Journal of The Electrochemical Society, 156 共12兲 B1369-B1375 共2009兲

B1369

0013-4651/2009/156共12兲/B1369/7/$25.00 © The Electrochemical Society

Evidence for Two Activation Mechanisms in LSM SOFC Cathodes M. Ali Haider* and Steven McIntosh**,z Department of Chemical Engineering, University of Virginia, Charlottesville, Virginia 22904-4741, USA Dense La0.8Sr0.2MnO3 共LSM兲 film electrodes with an average thickness of 600 nm were fabricated on yttria-stabilized zirconia and cerium gadolinium oxide by ultrasonic spray pyrolysis. LSM was studied for initial nonstationary behavior by activating with current density for short duration 共5 min兲 and long duration 共16 h兲. The polarization resistance at zero dc bias was reduced upon activation irrespective of the electrolyte, with the reduction more significant after long-duration activation. Short-duration activation was removed by deliberate introduction of La2Zr2O7 impurities into the LSM phase or by surface doping with La0.6Sr0.4FeO3 nanoparticles. However, long-duration activation still occurred in these samples. Scanning electron micrographs of short-durationactivated films showed no changes in morphology while long-duration activation resulted in a significant bulk pore formation in the LSM phase. Two distinct mechanisms for LSM activation in a solid oxide fuel cell 共SOFC兲 are proposed. Short-duration activation results in changes in the film surface chemistry while long-duration activation leads to the reconstruction of the LSM phase. © 2009 The Electrochemical Society. 关DOI: 10.1149/1.3231500兴 All rights reserved. Manuscript submitted May 7, 2009; revised manuscript received August 24, 2009. Published October 1, 2009.

Porous composite La0.8Sr0.2MnO3 共LSM兲/yttria-stabilized zirconia 共YSZ兲 cathodes are commonly utilized in solid oxide fuel cells. Many groups have reported on the complex nonstationary behavior of LSM, particularly the “activation” of LSM/YSZ electrodes upon the initial application of a cathode current.1-10 The initially high polarization resistance of the electrodes can be reduced by up to a factor of 50.3 This activation is reported as reversible. It occurs over short periods of polarization 共minutes兲 and reverts to the original state only after very long periods 共days兲 in an open circuit.3,4 Thus any proposed mechanism must account for this difference in time constant. Two primary theories have been proposed in the literature to explain this activation process: One based on changes in surface chemistry and the other based on changes in bulk structure. Jiang and co-workers performed several studies to indicate that the activation is due to the reincorporation of inhibiting SrO species upon application of cathodic polarization.1,5,6,8 This is in contrast with the work of la O’ et al., which indicated surface enrichment of Sr and Mn upon cathodic polarization10 of LSM, in agreement with the previous work of Baumann et al., which attributed the activation of La0.6Sr0.4Co0.8Fe0.2O3−␦ microelectrodes to SrO enrichment.11 Our group has demonstrated in a previous work that the activation is removed by surface addition of La2O3,4 suggesting that La enrichment is beneficial. While the nature of the change is debated, all of these studies indicate that the activation is directly linked to a change in surface chemistry. An alternative mechanism is proposed by several studies that report significant changes in electrode structure upon cathodic polarization. Bulk pore formation,12 changes in LSM crystal shape from platelike to spherical,13 pore/crack formation due to wetting/ dewetting of the underlying YSZ structure,9,14,15 and formation of surface nanoparticles at a high applied potential10 have all been observed. The mobility of reduced Mn species has been suggested to be responsible for these structural changes.16,17 As with shifts in surface chemistry, while the nature of this change varies among studies, changes in electrode microstructure are significant and influence polarization resistance. In the current study, we utilize dense LSM film cathodes with a thickness of 600 nm to determine the mechanism of electrode activation as a function of magnitude and duration of applied cathodic current, electrode/electrolyte interface, surface kinetics enhancement, and secondary phase formation. The dense films were deposited onto polycrystalline YSZ or Gd0.2Ce0.8O1.9 共CGO兲 electrolyte substrates by ultrasonic spray pyrolysis 共USP兲. The use of a dense

* Electrochemical Society Student Member. ** Electrochemical Society Active Member. z

