Activation and Deactivation Kinetics of Oxygen Reduction over a La0

Nov 4, 2008 - La(NO3)3·xH2O was first prepared into an aqueous solution around 1 M with its ... A post factum correction of the IR drop in the all ov...
0 downloads 0 Views 482KB Size
18690

J. Phys. Chem. C 2008, 112, 18690–18700

Activation and Deactivation Kinetics of Oxygen Reduction over a La0.8Sr0.2Sc0.1Mn0.9O3 Cathode Yao Zheng, Ran Ran, and Zongping Shao* State Key Laboratory of Materials, Oriented Chemical Engineering, College of Chemistry and Chemical Engineering, Nanjing UniVersity of Technology, No.5 Xin Mofan Road, Nanjing, 210009, People’s Republic of China ReceiVed: August 4, 2008; ReVised Manuscript ReceiVed: September 2, 2008

Electrochemical impedance spectroscopy, step current polarization, and cyclic voltammetry were applied to investigate the activation and deactivation kinetics of oxygen reduction over a novel La0.8Sr0.2Sc0.1Mn0.9O3 (LSSM) cathode material. Oxygen vacancies were created after cathodic polarization for a certain period of time. The generating rate was closely related with oxygen partial pressure of surrounding atmosphere (PO2), polarization time, temperature, and voltage. The in situ created oxygen vacancies could propagate both over the surface and into the bulk of the LSSM electrode after a high cathodic polarization. Both chemical oxidation by ambient air and electrochemical oxidation by anodic polarization were exploited to demonstrate the deactivation mechanism of these in situ created oxygen vacancies. The rate-determining step of oxygen reduction reaction over LSSM electrode before and after the activation was also investigated. It was by oxygen ion surface diffusion at 800 °C in air, while a steady change to an electron-transfer process was observed with decreasing temperature and PO2. 1. Introduction A solid-oxide fuel cell (SOFC) is an all solid state electrochemical device applying hydrogen or hydrocarbon as fuels with low emissions and high energy conversion efficiency. Because the oxygen electrocatalytic reduction over cathode typically has slower kinetics and higher activation energy than hydrogen electrocatalytic oxidation over anode, main cell polarization loss is frequently contributed from the cathode at lower operating temperatures. Therefore, developing cathode materials with high electrocatalytic activity for oxygen reduction and better understanding the mechanism of oxygen reduction reaction (ORR) are of great technological or scientific importance toward the commercialization of SOFC.1,2 In the past 30 years, many perovskite oxides have been exploited as cathodes of SOFCs.3-7 Among these materials, lanthanum strontium manganese (LSM) has been the most widely investigated since mid 1980s. These investigations mainly focused on its chemical composition,8-10 defect chemistry, nonstoichiometry,11-15 chemical and thermal matching with a classical electrolyte of yttria-stabilized zirconia (YSZ),16-18 electronic structure, electronic conductivity, Seebeck coefficient,19-21 mechanical and sintering capabilities,22,23 and so on. The mechanism of oxygen electrochemical reduction over LSM cathode has also been intensively exploited by isotope tracing,24,25 secondary ion mass spectrometer,26,27 electrochemical impedance spectroscopy (EIS) and step current polarization based on porous electrode, dense patterned thin-film,28,29 or fine geometrical microconfiguration electrode,30-32 or by theoretical simulation.33-37 EIS and step current polarization in combination with proper mathematic modes are the two most powerful tools to investigate ORR over cathode.38-48 The ORR over an oxide electrode involves many subprocesses, which are closely related with surface and bulk oxygen vacancy concentration of the oxide. LSM typically exhibits * To whom correspondence should be addressed. Phone: 86 25 83587722. Fax: 86 25 83365813. E-mail: [email protected].

