the medium grey spectra the segments in the middle of the MEA, and the black spectra the segments towards the cathode outlet. The first intersection with the real axis was at 0.01  cm2 and was more or less constant for all spectra. The second intersection changed with the relative position of the segment over the MEA. Table 1 summarizes selected current density values and fitted EIS model parameters. The spectra showed a high frequency arch and a low frequency arch. As oxygen was consumed towards the cathode outlet, the charge transfer resistance increased, and both arches became more visible. Because the cathode stoichiometry was relatively low, the low frequency loop may arise from oscillations in the gas partial pressure as explained above. This effect may be more pronounced at segments that are located closer to the cathode outlet.

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

349

I = 20 A / Ts = 160°C / Tf = 21°C cm2

RHF / m

cm2

RLF / m

100

100

80

80

1.3 /

c

4

c

2.5

1.3 /

c

2

a 1.1 /

c

1.4

a

60

60

a 1.3 /

40

40

a

20

20 0

0 S17 S15

a)

S11

S17 S15

b)

S10

S11

S10

Fig. 10. Fitted EIS model parameters for type II flow-field (co-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, and 20 A load current. High (a) and low frequency resistance (b).

Fig. 6 shows the fitted EIS model parameters for the spectra shown in Fig. 5(b). The high frequency resistance slightly changed over the MEA area. The low frequency resistance increased continuously from the cathode inlet to outlet. These results were in agreement with the current density measurements shown in Fig. 5(a). The Nernst resistance increased almost linearly from the cathode inlet to outlet. The low frequency capacity varied between 1.84 and 2.59 F over the entire area of the MEA. The high frequency capacity varied between 5.11 and 7.18 F over the entire area of the MEA and appeared to slightly increase towards the anode inlet.

under the middlemost gas channels) and the oxygen availability. The lowest values appeared in the lower part over the MEA near the cathode outlet. All four gas stoichiometries had somewhat higher current density distributions on the left-hand side over the MEA than the right-hand side. A possible explanation was that oxygen was more available in this region due to the number of straight channels in direct contact with the cathode inlet. When reducing the gas stoichiometry, the current density distribution became more inhomogeneous; however, the shape remained the same. Table 2 summarizes the influence of anode and cathode stoichiometry variations on selected current density values. As can be seen from Fig. 8, the first intersection of the RE axis was at 0.01  cm2 and was comparable to the value for type I flowfield. At stoichiometric flow rates of 1.3 (anode) and 4.0 (cathode) (Fig. 8(a)), the local spectrum of segment S17 showed the smallest charge transfer resistance, followed by the local spectra of segments S15, S11, and S10, which were almost the same size and shape.

3.3. Type II flow-field The measured current density distributions for the 4 different stoichiometric flow rates and gas co-flow configurations while drawing 20 A load current at 160 ◦ C are depicted in Fig. 7. The highest current densities were observed close to the cathode and anode inlets. The shape of the current density distributions was defined by the bad fluid-flow distribution (especially cin

ain S9

S27

cin S27

9

S9

j / A m-2 6,050-6,600

S9

9

5,500-6,050

8

8

8

7

7

7

4,400-4,950

6

6

6

3,850-4,400

5

3,300-3,850

4

2,750-3,300

5 S13

4

S13

4 3

3 S20

ain

S27

9

5 S13

cin

ain

S20

2

3 S20

2

2

4,950-5,500

2,200-2,750 1,650-2,200 1,100-1,650

1

1

S4 aout a)

S3

S2

S1 cout

S4 aout b)

S3

S2

S1 cout

1

S4 aout

S3

S2

S1 cout

550-1,100 0-550

c)

Fig. 11. Measured current density distribution for type II flow-field (co-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, 15 A load current, 0% CO (a); 2% CO (b); 4% CO (c).

350

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

Table 1 Selected current density values and fitted EIS model parameters for type I flow-field (air/hydrogen operation).

0.1 S27 15 A 1.3 / 2.5 S20 15 A 1.3 / 2.5

Stoichiometry 1.3 (anode)/1.4 (cathode) Lowest value

3092 (100%) 6.1 (100%) 22.25 (100%)

4423 (142.96%) (S1) 7.74 (122.55%) (S22) 28.57 (128.44%) (S36)

1550 (50.09%) (S36) 4.8 (78.77%) (S9) 15.57 (69.99%) (S1)

Table 2 Selected current density values for type II flow-field (air/hydrogen operation). Mean value

Highest value

Lowest value

Stoichiometry 1.3 (anode)/4 (cathode) 4123 (100%) 5327 (129.26%) (S27) j (A m−2 )

2707 (65.68%) (S11)

Stoichiometry 1.3 (anode)/2.5 (cathode) 4123 (100%) 5476 (133.02%) (S35) j (A m−2 )

