- Email: [email protected]
Challenge for lowering concentration polarization in solid oxide fuel cells
Journal of Power Sources 302 (2016) 53e60
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Challenge for lowering concentration polarization in solid oxide fuel cells Hiroyuki Shimada*, Toshio Suzuki, Toshiaki Yamaguchi, Hirofumi Sumi, Koichi Hamamoto, Yoshinobu Fujishiro Inorganic Functional Materials Research Institute, Department of Materials and Chemistry, National Institute of Advanced Industrial Science and Technology (AIST), 2266-98 Anagahora, Shimo-shidami, Moriyama-ku, Nagoya 463-8560, Japan
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Porous microstructure control technique using extrusion process was developed. A Ni-YSZ anode shows 53.6 vol.% porosity with a median pore diameter of 0.911 mm. Lowing concentration polarization is essential to achieve high power density. Reduced concentration polarization in SOFC resulted in 3.09 W cm2 at 800 C. Estimated concentration overpotential was less than 0.1 V at 10 A cm2.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 17 July 2015 Received in revised form 7 September 2015 Accepted 8 October 2015 Available online xxx
In the scope of electrochemical phenomena, concentration polarization at electrodes is theoretically inevitable, and lowering the concentration overpotential to improve the performance of electrochemical cells has been a continuing challenge. Electrodes with highly controlled microstructure, i.e., high porosity and uniform large pores are therefore essential to achieve high performance electrochemical cells. In this study, state-of-the-art technology for controlling the microstructure of electrodes has been developed for realizing high performance support electrodes of solid oxide fuel cells (SOFCs). The key is controlling the porosity and pore size distribution to improve gas diffusion, while maintaining the integrity of the electrolyte and the structural strength of actual sized electrode supports needed for the target application. Planar anode-supported SOFCs developed in this study realize 5 mm thick dense electrolyte (yttria-stabilized zirconia: YSZ) and the anode substrate (Ni-YSZ) of 53.6 vol.% porosity with a large median pore diameter of 0.911 mm. Electrochemical measurements reveal that the performance of the anode-supported SOFCs improves with increasing anode porosity. This Ni-YSZ anode minimizes the concentration polarization, resulting in a maximum power density of 3.09 W cm2 at 800 C using humidified hydrogen fuel without any electrode functional layers. © 2015 Elsevier B.V. All rights reserved.
Keywords: Solid oxide fuel cell (SOFC) Concentration polarization Planar anode-supported cell Anode porosity Extrusion process
1. Introduction * Corresponding author. E-mail address: [email protected] (H. Shimada). http://dx.doi.org/10.1016/j.jpowsour.2015.10.024 0378-7753/© 2015 Elsevier B.V. All rights reserved.
Electrochemical ceramic cells such as solid oxide fuel cells (SOFCs) are expected to apply to next generation power devices due
54
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
to their high energy conversion efficiency and fuel flexibility. For system minimization and cost reduction, high power density operation of SOFCs is needed, and thus improvement in single cell performance is essential. Planar anode-supported SOFCs are the most promising candidate type to achieve high power density because their ohmic resistance is very low due to thin electrolyte and short current pass in the electrodes [1e5]. Although the performance of planar anode-supported SOFCs is dominated by electrode polarization resistance with decreasing ohmic resistance, electrode performance is improved by using materials having high intrinsic catalytic activity [6e9] and also by using nano-size materials [10e15] to extend the electrochemically active area such as the triple-phase boundary (TPB), where electronic, ionic, and gas phases are in contact. Electrode polarization resistance is mainly attributed to two processes, namely, interfacial reaction and gas diffusion. The resistances due to these processes are respectively called activation polarization resistance and the concentration polarization resistance. Although the dominant resistance in electrodes is generally due to activation polarization, concentration polarization often notably affects electrode performance when the electrode is relatively thick, such as in anode-supported SOFCs [2,16e20]. In particular, the concentration polarization resistance is crucial to maximize cell performance at high temperature operation (700 C), in which the ohmic resistance of the electrolyte and the activation polarization resistance of the electrodes are significantly lowered although the concentration polarization resistance remains relatively constant or is slightly increased [2,16e23]. Concentration polarization, defined as the electrode polarization due to the difference in partial pressure of oxygen, pO2 , between the anode surface and the anode/electrolyte interface, of porous electrodes is a theoretically inevitable phenomenon. Therefore, achieving high power density in planar anode-supported SOFCs requires developing an anode substrate that has good gas diffusion properties such as high porosity and large pore size and can enhance catalytic activity. Although high porosity leads to lower concentration polarization, fabricating highly porous structured anodes with sufficient mechanical strength for use as actual sized substrate in SOFCs is difficult. Furthermore, to obtain the dense electrolyte and porous anode structure, the anode substrates must have the same sintering and shrinkage behaviors as the electrolyte because they undergo cosintering with the electrolyte. Thus, the development of an electrode substrate fabrication process that can finely control porous microstructure is required. Extrusion is a conventionally used process to fabricate ceramic substrates for tubular [5,9,24], flat-tube [25], and honeycomb SOFCs [26]. Porous ceramic substrates that possess high porosity and mechanical strength can be obtained by the addition of pore former and the optimization of the extrusion process. Also, this process is attractive for industrial manufacturing because it can yield ceramic substrates even without organic solvent and with only a relatively small amount of binder compared with the other wet ceramic processes such as tape casting. In the present study, we focus on the effect of the gas diffusion properties at nickel (Ni) and yttria-stabilized zirconia (YSZ) composite porous anodes on electrochemical performance in SOFCs, and report challenges to realize highly microstructure-controlled Ni-YSZ anodes with high porosity and uniform large pores. To achieve high porosity in the anodes while maintaining the mechanical strength that allows their use for actual size planar anodesupported SOFCs, we developed a microstructure control technique using the extrusion process. The porosity and pore size of extruded nickel oxide (NiO)-YSZ anode substrates were controlled by adding pore former and by adjusting the sintering temperature. The electrochemical performance of fabricated anode-supported SOFCs was
evaluated in the temperature range of 650e800 C, and then the impact of the concentration polarization on cell performance in high power density operation was discussed. 2. Experimental 2.1. Fabrication of planar anode substrates by extrusion process NiO powder (0.7 mm in median diameter, Sumitomo Metal Mining Co.) and YSZ powder (8 mol% Y2O3eZrO2, TZ-8YS, Tosoh) at a weight ratio of 60:40 were mixed with pore former and binder. The pore former was a mixture of graphite carbon (UF-G10, Showa Denko) and cellulose (TG-101, Asahi Kasei Chemicals) at a weight ratio of 2:1, and was added to the NiO-YSZ at three different composition ratios, namely, 11.3, 22.6, and 33.8 wt.% of the NiO and YSZ powders. These materials were stirred for 30 min in a vacuum chamber, and then after the addition of a proper amount of distilled water, the mixture was stirred again for 1 h. The mixture was then aged in ambient atmosphere for 15 h, yielding clay for extrusion. To produce NiO-YSZ green sheets (Fig. 1a), the clay was then extruded through a metal mold (0.7 mm thick, 65 mm wide) using a screw type extruder (Miyazaki Iron Works Co.). NiO-YSZ anode substrates were finally fabricated by cutting the green sheets into a desired size and then sintering them at 1250 C for 1 h in air. 2.2. Fabrication of anode-supported SOFCs A YSZ paste was prepared by mixing YSZ powder (TZ-8Y, Tosoh) in a-terpineol solvent (Kanto Chemical Co.) with ethyl cellulose 45 cp (Kishida Chemical Co.), a dispersant, and a plasticizer. The YSZ paste was screen-printed onto the prefabricated NiO-YSZ anode substrates. Then the anode substrates with YSZ were co-sintered at either 1350 or 1400 C for 3 h in air. The Ce0.9Gd0.1O1.95 (GDC) interlayer was prepared by screen-printing a GDC paste, made from GDC powder (40 m2 g1 in specific surface area, Anan Kasei Co.) and the same admixtures as in the YSZ paste, onto the YSZ electrolyte and sintering at 1300 C for 1 h in air. Note that 0.3 mol% Mn was added to the GDC paste to improve the sinterability of GDC. The detailed fabrication procedure has been presented elsewhere [27]. Also, an La0.6Sr0.4Co0.2Fe0.8O3d (LSCF)-GDC paste for the cathode was prepared using LSCF powder (0.7 mm in median diameter, AGC Seimi Chemical Co.) and the same GDC powder as the interlayer material at a weight ratio of 70:30 with the same admixtures as in the YSZ paste, and then screen-printed on the GDC interlayer and sintered at 950 C for 1 h in air. In electrochemical measurements, button cells with an effective cathode area of 0.283 cm2 were used (Fig. 1b). Four different anode-supported SOFC samples were fabricated with different pore former content and sintering temperature of the anode substrates, namely, 11.3 wt.% pore former and sintered at 1400 C (denoted as Cell-1), 11.3 wt.% pore former and sintered at 1350 C (Cell-2), 22.6 wt.% pore former and sintered at 1350 C (Cell-3), and 33.8 wt.% pore former and sintered at 1350 C (Cell-4). 2.3. Characterization and electrochemical measurements The microstructure of the fabricated samples was observed using scanning electron microscopy (SEM, JSM-5600, JEOL) and field emission scanning electron microscopy (FE-SEM, JSM-6330F, JEOL) after the electrochemical measurements. To estimate the relative density of the YSZ electrolyte and the GDC interlayer, SEM images were first converted into binary images where pore and material phases were separated, and then the discernible material phase area was numerically integrated. To prepare the Ni-YSZ substrate samples for the porosity and pore size distribution measurements,