apparent oxygen excess nonstoichiometry (δ < 0) in air and a weakly reducing atmosphere, which is significantly different from ionic-electronic mixed conductor perovskite as strontium cobalt based oxides.8,9 Because of its negligible bulk oxygen ionic conductivity, the active site for ORR over LSM electrode is mainly along the electrolyte-electrode-air triple-phase boundary (TPB) region under open circuit voltage (OCV). A large cathodic polarization resistance (Rp) is usually experienced at low operating temperature. Many strategies have been tried to improve the cathode performance of LSM at reduced temperature, such as introducing an ionic conducting phase YSZ into LSM to form a composite electrode.49-53 Such improvement is due to the extension of active sites from the traditional TPB into the whole bulk of composite electrode. The most direct and simplest method to activate LSM electrode is still via the cathodic polarization, which can create oxygen vacancy in the bulk and/or over the surface of electrode. A great extension of oxygen reduction sites and reduction in Rp is then expected. The generation of oxygen vacancy by cathodic polarization is fairly complicated. There are many literatures describing this process with focus on the generating site and diffusion path.42-44,54-56 Up to now, the investigations have been primarily conducted by static measurement while the electrode kinetics analysis was less involved. Previously, we have successfully demonstrated a new LSM-based cathode La0.8Sr0.2ScyMn1-yO3 which showed a better cathode performance than LSM.57-59 In the present work, EIS, cyclic voltammetry, and Tafel slope have been conducted to evaluate the formation kinetics of oxygen vacancy in this new La0.8Sr0.2Sc0.1Mn0.9O3 (LSSM) electrode and also the propagating path of these generated oxygen vacancies. The deactivation kinetics of these in situ created oxygen vacancies were also exploited via electrochemical oxidation by anodic polarization and chemical oxidation by ambient air.

10.1021/jp806941d CCC: $40.75  2008 American Chemical Society Published on Web 11/05/2008

Oxygen Reduction over a La0.8Sr0.2Sc0.1Mn0.9O3 Cathode

J. Phys. Chem. C, Vol. 112, No. 47, 2008 18691

2. Experimental Section LSSM oxide powder was synthesized by a combined EDTA-citrate complexing sol-gel process.60 La(NO3)3 · xH2O, Sr(NO3)3, Sc2O3, and Mn(Ac)2 · 4H2O, (all A.R. grade) were used as the cation sources. La(NO3)3 · xH2O was first prepared into an aqueous solution around 1 M with its precise concentration determined by standard EDTA titration technique. Sc2O3 was prepared in 1 M aqueous Sc(NO3)3 solution by dissolving in 6 M nitric acid under heating and then diluted with a proper amount of deionized water. Required amounts of La(NO3)3, Sr(NO3)2, Sc(NO3)3, and Mn(Ac)2 · 4H2O were then prepared into a mixed aqueous solution. EDTA-NH3 · H2O solution and citric acid solid at a mole ratio of total metal ions to EDTA to citric acid of 1:1:2 were added in sequence under stirring and heating. The water was evaporated from the solution by heating at 90 °c until a transparent gel was obtained, which was prefired at 250 °c and further calcinated at 950 °c in air for 5 h to get the final products with the desired lattice structure. Electrochemical characterization of LSSM was conducted based on a three-electrode configuration with homemade (Sc2O3)0.1(ZrO2)0.9 (ScSZ) as electrolyte. To fabricate a single three-electrode cell, ScSZ powders were pressed into diskshape pellets and then sintered in air at 1500 °C for 5 h to achieve dense electrolyte substrate with a diameter of ∼13 mm and a thickness of ∼0.3 mm. The as-prepared LSSM powders were dispersed in a premixed solution of glycerol, ethylene glycol, and isopropyl alcohol to form a colloidal suspension by a high-energy ball miller (Fritsch, Pulverisettle 6) at a rotational speed of 400 rpm for 1 h. The suspension slurry was sprayed onto one side of electrolyte substrate and then calcinated at 1100 °C in air for 2 h to perform as working electrode (WE); ethanolbased silver slurry was used as current collector. Pt paste (PEPt-7840, Guizhou, China) was applied to the other side of electrolyte as symmetrically as possible with WE and calcinated at 950 °C in air for 1 h to act as counter electrode (CE). Ag paste (DAD-87, Shanghai, China) was used as reference electrode, which was painted as a ring surrounding the CE. The gap between CE and RE was ∼4 mm; the area of the WE, CE,and RE was 0.26, 0.26, and 0.3 cm2, respectively. The cathodic polarization resistance was investigated by EIS using a Solartron 1260 Frequency Response Analyzer in combination with a Solartron 1287 Potentiostat. The frequency of EIS was ranged from 0.1 Hz to 1000 kHz, and signal amplitude was 10 mV. Samples were tested under OCV conditions and constant cathodic or anodic polarization voltages of (0.5 V. Data were collected using Z-View 2.9c software. The overall impedance data were fitted by a complex nonlinear least-squares (CNLS) fitting program via Z-Plot 2.9c software. The cathode overpotential was obtained by step current polarization measurement controlled by Corrware 2.9c software; polarization current density ranged from 0 the 1000 mA cm-2 at 10 mA S-1 per step. Cyclic voltammogram of LSSM electrode were also obtained through Corrware 2.9c software. The electrode was first scanned from 0 to +0.25 V to consume the possible oxygen vacancies in the bulk or over the surface of the oxide, a positive scan from +0.25 to -1.00 V and then reverse scan from -1.0 to 0 V were successively conducted. Each fitting data from the EIS, CV, and step current polarization is the average value of three successive measurements. A post factum correction of the IR drop in the all overpotential and CV measurements was conducted, where R can be obtained from EIS. The various PO2 were balanced by O2 (99.999%), N2 (99.999%) and 1 vol. % O2/Ar.