2577 (62.6%) (S11)

S13 15 A 1.3 / 2.5

0.08 cm2

Highest value

Abs. IM Impedance /

j (A m−2 ) RHF (m cm2 ) RLF (m cm2 )

Mean value

S9 15 A 1.3 / 2.5 H2 (0%CO) / Air

0.06

0.04

0.02

0

a)

0

Stoichiometry 1.3 (anode)/2 (cathode) 4123 (100%) 5936 (144.03%) (S27) j (A m−2 )

2227 (54.03%) (S20)

Stoichiometry 1.1 (anode)/1.4 (cathode) 6566 (159.08%) (S27) 4123 (100%) j (A m−2 )

1916 (46.42%) (S11)

0.02

0.04

0.06

Abs. RE Impedance /

0.08

0.1

cm2

0.1 S27 15 A 1.3 / 2.5 S20 15 A 1.3 / 2.5 S13 15 A 1.3 / 2.5

cm2 Abs. IM Impedance /

S9 15 A 1.3 / 2.5 H2 (2%CO) / Air

0.06

0.04

0.02

0

b)

0

0.02

0.04

0.06

Abs. RE Impedance / 0.1

0.08

0.1

cm2

S27 15 A 1.3 / 2.5 S20 15 A 1.3 / 2.5 S13 15 A 1.3 / 2.5

cm2

0.08

Abs. IM Impedance /

At stoichiometric flow rates of 1.3 (anode) and 2.5 (cathode), the local spectra for the segments differed; however, their shapes were similar (Fig. 8(b)). The local spectrum for segment S17 showed the smallest charge transfer resistance followed by segments S15, S11, and S10. It is remarkable that the local spectra for segments S17 and S15 looked similar in size and shape (as was the case for segments S11 and S10). The highest current density values were located near the cathode inlet, and the lowest was in the region under the middlemost gas channels close to the cathode outlet. Reducing the stoichiometric flow rates to 1.3 (anode) and 2.0 (cathode) increased the current density gradients. Again, the local spectra for segments S17 and S15 were similar (Fig. 9(a)). The local spectrum of segment S17 showed the smallest charge transfer resistance directly followed by segment S15. The local spectra of segments S11 and S10 also looked similar in size and shape but were significantly larger. Operating the cell at lower stoichiometric flow rates further increased the charge transfer resistance (Fig. 9(b)). The current density gradient increased. The highest value was found at the cathode inlet and the lowest value was found at the same position found for the previous measurements. The local spectrum of segment S17 showed the smallest charge transfer resistance followed by segment S15. The local spectra of segments S11 and S10 again looked similar in size and shape but were significantly larger. The fitted EIS model parameters for the high and low frequency resistances shown in Fig. 10 agreed with the current density measurements and local spectra data. The high and low frequency resistances increased for the segments located near both outlets. It was observed that reducing the cathode stoichiometry lead to an increase of the low frequency loop. The highest values were found for segments S10 and S11 at the low stoichiometric flow rates of 1.1 (anode) and 1.4 (cathode). For these measurements, the Nernst resistance also increased for segments located near both outlets at reduced stoichiometric flow rates. The high frequency capacity slightly increased under reduced stoichiometric flow rates for all 4 segments, whereas the low frequency capacity seemed to decrease slightly for all 4 segments. The fluid-flow distribution and oxygen availability had a major influence on the local parameters. Fig. 11 shows the results for the cell operating with stoichiometric flow rates of 1.3 (anode) and 2.5 (cathode) when drawing 15 A load current (gas co-flow). The operating temperature was 160 ◦ C.

0.08

S9 15 A 1.3 / 2.5 H2 (4%CO) / Air

0.06

0.04

0.02

0 c)

0

0.02

0.04

0.06

Abs. RE Impedance /

0.08

0.1

cm2

Fig. 12. Local spectra obtained from sectional EIS measurements for type II flowfield (co-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, 15 A load current, 0% CO (a); 2% CO (b); 4% CO (c).

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

351

I = 15 A / Ts = 160°C / Tf = 21°C RHF / m

cm2

cm2

RLF / m

100

100

80

80

60

60

40

40

20

20

0

H2 (0%CO) / Air H2 (2%CO) / Air H2 (4%CO) / Air

0 S27

S20

a)

S13

S27 b)

S9

S20

S13

S9

Fig. 13. Fitted EIS model parameters for type II flow-field (co-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, and 15 A load current. High (a) and low frequency resistance (b).