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
55
Fig. 1. Images for representative sample Cell-4: (a) extruded NiO-YSZ green sheet, (b) fabricated anode-supported SOFCs, (c) low magnification cross-sectional SEM image, and (d) high magnification cross-sectional SEM image. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
the extruded green NiO-YSZ sheets were first cut into proper size pieces for the measurements and sintered at either 1350 or 1400 C for 3 h in air. The sintered NiO-YSZ pieces were then heat-treated at 700 C for 4 h in 10 vol.% hydrogen (H2)-nitrogen to reduce NiO to Ni. The porosity and pore size distribution of the Ni-YSZ substrate samples were determined by the Archimedes method and mercury porosimetry, respectively. Current densityevoltage (ieV) measurements and electrochemical impedance spectroscopy (EIS) were carried out for the anode-supported SOFCs in the temperature range of 650e800 C using 3 vol.% humidified H2 as fuel at a feed rate of 70 standard temperature and pressure (STP) mL min1 and ambient air as oxidant at a feed rate of 140 STP mL min1. Before the electrochemical measurements, a gold paste (<1 mm thick, Tanaka Kikinzoku Kogyo Co.) was painted on the top surface of both anode and cathode to reduce the current collection loss. To measure the actual cell temperature as accurately as possible, a thermocouple was closely placed (approximately 1 mm from the cell surface) at the cathode side. A potentiostat/galvanostat with a frequency response analyzer (Autolab PGSTAT302, Metrohm) was used for the EIS measurements. The frequency range was 1 MHz to 0.1 Hz and the amplitude of applied voltage was 10 mV. EIS was performed under the open-circuit voltage (OCV) condition and the operating condition at 0.75 V. The impedance spectra were analyzed using ZView (Scribner Associates) with an equivalent circuit model. 3. Results 3.1. Structural properties of Ni-YSZ anodes Without any functional layers, our anode-supported SOFCs were simply composed of four layers, namely, NiO-YSZ anode, YSZ electrolyte, GDC interlayer, and LSCF-GDC cathode. As listed in Table 1, four different samples were fabricated with different pore former
content and sintering temperature of the anode substrates. Fig. 1c and d shows SEM images of a representative cell sample configuration (Cell-4) after the electrochemical measurements. The approximate thickness of the layers was 0.55 mm for the anode, 5 mm for the electrolyte, 3 mm for the interlayer, and 20 mm for the cathode. On the dense electrolyte (relative density of approximately 99%), the interlayer was prepared without severe cracks. The relative density of the interlayer was approximately 90% [27], suggesting that the interlayer was sufficiently functional to prevent formation of highly resistive phases such as lanthanum zirconate and strontium zirconate from the reaction between the cathode and electrolyte [28e31]. The cathode exhibited porous structure and good contact condition with the interlayer. Our cathode was designed to achieve a porosity between 40 and 50 vol.% so that gas diffusion was not inhibited while keeping the electronic and ionic conductivities and the electrochemical active area. FE-SEM images in Fig. 2 show cross-sections of the Ni-YSZ anodes for the different cells. All the Ni-YSZ anodes exhibited porous structure, and the porosity increased with decreasing sintering temperature and with increasing pore former content. The pores of all the anodes seemed to be evenly distributed within the anodes, suggesting that our fabrication process effectively produces finely mixed pore formers without excessive aggregation. This porous structure stably and effectively transported fuel gas to all the electrochemically active TPB regions. The grain size of network stems of Cell-2, Cell-3, and Cell-4 observed in Fig. 2 was around 0.5e1 mm, which is almost the same size as the raw powder materials of the anodes (0.7 mm), although that of Cell-1 was slightly larger than that of the other anodes. Although in general, high temperature sintering process (1350 C) promotes grain growth, the Ni and YSZ grains within Cell-2, Cell-3, and Cell-4 retained the approximate initial size after sintering at 1350 C. This result was considered to be due to the well dispersed NiO and YSZ by our mixing process successfully preventing their grain growth.
56
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
Table 1 Structural properties of Ni-YSZ anode substrates after reduction treatment fabricated by extrusion process. Sample
Pore former content (wt.%)
Sintering temperature ( C)
Porosity (vol.%)
Median pore diameter (mm)
Cell-1 Cell-2 Cell-3 Cell-4
11.3 11.3 22.6 33.8
1400 1350 1350 1350
26.9 34.1 44.1 53.6
0.447 0.541 0.743 0.911
Fig. 2. SEM images and pore size distribution of Ni-YSZ anodes after reduction treatment.
Fig. 2 also shows the pore size distribution of the Ni-YSZ anodes measured after reduction treatment. Although the two types of materials, namely, graphite carbon and cellulose, were simultaneously used as pore former for the Ni-YSZ in the present study, a single peak in the several hundred nanometer range was observed for all the Ni-YSZ anodes, indicating that uniform porous structure was achieved. Previous studies [32e34] reported that, in electrodes with the same porosity, the difference in pore size and distribution is crucial to electrode polarization. They also concluded that to achieve high performance electrodes, uniform pore size distribution is preferable because in anodes with uneven pore size distribution, the gas diffusion process is determined by the relatively small gaps among pores between Ni and YSZ particles. Table 1 shows the porosity and median pore diameter of the NiYSZ anodes after reduction treatment. The porosity increased with decreasing sintering temperature and with increasing pore former content, which agrees with the FE-SEM observation results. In addition, the median pore diameter increased with increasing pore former content. Eventually, with approximately 1 mm median pore diameter, Cell-4 reached a high porosity of 53.6 vol.%, which was more than half of the apparent volume of the Ni-YSZ anode. Porosity is the primary factor that determines the gas diffusion process. When the pore size is smaller than the mean free path of the gas molecules, namely, in the Knudsen region, another factor is pore size. The high porosity and the large pores with uniform pore size distribution for the Ni-YSZ anode of Cell-4 were therefore expected to minimize the concentration polarization resistance. 3.2. Electrochemical performance of anode-supported SOFCs Fig. 3 shows the ieV characteristics for the anode-supported
SOFCs with different porosity and pore size in the Ni-YSZ anodes using 3 vol.% humidified H2 as fuel and ambient air as oxidant. The measured OCVs for all the cell samples were similar to the theoretical values, indicating gas tightness of the electrolyte. The cell performance was improved with increasing porosity and pore size of the Ni-YSZ anodes. The power density of Cell-4 at 0.75 V was 1.95, 1.29, 0.79, and 0.44 W cm2 at 800, 750, 700, and 650 C, respectively, and the respective maximum power density reached 3.09, 1.91, 1.11, and 0.60 W cm2. Cell-4 particularly exhibited high performance at temperatures higher than 750 C, suggesting that concentration polarization dominated the overall cell performance and that the resistances due to other factors such as ohmic loss and activation polarization were sufficiently lowered in the high temperature region. At temperatures lower than 700 C, the difference in cell performance between Cell-4 and the other cell samples decreased compared with that at the higher temperatures. For example, although the maximum power density of Cell-4 was 2.1 (¼ 3.09/1.46) times higher than that of Cell-1 at 800 C, that of Cell4 decreased to 1.2 (¼ 0.60/0.49) times for that of Cell-1 at 650 C. These results were due to the increase in ohmic and activation polarization resistances with decreasing temperature, whereas the concentration polarization resistance was relatively independent of temperature. Fig. 4 shows the EIS results for the anode-supported SOFCs under the OCV condition. Here, the impedance spectra at 800 C are shown as representative results. At every temperature, the impedance spectra exhibited large arcs in the frequency range lower than 10 Hz, attributed to the gas diffusion process. These large arcs were mainly due to orders of magnitude change in pO2 at the anode side by the applied alternating current in the EIS measurements [2,16,35e38]. Fig. 5 shows the impedance spectra under
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
57
Fig. 3. ieV characteristics for anode-supported SOFCs with different porosity and pore size in Ni-YSZ anodes using 3 vol.% humidified H2 and ambient air at (a) 800 C, (b) 750 C, (c) 700 C, and (d) 650 C.
the operating condition at 0.75 V. Compared with the OCV condition, the arcs of the impedance spectra measured at 0.75 V were small, and those in the low frequency range were particularly diminished. The total resistance, determined by the intercept on the real axis at the lower frequency side (0.1 Hz), approximately coincided with the gradients of the ieV characteristics around 0.75 V (Fig. 3), indicating that the EIS results at 0.75 V can be used as an indicator to evaluate cell performance during this operating condition. 3.3. Concentration polarization resistance of anode-supported SOFCs The measured impedance spectra at 0.75 V seemed to be composed of three arcs. To separate the resistance factors, we therefore analyzed the impedance spectra by using an equivalent circuit model shown in Fig. 6, consisting of four resistances, namely,
Fig. 4. Impedance spectra for anode-supported SOFCs with different porosity and pore size in Ni-YSZ anodes using 3 vol.% humidified H2 and ambient air at 800 C under OCV condition.