Figure 1. Polarization currents responding to polarization time under sequential cathodic, anodic, cathodic polarization of (0.5 V at 800 °C in various oxygen partial pressures.

3. Results and Discussion Under discharge state, oxygen is electrocatalytically reduced to oxygen ion over the cathode surface or along the TPB region in a SOFC. By use of Kroger-Vink notation, the overall ORR can be written as

1 O + VO•• + 2e f OO× 2 2

(1)

where VO•• notes double charged oxygen vacancy in electrolyte lattice or bulk and/or surface of cathode; OO× is oxygen ion in a normal oxygen site. According to eq 1, the ORR is closely related with the oxygen vacancy concentration. We previously demonstrated that introdcution of a small amount of Sc3+ into the B site of LSM did not create intrinsic oxygen vacancy into the bulk of LSSM, i.e., the ORR still occurred at the TPB region in LSSM electrode under zero direct current passage. However, Sc3+ doping facilitated the creation of oxygen vacancy in LSSM under polarization condition.57 Figure 1 shows the dependence of polarization current on polarization time under sequential cathodic, anodic, cathodic polarization of -0.5, +0.5, -0.5 V at 800 °C under various atmospheres with different PO2. Under a cathodic polarization, the polarization current experienced a sharp increase in the initial stage and reached a steady value after a certain time under atmospheres with relative high PO2 (0.21 and 1 atm). This implies the resistance for oxygen activation decreased obviously after the polarization. It is generally believed that Mn4+ in a perovskite lattice can be partially reduced after a cathodic current passage in concomitance with the formation of oxygen vacancy, which can effectively facilitate the ORR. It also should be noted that the surface SrO segregation has turned out to be a normal phenomenon for LSM electrode. Its effect on cathodic polarization was detailed studied by Jiang et al.61,62 They reported that such surface enriched SrO would be incorporated into the LSM lattice during the cathodic polarization.61,62 As to LSSM, via the technique of water boiling of the oxide and then monitoring change of the pH value of the water,57 we have demonstrated previously that the SrO surface segregation was greatly suppressed because of the Sc3+ doping. Therefore, the SrO segregation was not considered in the activation and deactivation kinetics of oxygen reduction over LSSM cathode in this study. As can be seen from Figure 2, the cathodic resistance of LSSM did decrease obviously after the polarization. It reduced from an initial value of 1.45 Ω cm2 to 0.20 Ω cm2 at 800 °C in air after a cathodic polarization for 40 min. Under low PO2, for

18692 J. Phys. Chem. C, Vol. 112, No. 47, 2008

Zheng et al.

[

]

•• × •• d[VO,LSSM ] [OO,LSSM ][Mn•Mn]2[VO,ScSZ ] RcnFE exp ) k1 × × 2 dt RT [O ][Mn ] O,ScSZ

Mn

k2

•• [VO,LSSM ]pO0.52 [Mn×Mn]2 × [OO,LSSM ][Mn•Mn]2

(4)

Equation 4 can be simplified to

[

]

•• d[VO,LSSM ] RcnFE •• - k2[VO,LSSM ]pO0.52 ) k1 exp dt RT

(5)

where

k1 ) k1 Figure 2. Cathodic polarization resistance of LSSM tested in various conditions.

