Вы находитесь на странице: 1из 6

B36

Journal of The Electrochemical Society, 155 1 B36-B41 2008


0013-4651/2007/155 1 /B36/6/$23.00 The Electrochemical Society

Evaluation and Modeling of the Cell Resistance in Anode-Supported Solid Oxide Fuel Cells
A. Leonide,z V. Sonn, A. Weber, and E. Ivers-Tiffe*
Institut fr Werkstoffe der Elektrotechnik, Universitt Karlsruhe, Karlsruhe (TH), Germany The impedance of anode-supported single cells Ni/8 yttria-stabilized zirconia YSZ anode; La0.58Sr0.4Co0.2Fe0.8O3 cathode; 8YSZ electrolyte; area 1 cm2 was characterized in a broad measuring range of temperature and air/fuel gas composition. The data has been analyzed by calculating the distribution function of relaxation times DRTs . DRT computations enabled us to separate ve different loss mechanisms occurring inside the cathode and anode without the need of an equivalent circuit. Two processes exhibit a systematic dependency on changes in the oxygen partial pressure of the cathode gas and thus can be attributed to diffusional and electrochemical losses on the cathode side. The remaining three processes are very sensitive to changes in the fuel gas but are not affected by variations of the cathode gas. These resistances are classied as a gas diffusion polarization within the anodesubstrate and as an electro-oxidation reaction at the triple-phase boundary, respectively. 2007 The Electrochemical Society. DOI: 10.1149/1.2801372 All rights reserved. Manuscript submitted June 6, 2007; revised manuscript received September 26, 2007. Available electronically November 7, 2007.

In order to develop a physicochemical model of solid oxide fuel cell SOFC single cells and to rene their efciency and long-term stability, the performance-related polarization processes have to be identied and proven. Electrochemical impedance spectroscopy EIS is one of the most promising methods for unfolding complex electrochemical systems such as SOFCs. Commonly the obtained impedance spectra are analyzed by a complex nonlinear least-squares CNLS approximation to a model function represented by an equivalent circuit.1 In this case the equivalent circuit model needs to be dened a priori without any knowledge about the real number of polarization processes contributing to the overall polarization loss of the cell. This leads very often to a severe ambiguity of the adopted model.2 To overcome that disadvantage an alternative approach for analyzing impedance spectra is used in this work. The equivalent circuit model and the optimal starting parameters for the CNLS algorithm were obtained by a preidentication of the impedance response by calculating and analyzing the corresponding distribution function of relaxation times DRTs , as reported in detail in Ref. 3. Our DRT approach is particularly advantageous for the analysis of anodesupported SOFC single cells coupled to thin electrolytes thickness less than 20 m , where reference electrodes are not applicable for the separation of anode and cathode losses.4

lyte by a continuous CGO interlayer with lateral dimensions of 12 30 mm. The cell geometry and microstructure are shown in Fig. 1a and b. Cell testing. The single cells were mounted into ceramic housings. Cathode and anode were contacted by gold and nickel meshes, respectively. Gold rings were used for sealing. The cells were operated under ambient pressure with different N2 /O2 mixtures at the

Experimental Cell geometry. The SOFC single cells analyzed within this study were based on 50 50 mm anode substrates Ni/8 yttriastabilized zirconia YSZ with an average thickness of about 1.5 mm. On these substrates, an anode functional layer Ni/8YSZ, approx. 10 m and an electrolyte 8YSZ, approx. 10 m were deposited and cored at 1400C. A Ce0.8Gd0.2O2 CGO interlayer was screen-printed and sintered on the electrolyte approx. 7 m thick . This interlayer was used to prevent a chemical reaction between LSCF lanthanum strontium cobalt ferrite and 8YSZ, which otherwise forms an insulating layer of SrZrO3. On top of this interlayer a La0.58Sr0.4Co0.2Fe0.8O3 cathode was applied by screenprinting, resulting in a thickness of approx. 45 m after sintering. Details regarding the manufacturing procedures can be found elsewhere.5,6 The active area of the working cathode was 10 10 mm. Two auxiliary electrodes in gas ow direction in front of and behind the cathode were applied for open-circuit voltage OCV control OCV probes in Fig. 1b . The electrodes were separated from the electro-

* Electrochemical Society Active Member.


z

E-mail: andre.leonide@iwe.uni-karlsruhe.de

Figure 1. a A scanning electron micrograph of the polished cross section of a fractured cell showing part of the porous anode, the dense electrolyte, the CGO buffer layer, and the porous cathode. b Design of the working and auxiliary OCV probe electrodes at the cathode side of the anode-supported SOFC single characterized cell.