ohmic resistance, Rohm, and electrode polarization resistances in the high to low frequency ranges respectively numbered, R1, R2, and R3, and inductance, L, capacitance, C1, and two constant phase elements (CPEs), Q2 and Q3. In anode-supported SOFCs, proper resistance isolation between anode and cathode by three-electrode measurements is generally impossible because reference electrodes cannot be used in this configuration due to the very thin electrolyte [39e43]. In the present study, EIS was performed between the anode and cathode, and thus the arcs of impedance spectra comprised four electrode polarization resistances, namely, the activation and concentration polarization resistances of each electrode. We therefore attributed R1 and R2 to the merged interfacial activation polarization resistance of the anode and cathode, and R3 to the merged concentration polarization resistance of the anode and cathode. Note that the concentration polarization resistance in anode-supported SOFCs can be regarded as mainly due to the anode because this resistance directly depends on the electrode thickness (in the present study, the anode thickness was 0.55 mm and the cathode thickness was 20 mm). This interpretation about the electrode polarization resistances is in general agreement with previous studies where the contributions from anode and cathode were investigated by measuring EIS under various partial pressures and temperatures [41e46]. Table 2 summarizes the obtained area-specific resistances by fitting analysis using the equivalent circuit model for the anodesupported SOFCs at 0.75 V, where Rp is the total electrode polarization resistance (¼R1 þ R2 þ R3) and Rt is the total cell resistance (¼Rohm þ Rp). Fig. 7a shows the area-specific resistances of Cell-4, which exhibited the highest electrochemical performance among the four cells, as representative data of anode-supported SOFCs. Although the concentration polarization resistance is generally measured at low partial pressure of hydrogen, pH2 , conditions, R3 corresponding to the concentration polarization resistance was clearly observed even at high pH2 in our measurements because the
58
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
Fig. 5. Impedance spectra for anode-supported SOFCs with different porosity and pore size in Ni-YSZ anodes using 3 vol.% humidified H2 and ambient air under operating voltage of 0.75 V at (a) 800 C (b) 750 C, (c) 700 C, and (d) 650 C.
Fig. 6. Equivalent circuit model used for fitting analysis. Table 2 Area-specific resistances of anode-supported SOFCs estimated from measured impedance spectra at 0.75 V using equivalent circuit model analysis. Area-specific resistance (mU cm2)
Sample
Cell-1
Cell-2
Cell-3
Cell-4
800 750 700 650 800 750 700 650 800 750 700 650 800 750 700 650
C C C C C C C C C C C C C C C C
Rohm
R1
R2
R3
Rp
Rt
65.3 103.8 161.2 288.1 69.4 98.7 165.9 296.0 64.6 94.8 149.6 264.4 63.3 93.0 149.0 278.2
5.2 9.3 34.8 75.5 4.2 18.4 38.3 85.2 6.6 16.9 44.5 90.9 5.4 18.2 44.2 86.5
47.4 69.4 111.5 198.7 27.3 47.8 80.9 156.2 23.7 44.7 77.9 150.5 24.8 39.7 66.4 139.2
39.4 28.4 28.0 33.3 19.8 18.7 19.1 22.0 13.1 13.6 14.1 16.7 10.0 9.9 10.5 11.7
92.0 107.1 174.3 307.5 51.3 84.9 138.3 263.4 43.4 75.2 136.5 258.1 40.2 67.8 121.1 237.4
157.3 210.9 335.5 595.6 120.7 183.6 304.2 559.4 108.0 170.0 286.1 522.5 103.5 160.8 270.1 515.6
other resistance factor of activation polarization was sufficiently lowered in Cell-4. R3 was relatively constant over the entire temperature range, whereas Rohm, R1, and R2, depended on temperature. Such dependence of each area-specific resistance on
temperature in Cell-4 was also observed in the other anodesupported SOFCs, confirming that R1 and R2 represented activation polarization resistance and R3 represented the concentration polarization resistance. For Rohm, the measured value (Table 2) was higher than the value calculated from the theoretical conductivity of YSZ and GDC, because Rohm contained the ohmic contact resistances at the interface of each component [27,30,47e51], although Rohm was primarily due to the bulk resistance of the electrolyte and interlayer. Fig. 7b shows the area-specific resistances of the anodesupported SOFCs as a function of the porosity of the Ni-YSZ anodes at 800 C. Because the cathodes in all the anode-supported SOFCs were identically designed in the present study, difference in electrode polarization resistance can be attributed to the anodes. As can be seen in Fig. 7b, R3 clearly decreased from 39.4 to 10.0 mU cm2 with increasing anode porosity, indicating that higher anode porosity is crucial to improve gas diffusion in anodesupported SOFCs. In contrast, R1 did not depend on anode porosity. For all the anode-supported SOFCs, R1 was constant within a narrow range (4.2e6.6 mU cm2). In Fig. 7b, every R2 except that of Cell-1 (47.4 mU cm2) exhibited similar values (23.7e27.3 mU cm2). The higher R2 of Cell-1 might be due to the decreased TPB region with grain growth in the anode because the anode of only Cell-1 was sintered at a temperature higher (1400 C) than the other anodes (1350 C) during the fabrication process. Also, R2 at 650e750 C shown in Table 2 slightly decreased with increasing anode porosity, indicating that adding a large amount of pore former might lead to an increase in the resulting TPB region in the anode. These results suggest that R1, R2, and R3 were dominantly determined by the polarizations caused in the cathode, both electrodes, and anode, respectively. To clarify the impact of the concentration polarization in the anodes on the total electrode performance, we compared R3 and Rp. Fig. 7c shows that the ratio R3/Rp decreased with increasing anode porosity and also with decreasing temperature. For example, R3/Rp was 0.43 at 800 C with an anode porosity of 26.9 vol.% (Cell-1) and was 0.05 at 650 C with an anode porosity of 53.6 vol.% (Cell-4). R3 significantly affected Rp at 800 C because R1 and R2 exhibited relatively low values at higher temperature range. This indicates
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
59
4. Discussion To discuss in more detail the improved gas diffusion property in the Ni-YSZ anodes, we estimated the concentration overpotential, hc, and the limiting current density, iL, of the Ni-YSZ anodes from experimentally obtained R3 and the following equations [2,16e20]. Here, R3 was assumed to be determined only by the concentration polarization on the anode side and the contribution of the cathode was neglected because the cathode thickness was much thinner than the anode thickness. Also, in our experimental condition, 3 vol.% humidified H2 was supplied to the anode at sufficiently excessive feed rate, namely, 70 STP mL min1. For example, the highest fuel utilization during the ieV measurements was 22%. Thus, we did not consider the effect of the fuel utilization on pH2 . hc is caused by the difference in pO2 between the outside of the anode and at the anode/electrolyte interface and can be expressed by the Nernst equation: 0
pO RT ln 2 hc ¼ 4F pO2
!
RT pH2 pH2O ln 0 ¼ 2F pH2 pH2 O
0
(1)
where R, T, and F are the gas constant, temperature, and the Faraday constant, respectively, pO2 , pH2 , and pH2 O are pO2 , pH2 , and pH2 O in 0 0 0 the feed fuel, respectively, and pO2 , pH2 , and pH2O are pO2 , pH2 , and pH2 O near the anode/electrolyte interface, respectively. Assuming that the gradient of every gas composition across the anode is 0 0 linear, pH2 and pH2O , and are expressed as follows:
0 i pH2 ¼ pH2 1 iL
(2a)
0
pH2 O ¼ pH2 O þ
pH2 i iL
(2b)
Substituting Equations (2a) and (2b) into Equation (1), hc can be expressed as
! pH2 i RT i RT ln 1 þ þ hc ¼ ln 1 2F iL 2F pH2 O iL
Fig. 7. Area-specific resistances for (a) Cell-4 as a function of temperature and (b) anode-supported SOFCs at 800 C as a function of anode porosity. (c) Ratio of concentration polarization resistance to total electrode polarization resistance R3/Rp versus anode porosity for anode-supported SOFCs (open symbols: R3, closed symbols: R3/Rp).
that higher performance electrodes that have lower R1 and R2 due to higher catalytic activity than that used in the present study lead to further increase in R3/Rp. Although R3 at a low temperature and high anode porosity might be negligible (R3/Rp < 0.1), the obtained results revealed that decreasing R3 by increasing anode porosity is essential to achieve high performance in anode-supported SOFCs.