E-mail: [email protected]

electrode film limits performance by forcing an unfavorable bulk path for oxygen ion transport from the surface incorporation site to the LSM/electrolyte interface. This is particularly apparent at zero bias due to the low ionic conductivity of LSM; however, the bulk path can become more favorable at a higher dc bias due to the reduction in the LSM phase and concomitant enhancement of bulk ion conductivity.2 In this case, the use of well-defined films simplifies microstructural characterization and provides a well-defined surface for doping when compared to porous electrodes. Experimental YSZ and CGO electrolyte pellets were fabricated by pressing ⬃2 g of YSZ powder 共TZ-8Y, Tosoh, Grove City, OH兲 and ⬃2.5 g of CGO powder 共GDC 20 M, fuelcellmaterials.com, Ohio兲 in a uniaxial press at 140 bars in a 0.75 in. diameter die. The YSZ pellet was sintered in air at 1823 K, and the CGO pellet was sintered in air at 1723 K for 6 h, resulting in dense pellets with 20 mm diameter and ⬃1 mm thickness. Both sides of the electrolyte pellets were ground to a planar surface using a 0.05 ␮m diamond polishing paste 共MasterPrep polishing suspension, Buehler, Lake Bluff, IL兲. USP was employed to deposit LSM films onto the YSZ or CGO substrates. A precursor solution was prepared by mixing aqueous solutions of lanthanum, strontium, and manganese nitrates 共minimum of 99.97%, Alfa Aesar, Ward Hill, MA兲 in the 8:2:10 ratio. This solution was pumped into a nebulizing chamber where a 2.4 MHz ultrasonic nebulizer 共Sonaer Ultrasonics, Farmingdale, NY兲 created a fine mist. Dry compressed air was used to transport the precursor droplets out of the nebulizing chamber and onto the heated substrate surface. The substrate was heated to 573 K during deposition, resulting in the decomposition of the nitrate solution as the droplets impacted on the surface. The electrode geometry and placement were controlled using an aluminum mask, resulting in circular electrodes with an area of 0.4 cm2 aligned at the center of the YSZ substrate surfaces. The electrode was deposited in three cycles with 16 min deposition time each. After every cycle the electrode was sintered at 1373 K for 4 h with a heating and cooling rate of 3 K/min. After deposition on one side, the process was repeated on the opposite side of the substrate to yield a symmetric arrangement of two identical thin-film LSM electrodes. Following deposition, an LSM reference electrode with an area of approximately 0.01 cm2 was painted onto one side of the cell using LSM powder 共Praxair Specialty Ceramics, Woodinville, WA兲 mixed with glycerin. The reference electrode was placed close to the edge of the YSZ substrate at a distance from the thin-film electrode of at least 3 times the thickness of the YSZ substrate.18 La0.6Sr0.4FeO3 共LSF兲 nanoparticles were prepared using a precipitation method. A precursor solution of lanthanum, strontium, and

Downloaded 16 Oct 2009 to 128.143.72.95. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

B1370

Journal of The Electrochemical Society, 156 共12兲 B1369-B1375 共2009兲

iron nitrates was mixed in a molar ratio of 6:4:10, and an excess of 1 M NaOH was added dropwise under sonication, resulting in a metal hydroxide precipitate. The precipitate was centrifuged and washed three times with cold deionized water and dispersed in ethanol for 15 min under sonication. It was dried at 393 K overnight and calcined at 973 K for 1 h with a heating and cooling rate of 3 K/min. LSF nanoparticles 共4.5 ␮mol兲 were used for surface doping of LSM electrodes. The LSM electrode for doped and undoped samples was similarly deposited onto the YSZ substrate. Surface doping was performed after film deposition and annealing. Doped samples were annealed at 973 K for 1 h. A silver wire was attached with silver ink to the electrode surface for electrical contacts. A lithographically patterned Pt mesh was used as the electrical contact for surface-doped samples. A negative photoresist AZ4010 was applied on the electrode surface for 30 s using a spin coater at 4000 K. The sample was baked at 393 K for 2 min and was exposed to UV light for 1.5 min under a mask. An AZ400K was used to develop the printed photoresist mask on the surface of the electrode. A precision etching coating system 共Gatan, Pleasanton, CA兲 was used to deposit 300 nm of Pt. The remaining photoresist was removed by washing with acetone, yielding a square Pt grid with 0.1 mm spacing. X-ray diffraction 共XRD兲 共Scintag XRD, XDS 2000, Cupertino, CA兲 patterns were recorded in the 2␪ range of 20–65° using Cu K␣ radiation with 0.01° step size. Field emission scanning electron microscopy 共SEM, JEOL 6700F, Waterford, VA兲 was used to analyze the film surface morphology and fracture cross section. Before analyzing in SEM, samples were coated with a Au/Pd conductive layer of ⬃50 nm in the precision etching coating system. LSF nanoparticles were analyzed for particle size using tunneling electron microscopy 共JEOL 2000FX兲. Samples were prepared by dispersing the oxide particles in ethanol and by depositing onto a copper grid with a lacy carbon support film. Images were recorded with a slow scan charge-coupled device camera. Samples were characterized for their electrochemical performance by impedance spectroscopy 共Reference 600, Gamry Instruments, Malvern, PA兲. Impedance spectra were recorded in galvanostatic mode with an ac perturbation of 1 mA root-mean-square between 1 MHz and 0.01 Hz, with 10 points collected per decade of frequency. Measurements were performed both during application of a dc bias and at an open circuit after application of a dc bias in twoor three-electrode configurations, respectively, to characterize the entire cell and individual electrodes. All the samples were tested at 973 K in laboratory air. Impedance spectra were fitted in model resistance–capacitance 共RC兲 circuits using the Levenberg– Marquardt method 共Gamry Echem Analyst, version 5.5, Malvern, PA兲.