H2

a

1.3 / Air cout

ain

9 8

S14

S3

S3

S2

5,500-6,050

3.2-3.6

3,850-4,400

5

3,300-3,850

4

2,750-3,300

2.4-2.8

3.2

1.6

S1 cin

cin S1

a)

0-550

6

0.8 0.0 S4 S3 S2

1,100-1,650 550-1,100

ain

2.4

1,650-2,200

4

7

8

9

2.0-2.4 1.6-2.0 1.2-1.6 0.8-1.2

5

0.4-0.8

3

0.0-0.4

2 1

RLF / m

cm2

7.2-8.0 6.4-7.2 5.6-6.4 8.0

S1

0.015

ain

4.8

S9

3.2

S14

6

1.6 0.0 S4 S3 S2

S27

0.01

4.8-5.6

6.4

S3 cm2

2.8-3.2 4.0

2,200-2,750

1

cm2

3.6-4.0

0.02

Abs. IM Impedance /

2 / I = 20 A / Ts = 150°C / Tf = 21°C

6,050-6,600

4,400-4,950

2 S1

c

RHF / m

6

3

a)

1.3 / Air

4,950-5,500

7 S33

a

j / A m-2

S9

S27

S4 aout

H2

2 / I = 20 A / Ts = 150°C / Tf = 21°C

c

S33

b)

0.005

cin S1

4 3 2

5

7

8

9

4.0-4.8 3.2-4.0 2.4-3.2 1.6-2.4 0.8-1.6 0.0-0.8

1

Fig. 15. Fitted EIS model parameters for type III flow-field (counter-flow) at stoichiometric flow rates, a set-temperature of 150 ◦ C, and 20 A load current. High (a) and low frequency resistance (b).

0

b)

0

0.005

0.01

Abs. RE Impedance /

0.015

0.02

cm2

Fig. 14. Measured current density distribution (a) and local spectra obtained from sectional EIS measurements (b) for type III flow-field (counter-flow) at stoichiometric flow rates, a set-temperature of 150 ◦ C, and 20 A load current.

352

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355 Table 3 Selected current density values for type II flow-field (air/hydrogen (CO) operation).

The current density distribution was similar to the distribution discussed above for the same stoichiometric flow rates. Fig. 11(b) and (c) depicts the current density distribution for 15 A load current with a CO enriched anode gas. The shape of the current density imposed by the oxygen availability was overlapped by the hydrogen availability. Increasing the CO content decreased the current density near the anode outlet and increased it near the anode inlet. The distribution for 2% CO within the anode gas yielded higher values on the right side of the cell near the anode inlet. The lowest values were observed in the anode outlet region and near the middlemost gas channels. Fig. 11(c) shows the results for 4% CO within the anode gas. The oxygen and hydrogen availabilities now dictated the shape of the distribution. Under these operating conditions, the highest values were observed on the right hand of the cell and especially in the anode inlet region. In fact, the current density increased close to the anode inlet and decreased close to the anode outlet. Table 3 summarizes selected current density values. As can be seen from Fig. 12(a), when no CO was present within the anode gas, the local spectra were comparable to those shown in Fig. 8(a). Segment 27 was located very close to the cathode inlet, and the charge transfer resistance showed the lowest value. The local spectra of the segments located near the cathode outlet showed a slightly larger charge transfer resistance. Increasing the CO content within the anode gas to 2% using the same stoichiometric flow rates changed the size and shape of all 4 local spectra as shown in Fig. 12(b). The two arches can clearly be observed from the Nyquist plots. The local spectrum for segment S27 was similar to the local spectrum of segment S9. The largest charge transfer resistance was present at segment S20 because this segment was closest to the anode outlet. A higher CO content within the anode gas (4%) further increased the size of the local spectra and charge transfer resistance (Fig. 12(c)). Similar to what was observed before, the local spectrum for segment S27 was almost identical to the local spectrum of segment S9. The local spectra of segment S20 and segment S13 were much larger in shape and size, with the second arch being dominant. The fitted EIS model parameters values agreed with all of the measurements (Fig. 13). The high frequency resistance was nearly constant for these measurements. The low frequency resistance increased as soon as the CO enriched anode gas was used. The cout

cout

ain

S36

S9

9

S28

S1

Mean value

a)

S3

S2

3890 (125.44%) (S18)

1843 (59.43%) (S11)

CO content 2% j (A m−2 ) 3092 (100%)

4277 (138.44%) (S27)

1931 (62.51%) (S20)

CO content 4% 3092 (100%) j (A m−2 )

4479 (144.43%) (S18)