(3)
This equation indicates that hc is determined by only iL. In the present study, pH2 and pH2 O in the actually used feed fuel were 0.97 and 0.03 atm, respectively. i at 0.75 V, i0.75V, was estimated from the ieV characteristics. Here, we used the experimental values at 800 C, namely, i0.75V ¼ 1.52, 2.01, 2.41, 2.60 A cm2 for Cell-1, Cell2, Cell-3, and Cell-4, respectively. By using Equation (3), we finally estimated iL by fitting dhc/di at i0.75V to R3. Fig. 8 shows the estimated hc and iL at 800 C. As expected from the experimental results, Cell-1, in which the apparent diffusion limitation was observed in the ieV characteristics (Fig. 3a), exhibited very high hc. In addition, hc for Cell-2 was higher than 0.1 V at 4 A cm2, although the diffusion limitation did not appear in the ieV characteristics (Fig. 3a). Because the potential loss in cell terminal voltage due to hc is theoretically inevitable, the high hc observed in Cell-1 and Cell-2 leads to hardly achieve a high power density over 3 W cm2 hc decreased with increasing iL; in particular, Cell-4 exhibited a high iL (75.7 A cm2), resulting in a very low hc (approximately 0.1 V) even in a high i (>10 A cm2). In summary, both the above calculation results and the electrochemical measurement results show that a decrease in the concentration polarization is essential to achieve high power
60
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
Fig. 8. Concentration overpotential hc and limiting current density iL for anodesupported SOFCs at 800 C using 3 vol.% humidified H2 and ambient air.
density operation. The high porosity Ni-YSZ anodes fabricated by the extrusion process used in the present study enabled the concentration polarization to significantly decrease, indicating that the present anodes are promising substrate materials for much higher performance SOFCs. 5. Conclusions High porosity Ni-YSZ anodes for planar SOFCs fabricated by an extrusion process showed reduced concentration polarization, resulting in high power density operation. The present fabrication process yielded highly distributed pores with uniform pore size in the Ni-YSZ anodes. By controlling the anode microstructure, the porosity and median pore diameter respectively reached 53.6 vol.% and 0.911 mm, thus accelerating the gas diffusion process. The performance of the anode-supported SOFCs with the Ni-YSZ anodes improved with increasing the anode porosity; the highest maximum power density was 3.09 W cm2 at 800 C using humidified H2 fuel. EIS measurements and fitting analysis also revealed a decrease in concentration polarization resistance from 39.4 mU cm2 at the lowest porosity of 26.9 vol.% to 10.0 mU cm2 at the highest porosity of 53.6 vol.% at 800 C. The impact of this improvement on the overall cell performance was notable, particularly in higher temperature range, where the other resistances such as ohmic and activation polarization resistances relatively decreased depending on temperature. The estimated limiting current density of the high porosity anode were 75.7 A cm2, resulting in a very low concentration overpotential of less than 0.1 V at 10 A cm2. Concentration overpotential is theoretically inevitable, and high porosity and large pores with uniform pore size in the anodes are therefore essential to achieve high power density SOFCs. References [1] S. Souza, S.J. Visco, L.C.D. Jonghe, J. Electrochem. Soc. 144 (1997) L35. [2] J.-W. Kim, A.V. Virkar, K.-Z. Fung, K. Mehta, S.C. Singhal, J. Electrochem. Soc. 146 (1999) 69. [3] B.C.H. Steele, A. Heinzel, Nature 414 (2001) 345. [4] Z. Wang, M. Cheng, Y. Dong, M. Zhang, H. Zhang, Solid State Ionics 176 (2005) 2555. [5] T. Suzuki, Z. Hasan, Y. Funahashi, T. Yamaguchi, Y. Fujishiro, M. Awano,
Science 325 (2009) 852. [6] A. Atkinson, S. Barnett, R.J. Gorte, J.T.S. Irvine, A.J. Mcevoy, M. Mogensen, S.C. Singhal, J. Vohs, Nat. Mater. 3 (2004) 17. [7] Y. Jin, H. Saito, K. Yamahara, M. Ihara, Electrochem. Solid-State Lett. 12 (2009) B8. [8] P.I. Cowin, C.T.G. Petit, R. Lan, J.T.S. Irvine, S. Tao, Adv. Energy Mater. 1 (2011) 314. [9] T. Suzuki, T. Yamaguchi, K. Hamamoto, Y. Fujishiro, M. Awano, N. Sammes, Energy Environ. Sci. 4 (2011) 940. [10] W.G. Wang, Y.-L. Liu, R. Barfod, S.B. Schougaard, P. Gordes, S. Ramousse, P.V. Hendriksen, M. Mogensen, Electrochem. Solid-State Lett. 8 (2005) A619. [11] S. Suzuki, H. Uchida, M. Watanabe, Solid State Ionics 177 (2006) 359. [12] H. Shimada, E. Takami, K. Takizawa, A. Hagiwara, M. Ihara, Solid State Ionics 193 (2001) 43. [13] H. Shimada, E. Takami, F. Ohba, C. Takei, A. Hagiwara, M. Ihara, J. Electrochem. Soc. 158 (2011) B1341. [14] B.S. Prakash, S.S. Kumar, S.T. Aruna, Renew. Sus. Energy Rev. 36 (2014) 149. [15] Z. Gao, E.C. Miller, S.A. Barnett, Adv. Funct. Mater. 24 (2014) 5703. [16] K. Eguchi, Y. Kunisa, K. Adachi, H. Arai, J. Electrochem. Soc. 143 (1996) 3699. [17] S. Primdahl, M. Mogensen, J. Electrochem. Soc. 146 (1999) 2827. [18] Y. Patcharavorachot, A. Arpornwichanop, A. Chuachuensuk, J. Power Sources 177 (2008) 254. [19] W. He, B. Wang, Adv. Energy Mater. 2 (2012) 329. [20] L. Zhu, L. Zhang, A.V. Virkar, J. Power Sources 291 (2015) 138. [21] H. Kishimoto, K. Yamaji, T. Horita, Y.-P. Xiong, N. Sakai, M.E. Brito, H. Yokokawa, J. Electrochem. Soc. 153 (2006) A982. [22] W.G. Bessler, J. Electrochem. Soc. 153 (2006) A1492. e, J. Electrochem. Soc. 155 (2008) B675. [23] V. Sonn, A. Leonide, E. Ivers-Tiffe [24] T. Yamaguchi, S. Shimizu, T. Suzuki, Y. Fujishiro, M. Awano, Electrochem. Commun. 10 (2008) 1381. [25] T. Suzuki, T. Yamaguchi, H. Sumi, K. Hamamoto, H. Shimada, Y. Fujishiro, J. Power Sources 272 (2014) 730. [26] T. Yamaguchi, S. Shimizu, T. Suzuki, Y. Fujishiro, M. Awano, Mater. Lett. 63 (2009) 2577. [27] H. Shimada, A. Hagiwara, Solid State Ionics 258 (2014) 38. [28] J.M. Ralph, C. Rossignol, R. Kumar, J. Electrochem. Soc. 150 (2003) A1518. [29] T. Kawada, N. Sakai, H. Yokokawa, M. Dokiya, I. Anzai, Solid State Ionics 50 (1992) 189. [30] T.L. Nguyen, K. Kobayashi, T. Honda, Y. Iimura, K. Kato, A. Neghisi, K. Nozaki, F. Tappero, K. Sasaki, H. Shirahama, K. Ota, M. Dokiya, T. Kato, Solid State Ionics 174 (2004) 163. € ver, Solid State Ionics [31] A. Mai, V.A.C. Haanappel, S. Uhlenbruck, F. Tietz, D. Sto 176 (2005) 1341. [32] H. Moon, S.-D. Kim, S.-H. Hyun, H.-S. Kim, Int. J. Hydrogen Energy 33 (2008) 1758. [33] H. Sumi, T. Yamaguchi, K. Hamamoto, T. Suzuki, Y. Fujishiro, J. Am. Ceram. Soc. 96 (2013) 3584. [34] H. Sumi, T. Yamaguchi, T. Suzuki, H. Shimada, K. Hamamoto, Y. Fujishiro, Electrochem. Commun. 49 (2014) 34. [35] N. Wagner, W. Schnurnberger, B. Müller, M. Lang, Electrochim. Acta 43 (1998) 3785. [36] A.J. McEvoy, Solid State Ionics 135 (2000) 331. [37] W. Jiang, R. Fang, J.A. Khan, R.A. Dougal, J. Power Sources 162 (2006) 316. [38] H. Nakajima, T. Kitahara, T. Konomi, J. Electrochem. Soc. 157 (2010) B1686. [39] J. Winkler, P.V. Hendriksen, N. Bonanos, M. Mogensen, J. Electrochem. Soc. 145 (1998) 1184. [40] S.B. Adler, J. Electrochem. Soc. 149 (2002) E166. [41] S. McIntosh, J.M. Vohs, R.J. Gorte, J. Electrochem. Soc. 150 (2003) A1305. [42] D.A. Cui, C.H. Yang, K. Huang, F.L. Chen, Int. J. Hydrogen Energy 35 (2010) 10495. [43] H. Finklea, X. Chen, K. Gerdes, S. Pakalapati, I. Celik, J. Electrochem. Soc. 160 (2013) F1055. [44] H. Sumi, R. Kishida, J.-Y. Kim, H. Muroyama, T. Matsui, K. Eguchi, J. Electrochem. Soc. 157 (2010) B1747. [45] H. Sumi, T. Yamaguchi, K. Hamamoto, T. Suzuki, Y. Fujishiro, T. Matsui, K. Eguchi, Electrochim. Acta 67 (2012) 159. [46] H. Sumi, T. Yamaguchi, K. Hamamoto, T. Suzuki, Y. Fujishiro, J. Power Sources 226 (2013) 354. [47] K. Eguchi, N. Akasaka, H. Mitsuyasu, Y. Nonaka, Solid State Ionics 135 (2000) 589. [48] A. Tsoga, A. Gupta, A. Naoumidis, P. Nikolopoulos, Acta Mater. 48 (2000) 4709. [49] T. Kenjo, T. Nakagawa, J. Electrochem. Soc. 143 (1996) L92. [50] F.H. van Heuveln, H.J.M. Bouwmeester, F.P.F. van Berkel, J. Electrochem. Soc. 144 (1997) 126. [51] H. Shimada, K. Takizawa, H. Michibata, A. Hagiwara, M. Ihara, J. Electrochem. Soc. 162 (2015) F40.