Downloaded 23 Feb 2012 to 131.215.51.245. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Journal of The Electrochemical Society, 155 1 B36-B41 2008

B37

Figure 2. KramersKronig test residuals of a typical impedance spectrum.

cathode side and varying H2O/H2 mixtures at the anode side. High contents of water vapor could be realized by feeding oxygen into an upstream combustion chamber. The total anodic and cathodic gas ow rates were maintained at a constant value of 250 mL/min during all experiments. The cells were tested over a range of temperatures between 650 and 850C. Impedance measurements were carried out with a Solartron 1260 frequency response analyzer in a frequency range from 100 mHz to 1 MHz. The amplitude of the current stimulus was chosen in order to achieve a voltage response not higher than 12 mV. All experiments within this study were conducted under open-circuit conditions OCC . To investigate the parameter dependence of each single polarization process, a series of impedance measurements was carried out in which only one cell parameter at a time was varied oxygen partial pressure, water partial pressure, temperature .

validation for a typical impedance spectrum. For the most part of the spectrum the relative errors of both real and imaginary data lie below an acceptable value of 0.4%. As can be envisaged from the innite integration range in Eq. 2, the calculation of the distribution function of relaxation times requires that the analyzed impedance data are measured over an innite range of frequencies from f = 0 to . However, in reality the frequency range of measurement data is xed by the experimental conditions noise, limited frequency range of the frequency response analyzer FRA , etc. . Therefore, in order to reduce numerical errors due to this nite frequency range, an extrapolation of the data in this case the imaginary part Z of the impedance at each end of the frequency was extrapolated by a spectrum is necessary. In our case Z straight line in the plot of log Z vs log f .3 This can be done on the assumption that on the edges of the spectrum only one polarization process is still active.10 The theoretical distribution function g f of an ideal resistance capacitance RC element is represented by a Dirac peak f-f max at the respective summit frequency max = 2 f max. For a subcircuit resistor R connected in parallel with a constant phase element Q RQ the peak broadens out with decreasing exponent nRQ Fig. 3b . Another impedance element often encountered in impedance studies, describing diffusion-related processes, is the generalized nite length Warburg element G-FWS .11 The impedance expression for the G-FWS element is given by ZGFWS = RW tanh j TW j TW 3

Theoretical Background Distribution of relaxation times. The distribution function of relaxation times is associated with the complex impedance Z by the following expression3 Z = R0 + Zpol = R0 + Rpol
0

For a perfect one-dimensional diffusion limitation is equal to 0.5 and TW corresponds to l2 /Di with ld the effective diffusion thickness d and Di the effective diffusion coefcient of the diffusing species i. RW denotes the dc diffusion resistance. In this study, values of ca. 0.46 were obtained. In Fig. 3c and d the Nyquist plot and the corresponding theoretical DRT of a simulated G-FWS element with RW = 21.8 m and TW = 0.0783 s and = 0.465 are shown. In this case the theoretical distribution function is characterized by a large peak located at the characteristic frequency followed by smaller peaks at higher frequencies.

1+j

d Results and Discussion 1 Equivalent circuit analysis. Figure 4a shows a classic example for the impedance spectrum of an anode-supported single cell and Fig. 4b the corresponding DRT computation result. Whereas the polarization processes overlap in the impedance curve, processes can be clearly distinguished in the distribution function of relaxation times. From an accurate analysis of all recorded impedance data and their corresponding DRTs an equivalent circuit model composed of ve impedance elements connected in series was developed Fig. 5a . The processes P1C, P2A, and P3A are modeled by a RQ element, whereas P1A and P2C are modeled by a G-FWS element and a Gerischer element, respectively. Special attention needs to be given to the two peaks at around 10 and 100 Hz in Fig. 4b denoted as P1A + P2C . We believe that these two peaks are both ascribable to the Warburg process P1A compare with Fig. 3d . However, as we see later, depending on the operating conditions a further process P2C can overlap with the two peaks caused by P1A. Figure 5b shows the CNLS t applied to the imaginary part of the impedance curve depicted in Fig. 4b. The residuals relative errors are distributed uniformly around the frequency axis, showing no systematic deviation Fig. 5c . For most of the spectrum the relative errors lie below an absolute value of 0.25%, thus demon-

with
0

d =1

the polarization part of Here R0 represents the ohmic and Zpol Z . Rpol is the total dc polarization resistance and j the imaginary unit. The term / 1+j d species the fraction of the overall polarization with relaxation times between and + d . The real and imaginary parts of the impedance data of a linear, time-invariant system are connected by KramersKronig KK transformations.7 Therefore, it is sufcient to consider the imaginary part of the impedance only. This gives rise to the following equation Im Z =Z = Rpol
0