1645 (53.04%) (S28)

highest value was found for segment S20 followed by segment S13. The low frequency resistance had high values for the CO enriched anode gas, especially in segments S20 and S13. As stated in [20], a possible explanation was that the active sites within the reaction layer and especially in the channel direction were occupied by CO. Thus, hydrogen had to diffuse further before reaching a free site, which increased the local hydrogen concentration amplitudes and influenced the current amplitude. Similarly to the local oscillations in the oxygen partial pressure at the cathode side, it was possible that local oscillations in the hydrogen partial pressure at the anode side increased the low frequency resistance. This effect was more pronounced near the anode outlet because the hydrogen was continuously consumed. The model parameter fitting showed that the Nernst resistance increased in a similar way as the low frequency resistance did. For all 4 segments, the high frequency capacity increased with 2% CO within the anode gas and decreased with 4% CO. The low frequency capacity of the 4 segments continuously decreased with higher CO content within the anode gas. 3.4. Type III flow-field The cell with type III flow-field was operated using hydrogen and air at 150 ◦ C while drawing 20 A load current. The stoichiometric flow rates were 1.3 (anode) and 2 (cathode). The current density distributions showed the typical shape and characteristics dictated by the oxygen availability as observed previously (Fig. 14(a)). Table 4 summarizes selected current density values and fitted EIS model parameters for the measurements shown in Fig. 14.

S9

cout

9

ain

S36

S9

j / A m-2 6,050-6,600 9

5,500-6,050

8

8

8

4,950-5,500

7

7

7

4,400-4,950

6

6

6

3,850-4,400

5

5

5

3,300-3,850

4

4

4

2

3

S20 S28

2

S1

1 S4 aout

Lowest value

ain

S36

3

S20

Highest value

CO content 0% 3092 (100%) j (A m−2 )

3

S20 S28

S1

1

1

S1 cin

S4 aout b)

S3

S2

2

S4 aout

S1 cin

S3

S2

S1 cin

2,750-3,300 2,200-2,750 1,650-2,200 1,100-1,650 550-1,100 0-550

c)

Fig. 16. Measured current density distribution for type III flow-field (counter-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, 20 A load current, 0% CO (a); 2% CO (b); 4% CO (c).

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355 Table 4 Selected current density values and fitted EIS model parameters for type III flow-field (air/hydrogen operation).

353

0.04 S1 20 A 1.3 / 2.5 S9 20 A 1.3 / 2.5

Stoichiometry 1.3 (anode)/2 (cathode) Lowest value

4123 (100%) 3.32 (100%) 5.88 (100%)

5106 (123.21%) (S2) 3.45 (104.11%) (S16) 7.09 (120.49%) (S36)

3044 (73.45%) (S36) 3.13 (94.42%) (S4) 5.04 (85.74%) (S3)

Table 5 Selected current density values for type III flow-field (air/hydrogen (CO) operation). Mean value

Highest value

Lowest value

CO content 0% j (A m−2 )

4123 (100%)

4906 (118.99%) (S2)

3374 (81.83%) (S36)

CO content 2% j (A m−2 )

4123 (100%)

4731 (114.74%) (S4)

3681 (89.27%) (S30)

CO content 4% j (A m−2 )

4123 (100%)

4634 (112.39%) (S7)

3571 (86.61%) (S28)

cm2

Highest value

Abs. IM Impedance /

j (A m−2 ) RHF (m cm2 ) RLF (m cm2 )

S20 20 A 1.3 / 2.5

Mean value

0.03

S28 20 A 1.3 / 2.5 S36 20 A 1.3 / 2.5 H2 (0%CO) / Air

0.02

0.01

0

0

a)

0.01

0.02

0.03

Abs. RE Impedance /

0.04

cm2

0.04 S1 20 A 1.3 / 2.5 S9 20 A 1.3 / 2.5

Abs. IM Impedance /

cm2

S20 20 A 1.3 / 2.5 0.03

S28 20 A 1.3 / 2.5 S36 20 A 1.3 / 2.5 H2 (2%CO) / Air

0.02

0.01

0

0

b)