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Challenge for lowering concentration polarization in solid oxide fuel cells Hiroyuki Shimada*, Toshio Suzuki, Toshiaki Yamaguchi, Hirofumi Sumi, Koichi Hamamoto, Yoshinobu Fujishiro Inorganic Functional Materials Research Institute, Department of Materials and Chemistry, National Institute of Advanced Industrial Science and Technology (AIST), 2266-98 Anagahora, Shimo-shidami, Moriyama-ku, Nagoya 463-8560, Japan
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Porous microstructure control technique using extrusion process was developed. A Ni-YSZ anode shows 53.6 vol.% porosity with a median pore diameter of 0.911 mm. Lowing concentration polarization is essential to achieve high power density. Reduced concentration polarization in SOFC resulted in 3.09 W cm2 at 800 C. Estimated concentration overpotential was less than 0.1 V at 10 A cm2.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 17 July 2015 Received in revised form 7 September 2015 Accepted 8 October 2015 Available online xxx
In the scope of electrochemical phenomena, concentration polarization at electrodes is theoretically inevitable, and lowering the concentration overpotential to improve the performance of electrochemical cells has been a continuing challenge. Electrodes with highly controlled microstructure, i.e., high porosity and uniform large pores are therefore essential to achieve high performance electrochemical cells. In this study, state-of-the-art technology for controlling the microstructure of electrodes has been developed for realizing high performance support electrodes of solid oxide fuel cells (SOFCs). The key is controlling the porosity and pore size distribution to improve gas diffusion, while maintaining the integrity of the electrolyte and the structural strength of actual sized electrode supports needed for the target application. Planar anode-supported SOFCs developed in this study realize 5 mm thick dense electrolyte (yttria-stabilized zirconia: YSZ) and the anode substrate (Ni-YSZ) of 53.6 vol.% porosity with a large median pore diameter of 0.911 mm. Electrochemical measurements reveal that the performance of the anode-supported SOFCs improves with increasing anode porosity. This Ni-YSZ anode minimizes the concentration polarization, resulting in a maximum power density of 3.09 W cm2 at 800 C using humidified hydrogen fuel without any electrode functional layers. © 2015 Elsevier B.V. All rights reserved.
Keywords: Solid oxide fuel cell (SOFC) Concentration polarization Planar anode-supported cell Anode porosity Extrusion process
1. Introduction * Corresponding author. E-mail address: [email protected] (H. Shimada). http://dx.doi.org/10.1016/j.jpowsour.2015.10.024 0378-7753/© 2015 Elsevier B.V. All rights reserved.
Electrochemical ceramic cells such as solid oxide fuel cells (SOFCs) are expected to apply to next generation power devices due
54
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
to their high energy conversion efficiency and fuel flexibility. For system minimization and cost reduction, high power density operation of SOFCs is needed, and thus improvement in single cell performance is essential. Planar anode-supported SOFCs are the most promising candidate type to achieve high power density because their ohmic resistance is very low due to thin electrolyte and short current pass in the electrodes [1e5]. Although the performance of planar anode-supported SOFCs is dominated by electrode polarization resistance with decreasing ohmic resistance, electrode performance is improved by using materials having high intrinsic catalytic activity [6e9] and also by using nano-size materials [10e15] to extend the electrochemically active area such as the triple-phase boundary (TPB), where electronic, ionic, and gas phases are in contact. Electrode polarization resistance is mainly attributed to two processes, namely, interfacial reaction and gas diffusion. The resistances due to these processes are respectively called activation polarization resistance and the concentration polarization resistance. Although the dominant resistance in electrodes is generally due to activation polarization, concentration polarization often notably affects electrode performance when the electrode is relatively thick, such as in anode-supported SOFCs [2,16e20]. In particular, the concentration polarization resistance is crucial to maximize cell performance at high temperature operation (700 C), in which the ohmic resistance of the electrolyte and the activation polarization resistance of the electrodes are significantly lowered although the concentration polarization resistance remains relatively constant or is slightly increased [2,16e23]. Concentration polarization, defined as the electrode polarization due to the difference in partial pressure of oxygen, pO2 , between the anode surface and the anode/electrolyte interface, of porous electrodes is a theoretically inevitable phenomenon. Therefore, achieving high power density in planar anode-supported SOFCs requires developing an anode substrate that has good gas diffusion properties such as high porosity and large pore size and can enhance catalytic activity. Although high porosity leads to lower concentration polarization, fabricating highly porous structured anodes with sufficient mechanical strength for use as actual sized substrate in SOFCs is difficult. Furthermore, to obtain the dense electrolyte and porous anode structure, the anode substrates must have the same sintering and shrinkage behaviors as the electrolyte because they undergo cosintering with the electrolyte. Thus, the development of an electrode substrate fabrication process that can finely control porous microstructure is required. Extrusion is a conventionally used process to fabricate ceramic substrates for tubular [5,9,24], flat-tube [25], and honeycomb SOFCs [26]. Porous ceramic substrates that possess high porosity and mechanical strength can be obtained by the addition of pore former and the optimization of the extrusion process. Also, this process is attractive for industrial manufacturing because it can yield ceramic substrates even without organic solvent and with only a relatively small amount of binder compared with the other wet ceramic processes such as tape casting. In the present study, we focus on the effect of the gas diffusion properties at nickel (Ni) and yttria-stabilized zirconia (YSZ) composite porous anodes on electrochemical performance in SOFCs, and report challenges to realize highly microstructure-controlled Ni-YSZ anodes with high porosity and uniform large pores. To achieve high porosity in the anodes while maintaining the mechanical strength that allows their use for actual size planar anodesupported SOFCs, we developed a microstructure control technique using the extrusion process. The porosity and pore size of extruded nickel oxide (NiO)-YSZ anode substrates were controlled by adding pore former and by adjusting the sintering temperature. The electrochemical performance of fabricated anode-supported SOFCs was
evaluated in the temperature range of 650e800 C, and then the impact of the concentration polarization on cell performance in high power density operation was discussed. 2. Experimental 2.1. Fabrication of planar anode substrates by extrusion process NiO powder (0.7 mm in median diameter, Sumitomo Metal Mining Co.) and YSZ powder (8 mol% Y2O3eZrO2, TZ-8YS, Tosoh) at a weight ratio of 60:40 were mixed with pore former and binder. The pore former was a mixture of graphite carbon (UF-G10, Showa Denko) and cellulose (TG-101, Asahi Kasei Chemicals) at a weight ratio of 2:1, and was added to the NiO-YSZ at three different composition ratios, namely, 11.3, 22.6, and 33.8 wt.% of the NiO and YSZ powders. These materials were stirred for 30 min in a vacuum chamber, and then after the addition of a proper amount of distilled water, the mixture was stirred again for 1 h. The mixture was then aged in ambient atmosphere for 15 h, yielding clay for extrusion. To produce NiO-YSZ green sheets (Fig. 1a), the clay was then extruded through a metal mold (0.7 mm thick, 65 mm wide) using a screw type extruder (Miyazaki Iron Works Co.). NiO-YSZ anode substrates were finally fabricated by cutting the green sheets into a desired size and then sintering them at 1250 C for 1 h in air. 2.2. Fabrication of anode-supported SOFCs A YSZ paste was prepared by mixing YSZ powder (TZ-8Y, Tosoh) in a-terpineol solvent (Kanto Chemical Co.) with ethyl cellulose 45 cp (Kishida Chemical Co.), a dispersant, and a plasticizer. The YSZ paste was screen-printed onto the prefabricated NiO-YSZ anode substrates. Then the anode substrates with YSZ were co-sintered at either 1350 or 1400 C for 3 h in air. The Ce0.9Gd0.1O1.95 (GDC) interlayer was prepared by screen-printing a GDC paste, made from GDC powder (40 m2 g1 in specific surface area, Anan Kasei Co.) and the same admixtures as in the YSZ paste, onto the YSZ electrolyte and sintering at 1300 C for 1 h in air. Note that 0.3 mol% Mn was added to the GDC paste to improve the sinterability of GDC. The detailed fabrication procedure has been presented elsewhere [27]. Also, an La0.6Sr0.4Co0.2Fe0.8O3d (LSCF)-GDC paste for the cathode was prepared using LSCF powder (0.7 mm in median diameter, AGC Seimi Chemical Co.) and the same GDC powder as the interlayer material at a weight ratio of 70:30 with the same admixtures as in the YSZ paste, and then screen-printed on the GDC interlayer and sintered at 950 C for 1 h in air. In electrochemical measurements, button cells with an effective cathode area of 0.283 cm2 were used (Fig. 1b). Four different anode-supported SOFC samples were fabricated with different pore former content and sintering temperature of the anode substrates, namely, 11.3 wt.% pore former and sintered at 1400 C (denoted as Cell-1), 11.3 wt.% pore former and sintered at 1350 C (Cell-2), 22.6 wt.% pore former and sintered at 1350 C (Cell-3), and 33.8 wt.% pore former and sintered at 1350 C (Cell-4). 2.3. Characterization and electrochemical measurements The microstructure of the fabricated samples was observed using scanning electron microscopy (SEM, JSM-5600, JEOL) and field emission scanning electron microscopy (FE-SEM, JSM-6330F, JEOL) after the electrochemical measurements. To estimate the relative density of the YSZ electrolyte and the GDC interlayer, SEM images were first converted into binary images where pore and material phases were separated, and then the discernible material phase area was numerically integrated. To prepare the Ni-YSZ substrate samples for the porosity and pore size distribution measurements,