Figure 1. XRD patterns of the LSM film deposited onto a YSZ substrate sintered at 1373 共䉲兲 and 1573 K, showing the formation of the La2Zr2O7 共〫兲 phase.

average size of ⫾400 nm. The LSM film deposited onto YSZ shows a relatively nonuniform grain size as compared to the film deposited onto CGO. If implemented correctly, a symmetric cell system to separate electrode response should yield identical impedance spectra for both electrodes when measured at zero bias before any polarization.4 Both the ohmic and polarization resistances, R⍀ and RP, should be identical. Such accuracy can only be achieved by geometric symmetry and correct placement of the reference electrode.18,21 Figure 3 shows the open-circuit impedance spectra measured for each electrode and the reference electrode for two LSM films deposited onto

a)

c)

Results Figure 1 shows the XRD pattern of an LSM film deposited onto the YSZ substrate and calcined at 1373 and 1573 K. The peaks of ¯ c and Fm3 ¯ m space the LSM and YSZ phases were indexed to R3 groups, respectively, with refined lattice parameters of 5.16 Å for YSZ and a = b = 5.517 Å and c = 13.41 Å for LSM, confirming the formation of a pure perovskite phase at 1373 K. As observed in previous studies,19,20 calcination at 1573 K led to the formation of a secondary La2Zr2O7 phase due to a reaction between YSZ and LSM. The intensities of the La2Zr2O7 peaks are relatively higher than the pure LSM peaks, suggesting that the phase was dominant in the entire film. The structure of the pure LSM films sintered at 1373 K was analyzed using SEM. Figure 2a shows a fracture cross section of the LSM film deposited onto the CGO substrate. The LSM film was dense, showing no openings or pores and having an average thickness of 600 nm. Similar film structures were observed in films deposited onto YSZ 共Fig. 2c兲. Figure 2b and d shows the finegrained nature of the LSM film deposited onto CGO and YSZ, respectively. The film deposited onto CGO had uniform grains with an

500 nm

500 nm

b)

500 nm

d)

500 nm

Figure 2. 共a兲 Fracture cross section and 共b兲 surface morphology of the dense LSM film deposited onto CGO substrate. 共c兲 Fracture cross section and 共d兲 surface morphology of the similarly deposited LSM film onto YSZ substrate.

Downloaded 16 Oct 2009 to 128.143.72.95. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Journal of The Electrochemical Society, 156 共12兲 B1369-B1375 共2009兲

B1371

a)

1

0 1

0

0 1 0

1

b)

1

0

Figure 3. Impedance spectra measured between reference → electrode 1 共䊊兲 and reference → electrode 2 共쎲兲 in air at 973 K for a symmetrical cell with the LSM film electrode deposited onto either side of the YSZ disk electrolyte. Numbers indicate frequency in 10n Hz.

0 either side of the YSZ electrolyte. The total cell R⍀, 4.85 ⍀ cm2, was split equally to give R⍀ of 2.42 ⍀ cm2 for each electrode spectrum. The total cell R⍀ agrees with that predicted for YSZ at 973 K, demonstrating that there was no significant ohmic resistance from the LSM film or contacts. The near identical polarization resistance and impedance response of the two electrodes demonstrate the reproducibility of the deposited film and the accuracy of the measurement. This measurement was performed as an accuracy check for each cell before performing any further experiments. The individual electrode impedance spectra reported here are, therefore, representative of individual electrode response. Before any polarization treatment, the single electrode RP was 92.2 ⍀ cm2. The impedance spectrum is dominated by a large low frequency arc with a 73.44 ⍀ cm2 span. The impedance spectra, as shown in Fig. 3, represent a suppressed semicircular response, which can be deconvoluted with the help of an analogous parallel RC circuit. Capacitors in this study were replaced by a constant phase element 共CPE兲 to represent the suppressed shape of the arc. Three parallel resistance CPE subcircuits in series were used. The lowest frequency arc, with a peak frequency of 0.99 Hz and a span of 73.44 ⍀ cm2, was fitted to a power 1 CPE, yielding a converted true capacitance value of 1.9 mF/cm2. The smaller, higher frequency arc, with a peak frequency of 19.86 Hz and a span of 17.44 ⍀ cm2, was fitted to a power 1 CPE, yielding a converted true capacitance value of 1.2 mF/cm2. The highest frequency arc was fitted to a span of 1.32 ⍀ cm2, yielding a true capacitance value of 1 mF/cm2. This highest frequency was of a very small span as compared to the other two arcs in all impedance measurements. Influence of the cathode/electrolyte interface.— Both YSZ and CGO electrolyte substrates were utilized to investigate the influence of electrode/electrolyte interface on electrode activation. One cell was tested following the deliberate formation of a La2Zr2O7 interfacial phase on the YSZ substrate by calcining at 1573 K, as shown in Fig. 1. Figure 4a shows the impedance spectra of an LSM film electrode on the YSZ electrolyte. The measurement was performed at an open circuit after applying cathodic currents for 5 min in increasing order up to a value of 200 mA/cm2 across the whole cell. RP decreased from a value of 92.2–81.6 ⍀ cm2 on an applying current of 10 mA/cm2 for 5 min. The span of the lowest frequency arc decreased to 62.84 ⍀ cm2 while the peak frequency increased