example, 0.01 atm, the change of polarization current with respect to polarization time was not so obvious. After the cathodic polarization, the sample was immediately switched to an anodic polarization of 0.5 V for a same period of time. Different from the cathodic polarization, the anodic current decreased quickly in the initial period and reached an equilibrium value for 10-30 min. At lower PO2, such a change was more significant and required a longer period to reach its steady state. The decrease in polarization current under anodic polarization indicates that the oxygen activation was blocked due to the decrease in oxygen vacancy concentration. When a second cathodic polarization was applied after the anodic polarization, the current curve matched well with the previous one. This implies the generation of oxygen vacancy by cathodic polarization has a highly reversible kinetics. 3.1. Oxygen Vacancy Formation Kinetics via DC Polarization. By application of cathodic polarization voltage of -0.5 V (PO2 ) 2.2 × 10-10 atm) to LSSM, Mn4+ in LSSM lattice was partial reduced to Mn3+. To sustain the macroscopic electrical neutrality, the concomitant generation of oxygen vacancy at the electrode/electrolyte interface region was happened. Such electrochemical reaction can be written as k1

× •• OO,LSSM + 2Mn•Mn + VO,ScSZ + 2e ′ 98 2Mn×Mn + •• × + OO,ScSZ (2) VO,LSSM • × denote Mn4+ and Mn3+, O× where MnMn and MnMn O,LSSM and × OO,ScSZ are oxygen ions in LSSM and ScSZ lattice sites, V••O,LSSM •• and VO,ScSZ are oxygen vacancies in LSSM and ScSZ, respectively. On the other hand, these formed oxygen vacancies were also consumed from chemical oxidation by ambient oxygen via the reaction

k2 1 •• × VO,LSSM + O2 + 2Mn×Mn 98 OO,LSSM + 2Mn•Mn 2

(3)

The generating rate of oxygen vacancy under cathodic polarization is the difference between the oxygen vacancy creating rate and eliminating rate, which can be described as54

× •• [OO,LSSM ][Mn•Mn]2[VO,ScSZ ] × [OO,ScSZ ][Mn×Mn]2

k2 ) k2

Interation of eq 5 then makes

•• [VO,LSSM ])

[

•• [VO,LSSM ][Mn×Mn]2 × [OO,LSSM ][Mn•Mn]2

]

RcnFE  0.5 RT ( 1 - e-k2pO2t)  0.5 k2pO2

k1 exp

(6)

On the basis of eq 6, the dependences of oxygen vacancy concentration on temperature, polarization time, and PO2 of the surrounding are •• log([VO,LSSM ]) ∝ log(t) •• log([VO,LSSM ]) ∝ log •• log([VO,LSSM ]) ∝

(7)

(PO )

(8)

1 T

(9)

2

Under the steady state, various cations and lattice oxygen will get their equilibrium concentrations. The maximum oxygen vacancy concentration is

•• [VO,max ])

[

RcnFE RT  0.5 k2pO2

k1 exp

]

(10)

It was reached at the limiting time of

t)

1 -0.5 pO k2 2

(11)

which depends only on surrounding PO2. On the basis of eqs 7-11, those phenomena observed under cathodic polarization with various testing conditions as shown in Figure 1 can be well explained. Parts a-c of Figure 3 show the typical CV of LSSM electrode under various testing conditions. Every CV exhibits a typical hysteresis loop (counterclockwise response). The area of loop can be regarded as the concentration of in situ created oxygen vacancy from the cathodic polarization.54,55 In Figure 3, the higher temperature or PO2 or the longer polarization time, the more amount of oxygen vacancies were generated. Furthermore, a linear response of hysteresis loop area to PO2 and scanning rate and an idempotent relationship with temperature were observed. Such phenomena are in well agreement with eqs 7-9. 3.2. Propagation of the in Situ Created Oxygen Vacancies. In eq 2, the in situ created oxygen vacancies are all congregated at the electrode/electrolyte interface. Because of their low migration energy (∼0.7 eV),63 these highly mobile oxygen vacancies may propagate to both surface and bulk of the electrode. There are generally two mechanisms to describe the

Oxygen Reduction over a La0.8Sr0.2Sc0.1Mn0.9O3 Cathode

J. Phys. Chem. C, Vol. 112, No. 47, 2008 18693

Figure 3. Typical CV of LSSM under various testing conditions. (a) Scan rate ) 20 mV S1-, T ) 800 °C, under various PO2. (b) PO2 ) 0.21 atm, T ) 800 °C, with various scan rates. (c) PO2 ) 0.21 atm, scan rate ) 20 mV S1-, under various temperatures. Inset is logarithmic loop area depending on various testing parameters.