0.01

0.02

0.03

Abs. RE Impedance /

0.04

cm2

0.04 S1 20 A 1.3 / 2.5 S9 20 A 1.3 / 2.5 cm2

S20 20 A 1.3 / 2.5

Abs. IM Impedance /

Fig. 14(b) depicts the local spectra of 6 selected segments. The first intersection with the real axis was somewhat above 0.01  cm2 . It was nearly constant and comparable to previous measurements. The second intersection changed with the position of the segment. Segment S1 was located directly at the cathode inlet and possessed a low charge transfer resistance. As oxygen was consumed towards the cathode outlet, the charge transfer resistance increased, and the two arches became slightly more visible. Consequently, the local spectrum of segment S36 showed the largest low frequency arch. Fig. 15 shows the fitted EIS model parameters for these spectra. The high frequency resistance was more or less constant over the MEA area. The low frequency resistance increased continuously from the cathode inlet to outlet. These results were in agreement with the current density measurements shown in Fig. 14(a). For the other parameters, the Nernst resistance followed the low frequency resistance, and the lowest values were obtained close to the cathode inlet with the highest values near the cathode outlet. The high frequency capacity was almost constant over the area of the MEA. The low frequency capacity decreased towards the cathode outlet. The highest value occurred near the cathode inlet with the lowest value at the cathode outlet. Fig. 16 depicts the situation for type III flow-field when drawing 20 A load current from the cell using stoichiometric flow rates of 1.3 (anode) and 2.5 (cathode). The operating temperature was 160 ◦ C. The current density distribution was very similar to the one observed in Fig. 14(a). Table 5 summarizes selected current density values for different CO contents within the anode gas. As for the sectional EIS measurements, the different local spectra were all similar in size and shape when no CO was present in the anode gas. A closer look indicates the local spectra of the segments closer to the cathode inlet were somewhat smaller in size and shape (Fig. 17(a)). Increasing the CO content of the anode gas to 2% changed the situation, with a second arch becoming visible (Fig. 17(b)). The charge transfer resistance increased especially for segment S28. Indeed, this segment was located at the anode outlet. The charge transfer resistance slightly increased in all local spectra. A similar current density shift from the anode outlet to the anode inlet was observed as previously discussed for the other types of flow-fields. The local spectra for 4% CO in the anode gas are shown in Fig. 17(c). The charge transfer resistances increased for all segments, and the second arch was clearly visible. Segment S9 was located directly at the anode inlet, and its local spectrum showed the smallest charge in the transfer resistance. Similar to the values

0.03

S28 20 A 1.3 / 2.5 S36 20 A 1.3 / 2.5 H2 (4%CO) / Air

0.02

0.01

0 c)

0

0.01

0.02

Abs. RE Impedance /

0.03

0.04

cm2

Fig. 17. Local spectra obtained from sectional EIS measurements for type III flowfield (counter-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, 20 A load current, 0% CO (a); 2% CO (b); 4% CO (c).

354

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355

I = 20 A / Ts = 160°C / Tf = 21°C RHF / m

cm2

RLF / m

50

50

40

40

30

30

20

20

10

10

H2 (0%CO) / Air H2 (2%CO) / Air H2 (4%CO) / Air

0

0 a)

cm2

S1 S9 S20 S28 S36

b)

S1 S9 S20 S28 S36

Fig. 18. Fitted EIS model parameters for type III flow-field (counter-flow) at stoichiometric flow rates, a set-temperature of 160 ◦ C, and 20 A load current. High (a) and low frequency resistance (b).

reported for 2% CO in the anode gas, the current density decreased near the anode outlet and increased near the anode inlet. Fig. 18 shows the fitted EIS model parameters for the spectra shown in Fig. 17. The high frequency resistance was more or less constant over the MEA area but seemed to decrease slightly with increasing CO content in the anode gas. The low frequency resistance increased with increasing CO content. When no CO was present within the anode gas, the charge transfer resistance was more or less constant for all segments. As soon as a CO enriched anode gas was used, the low frequency charge transfer resistance increased for all segments especially for those near the anode outlet. These results were in agreement with the current density measurements shown in Fig. 16. It is worth noting that local oscillations in the hydrogen partial pressure on the anode side could increase the low frequency resistance, which was more pronounced for segments near the anode outlet.

4. Conclusions The presented prototype setup allowed for segmented solidphase temperature measurements, segmented current density measurements, and sectional EIS measurements. For all measurements, type I flow-field returned the best I–V-performance, directly followed by type III flow-field. Type II flow-field returned a lower performance. Under most operating conditions, the current densities of the three types of flow-fields showed the highest values in the regions close to the cathode inlet with the lowest values occurring near the cathode outlet. For type II flow-fields, the influence of the poor fluid-flow distribution significantly overlapped the distribution in the middlemost gas channel region especially near the outlet. In summary, the oxygen availability defined the shape of the current density distribution for both gas counter-flow and coflow configurations. These findings changed once a CO enriched gas was used at the anode side. In this case, the highest current density shifted towards the anode inlet and significantly overlapped the shape provided by the oxygen availability. Therefore, using a CO enriched gas with a counter-flow configuration yielded a flatter current density distribution than for gas co-flow configuration. These measurements were performed for CO contents of up to several percent, and the effects were less pronounced at lower CO percentages. For a few selected operating conditions and segments, sectional EIS measurements supported the observed trends. The local spectra were evaluated using equivalent circuit modeling to discuss the parameters. For all of the measurements, the high frequency resistance remained almost constant for the local spectra and did not change much with the segment position. Operating