1 0 c)

1

0

0

1

Figure 4. Impedance spectra of the LSM film electrode measured at open circuit after application of currents across the whole cell at 0 共䊐兲, 10 共䊏兲, 20 共䉭兲, 40 共䉱兲, 80 共〫兲, 160 共⽧兲, and 200 共䉰兲 mA/cm2 for 5 min and at 160 共䉮兲 and 200 共䊊兲 mA/cm2 for 16 h in air at 973 K. The film was deposited onto 共a兲 YSZ sintered at 1373 K, 共b兲 CGO sintered at 1373 K, and 共c兲 YSZ sintered at 1573 K.

to 1.26 Hz. This was associated with an increase in the fitted true capacitance from 1.9 to 2.45 mF/cm2. However, the magnitude, peak frequency, and hence capacitance of the high frequency arc remained constant at 1.2 mF/cm2. The impedance response at an open circuit remained relatively unchanged when increasing the ap-

Downloaded 16 Oct 2009 to 128.143.72.95. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

B1372

Journal of The Electrochemical Society, 156 共12兲 B1369-B1375 共2009兲

plied current density for 5 min from 10 to 200 mA/cm2. This suggests an upper limit of activation achieved by short-term applications of the cathodic current. The LSM electrode was then polarized for 16 h at 200 mA/cm2. The impedance response at zero dc bias following this treatment is shown in the inset of Fig. 4a. RP decreased significantly to 10.8 ⍀ cm2, with the low frequency arc reduced to 5.44 ⍀ cm2 and the peak frequency increased to 6.3 Hz. The peak frequency of the high frequency arc remained constant at 19.86 Hz. The distinct change in impedance spectrum after long-term activation suggests a shift in controlling mechanism of the oxygen reduction reaction. Figure 4b shows the impedance spectra at zero dc bias of an LSM electrode on the CGO substrate at 973 K after applying the current in an increasing order for 5 min. All the impedance spectra were dominated by a single low frequency arc. Cell performance was improved with a reduction in polarization resistance from a value of 56.2–30 ⍀ cm2 on an applying current with a magnitude of 160 mA/cm2 for 5 min. This was accompanied by an increase in peak frequency from 0.4 to 1.26 Hz of the low frequency arc. The higher frequency arc had a peak frequency of 158.4 Hz and was constant during all of these measurements. Although the trend is the same, the RP observed was lower compared to those of the YSZ substrate. This is attributed to the differences in grain size and morphology between films deposited onto different substrates 共Fig. 2兲. The film structure depends strongly on the interaction of the precursor solution sprayed on the substrate.22 The activation observed in LSM on YSZ was more pronounced on CGO; however, the activation showed a similar upper limit, after which no further activation was observed when increasing the current value. Long-term polarization of the cell was for 16 h with a current magnitude of 160 mA/cm2, resulting in a similar considerable decrease in RP 共Fig. 4b兲. RP decreased from an initial value of 56.2–10 ⍀ cm2, with an increase in peak frequency of the low frequency arc to 6.31 Hz. The peak frequency of the high frequency arc was constant at 158.4 Hz after long-term polarization. The total RP and peak frequencies of the low frequency arc are of similar values for LSM tested on CGO or YSZ after long-term polarization. To investigate the role of secondary phases on the initial activation process, cells with a La2Zr2O7/LSM phase were tested using a similar procedure. No activation occurred after 5 min of applying the current 共Fig. 4c兲. The impure LSM electrode performance was lower than the pure LSM film with an initial RP of 125 ⍀ cm2 and a low frequency arc peak frequency of 0.5 Hz. In contrast, RP decreased to 12 ⍀ cm2 after application of 200 mA/cm2 for 16 h. The polarization resistances of the impure LSM film as well as the pure LSM film deposited onto YSZ or CGO are of similar values after long-term activation. This suggests that the governing mechanism after the application of the long-term current is independent of the electrode/electrolyte interface. Figure 5a shows the SEM image of the fractured cross section of a pure LSM film deposited onto CGO. The cells were quenched to room temperature by removing them from the test furnace and cooling in laboratory air. The image was taken after the initial activation of the cathode by application of a current up to 200 mA/cm2 for 5 min. The film’s dense structure is identical to the initially deposited film, with no openings or pores. The effect of long-term polarization of the film is shown in Fig. 5b. The SEM micrograph represents the cross section of a cathode film after the long-term activation of 200 mA/cm2 for 16 h. This film shows a significant change in microstructure with the formation of pores at the interface. Similar results were obtained for multiple samples. To make sure that the long-term activation is not dependent on a critical current value, the applied maximum current density for short term as well as long term was increased from 160 to 200 mA/cm2 for LSM on CGO. The impedance spectra measured were the same 共Fig. 4b兲. Operation under dc cathodic bias.— Figure 6a shows the impedance response of the pure LSM film on the YSZ electrolyte during the application of a dc cathodic current at 973 K in air. These