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
55
Fig. 1. Images for representative sample Cell-4: (a) extruded NiO-YSZ green sheet, (b) fabricated anode-supported SOFCs, (c) low magnification cross-sectional SEM image, and (d) high magnification cross-sectional SEM image. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
the extruded green NiO-YSZ sheets were first cut into proper size pieces for the measurements and sintered at either 1350 or 1400 C for 3 h in air. The sintered NiO-YSZ pieces were then heat-treated at 700 C for 4 h in 10 vol.% hydrogen (H2)-nitrogen to reduce NiO to Ni. The porosity and pore size distribution of the Ni-YSZ substrate samples were determined by the Archimedes method and mercury porosimetry, respectively. Current densityevoltage (ieV) measurements and electrochemical impedance spectroscopy (EIS) were carried out for the anode-supported SOFCs in the temperature range of 650e800 C using 3 vol.% humidified H2 as fuel at a feed rate of 70 standard temperature and pressure (STP) mL min1 and ambient air as oxidant at a feed rate of 140 STP mL min1. Before the electrochemical measurements, a gold paste (<1 mm thick, Tanaka Kikinzoku Kogyo Co.) was painted on the top surface of both anode and cathode to reduce the current collection loss. To measure the actual cell temperature as accurately as possible, a thermocouple was closely placed (approximately 1 mm from the cell surface) at the cathode side. A potentiostat/galvanostat with a frequency response analyzer (Autolab PGSTAT302, Metrohm) was used for the EIS measurements. The frequency range was 1 MHz to 0.1 Hz and the amplitude of applied voltage was 10 mV. EIS was performed under the open-circuit voltage (OCV) condition and the operating condition at 0.75 V. The impedance spectra were analyzed using ZView (Scribner Associates) with an equivalent circuit model. 3. Results 3.1. Structural properties of Ni-YSZ anodes Without any functional layers, our anode-supported SOFCs were simply composed of four layers, namely, NiO-YSZ anode, YSZ electrolyte, GDC interlayer, and LSCF-GDC cathode. As listed in Table 1, four different samples were fabricated with different pore former
content and sintering temperature of the anode substrates. Fig. 1c and d shows SEM images of a representative cell sample configuration (Cell-4) after the electrochemical measurements. The approximate thickness of the layers was 0.55 mm for the anode, 5 mm for the electrolyte, 3 mm for the interlayer, and 20 mm for the cathode. On the dense electrolyte (relative density of approximately 99%), the interlayer was prepared without severe cracks. The relative density of the interlayer was approximately 90% [27], suggesting that the interlayer was sufficiently functional to prevent formation of highly resistive phases such as lanthanum zirconate and strontium zirconate from the reaction between the cathode and electrolyte [28e31]. The cathode exhibited porous structure and good contact condition with the interlayer. Our cathode was designed to achieve a porosity between 40 and 50 vol.% so that gas diffusion was not inhibited while keeping the electronic and ionic conductivities and the electrochemical active area. FE-SEM images in Fig. 2 show cross-sections of the Ni-YSZ anodes for the different cells. All the Ni-YSZ anodes exhibited porous structure, and the porosity increased with decreasing sintering temperature and with increasing pore former content. The pores of all the anodes seemed to be evenly distributed within the anodes, suggesting that our fabrication process effectively produces finely mixed pore formers without excessive aggregation. This porous structure stably and effectively transported fuel gas to all the electrochemically active TPB regions. The grain size of network stems of Cell-2, Cell-3, and Cell-4 observed in Fig. 2 was around 0.5e1 mm, which is almost the same size as the raw powder materials of the anodes (0.7 mm), although that of Cell-1 was slightly larger than that of the other anodes. Although in general, high temperature sintering process (1350 C) promotes grain growth, the Ni and YSZ grains within Cell-2, Cell-3, and Cell-4 retained the approximate initial size after sintering at 1350 C. This result was considered to be due to the well dispersed NiO and YSZ by our mixing process successfully preventing their grain growth.
56
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
Table 1 Structural properties of Ni-YSZ anode substrates after reduction treatment fabricated by extrusion process. Sample
Pore former content (wt.%)
Sintering temperature ( C)
Porosity (vol.%)
Median pore diameter (mm)
Cell-1 Cell-2 Cell-3 Cell-4
11.3 11.3 22.6 33.8
1400 1350 1350 1350
26.9 34.1 44.1 53.6
0.447 0.541 0.743 0.911
Fig. 2. SEM images and pore size distribution of Ni-YSZ anodes after reduction treatment.
Fig. 2 also shows the pore size distribution of the Ni-YSZ anodes measured after reduction treatment. Although the two types of materials, namely, graphite carbon and cellulose, were simultaneously used as pore former for the Ni-YSZ in the present study, a single peak in the several hundred nanometer range was observed for all the Ni-YSZ anodes, indicating that uniform porous structure was achieved. Previous studies [32e34] reported that, in electrodes with the same porosity, the difference in pore size and distribution is crucial to electrode polarization. They also concluded that to achieve high performance electrodes, uniform pore size distribution is preferable because in anodes with uneven pore size distribution, the gas diffusion process is determined by the relatively small gaps among pores between Ni and YSZ particles. Table 1 shows the porosity and median pore diameter of the NiYSZ anodes after reduction treatment. The porosity increased with decreasing sintering temperature and with increasing pore former content, which agrees with the FE-SEM observation results. In addition, the median pore diameter increased with increasing pore former content. Eventually, with approximately 1 mm median pore diameter, Cell-4 reached a high porosity of 53.6 vol.%, which was more than half of the apparent volume of the Ni-YSZ anode. Porosity is the primary factor that determines the gas diffusion process. When the pore size is smaller than the mean free path of the gas molecules, namely, in the Knudsen region, another factor is pore size. The high porosity and the large pores with uniform pore size distribution for the Ni-YSZ anode of Cell-4 were therefore expected to minimize the concentration polarization resistance. 3.2. Electrochemical performance of anode-supported SOFCs Fig. 3 shows the ieV characteristics for the anode-supported
SOFCs with different porosity and pore size in the Ni-YSZ anodes using 3 vol.% humidified H2 as fuel and ambient air as oxidant. The measured OCVs for all the cell samples were similar to the theoretical values, indicating gas tightness of the electrolyte. The cell performance was improved with increasing porosity and pore size of the Ni-YSZ anodes. The power density of Cell-4 at 0.75 V was 1.95, 1.29, 0.79, and 0.44 W cm2 at 800, 750, 700, and 650 C, respectively, and the respective maximum power density reached 3.09, 1.91, 1.11, and 0.60 W cm2. Cell-4 particularly exhibited high performance at temperatures higher than 750 C, suggesting that concentration polarization dominated the overall cell performance and that the resistances due to other factors such as ohmic loss and activation polarization were sufficiently lowered in the high temperature region. At temperatures lower than 700 C, the difference in cell performance between Cell-4 and the other cell samples decreased compared with that at the higher temperatures. For example, although the maximum power density of Cell-4 was 2.1 (¼ 3.09/1.46) times higher than that of Cell-1 at 800 C, that of Cell4 decreased to 1.2 (¼ 0.60/0.49) times for that of Cell-1 at 650 C. These results were due to the increase in ohmic and activation polarization resistances with decreasing temperature, whereas the concentration polarization resistance was relatively independent of temperature. Fig. 4 shows the EIS results for the anode-supported SOFCs under the OCV condition. Here, the impedance spectra at 800 C are shown as representative results. At every temperature, the impedance spectra exhibited large arcs in the frequency range lower than 10 Hz, attributed to the gas diffusion process. These large arcs were mainly due to orders of magnitude change in pO2 at the anode side by the applied alternating current in the EIS measurements [2,16,35e38]. Fig. 5 shows the impedance spectra under
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
57
Fig. 3. ieV characteristics for anode-supported SOFCs with different porosity and pore size in Ni-YSZ anodes using 3 vol.% humidified H2 and ambient air at (a) 800 C, (b) 750 C, (c) 700 C, and (d) 650 C.
the operating condition at 0.75 V. Compared with the OCV condition, the arcs of the impedance spectra measured at 0.75 V were small, and those in the low frequency range were particularly diminished. The total resistance, determined by the intercept on the real axis at the lower frequency side (0.1 Hz), approximately coincided with the gradients of the ieV characteristics around 0.75 V (Fig. 3), indicating that the EIS results at 0.75 V can be used as an indicator to evaluate cell performance during this operating condition. 3.3. Concentration polarization resistance of anode-supported SOFCs The measured impedance spectra at 0.75 V seemed to be composed of three arcs. To separate the resistance factors, we therefore analyzed the impedance spectra by using an equivalent circuit model shown in Fig. 6, consisting of four resistances, namely,
Fig. 4. Impedance spectra for anode-supported SOFCs with different porosity and pore size in Ni-YSZ anodes using 3 vol.% humidified H2 and ambient air at 800 C under OCV condition.