1+

A logarithmic variable substitution transforms expression 2 into a simple convolution equation which can be solved by Fourier transformation. From this the distribution function g f with f = 2 / =2 of the measured impedance data can be computed.3 The necessary compliance of the measured data with the KramersKronig transformation rules is veried by using the KK test for Windows software.8,9 Figure 2 shows a KramersKronig

Downloaded 23 Feb 2012 to 131.215.51.245. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

B38

Journal of The Electrochemical Society, 155 1 B36-B41 2008

Figure 3. a Nyquist plot of an RQ element for ve different nRQ values and b corresponding theoretical distributions of relaxation times max = 2 f max . c Nyquist plot of a G-FWS element and d corresponding theoretical distribution of relaxation times. The distributions were calculated from the imaginary part of the impedance using the formula derived in Ref. 13.

strating the validity of the model. Only from about 300 kHz upward do inductive artifacts caused by the electrical wiring become noticeable. The CNLS t of the impedance data was carried out with the commercial program ZView.12 Cathodic oxygen partial pressure. The oxygen content of the gas mixture supplied to the cathode was varied between 21 air and 1% balance nitrogen , whereas the composition of the fuel gas was kept constant at a ratio of 62.5% H2O to 37.5% H2. The high water content was used to reduce the anodic polarization losses to a minimum, thereby making the deconvolution of the cathodic contributions simpler. Figure 6a shows the inuence of the oxygen partial pressure pO2 cathode on the distribution of relaxation times. It is clearly visible how a process evolves in the frequency range below 10 Hz denoted as P1C at oxygen contents 0.05 atm. At the same time a second process here denoted as P2C becomes visible between 100 and 10 Hz, partly overlapping the low-frequency peak related to process P1A see arrows in Fig. 6a . This observation conrms that the overall polarization loss caused by the cathode is strongly related to the two processes P1C and P2C. However, P1A, P2A, and P3A show no dependency on the pO2 cathode , demonstrating that these three losses are ascribable to the anode. The polarization resistances obtained by the CNLS t using the equivalent circuit depicted in Fig. 5a are plotted over the partial pressure of oxygen in Fig. 6b. In this case the resistance R1A was kept xed during the entire t procedures. This approach was essential because otherwise the similar summit frequencies of P1A and P2C at lower pO2 cathode would have destabilized the t algorithm. As expected, the losses R2A and R3A are independent of the change in O2, whereas the two resistances R1C and R2C both show an almost linear trend in the double-logarithmic plane with slopes of

1.08 and 0.26, respectively. The high-frequency process of the cathode P2C is probably associated with the oxygen surface exchange kinetics of LSCF as well as the diffusivity of oxygen ions through the LSCF bulk. The low-frequency process P1C probably reects the mass-transfer resistance caused by the gas-phase diffusion in the pores of the LSCF electrode.14 In the following discussions we focus our attention on process P1C. It can easily be shown that the cathodic gas diffusion resistance can be described by the following equation15 RD cathode = RT 4F
2

lc

1 1 c 1 DO2,N2 VV,c pO2 cathode 105 Pa atm


1

1.0133

where lc is the cathode thickness, DO2,N2 is the binary diffusion coefcient for a mixture of oxygen and nitrogen, c is the tortuosity factor, and VV,c is the volume fraction porosity related to the porous cathode structure. According to the ChapmanEnskog theory, the binary diffusion coefcient can be estimated by16 Djn = 1.858 103T1.5 M j + M n /M j M n P
2 jn D 1/2

where D is the collision integral dimensionless , jn is the average collision diameter in Angstroms , M j and M n are the molecular weights of component j and n, respectively, and P is the total pressure in atm . Using D and jn data from Ref. 16, DO2,N2 T = 800C = 1.7855 cm2 /s is estimated. The model Eq. 4 was tted to the experimental data with the porosity-to-tortuosity factor c = VV,c / c representing the unknown variable lc was set to 45 m . The t result is shown in Fig. 6b

Downloaded 23 Feb 2012 to 131.215.51.245. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Journal of The Electrochemical Society, 155 1 B36-B41 2008

B39

Figure 4. a Impedance spectra of a SOFC single cell recorded at T = 800C, pO2 cathode = 0.01 atm, pH2O anode = 0.625 atm, and b corresponding DRTs. Unlike the Nyquist plot, at least ve processes are visible in the distribution curve.