propagating path of oxygen vacancies involved during the cathodic polarization. Hammouche and Siebert et al. proposed that a new pathway for the ORR was created based 2on the bulk diffusion in electrode.56,66 Oad, Oad , or Oad formed over the electrode surface can easily transfer in these pathways inside the electrode bulk to the electrode/electrolyte interface and then incorporate into the electrolyte via an ion-charge transfer process. According to another mechanism, the formed oxygen vacancies spread over the electrode surface mostly by surface or grain-boundary diffusion with the concomitant enlargement of TPB length.54,55 Van Heuveln and Bouwmeester further demonstrated that the bulk diffusion pathway was predominant under high cathodic overpotential while the surface diffusion pathway was significant only at low cathodic overpotential.42 Figure 4 is the dependence of Rp on cathodic polarization voltage. In the low voltage zone ranging from 0 to 50 mV (equivalent PO2 from 0.21 to 0.024 atm), denoted as zone A, Rp kept almost unchanged and approximated to the original value ∼1.45 Ω cm2 before polarization. This implies Mn4+ in LSSM was not reduced to Mn3+ under such a low voltage. In other words, not any oxygen vacancy was generated at the electrode/electrolyte interface, and the active site for ORR was still restricted to the TPB region. With increasing cathodic polarization voltage, the reduction of Mn4+ in LSSM to Mn3+ was initialized in concomitance with the formation of highly mobile oxygen vacancies. It then led to a sharp decrease of Rp as shown in zone B. To maintain electrical neutrality the oxygen vacancy in LSSM should be compensated by the decrease of oxidation state of Mn4+ in the B site of LSSM. The average valence of Mn in LSSM is 3.187 at 800 °C determined by quench the powder and then subjected to an iodometric titration technique in the room temperature.

Figure 4. The dependence of polarization resistance on cathodic polarization voltage at 800 °C.

As the cathodic polarization voltage high than 500 mV, Rp got a constant value about 0.25 Ω cm2. This can be explained by the following aspect. Under high cathodic voltage with low PO2, Mn4+ (even though not the whole) was reduced to Mn3+ or Mn2+ in concomitance with the steadily increasing concentration of oxygen vacancy, resulting in a decreasing Rp. However, the higher the oxygen vacancy concentration, the more likely oxygen vacancies will be ordered via the formation of defect association. This reaction can be described as

2Mn×Mn + OO× f (MnMn - V••O - MnMn) + 0.5O2 ′ MnMn

Mn2+.

(12)

where denotes Such cluster-defect has a negative effect on the mobility of oxygen vacancy and then makes Rp

18694 J. Phys. Chem. C, Vol. 112, No. 47, 2008

Figure 5. EIS of LSSM testing under three typical cathodic voltages at 800 °C.

decrease much more slowly. Therefore, when the polarization voltage received a certain value, in this case around 500 mV, the polarization resistance was leveled off or even increased with the further increase in polarization voltage, as shown in zone C. Figure 5 shows the typical EIS curves of LSSM cathode tested under various cathodic voltages at 800 °C. Under -10 and -100 mV cathodic polarization voltages, the low-frequency arcs of EIS decreased rapidly, while high frequency arcs kept almost unchanged. It means the in situ created oxygen vacancies spread mostly over the surface of electrode or along the TPB region under these cathodic voltages, corresponding to zone A in Figure 4. Under a cathodic polarization voltage of -0.5 V, the size of

Zheng et al. both low and high frequency arcs was reduced, implying the oxygen vacancy was successfully penetrated into the bulk of electrode to form more active sites for ORR, corresponding to zone B in Figure 4. Above results also supported the model proposed by Van Heuveln and Bouwmeester for the ORR and propagation path of the oxygen vacancy.42 3.3. Deactivation Kinetics of in Situ Created Oxygen Vacancies. As demonstrated previously, both surface and bulk oxygen vacancies could be generated after the polarization at a proper voltage for a limited period of time. However, the oxygen vacancy could be slowly eliminated once the polarization was stopped due to the reoxidation by surrounding atmosphere via reaction as expressed in eq 3. Indeed it has been demonstrated that the Rp relied strongly on the activation and deactivation processes, as shown in Figure 2. To exploit the elimination mechanism of these oxygen vacancies, two different oxidization techniques were applied, i.e., chemical oxidation by gas phase oxygen and electrochemical oxidation via the help of anodic current. After the LSSM electrode was activated by cathodic polarization for a period of time, it was subjected for relaxation under zero cathodic or anodic current passage. The in situ created oxygen vacancies would slowly react with oxygen from atmosphere over the electrode surface as

1 •• × VO,LSSM + O2 + e- f OO,LSSM 2

(13)

where e- was supplied by Mn3+ f Mn4+. The oxidizing rate of oxygen vacancy can be written as

Figure 6. The value of relaxing Rp depending on time in various states (a) 0.01 atm; (b) 0.21 atm; (c) 1 atm. Insets are the linear relationship between Rp and t1/2. The values of Rp in original and after anode polarization are also shown as fixed values.

Oxygen Reduction over a La0.8Sr0.2Sc0.1Mn0.9O3 Cathode •• •• d[VO,LSSM ] [VO,LSSM ] 0.5 PO2 ) -k3 × dt O

J. Phys. Chem. C, Vol. 112, No. 47, 2008 18695

(14)

O,LSSM

Equation 14 can be simplified and integrated as •• [VO,LSSM ])-

 0.5

e1-k3PO2t k3PO0.52

(15)

× where k′3 ) k3/[OO,LSSM ]. From eq 15, there is also a limiting time as

tL )

1 -0.5 pO k3 2

(16)

This limiting time relates only with the surrounding PO2. The dependence of Rp on relaxation time (t) at different PO2 are shown in parts a-c of Figure 6. The Rp initially increased rapidly and then reached a steady value after a limited period of time at PO2 ) 0.21 and 1 atm. The increase in Rp with relaxation time suggests that the surface oxygen vacancy concentration were steadily reduced until an equilibrium concentration was reached. Limiting time tL also decreased with increasing PO2, which was in good agreement with eq 16. As shown in the insets of Figure 6, a linear response of Rp to t0.5 at the initial stage of relaxation was observed. This implies that such oxidization process was controlled by diffusion processes.55,67 It is well known that for the LSM electrode the dominant defect under normal partial pressure of oxygen, e.g., 0.21 atm and SOFC temperatures, is cation vacancy while oxygen vacancy defect only occurs under low partial pressure of oxygen and high polarization conditions. Therefore, a very slow relaxation of Rp (about 60 h) was needed for Rp to get its balance value,61,62,64 while for LSSM it needs only ∼3 h. This implies the defect chemistry of LSSM may be different from LSM. It is also well-known the movement of cation vacancies is a rather slower process than oxygen vacancies.63,65 We then believe the main defect in LSSM is oxygen vacancy instead of cation vacancy. Such an assumption was strongly supported by the nonstoichiometry of LSSM before polarization. As mentioned previously, the average valence of Mn in LSSM is 3.187 at 800 °C, and the oxygen nonstoichiometry is 2.984. It suggests that the LSSM oxide is very slightly oxygen deficient before polarization, i.e., the cation vacancy in the LSSM at 800 °C is negligible. Thereby, it is likely that only the surface oxygen vacancies were oxidized by ambient air during the relaxation. Such a viewpoint was further supported by investigating the dependence of EIS curves on relaxing-time in air. As shown in Figure 7, the overall shape of EIS was similar during the different stage of the relaxation process, implying that the ORR mechanism and rate-determining step may not alter during the relaxation process. Low-frequency arcs which were related with surface diffusion increased steadily, while high-frequency arcs kept almost unchanged with relaxation time. It suggests the concentration of surface oxygen vacancy decreased while the concentration of the oxygen vacancy in the bulk of electrode and at interface of electrode/electrolyte was not affected. When an anodic current was applied to the LSSM electrode (WE), there was a reverse current as cathodic current to the CE of Pt. Since Pt is an excellent electrocatalyst for oxygen reduction at high temperature,38 a large amount of oxygen molecules were reduced to lattice oxygen at the TPB of CE. Such lattice oxygen would be pumped by the anodic current through the oxygen vacancy in ScSZ to the interface of LSSM

Figure 7. The EIS curves of LSSM depending on various relaxing times in air at 800 °C.

WE and electrolyte. Consequentially, the lattice oxygen reacted with the oxygen vacancies and electronic holes (h•) over the LSSM WE as k4

•• 2OO× + 2h• - VO,LSSM 98 O2

(17)

The h• notes as a reverse electron described by

h• + e- ) null

(18)

Different from the oxidization by oxygen from the surrounding atmosphere which occurred only over the electrode surface, the oxidation of oxygen vacancy induced by anodic current could penetrate into the electrode bulk until all surface and bulk oxygen vacancies were totally consumed up via the reaction as described in eq 17. According to Figure 2, the value of Rp after the second oxidization process was higher than the first oxidation. It may indicate that some oxygen vacancies at the electrode/electrolyte interface and bulk of electrode were not oxidized during the first oxidation process. Even after the second oxidization process, Rp still did not recover to its original value. Such a phenomenon was also observed by other investigators.22,68,69 They proposed that the microstructure of electrode was optimized after a high cathodic polarization and resulted in an extension of TPB. On the basis of the above analysis, mechanisms about activation and deactivation of LSSM electrode were proposed as shown in parts a-d of Figure 8. Before the cathodic polarization, there was not any oxygen vacancy over the surface nor in the bulk of LSSM electrode; the ORR happened only along the typical TPB region, as shown in Figure 8a. After a cathodic polarization, a large number of in situ created oxygen vacancies were formed, which spread to the whole electrode surface and also penetrated into bulk of the electrode. Thereby, much more active sites for the ORR were created, which then effectively reduced the Rp (Figure 8b). After the relaxation for a period of time, the in situ created surface oxygen vacancies during the cathodic polarization were oxidized while those in the bulk of electrode and at the electrode/electrolyte interface were still available and continued to provide for the ORR (Figure 8c). When an anodic current was applied to the electrode, the oxygen vacancies in the bulk of electrode and the interface of electrode/electrolyte were thoroughly oxidized by lattice oxygen generated by anodic current, the active site for ORR returned to near the original TPB region again (Figure 8d).

18696 J. Phys. Chem. C, Vol. 112, No. 47, 2008

Zheng et al.

Figure 8. Scheme of activation and reactivation of ORR (a) before polarization, (b) after cathodic polarization, (c) after cathodic polarization and then relaxed in ambient air for a period of time, and (d) after cathodic and then anodic polarization. Thick red lines denote the active sites for ORR.

3.4. Effect of Activation and Deactivation on the ORR. As mentioned previously, the ORR process may involve many substeps as gas-phase diffusion, surface adsorption, dissociation, surface diffusion, and charge transfer (both electron and oxygen ion). Any of above steps may cause a significant polarization resistance toward the ORR. Van Heuveln et al. considered that ORR over LSM electrode was a two-step process and oxygen ion surface diffusion was the rds at low overpotential.42 Sibert et al. believed that the rate-determining step for ORR was a dissociative adsorption process,66 such viewpoint was also approved by Murray et al.53 Ostergard and Mogensen considered that three processes were available during the whole ORR, and each one could be the rate-determining step depending on the reaction conditions.39 Co et al. synoptically proposed four possible mechanisms based on Langmuir and Temkin adsorption condition by electrode kinetics analysis.40 Above literature results suggest that the ORR over oxide electrode is really complicated. A better understanding of ORR mechanism is then of great importance. Since the polarization could result in a change of oxygen vacancy concentration in the electrode, it may induce an alteration of the ORR mechanism. The ORR over LSSM electrode before and after a cathodic polarization was then investigated in detail. The inset of Figure 9 is the dependence of polarization impedance on temperature under various PO2 before cathodic polarization. The corresponding activation energies (Ea) are 178.2, 170.8, and 169.0 kJ mol-1 at PO2 ) 0.01, 0.21, and 1 atm, respectively, which are in the typical range for a LSM electrode.38,39,53 The weak dependence of Ea on surrounding PO2 suggests that the activity of ORR before polarization was related only with some intrinsic properties of the electrode. However, the Ea of ORR at PO2 ) 0.21 atm decreased sharply from 178.1 to 134.44 kJ mol-1 after a cathodic polarization as shown in Figure 9. It suggests the oxygen reduction process was optimized after the polarization on the one hand, and a possible alteration of the mechanism and ratedetermining step of the ORR was induced on the other hand.

Figure 9. Activation energies (Ea) of impedance polarization with respect to temperature. Inset is Ea before polarization under various PO 2 .