a)

500 nm

b)

500 nm

Figure 5. Fracture cross section of the LSM film deposited onto CGO after the application of current across the whole cell of 共a兲 200 mA/cm2 for 5 min and 共b兲 200 mA/cm2 for 16 h.

were measured on electrodes following the short-term 共5 min兲 activation but before the long-term 共16 h兲 activation. As previously reported for porous film electrodes,4 the dense film’s performance is nonlinear with current. RP decreased to 7.4, 3.2, and 2.1 ⍀ cm2 under dc cathodic biases of 10, 20, and 30 mA/cm2, respectively. The impedance response of the LSM electrode measured during the application of current is dominated by one low frequency arc of the order of 10 Hz. The true capacitance values of this arc, as extracted by an equivalent RC circuit, were 3.75, 5, and 6 mF/cm2 on applying currents of 10, 20, and 30 mA/cm2, respectively. The pure capacitance associated with the high frequency arc with a peak frequency of 49.87 Hz remained unchanged at 0.8 mF/cm2 when changing current density. The impedance spectra of the LSM electrode at CGO were measured while passing current in an identical manner 共Fig. 6b兲. The performance of the electrode was improved to cathode polarization resistances of 17.1, 8.1, and 2.8 ⍀ cm2 on applying current densities of 10, 20, and 40 mA/cm2, respectively. The peak frequency of

Downloaded 16 Oct 2009 to 128.143.72.95. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Journal of The Electrochemical Society, 156 共12兲 B1369-B1375 共2009兲

B1373

a)

0 1 1

2 1

b) Figure 7. Impedance spectra of the LSM film electrode deposited onto YSZ with Pt mesh 共䊐兲 and Pt ink 共䊏兲 as electrical contact.

The impure LSM/La2Zr2O7 film showed a similar nonlinear performance 共Fig. 6c兲. RP decreased to 6.4, 3.1, and 2.1 ⍀ cm2 on increasing currents from 10, 20, and 30 mA/cm2, respectively. These values are similar to those obtained from a pure film. The impedance spectra were dominated by an arc with a peak frequency of 39.72 Hz.

1

2 1

c)

2

1 1

Figure 6. Impedance spectra of the LSM film electrode measured in air during application of dc cathodic currents at 10 共䊐兲, 20 共䊏兲, 30 共䉭兲, and 40 共䊊兲 mA/cm2. The film was deposited onto 共a兲 YSZ sintered at 1373 K, 共b兲 CGO sintered at 1373 K, and 共c兲 YSZ sintered at 1573 K.

the low frequency arc remained constant at 7.94 Hz during the application of current. Although the trend is similar, the RP values were relatively higher on the CGO substrate as compared to the YSZ substrate. This is attributed to the differences in morphology and structure of the deposited LSM film.

Influence of surface rate enhancement.— Vance and McIntosh demonstrated a significant change in electrode activation and performance upon surface doping,4 with the complete removal of shortterm activation upon surface doping with La2O3. To provide a clear, exposed electrode surface for doping in this study, we utilized a lithographically patterned Pt mesh collector with a line spacing of 0.1 mm instead of a Pt ink covering. Figure 7 shows the impedance response of an LSM electrode on YSZ tested using the Pt mesh at 973 K in air without applying any dc bias. The impedance response looks similar to the impedance spectra obtained from earlier tests when the electrode surface was painted with silver ink for electrical contacts. The cell tested using a Pt mesh was cooled and tested again after painting the electrode surface with Pt ink. The impedance spectra perfectly overlap, demonstrating that the use of the Pt mesh does not influence the electrode mechanism. Only a slight shift in the electrode ohmic resistance, ⫾0.5 ⍀ cm2, was observed. This is attributed to the additional path length for electron transport from the mesh contact as compared with a covering of Pt. The LSF particles have average sizes of 20–30 nm with clearly defined edges and crystalline structure, as observed via transmission electron microscopy and shown in Fig. 8a. There was a negligible sintering among particles. The powder XRD pattern of the nanoparticles, shown in Fig. 8b, was indexed to the cubic perovskite struc¯ m. Figure 9a shows the impedance spectra of an LSM film ture Pm3 on YSZ, surface doped with LSF nanoparticles. The impedance spectra measured at an open circuit before polarization for the doped and undoped cells are very similar. The peak frequency of the high and low frequency arcs remained the same at 19.86 and 3.15 Hz, respectively. Impedance spectra were also recorded at zero dc bias after applying currents of increasing magnitudes from 10 to 160 mA/cm2 for 5 min. No change in the open-circuit impedance was observed 共Fig. 9a兲. The performance of the LSM cathode surface doped with nanoparticles was tested under current 共Fig. 9b兲. When applying dc cathodic currents of 10, 20, 30, 40, and 50 mA/cm2, the RP of the electrode were reduced to 8.6, 3.3, 2, 1.5, and 1.3 ⍀ cm2, respec-

Downloaded 16 Oct 2009 to 128.143.72.95. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

B1374

Journal of The Electrochemical Society, 156 共12兲 B1369-B1375 共2009兲

a) 0 1 0

20 nm

(a)

b)

1 2

(b) Figure 8. 共a兲 Electron micrograph and 共b兲 XRD pattern of LSF nanoparticles sintered at 973 K.

tively, indicating an improvement in the performance. These values are the same as those obtained from undoped cells tested under similar conditions. Discussion As demonstrated in a previous study, USP is an effective technique for fabrication of LSM film electrodes.4 In contrast with this previous work, which utilized porous nanostructured electrodes, here we studied dense film structures. The change in film morphology was controlled by changing the deposition parameters.22 The resulting dense LSM films were polycrystalline with grain sizes of 500 nm or less. A difference in grain size was observed between films deposited onto YSZ and CGO electrolyte pellets. This may explain the differences in impedance spectra obtained from these two electrolytes. Before discussing the electrode activation, it is necessary to interpret the polarization resistance in these cells. With only a limited triple-phase boundary 共TPB兲 area, oxygen incorporation on our dense film electrodes is forced to occur by a reduction in molecular oxygen on the film surface followed by oxygen anion transport through the film bulk. Bouwmeester et al. proposed a characteristic thickness lc to help ascertain when either of these two processes

0 1

0

Figure 9. 共a兲 Impedance spectra of the LSM film electrode measured at open circuit before applying current 共䊏兲 and with LSF nanoparticles on the surface measured at open circuit after application of currents across the whole cell at 0 共䊐兲, 10 共䉭兲, and 160 共䊊兲 mA/cm2 for 5 min in air at 973 K. 共b兲 Impedance spectra of the LSM film electrode with LSF nanoparticle on the surface measured in air at 973 K during application of dc cathodic currents at 10 共䊐兲, 20 共䊏兲, 30 共䉭兲, 40 共䉱兲, and 50 共䊊兲 mA/cm2.

dominate performance.23 lc is the ratio of the surface exchange coefficient k and the bulk oxygen anion self-diffusivity D of the material. If an electrode is thicker than this characteristic length, the bulk diffusion pathway may be considered to be the controlling mechanism. lc for LSM is around 3.1 nm,24 significantly lower than the electrode thickness employed in this study 共⬎500 nm兲 and hence, we can consider that the bulk transport dominates the polarization resistance. This is supported by the relatively small changes in impedance spectra resulting from changing the film substrate 共Fig. 4a and b兲 共suggesting that the interfacial transport is not dominant兲 or surface doping with more electrocatalytically active LSF nanoparticles 共Fig. 9兲 共suggesting that the surface reaction is a small contribution兲. Therefore we assign the large, lower frequency arc and corresponding RP to the bulk oxygen anion transport through the film. The decrease in RP of this low frequency arc, with corresponding increase in capacitance, suggests an increase in the rate of this bulk transport step with an increasing dc cathodic current. This agrees with the view that LSM is reduced under a strong cathodic bias,4,25,26 leading to the formation of bulk oxygen vacancies and concomitant enhancement of bulk oxygen anion transport. As discussed above, LSM-based cathode activation upon initial cathodic polarization has been observed previously in several stud-

Downloaded 16 Oct 2009 to 128.143.72.95. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp

Journal of The Electrochemical Society, 156 共12兲 B1369-B1375 共2009兲 ies. Our results further confirm this phenomenon. RP measured at zero bias decreased, following the application of cathodic current for 5 min 共Fig. 4a and b兲. The zero bias RP is reduced on both YSZ and CGO electrolytes, although the absolute improvement differs. RP continues to decrease with increasing current density up to a maximum limit. Beyond this limit, no further reduction in RP occurs when the current is applied for 5 min. These differences and limits were reproducible and are ascribed to differences in film structure between the two electrolytes. This moderate activation agrees with the previous work4 but does not occur in La2Zr2O7-contaminated films or films with surface doping of LSF nanoparticles. The controlling factor in a further activation and a more significant reduction in RP is the duration of the dc cathodic current treatment. This was found to occur in both pure LSM films and those with a significant La2Zr2O7 contaminant phase 共Fig. 4a-c兲. This activation occurs at different values of applied dc current, for example, 160 mA/cm2 for LSM on CGO 共Fig. 4b兲 and 200 mA/cm2 for LSM on YSZ 共Fig. 4a兲 in this study. This importance of time over increase polarization is further emphasized in the report of la O’ et al. that showed no pore formation in LSM films even when cathodically polarized at 5 V.10 Based on the observed differences in magnitude and occurrence, we suggest that two activation processes occur: One upon short-duration application of current and the other upon long-duration application of current. Both phenomena were observed in CGO and YSZ electrolyte substrates, indicating that they are both independent of the electrolyte material and electrode/ electrolyte charge transfer. The SEM images shown in Fig. 5b demonstrate that the longduration activation leads to significant changes in bulk film structure. No morphological changes were observed upon short-duration polarization, suggesting a different mechanism. The pore formation significantly increases the active electrode area and creates gas/ electrolyte/LSM TPB points for oxygen anion incorporation in the previous dense film structure. The combination of these changes leads to the observed large decrease in RP. Although reported in numerous other studies, there is currently no agreed mechanism for these structural changes. Kuzencov et al. suggested that the pore formation occurs due to the local oxygen vacancy concentration falling below that for bulk decomposition due to cathodic polarization.12 This agrees with our observed shift in electrode mechanism upon polarization and our observed pore formation throughout the thin film, but it does not explain why long durations are required. Huang et al.15 suggested that wetting/dewetting upon reduction can lead to significant structural changes in electrodes. This motion of cations may be expected to be slow at 973 K. The reduction in the film results in the generation of stress due to increasing lattice parameter through chemical expansion.27 We can speculate that dissipation of this stress may be a driving force for structural changes but this important restructuring phenomenon requires significant further study. Short-duration activation was removed by two techniques: Formation of La2Zr2O7 by solid-state reaction with the YSZ electrolyte 共Fig. 4c兲 and surface doping with LSF nanoparticles 共Fig. 9兲. This second parameter in particular strongly suggests that the shortduration activation is a surface chemistry phenomenon. Clearly, the surface chemistry of the La2Zr2O7-rich film is significantly different from a pure LSM electrode. For the LSF nanoparticle-doped electrodes, it is anticipated that the surface reaction rate is dominated by these electrocatalytically active particles rather than the remaining LSM surface. The surface exchange coefficient of LSF is 10−3.5 cm/s 28 compared to 1.01 ⫻ 10−9 cm/s for LSM24 at 973 K. Therefore, any change that occurs in the LSM surface composition does not influence the cell impedance, and therefore the shortduration activation is removed. There is only a small reduction in RP upon doping 共Fig. 9a兲, further confirming that the bulk transport resistance dominates the polarization resistance of the dense film. A surface process is also compatible with the short durations required to activate the electrode as evolution or reincorporation of surface species occurs over significantly shorter length scales of bulk pore

B1375

formation. As discussed above, numerous groups have reported the role of surface species in activation, although the exact surface chemistry and driving forces involved still require further study. One remaining question is whether the observed activation processes occur in higher performance porous LSM/YSZ electrodes, where the primary reaction site is close to the gas–LSM–YSZ TPB. The use of dense electrodes shifts the reaction mechanism to force bulk transport of oxygen, and this shift in mechanism may change the activation; however, we and other researchers have previously observed both short-term activation with porous electrodes1,3-6,8 and bulk pore formation in LSM.12 It is possible that the time constants and dc bias at which activation is promoted differ between dense and porous electrodes, but the activation processes appear to be the same. Conclusions Two separate activation phenomena occur in LSM film electrodes. Changes in surface chemistry occur upon short-duration 共5 min in this study兲 application of dc cathodic current while bulk pore formation occurs over a longer time 共16 h in this study兲. While the driving forces and mechanism for these changes are still poorly understood, the present results aid in reconciling previous reports on these activation mechanisms. The performance of 600 nm thick LSM film electrodes is limited by bulk ion transport with minimal enhancement observed upon surface doping with active electrocatalyst particles. Acknowledgments The authors are grateful to the University of Virginia Energy Seed Fund for the financial support. University of Virginia assisted in meeting the publication costs of this article.

References 1. S. P. Jiang, J. G. Love, J. P. Zhang, M. Hoang, Y. Ramprakash, A. E. Hughes, and S. P. S. Badwal, Solid State Ionics, 121, 1 共1999兲. 2. S. B. Adler, Chem. Rev. (Washington, D.C.), 104, 4791 共2004兲. 3. S. McIntosh, S. B. Adler, J. M. Vohs, and R. J. Gorte, Electrochem. Solid-State Lett., 7, A111 共2004兲. 4. A. A. Vance and S. McIntosh, J. Electrochem. Soc., 155, B1 共2008兲. 5. S. P. Jiang, J. Solid State Electrochem., 11, 93 共2007兲. 6. S. P. Jiang and J. G. Love, Solid State Ionics, 138, 183 共2001兲. 7. J. Fleig, Annu. Rev. Mater. Res., 33, 361 共2003兲. 8. W. Wang and S. P. Jiang, Solid State Ionics, 177, 1361 共2006兲. 9. L. Y. Woo, R. S. Glass, R. J. Gorte, C. A. Orme, and A. J. Nelson, J. Electrochem. Soc., 156, B602 共2009兲. 10. G. J. la O’, R. F. Savinell, and Y. Shao-Horn, J. Electrochem. Soc., 156, B771 共2009兲. 11. F. S. Baumann, J. Fleig, M. Konuma, U. Starke, H. U. Habermeier, and J. Maier, J. Electrochem. Soc., 152, A2074 共2005兲. 12. M. Kuznecov, P. Otschik, P. Obenaus, K. Eichler, and W. Schaffrath, Solid State Ionics, 157, 371 共2003兲. 13. S. P. Jiang and J. G. Love, Solid State Ionics, 158, 45 共2003兲. 14. Y. Y. Huang, J. M. Vohs, and R. J. Gorte, J. Electrochem. Soc., 152, A1347 共2005兲. 15. Y. Huang, J. M. Vohs, and R. J. Gorte, Electrochem. Solid-State Lett., 9, A237 共2006兲. 16. G. Corre, G. Kim, M. Cassidy, J. M. Vohs, R. J. Gorte, and J. T. S. Irvine, Chem. Mater., 21, 1077 共2009兲. 17. M. Backhaus-Ricoult, K. Adib, T. St.Clair, B. Luerssen, L. Gregoratti, and A. Barinov, Solid State Ionics, 179, 891 共2008兲. 18. S. B. Adler, J. Electrochem. Soc., 149, E166 共2002兲. 19. K. Yang, J.-H. Shen, K.-Y. Yang, I. M. Hung, K.-Z. Fung, and M.-C. Wang, J. Power Sources, 159, 63 共2006兲. 20. W.-C. J. Wei, A. Roosen, and C.-C. T. Yang, J. Am. Ceram. Soc., 87, 1110 共2004兲. 21. S. McIntosh, J. M. Vohs, and R. J. Gorte, J. Electrochem. Soc., 150, A1305 共2003兲. 22. D. Beckel, U. P. Muecke, T. Gyger, G. Florey, A. Infortuna, and L. J. Gauckler, Solid State Ionics, 178, 407 共2007兲. 23. H. J. M. Bouwmeester, H. Kruidhof, and A. J. Burggraaf, Solid State Ionics, 72, 185 共1994兲. 24. S. Jiang, J. Mater. Sci., 43, 6799 共2008兲. 25. J. Mizusaki, T. Saito, and H. Tagawa, J. Electrochem. Soc., 143, 3065 共1996兲. 26. A. Grundy, M. Chen, L. Gauckler, and B. Hallstedt, J. Phase Equilib. Diffus., 26, 131 共2005兲. 27. S. McIntosh, J. F. Vente, W. G. Haije, D. H. A. Blank, and H. J. M. Bouwmeester, Chem. Mater., 18, 2187 共2006兲. 28. J. E. ten Elshof, M. H. R. Lankhorst, and H. J. M. Bouwmeester, J. Electrochem. Soc., 144, 1060 共1997兲.

Downloaded 16 Oct 2009 to 128.143.72.95. Redistribution subject to ECS license or copyright; see http://www.ecsdl.org/terms_use.jsp