the cell with pure hydrogen as the anode gas increased the low frequency resistance for those segments near the cathode outlet especially for type I and type III flow-fields. For type II flow-field, the charge transfer resistance further increased in regions where the fluid-flow distribution was bad as was the case for the middlemost gas channels. For the three flow-field types, it was observed that reducing the cathode stoichiometric flow rate increased the charge transfer resistance. Higher cathode stoichiometric flow rates decreased the charge transfer resistance until a minimum was reached. Beyond this value, further increasing the stoichiometric flow rates may increase the charge transfer resistance. For all of the flow-field types, a low cathode stoichiometric flow rate produced a second arch in the spectra, which may arise from oscillations in the oxygen partial pressure on the cathode side. In accordance with the current density distributions, the variations in the anode stoichiometric flow rate did not measurably change the local spectra. Once a CO enriched anode gas was used, this situation changed. The local spectra significantly increased in size and shape, and a second arch appeared. This second arch may come from local oscillations in the hydrogen partial pressure on the anode side. Locally, the largest charge transfer resistance was present near the anode outlet and the lowest charge transfer resistance near the anode inlet. The presented results should be confirmed by additional measurements using a more sophisticated setup. The effect of the gas channel structure, the possible oscillations at low stoichiometric flow rates, and the influence of porous media on the local spectra in an HTPEM fuel cell are especially interesting topics for future investigations. This work will help to design flow-fields for HTPEM fuel cells and layout the optimal positioning of the inlet and outlet to achieve more homogeneous quantities distributions.

Acknowledgement This work was supported by ‘LE GOUVERNEMENT DU GRANDDUCHÉ DE LUXEMBOURG, MCESR Recherche et Innovation’, Grant No.: AFR07/007.

References [1] The fuel cell today industry review 2012—http://www.fuelcelltoday.com (last accessed February 2013). [2] J.S. Wainright, J.T. Wang, D. Weng, R.F. Savinell, M. Litt, Acid-doped polybenzimidazoles: a new polymer electrolyte, J. Electrochem. Soc. 142 (1995) L121.

C. Siegel et al. / Electrochimica Acta 112 (2013) 342–355 [3] S.R. Samms, S. Wasmus, R.F. Savinell, Thermal stability of proton conducting acid doped polybenzimidazole in simulated fuel cell environments, J. Electrochem. Soc. 143 (1996) 1225. [4] J.T. Wang, R.F. Savinell, J.S. Wainright, M. Litt, H. Yu, A H2 /O2 fuel cell using acid doped polybenzimidazole as a polymer electrolyte, Electrochim. Acta 41 (1996) 193. [5] http://www.fuel-cell.basf.com/ca/internet/Fuel Cell/ (last accessed February 2013). [6] http://www.adventech.gr/ (last accessed February 2013). [7] http://www.elcore.com/home.html (last accessed February 2013). [8] http://www.clearedgepower.com (last accessed February 2013). [9] http://www.fumatech.com/Startseite/ (last accessed February 2013). [10] http://www.serenergy.com/ (last accessed February 2013). [11] C. Wannerk, High-temperature PEM fuel cells: electrolytes, cells, and stacks, in: D. Stolten (Ed.), Hydrogen and Fuel Cells—Fundamentals, Technologies and Applications, Wiley-VCH, Weinheim, 2010, p. 17, Chapter 2. [12] Q. Li, R. He, J.O. Jensen, N.J. Bjerrum, PBI-based polymer membranes for high temperature fuel cells—preparation, characterization and fuel cell operation, Fuel Cells 4 (2004) 147. [13] Q. Li, J.O. Jensen, R.F. Savinell, N.J. Bjerrum, High temperature proton exchange membranes based on polybenzimidazoles for fuel cells, Prog. Polym. Sci. 34 (2009) 449. [14] L. Xiao, H. Zhang, E. Scanlon, S. Ramanathan, E.-W. Choe, D. Rogers, T. Apple, B.C. Benicewicz, High-temperature polybenzimidazoles for fuel cell membranes via a sol–gel process, Chem. Mater. 17 (2005) 5328. [15] G. Bandlamudi, Systematic Characterization of HT PEMFCs Containing PBI/H3 PO4 Systems, Logos Verlag, Berlin, Germany, 2011. [16] X. Yuan, H. Wang, J.C. Sun, J. Zhang, AC impedance technique in PEM fuel cell diagnosis—a review, Int. J. Hydrogen Energy 32 (2007) 4365. [17] J. Wu, X.Z. Yuan, H. Wang, M. Blanco, J.J. Martin, J. Zhang, Diagnostic tools in PEM fuel cell research: part I Electrochemical techniques, Int. J. Hydrogen Energy 33 (2008) 1735. [18] N.H. Jalani, M. Ramani, K. Ohlsson, S. Buelte, G. Pacifico, R. Pollard, R. Staudt, R. Datta, Performance analysis and impedance spectral signatures of high temperature PBI—phosphoric acid gel membrane fuel cells, J. Power Sources 160 (2006) 1096. [19] J. Zhang, Y. Tang, C. Song, J. Zhang, Polybenzimidazole-membrane-based PEM fuel cell in the temperature range of 120–200 ◦ C, J. Power Sources 172 (2007) 163. [20] S.J. Andreasen, J.R. Vang, S.K. Kær, High temperature PEM fuel cell performance characterisation with CO and CO2 using electrochemical impedance spectroscopy, Int. J. Hydrogen Energy 36 (2012) 9815. [21] S.J. Andreasen, J.L. Jespersen, E. Schaltz, S.K. Kær, Characterisation and modelling of a high temperature PEM fuel cell stack using electrochemical impedance spectroscopy, Fuel Cells 9 (2009) 463. [22] M. Mamlouk, K. Scott, Analysis of high temperature polymer electrolyte membrane fuel cell electrodes using electrochemical impedance spectroscopy, Electrochim. Acta 56 (2011) 5493. [23] J. Hu, H. Zhang, Y. Zhai, G. Liu, J. Hu, B. Yi, Performance degradation studies on H3 PO4 high temperature PEMFC and one-dimensional numerical analysis, Electrochim. Acta 52 (2006) 394. [24] J. Hu, H. Zhang, L. Gang, Diffusion-convection/electrochemical model studies on polybenzimidazole (PBI) fuel cell based on AC impedance technique, Energy Convers. Manage. 49 (2008) 1019. [25] E. Schaltz, J.L. Jespersen, P.O. Rasmussen, Development of a 400 W high temperature PEM fuel cell power pack: equivalent circuit modelling, in: Fuel Cell Seminar 2006. Proceedings of the fuel cell seminar 2006 conference, Honolulu, U.S.A., 13–17 November 2006, 2006. [26] J.L. Jespersen, E. Schaltz, S.K. Kær, Electrochemical characterization of a PBIbased HT-PEM unit cell, J. Power Sources 191 (2009) 289. [27] T. Kurz, Entwicklung und Charakterizierung eines portable HochtemperaturPEM-Brennstoffzellensystems, Fraunhofer Verlag, Stuttgart, Germany, 2011. [28] B. Andreaus, A.J. McEvoy, G.G. Scherer, Analysis of performance losses in polymer electrolyte fuel cells at high current densities by impedance spectroscopy, Electrochim. Acta 47 (2002) 2223. [29] D.J.L. Brett, S. Atkins, N.P. Brandon, V. Vesovic, N. Vasileiadis, A.R. Kucernak, Localized impedance mesurements along a single channel of a solid polymer fuel cell, Electrochem. Solid-State Lett. 6 (2003) A63. [30] A. Hakenjos, C. Hebling, Spatially resolved measurement of PEM fuel cells, J. Power Sources 145 (2005) 307. [31] A. Hakenjos, M. Zobel, J. Clausnitzer, C. Hebling, Simultaneous electrochemical impedance spectroscopy of single cells in a PEM fuel cell stack, J. Power Sources 154 (2006) 360. [32] I.A. Schneider, H. Kuhn, A. Wokaun, G.G. Scherer, Study of water balance in a polymer electrolyte fuel cell by locally resolved impedance spectroscopy, J. Electrochem. Soc. 152 (2005) A2383. [33] I.A. Schneider, H. Kuhn, A. Wokaun, G.G. Scherer, Fast locally resolved electrochemical impedance spectroscopy in polymer electrolyte fuel cells, J. Electrochem. Soc. 152 (2005) A2092.