To have a better understanding the effect of polarization on the essential and rds of ORR, the EIS fore LSSM electrode at various PO2 before and after the polarization were fitted to the equivalent circuit L1Rs(RQ)1(RQ)2 in Figure 10c and L1Rs(RQ)H(RQ)1(RQ)2 in Figure 10d, respectively. The inductance Lo is primarily ascribed to leads and Ag current collectors on WE. Rs is electrolyte resistance and uncompensated resistance between RE and WE. A constant phase element (CPE) represents a nonideal capacitor of the double layer at a nonplanar TPB. For the electrode before polarization, an additional extrahigh frequency arc appeared independent of PO2, noted as RH, which corresponds to the incorporation of oxygen ion into electrolyte layer. RE1 denotes mainly ion and electron chargetransfer processes in high-frequency arc. RE2 denotes nonchargetransfer processes which are the sum of oxygen surface adsorption, dissociation, oxygen molecular, or ion surface diffusion in low-frequency arc. Parts a and b of Figure 10 show the fitting results of RE1 and RE2 depending on PO1/n2 for the

Oxygen Reduction over a La0.8Sr0.2Sc0.1Mn0.9O3 Cathode

J. Phys. Chem. C, Vol. 112, No. 47, 2008 18697

Figure 11. Fitting values of Rp depending on PO2 under (a) -0.5 V cathodic and (b) +0.5 V anodic polarizations, respectively.

Figure 10. Fitting values of Rp (a) before and (b) after polarization in various PO2 at 800 °C via the models in parts c and d, respectively.

electrode before and after the polarization, respectively. To interpret the relationship of 1/R with P1/n O2 , an appropriate model proposed by van Heuveln and Bouwmeester was adopted as42

O2 f 2Oad

(Step1)

Oad + e f Oad

(Step2)

Oad f OTPB

(Step3)

2OTPB + e f OTPB

(Step4)

2OTPB + VO•• f OO×

(Step5)

Generally, different values of n correspond to different reaction steps, for example, n ) 1/4, 3/8, 1/2, 0, and 1 reflects steps 3, 2, 1, 5, and oxygen molecule diffusion in cathode pores, respectively.42,49,70 In our study, the values of n for RE2 (lowfrequency arc) kept constant at ∼0.25 both before and after the polarization. This implies the process associated with RE2 was

not altered by cathodic polarization. This process could be the 2diffusion of Oad or Oad along the LSSM surface to the TPB region, supported by the fact that the values of capacitance and fmax (not shown here) were also within the typical range for such a process.38,39 The value of n for RE1 (high-frequency arc) increased from ∼0.25 before the polarization to ∼0.375 after the polarization. Thereby, a transition from a diffusion process of possible Oad to an electron charge-transfer process was likely what happened. This issue will be discussed later. As shown in parts a and b of Figure 11, the value of n for the whole ORR kept at approximately 0.25 under either cathodic or anodic polarization of 0.5 V. This suggests the polarization did not lead to a change of the rate-determining step of ORR at certain ranges of temperature and PO2, i.e., the rate-determining step for ORR over LSSM electrode was still the diffusion of Oad or O2ad from electrode surface to the TPB region. The transition from a surface diffusion process to an electron charge-transfer process for the high-frequency arc after a high cathodic polarization can be explained as follows. After the cathodic polarization, a high concentration of oxygen vacancy was created over the electrode surface and in the electrode bulk; these in situ created oxygen vacancies enhanced the oxygen dissociation and surface diffusion kinetics, on the one hand. However, it also decreased the electronic conductivity, which is detrimental to electron transfer process. Such transition can be further explained via electrode kinetics by analyzing the Tafel slopes. The Butler-Volmer equation has been adapted for series of consecutive steps electrode kinetics as71

18698 J. Phys. Chem. C, Vol. 112, No. 47, 2008

[ ( ) (

i ) i0 exp

RaFη -RcFη - exp RT RT

)]

Zheng et al.

(19)

It is derived based on the assumption that there is single ratedetermining step and mass transfer of the reactant and product is not the limiting step. i0 is exchange current density which reflects the inherent rate of the rate-determining step and only depends on the concentration of reactants and reacting temperature, η is overpotential, Ra and Rc are apparent anodic and cathodic charge-transfer coefficients, respectively. Parameters R, T, and F have their usual meaning. By assumption that the charge-transfer process is not the rate-determining step, then

ra + (1 - β)r ν rc Rc ) + βr ν

Ra )

(20) (21)

where ra and rc are the number of electrons that transferred before and after the rate-determining step, β is the symmetry coefficient usually with the value of 0.5, and r is the number of electrons in the rate-determining step. The stoichiometric number ν is the number of times that the rate-determining step occurs for overall reaction with

Ra + Rc )

n ν

(22)

where n is the total number of electrons transferred in the reaction. For the overall ORR, n ) 4. At low overpotential (