ohmic resistance, Rohm, and electrode polarization resistances in the high to low frequency ranges respectively numbered, R1, R2, and R3, and inductance, L, capacitance, C1, and two constant phase elements (CPEs), Q2 and Q3. In anode-supported SOFCs, proper resistance isolation between anode and cathode by three-electrode measurements is generally impossible because reference electrodes cannot be used in this configuration due to the very thin electrolyte [39e43]. In the present study, EIS was performed between the anode and cathode, and thus the arcs of impedance spectra comprised four electrode polarization resistances, namely, the activation and concentration polarization resistances of each electrode. We therefore attributed R1 and R2 to the merged interfacial activation polarization resistance of the anode and cathode, and R3 to the merged concentration polarization resistance of the anode and cathode. Note that the concentration polarization resistance in anode-supported SOFCs can be regarded as mainly due to the anode because this resistance directly depends on the electrode thickness (in the present study, the anode thickness was 0.55 mm and the cathode thickness was 20 mm). This interpretation about the electrode polarization resistances is in general agreement with previous studies where the contributions from anode and cathode were investigated by measuring EIS under various partial pressures and temperatures [41e46]. Table 2 summarizes the obtained area-specific resistances by fitting analysis using the equivalent circuit model for the anodesupported SOFCs at 0.75 V, where Rp is the total electrode polarization resistance (¼R1 þ R2 þ R3) and Rt is the total cell resistance (¼Rohm þ Rp). Fig. 7a shows the area-specific resistances of Cell-4, which exhibited the highest electrochemical performance among the four cells, as representative data of anode-supported SOFCs. Although the concentration polarization resistance is generally measured at low partial pressure of hydrogen, pH2 , conditions, R3 corresponding to the concentration polarization resistance was clearly observed even at high pH2 in our measurements because the
58
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
Fig. 5. Impedance spectra for anode-supported SOFCs with different porosity and pore size in Ni-YSZ anodes using 3 vol.% humidified H2 and ambient air under operating voltage of 0.75 V at (a) 800 C (b) 750 C, (c) 700 C, and (d) 650 C.
Fig. 6. Equivalent circuit model used for fitting analysis. Table 2 Area-specific resistances of anode-supported SOFCs estimated from measured impedance spectra at 0.75 V using equivalent circuit model analysis. Area-specific resistance (mU cm2)
Sample
Cell-1
Cell-2
Cell-3
Cell-4
800 750 700 650 800 750 700 650 800 750 700 650 800 750 700 650
C C C C C C C C C C C C C C C C
Rohm
R1
R2
R3
Rp
Rt
65.3 103.8 161.2 288.1 69.4 98.7 165.9 296.0 64.6 94.8 149.6 264.4 63.3 93.0 149.0 278.2
5.2 9.3 34.8 75.5 4.2 18.4 38.3 85.2 6.6 16.9 44.5 90.9 5.4 18.2 44.2 86.5
47.4 69.4 111.5 198.7 27.3 47.8 80.9 156.2 23.7 44.7 77.9 150.5 24.8 39.7 66.4 139.2
39.4 28.4 28.0 33.3 19.8 18.7 19.1 22.0 13.1 13.6 14.1 16.7 10.0 9.9 10.5 11.7
92.0 107.1 174.3 307.5 51.3 84.9 138.3 263.4 43.4 75.2 136.5 258.1 40.2 67.8 121.1 237.4
157.3 210.9 335.5 595.6 120.7 183.6 304.2 559.4 108.0 170.0 286.1 522.5 103.5 160.8 270.1 515.6
other resistance factor of activation polarization was sufficiently lowered in Cell-4. R3 was relatively constant over the entire temperature range, whereas Rohm, R1, and R2, depended on temperature. Such dependence of each area-specific resistance on
temperature in Cell-4 was also observed in the other anodesupported SOFCs, confirming that R1 and R2 represented activation polarization resistance and R3 represented the concentration polarization resistance. For Rohm, the measured value (Table 2) was higher than the value calculated from the theoretical conductivity of YSZ and GDC, because Rohm contained the ohmic contact resistances at the interface of each component [27,30,47e51], although Rohm was primarily due to the bulk resistance of the electrolyte and interlayer. Fig. 7b shows the area-specific resistances of the anodesupported SOFCs as a function of the porosity of the Ni-YSZ anodes at 800 C. Because the cathodes in all the anode-supported SOFCs were identically designed in the present study, difference in electrode polarization resistance can be attributed to the anodes. As can be seen in Fig. 7b, R3 clearly decreased from 39.4 to 10.0 mU cm2 with increasing anode porosity, indicating that higher anode porosity is crucial to improve gas diffusion in anodesupported SOFCs. In contrast, R1 did not depend on anode porosity. For all the anode-supported SOFCs, R1 was constant within a narrow range (4.2e6.6 mU cm2). In Fig. 7b, every R2 except that of Cell-1 (47.4 mU cm2) exhibited similar values (23.7e27.3 mU cm2). The higher R2 of Cell-1 might be due to the decreased TPB region with grain growth in the anode because the anode of only Cell-1 was sintered at a temperature higher (1400 C) than the other anodes (1350 C) during the fabrication process. Also, R2 at 650e750 C shown in Table 2 slightly decreased with increasing anode porosity, indicating that adding a large amount of pore former might lead to an increase in the resulting TPB region in the anode. These results suggest that R1, R2, and R3 were dominantly determined by the polarizations caused in the cathode, both electrodes, and anode, respectively. To clarify the impact of the concentration polarization in the anodes on the total electrode performance, we compared R3 and Rp. Fig. 7c shows that the ratio R3/Rp decreased with increasing anode porosity and also with decreasing temperature. For example, R3/Rp was 0.43 at 800 C with an anode porosity of 26.9 vol.% (Cell-1) and was 0.05 at 650 C with an anode porosity of 53.6 vol.% (Cell-4). R3 significantly affected Rp at 800 C because R1 and R2 exhibited relatively low values at higher temperature range. This indicates
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
59
4. Discussion To discuss in more detail the improved gas diffusion property in the Ni-YSZ anodes, we estimated the concentration overpotential, hc, and the limiting current density, iL, of the Ni-YSZ anodes from experimentally obtained R3 and the following equations [2,16e20]. Here, R3 was assumed to be determined only by the concentration polarization on the anode side and the contribution of the cathode was neglected because the cathode thickness was much thinner than the anode thickness. Also, in our experimental condition, 3 vol.% humidified H2 was supplied to the anode at sufficiently excessive feed rate, namely, 70 STP mL min1. For example, the highest fuel utilization during the ieV measurements was 22%. Thus, we did not consider the effect of the fuel utilization on pH2 . hc is caused by the difference in pO2 between the outside of the anode and at the anode/electrolyte interface and can be expressed by the Nernst equation: 0
pO RT ln 2 hc ¼ 4F pO2
!
RT pH2 pH2O ln 0 ¼ 2F pH2 pH2 O
0
(1)
where R, T, and F are the gas constant, temperature, and the Faraday constant, respectively, pO2 , pH2 , and pH2 O are pO2 , pH2 , and pH2 O in 0 0 0 the feed fuel, respectively, and pO2 , pH2 , and pH2O are pO2 , pH2 , and pH2 O near the anode/electrolyte interface, respectively. Assuming that the gradient of every gas composition across the anode is 0 0 linear, pH2 and pH2O , and are expressed as follows:
0 i pH2 ¼ pH2 1 iL
(2a)
0
pH2 O ¼ pH2 O þ
pH2 i iL
(2b)
Substituting Equations (2a) and (2b) into Equation (1), hc can be expressed as
! pH2 i RT i RT ln 1 þ þ hc ¼ ln 1 2F iL 2F pH2 O iL
Fig. 7. Area-specific resistances for (a) Cell-4 as a function of temperature and (b) anode-supported SOFCs at 800 C as a function of anode porosity. (c) Ratio of concentration polarization resistance to total electrode polarization resistance R3/Rp versus anode porosity for anode-supported SOFCs (open symbols: R3, closed symbols: R3/Rp).
that higher performance electrodes that have lower R1 and R2 due to higher catalytic activity than that used in the present study lead to further increase in R3/Rp. Although R3 at a low temperature and high anode porosity might be negligible (R3/Rp < 0.1), the obtained results revealed that decreasing R3 by increasing anode porosity is essential to achieve high performance in anode-supported SOFCs.