355

[34] I.A. Schneider, S. von Dahlen, A. Wokaun, G.G. Scherer, A segmented microstructured flow field approach for submillimeter resolved local current measurement in channel and land areas of a PEFC, J. Electrochem. Soc. 157 (2010) B338. [35] I.A. Schneider, M.H. Bayer, S. von Dahlen, Locally resolved electrochemical impedance spectroscopy in channel and land areas of a differential polymer electrolyte fuel cell, J. Electrochem. Soc. 158 (2011) B343. [36] W.H.J. Hogarth, J. Steiner, J.B. Benziger, A. Hakenjos, Spatially-resolved current and impedance analysis of a stirred tank reactor and serpentine fuel cell flow-field at low relative humidity, J. Power Sources 164 (2007) 464. [37] D. Gerteisen, W. Mérida, T. Kurz, P. Lupotto, M. Schwager, C. Hebling, Spatially resolved voltage, current and electrochemical impedance spectroscopy measurements, Fuel Cells 11 (2011) 339. [38] D. Gerteisen, N. Zamel, C. Sadeler, F. Geiger, V. Ludwig, C. Hebling, Effect of operating conditions on current density distribution and high frequency resistance in a segmented PEM fuel cell, Int. J. Hydrogen Energy 37 (2012) 7736. [39] I.A. Schneider, S.A. Freunberger, D. Kramer, A. Wokaun, G.G. Scherer, Oscillations in the gas channels: part I. The forgotten player in impedance spectroscopy in polymer electrolyte fuel cells, J. Electrochem. Soc. 154 (2007) B383–B388. [40] I.A. Schneider, D. Kramer, A. Wokaun, G.G. Scherer, Oscillations in gas channels II. Unraveling the characteristics of the low frequency loop in air-fed PEFC impedance spectra, J. Electrochem. Soc. 154 (2007) B770–B782. [41] A.A. Kulikovsky, A model for local impedance of the cathode side of PEM fuel cell with segmented electrodes, J. Electrochem. Soc. 159 (2012) F294–F300. [42] T.V. Reshetenko, G. Bender, K. Bethune, R. Rocheleau, Systematic study of back pressure and anode stoichiometry effects on spatial PEMFC performance distribution, Electrochim. Acta 56 (2011) 8700. [43] T.V. Reshetenko, G. Bender, K. Bethune, R. Rocheleau, Systematic studies of the gas humidification effects on spatial PEMFC performance distributions, Electrochim. Acta 69 (2012) 220. [44] T.V. Reshetenko, G. Bender, K. Bethune, R. Rocheleau, Effects of local variations of the gas diffusion layer properties on PEMFC performance using a segmented cell system, Electrochim. Acta 80 (2012) 368. [45] T.V. Reshetenko, K. Bethune, R. Rocheleau, Spatial proton exchange membrane fuel cell performance under carbon monoxide poisoning at a low concentration using a segmented cell system, J. Power Sources 218 (2012) 412. [46] T.V. Reshetenko, G. Bender, K. Bethune, R. Rocheleau, A segmented cell approach for studying the effects of serpentine flow field parameters on PEMFC current distribution, Electrochim. Acta 88 (2013) 571. [47] A. Bergmann, T. Kurz, D. Gerteisen, C. Hebling, Spatially resolved impedance spectroscopy in PEM fuel cells up to 200 ◦ C, in: 18th World hydrogen energy conference 2010 (WHEC). Parallel Sessions Book 1: Fuel Cell Basics/Fuel Infrastructures, Essen, German, 2010, 16–21 May 2010, Jülich, Forschungszentrum Jülich GmbH, Zentralbibliothek, 2010. [48] C. Siegel, I. Buder, G. Bandlamudi, A. Heinzel, Electrochemical measurements using a segmented HTPEM fuel cell, in: Electrochemistry 2010 (GDCh)—from microscopic understanding to global impact, Bochum, Germany, 13–15 September 2010, 2010, Book of Abstracts. [49] B. Schaar, Simulation einer Hochtemperatur-PEM-Brennstoffzelle, AutoUni–Schriftreihe, Logos Verlag, Berlin, Germany, 2008. [50] http://www.splusplus.de (last accessed February 2013). [51] H. Moser, E. Wallnöfer, V. Hacker, High temperature proton exchange membrane fuel cells—the impact of fuel contaminants and temperature on fuel cell performance, in: A3PS Conference—Alternative propulsion systems and energy carriers, Vienna, Austria, 16, October 2009, 2009, Oral presentation. [52] J. Lobato, P. Cãnizares, M.A. Rodrigo, F.J. Pinar, D. Úbeda, Study of flow channel geometry using current distribution measurement in a high temperature PEM fuel cell, J. Power Sources 196 (2011) 4209. [53] J. Lobato, P. Cãnizares, M.A. Rodrigo, J.P. Pinar, E. Mena, D. Úbeda, Threedimensional model of a 50 cm2 high temperature PEM fuel cell, Study of the flow channel geometry influence, Int. J. Hydrogen Energy 35 (2010) 5510. [54] J. Lebæk, S.T. Ali, P. Møller, C. Mathiasen, L.P. Nielsen, S.K. Kær, Quantification of in situ temperature measurements on a PBI-based high temperature PEMFC unit cell, Int. J. Hydrogen Energy 35 (2010) 9943. [55] C. Siegel, G. Bandlamudi, P. Beckhaus, J. Burfeind, F. Filusch, A. Heinzel, Segmented current and temperature measurement in a HTPEM fuel cell, in: 6th symposium on fuel cell modelling and experimental validation, Bad Herrenalb/Karlsruhe, Germany, 25–26 March 2009, 2009, Book of Abstracts. [56] C. Siegel, G. Bandlamudi, A. Heinzel, Locally resolved measurements in a segmented HTPEM fuel cell with straight flow-fields, Fuel Cells 11 (2011) 489. [57] http://www.zahner.de (last accessed February 2013). [58] W. Strunz, Personal communication, Zahner, Germany, 2012.