dashed line . As can be seen, the agreement between the model and the experimentally obtained resistance R1C is quite good. From the t a porosity-to-tortuosity factor c = 0.014 is estimated. This value is in accordance with values reported in the literature for this type of cathode structure.17 Anodic partial pressure of water vapor. In order to analyze the dependency of each anodic process on the partial pressure of water pH2O anode in the fuel gas, the H2O content was varied stepwise between 4.88 and 62.5%. Air was used as the cathode gas. Figure 7a shows the DRTs computed from the impedance spectra recorded at four different pH2O anode values. The processes P1A and P2A both show a signicant dependency on the water content in the fuel gas, whereas process P3A is characterized by a minor dependency. In Fig. 7b the polarization resistances obtained from the CNLS t are plotted over the water partial pressure. During this t procedure the resistance R2C was kept xed in order to ensure a stable t. The polarization contribution caused by the gas-phase diffusion in the pores of the LSCF electrode can be neglected when air is used as cathode gas; thus, the resistance R1C was set xed to zero. The two resistances R2A and R3A both show an almost linear trend in the double-logarithmic plane with slopes of 0.44 and 0.20, respectively. R1A shows the highest dependence on changes in the water content. This resistance is characterized by a negligible thermal activation compare next subchapter and shows attributes that are expected to be seen by gas diffusion processes.18

Figure 5. a Equivalent circuit model used for the CNLS t of impedance data. b CNLS t of the imaginary part of the impedance spectra shown in Fig. 3. c Residual pattern of the t.

The two high-frequency processes P2A 1000 Hz and P3A 10 kHz are associated with the reaction at the triple-phase boundary involving the charge-transfer reaction.19 In the following discussions we focus our attention on process P1A. We assume that R1A describes the mass-transfer resistance caused by the gas-phase diffusion in the pores of the Ni/YSZ-anode substrate. It can easily be shown that the resistance caused by diffusion limitations in a porous anode structure can be described by the following equation15 RD anode = RT 2F
2

la

1 1 1 a + DH2O,H2 VV,a pH2 anode pH2O anode 105 Pa atm


1

1.0133

where la is the anode thickness, DH2O,H2 is the binary diffusion coefcient for a mixture of water and hydrogen, a is the tortuosity factor, and VV,a is the volume fraction porosity related to the porous

Downloaded 23 Feb 2012 to 131.215.51.245. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

B40

Journal of The Electrochemical Society, 155 1 B36-B41 2008

Figure 6. Color online a Series of distribution curves at four different pO2 cathode values. b Characteristic dependence of the tted equivalent circuit elements on the cathodic oxygen partial pressure. The dashed line indicates the model prediction according to Eq. 3 pH2O anode = 0.625 atm balance H2 , T = 800C .

Figure 7. Color online a Series of distribution curves at four differen. pH2O anode values. b Characteristic dependence of the tted equivalent circuit elements on the anodic partial pressure of water pO2 cathode = 0.21 atm air , T = 757C .

anode substrate. Using Eq. 5 with D and jn data from Ref. 16, DH2O,H2 T = 757C = 7.1659 cm2 /s is estimated. The model Eq. 6 was tted to the experimental data with the porosity-to-tortuosity factor a = VV,a / a representing the unknown variable la was set to 1.5 mm . The t result is shown in Fig. 7b dashed line . As can be seen, the model approximates very well the trend of the experimentally obtained resistance R1A. From the t a porosity-to-tortuosity factor a = 0.0575 is estimated. Thermal activation. The cells were tested at temperatures between 650 and 850C in steps of 10 K with a fuel gas composition of 62.5% H2O and 37.5% H2. Air was used as cathode gas. The DRT of the cell impedance at four different temperatures is shown in Fig. 8a. It is evident that the processes P2C, P2A, and P3A are all characterized by a pronounced thermal activation. Process P1A shows, by contrast, a negligible dependency on the operating temperature. The Arrhenius plots of each single polarization resistance, R1A, R2A, R3A, and R2C, obtained by the CNLS t of the impedance curves, are represented in Fig. 8b. In order to facilitate the t procedure the resistance R1A was kept xed constant . The resistances R2C and R3A can be approximated well with a linear t, demonstrating good Arrhenius behavior. From the slope of the tted lines the activation energies Ea,2C = 1.39 eV and Ea,3A = 1.30 eV are obtained.

The Arrhenius plot of process P2A is a more complicated one, showing two ranges of different slopes with a discontinuity between 750 and 760C. Thus, these two ranges were tted separately. For the low-temperature range 730750C an activation energy of 2.05 eV was calculated, whereas the higher temperature range 760850C is characterized by an activation energy of 0.35 eV.