(3)
This equation indicates that hc is determined by only iL. In the present study, pH2 and pH2 O in the actually used feed fuel were 0.97 and 0.03 atm, respectively. i at 0.75 V, i0.75V, was estimated from the ieV characteristics. Here, we used the experimental values at 800 C, namely, i0.75V ¼ 1.52, 2.01, 2.41, 2.60 A cm2 for Cell-1, Cell2, Cell-3, and Cell-4, respectively. By using Equation (3), we finally estimated iL by fitting dhc/di at i0.75V to R3. Fig. 8 shows the estimated hc and iL at 800 C. As expected from the experimental results, Cell-1, in which the apparent diffusion limitation was observed in the ieV characteristics (Fig. 3a), exhibited very high hc. In addition, hc for Cell-2 was higher than 0.1 V at 4 A cm2, although the diffusion limitation did not appear in the ieV characteristics (Fig. 3a). Because the potential loss in cell terminal voltage due to hc is theoretically inevitable, the high hc observed in Cell-1 and Cell-2 leads to hardly achieve a high power density over 3 W cm2 hc decreased with increasing iL; in particular, Cell-4 exhibited a high iL (75.7 A cm2), resulting in a very low hc (approximately 0.1 V) even in a high i (>10 A cm2). In summary, both the above calculation results and the electrochemical measurement results show that a decrease in the concentration polarization is essential to achieve high power
60
H. Shimada et al. / Journal of Power Sources 302 (2016) 53e60
Fig. 8. Concentration overpotential hc and limiting current density iL for anodesupported SOFCs at 800 C using 3 vol.% humidified H2 and ambient air.
density operation. The high porosity Ni-YSZ anodes fabricated by the extrusion process used in the present study enabled the concentration polarization to significantly decrease, indicating that the present anodes are promising substrate materials for much higher performance SOFCs. 5. Conclusions High porosity Ni-YSZ anodes for planar SOFCs fabricated by an extrusion process showed reduced concentration polarization, resulting in high power density operation. The present fabrication process yielded highly distributed pores with uniform pore size in the Ni-YSZ anodes. By controlling the anode microstructure, the porosity and median pore diameter respectively reached 53.6 vol.% and 0.911 mm, thus accelerating the gas diffusion process. The performance of the anode-supported SOFCs with the Ni-YSZ anodes improved with increasing the anode porosity; the highest maximum power density was 3.09 W cm2 at 800 C using humidified H2 fuel. EIS measurements and fitting analysis also revealed a decrease in concentration polarization resistance from 39.4 mU cm2 at the lowest porosity of 26.9 vol.% to 10.0 mU cm2 at the highest porosity of 53.6 vol.% at 800 C. The impact of this improvement on the overall cell performance was notable, particularly in higher temperature range, where the other resistances such as ohmic and activation polarization resistances relatively decreased depending on temperature. The estimated limiting current density of the high porosity anode were 75.7 A cm2, resulting in a very low concentration overpotential of less than 0.1 V at 10 A cm2. Concentration overpotential is theoretically inevitable, and high porosity and large pores with uniform pore size in the anodes are therefore essential to achieve high power density SOFCs. References [1] S. Souza, S.J. Visco, L.C.D. Jonghe, J. Electrochem. Soc. 144 (1997) L35. [2] J.-W. Kim, A.V. Virkar, K.-Z. Fung, K. Mehta, S.C. Singhal, J. Electrochem. Soc. 146 (1999) 69. [3] B.C.H. Steele, A. Heinzel, Nature 414 (2001) 345. [4] Z. Wang, M. Cheng, Y. Dong, M. Zhang, H. Zhang, Solid State Ionics 176 (2005) 2555. [5] T. Suzuki, Z. Hasan, Y. Funahashi, T. Yamaguchi, Y. Fujishiro, M. Awano,
Science 325 (2009) 852. [6] A. Atkinson, S. Barnett, R.J. Gorte, J.T.S. Irvine, A.J. Mcevoy, M. Mogensen, S.C. Singhal, J. Vohs, Nat. Mater. 3 (2004) 17. [7] Y. Jin, H. Saito, K. Yamahara, M. Ihara, Electrochem. Solid-State Lett. 12 (2009) B8. [8] P.I. Cowin, C.T.G. Petit, R. Lan, J.T.S. Irvine, S. Tao, Adv. Energy Mater. 1 (2011) 314. [9] T. Suzuki, T. Yamaguchi, K. Hamamoto, Y. Fujishiro, M. Awano, N. Sammes, Energy Environ. Sci. 4 (2011) 940. [10] W.G. Wang, Y.-L. Liu, R. Barfod, S.B. Schougaard, P. Gordes, S. Ramousse, P.V. Hendriksen, M. Mogensen, Electrochem. Solid-State Lett. 8 (2005) A619. [11] S. Suzuki, H. Uchida, M. Watanabe, Solid State Ionics 177 (2006) 359. [12] H. Shimada, E. Takami, K. Takizawa, A. Hagiwara, M. Ihara, Solid State Ionics 193 (2001) 43. [13] H. Shimada, E. Takami, F. Ohba, C. Takei, A. Hagiwara, M. Ihara, J. Electrochem. Soc. 158 (2011) B1341. [14] B.S. Prakash, S.S. Kumar, S.T. Aruna, Renew. Sus. Energy Rev. 36 (2014) 149. [15] Z. Gao, E.C. Miller, S.A. Barnett, Adv. Funct. Mater. 24 (2014) 5703. [16] K. Eguchi, Y. Kunisa, K. Adachi, H. Arai, J. Electrochem. Soc. 143 (1996) 3699. [17] S. Primdahl, M. Mogensen, J. Electrochem. Soc. 146 (1999) 2827. [18] Y. Patcharavorachot, A. Arpornwichanop, A. Chuachuensuk, J. Power Sources 177 (2008) 254. [19] W. He, B. Wang, Adv. Energy Mater. 2 (2012) 329. [20] L. Zhu, L. Zhang, A.V. Virkar, J. Power Sources 291 (2015) 138. [21] H. Kishimoto, K. Yamaji, T. Horita, Y.-P. Xiong, N. Sakai, M.E. Brito, H. Yokokawa, J. Electrochem. Soc. 153 (2006) A982. [22] W.G. Bessler, J. Electrochem. Soc. 153 (2006) A1492. e, J. Electrochem. Soc. 155 (2008) B675. [23] V. Sonn, A. Leonide, E. Ivers-Tiffe [24] T. Yamaguchi, S. Shimizu, T. Suzuki, Y. Fujishiro, M. Awano, Electrochem. Commun. 10 (2008) 1381. [25] T. Suzuki, T. Yamaguchi, H. Sumi, K. Hamamoto, H. Shimada, Y. Fujishiro, J. Power Sources 272 (2014) 730. [26] T. Yamaguchi, S. Shimizu, T. Suzuki, Y. Fujishiro, M. Awano, Mater. Lett. 63 (2009) 2577. [27] H. Shimada, A. Hagiwara, Solid State Ionics 258 (2014) 38. [28] J.M. Ralph, C. Rossignol, R. Kumar, J. Electrochem. Soc. 150 (2003) A1518. [29] T. Kawada, N. Sakai, H. Yokokawa, M. Dokiya, I. Anzai, Solid State Ionics 50 (1992) 189. [30] T.L. Nguyen, K. Kobayashi, T. Honda, Y. Iimura, K. Kato, A. Neghisi, K. Nozaki, F. Tappero, K. Sasaki, H. Shirahama, K. Ota, M. Dokiya, T. Kato, Solid State Ionics 174 (2004) 163. € ver, Solid State Ionics [31] A. Mai, V.A.C. Haanappel, S. Uhlenbruck, F. Tietz, D. Sto 176 (2005) 1341. [32] H. Moon, S.-D. Kim, S.-H. Hyun, H.-S. Kim, Int. J. Hydrogen Energy 33 (2008) 1758. [33] H. Sumi, T. Yamaguchi, K. Hamamoto, T. Suzuki, Y. Fujishiro, J. Am. Ceram. Soc. 96 (2013) 3584. [34] H. Sumi, T. Yamaguchi, T. Suzuki, H. Shimada, K. Hamamoto, Y. Fujishiro, Electrochem. Commun. 49 (2014) 34. [35] N. Wagner, W. Schnurnberger, B. Müller, M. Lang, Electrochim. Acta 43 (1998) 3785. [36] A.J. McEvoy, Solid State Ionics 135 (2000) 331. [37] W. Jiang, R. Fang, J.A. Khan, R.A. Dougal, J. Power Sources 162 (2006) 316. [38] H. Nakajima, T. Kitahara, T. Konomi, J. Electrochem. Soc. 157 (2010) B1686. [39] J. Winkler, P.V. Hendriksen, N. Bonanos, M. Mogensen, J. Electrochem. Soc. 145 (1998) 1184. [40] S.B. Adler, J. Electrochem. Soc. 149 (2002) E166. [41] S. McIntosh, J.M. Vohs, R.J. Gorte, J. Electrochem. Soc. 150 (2003) A1305. [42] D.A. Cui, C.H. Yang, K. Huang, F.L. Chen, Int. J. Hydrogen Energy 35 (2010) 10495. [43] H. Finklea, X. Chen, K. Gerdes, S. Pakalapati, I. Celik, J. Electrochem. Soc. 160 (2013) F1055. [44] H. Sumi, R. Kishida, J.-Y. Kim, H. Muroyama, T. Matsui, K. Eguchi, J. Electrochem. Soc. 157 (2010) B1747. [45] H. Sumi, T. Yamaguchi, K. Hamamoto, T. Suzuki, Y. Fujishiro, T. Matsui, K. Eguchi, Electrochim. Acta 67 (2012) 159. [46] H. Sumi, T. Yamaguchi, K. Hamamoto, T. Suzuki, Y. Fujishiro, J. Power Sources 226 (2013) 354. [47] K. Eguchi, N. Akasaka, H. Mitsuyasu, Y. Nonaka, Solid State Ionics 135 (2000) 589. [48] A. Tsoga, A. Gupta, A. Naoumidis, P. Nikolopoulos, Acta Mater. 48 (2000) 4709. [49] T. Kenjo, T. Nakagawa, J. Electrochem. Soc. 143 (1996) L92. [50] F.H. van Heuveln, H.J.M. Bouwmeester, F.P.F. van Berkel, J. Electrochem. Soc. 144 (1997) 126. [51] H. Shimada, K. Takizawa, H. Michibata, A. Hagiwara, M. Ihara, J. Electrochem. Soc. 162 (2015) F40.