Conclusion A high-resolution impedance study of anode-supported single cells with thin electrolytes 20 m was presented. The cells were characterized at OCV over a broad range of operating conditions, including different temperatures and various cathode and anode gas compositions. Calculating the DRT allowed us to identify up to ve different processes contributing to the total polarization loss of a complete anode-supported cell. The high resolution of the DRT combined with the numeric accuracy of the CNLS t enabled us to identify oxygen surface exchange kinetics, diffusivity of oxygen ions in the bulk, and gas-phase diffusion in the mixed conducting LSCF cathode. At the anode side three limiting processes were identied. One process is attributed to the gas diffusion within the anode substrate, whereas the other two are related to the electro-oxidation, including the charge-transfer reaction and the ionic transport in the Ni/YSZ anode structure. To the best of our knowledge, these are unique results reported for the rst time in the literature, demonstrating that an unambigu-

Downloaded 23 Feb 2012 to 131.215.51.245. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Journal of The Electrochemical Society, 155 1 B36-B41 2008

B41

ous and detailed electrochemical characterization of each single electrode of an anode-supported cell is possible, even if no reference electrode can be used.

Acknowledgments The authors thank Dr. N. H. Menzler of the Forschungszentrum Jlich for the generous supply of cells. Funding by the Umweltministerium Baden Wrttemberg Z04B 26013 is gratefully acknowledged.
Universitt Karlsruhe (TH) assisted in meeting the publication costs of this article.

References
1. B. A. Boukamp, Solid State Ionics, 20, 31 1986 . 2. J. R. MacDonald, Impedance Spectroscopy, p. 95, Wiley Interscience, New York 1987 . 3. H. Schichlein, A. C. Mller, M. Voigts, A. Krgel, and E. Ivers-Tiffe, J. Appl. Electrochem., 32, 875 2002 . 4. M. Mongensen and P. V. Hendriksen, in High Temperature Solid Oxide Fuel Cells, Fundamentals, Design and Applications, S. C. Singhal and K. Kendall, Editors, Elsevier Advanced Technology, New York 2003 . 5. H. P. Buchkremer, U. Diekmann, and D. Stver, in 2nd European Solid Oxide Fuel Cell Forum, B. Thorstensen, Editor, p. 221, European Fuel Cell Forum, Oslo, Norway, May 610, 1996. 6. A. Mai, V. A. C. Haanappel, S. Uhlenbruck, F. Tietz, and D. Stver, Solid State Ionics, 176, 1341 2005 . 7. D. L. Misell and R. J. Sheppard, J. Phys. D, 6, 379 1973 . 8. B. A. Boukamp, J. Electrochem. Soc., 142, 1885 1995 . 9. B. A. Boukamp, Solid State Ionics, 169, 65 2004 . 10. B. A. Boukamp, Solid State Ionics, 62, 131 1993 . 11. V. Brichzin, J. Fleig, H.-U. Habermeier, G. Cristiani, and J. Maier, Solid State Ionics, 152, 499 2002 . 12. D. Johnson, Zplot, ZView Electrochemical Impedence Software, version 2.3b, Scriber Associates, Inc., Southern Pines, NC, USA 2000 . 13. R. M. Fuoss and J. G. Kirkwood, J. Am. Chem. Soc., 63, p. 385 1941 . 14. S. B. Adler, J. A. Lane, and B. C. H. Steele, J. Electrochem. Soc., 143, 3554 1996 . 15. J. Kim, A. V. Virkar, K. Fung, K. Mehta, and S. C. Singhal, J. Electrochem. Soc., 146, 69 1999 . 16. R. Reid, J. Prausnitz, and T. Sherwood, The Properties of Gases and Liquids, 3rd ed., p. 548, McGraw-Hill, New York 1977 . 17. T. Ackmann, L. G. J. de Haart, W. Lehnert, and D. Stolten, J. Electrochem. Soc., 150, 783 2003 . 18. S. Prindahl and M. Morgensen, J. Electrochem. Soc., 146, 2827 1999 . 19. V. Sonn, A. Leonide, and E. Ivers-Tiffe, J. Electrochem. Soc., Submitted.

Figure 8. Color online a Series of distribution curves at four temperatures. b Characteristic dependence of the tted equivalent circuit elements on the operating temperature pH2O anode = 0.625 atm balance H2 , pO2 cathode = 0.21 atm air .

Downloaded 23 Feb 2012 to 131.215.51.245. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp